Abstract
Citrus products are rich sources of furocoumarins, a class of photoactive compounds. Certain furocoumarins combined with ultraviolet radiation can induce skin cancer. We examined the relationship between citrus consumption and cutaneous melanoma risk among 56,205 Caucasian postmenopausal women in the Women’s Health Initiative. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of melanoma by citrus intake level. During a mean follow-up of 15.7 years, 956 incident melanoma cases were documented. In multivariable adjusted models, the HR (95% CI) for melanoma was 1.12 (0.91, 1.37) among the highest citrus consumers (1.5+ servings/day of fruit or juice) versus the lowest (<2 servings/week), 0.95 (0.76, 1.20) among the highest citrus fruit consumers (5+ servings/week) versus non-consumers, and was 1.13 (0.96, 1.32) for the highest citrus juice consumers (1+ servings/day) versus the lowest (<1 serving/week). In stratified analyses, an increased melanoma risk associated with citrus juice intake was observed among women who spent the most time outdoors in summer as adults; the HR for the highest versus lowest intake was 1.22 (1.02, 1.46) (p-trend = 0.03). Further research is needed to explore the association of melanoma with citrus juices among women with high sun exposure.
Keywords: vegetables and fruit, diet, cohort study, epidemiology, cancer risk
Introduction:
The incidence of melanoma, the most dangerous form of skin cancer, has steadily risen in the US over the past 30 years (1). Exposure to ultraviolet (UV) radiation is the only well-established modifiable risk factor for melanoma (2). The association between UV and melanoma is complex (3), and the role of UV in melanoma pathogenesis is modified by other factors such as skin type, hair color, genetic factors, geographic location (3, 4), and even certain dietary factors (5–7). In our recent work we found that greater consumption of citrus products was associated with greater risk of incident basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) (8) as well as melanoma (9) among men and women of the Health Professionals Follow-up Study (HPFS) and the Nurses’ Health Study (NHS). This association was not observed with consumption of non-citrus fruit and juice, however, suggesting that citrus products uniquely conferred this risk. One distinguishing characteristic common among many citrus products, and rather rare among most other fruits and vegetables, is their contents of furocoumarins.
Furocoumarins are a class of organic compounds produced by several plant species in response to stress and as a defense against predators (10). These photoactive compounds have been utilized in combination with UVA radiation in a phototherapy known as PUVA (psoralen and UVA) to treat proliferative skin diseases such as psoriasis. Upon photoactivation, furocoumarins may form cross-links with DNA, thereby disrupting cellular replication (11). While PUVA is effective for clearing psoriasis, it has been classified as a carcinogen (12), as its use has been shown to cause dose-dependent increases in the risk of BCC, SCC (13), and melanoma (14).
It is generally believed that citrus products are the primary source of furocoumarins in most American diets, as many citrus products are particularly rich sources of furocoumarins (15), and are commonly consumed as nutrient-dense food choices. Therefore, given the photocarcinogenic potential of furocoumarins, it has been hypothesized that citrus consumption may be associated with melanoma risk. Our preliminary investigation in men and women of the HPFS and NHS supports this hypothesis and suggested a synergistic effect between furocoumarins and UV radiation. However, the generalizability of these findings may be limited, as the populations represented in these cohorts are well-educated health professionals whose dietary habits and other health-related behaviors may differ from the general public. Therefore, additional investigation is needed to examine this hypothesis. Thus, the objective of this study was to evaluate the association of baseline citrus consumption with melanoma risk in the Women’s Health Initiative (WHI), a large prospective cohort of postmenopausal women.
Methods:
Study population
The study population consisted of participants from the WHI Observational Study (OS). The overall study design of the WHI has been documented elsewhere (16). The WHI-OS enrolled postmenopausal women aged 50–79 years at baseline from 40 clinical centers throughout the United States between 1993 and 1998 (17). While the initial study period ended in 2005, WHI Extension Studies have continued follow-up of all women with consent. For this study, follow-up data were available through February 2017. Given their low incidence of melanoma, non-white participants (n = 12,565) were excluded from this analysis. Those with a prior history of melanoma (n = 1,211), any cancer at baseline other than non-melanoma skin cancer (NMSC) (n = 11,022), missing food frequency questionnaire (FFQ) data (n = 3,239), and missing information on sun exposure variables (n = 9,434) were also excluded, yielding an analytic cohort of 56,205 women.
Assessment of dietary consumption and other risk factors
In baseline FFQs, participants were asked how often they consumed various foods during the past 3 months. Participants were asked how often (in 9 categories ranging from never to 2+ servings per day) they had consumed “oranges, grapefruits, and tangerines” (1 orange or ½ grapefruit), which was used to assess citrus fruit consumption in this study. Participants were also asked how often they consumed “orange juice and grapefruit juice” (6 ounce glass), which was used to assess citrus juice consumption. Total citrus was calculated as the sum of reported frequencies of consuming citrus fruit and citrus juices. Total non-citrus fruit and juice consumption was estimated by summing the reported frequencies for 12 other items on the FFQ that inquired about non-citrus fruit products. Information on other dietary factors such as total energy intake, coffee consumption, and alcohol intake were also collected using the FFQ. Dietary intake collected using the FFQ has previously been validated against 24-hour dietary recalls and 4-day food records for estimation of 30 nutrients in a subset of participants (18). For most nutrients, mean intake estimated by the FFQ was within 10% of that estimated by food records or recalls.
In baseline questionnaires participants were asked about lifestyle factors including cigarette smoking, physical activity, and postmenopausal hormone use. Data on several variables related to skin cancer risk were also collected including skin reaction to the sun (tan vs. burn), use of sunscreen, and average daily time spent outdoors in summer as a child and currently. Sun exposure in Langleys of the participant’s clinical center was also recorded. These potential confounders as well as dietary habits were defined using baseline values.
Ascertainment of melanoma cases
At the time of entry into the study, each woman was asked whether she had ever been diagnosed with cancer, and if so, what type(s) of cancer. Women were also mailed questionnaires annually to report new health information including new cancer diagnoses. Reports of melanoma (and other cancers except for NMSC) were confirmed by physicians through medical record review in a central adjudication process, as has been described elsewhere (19, 20).
Statistical analysis
SAS software version 9.2 (SAS Institute, Cary, NC) was used for all statistical analyses. Chi-square tests of association for categorical variables and one-way ANOVA for continuous variables were used to examine differences in variables of interest between participants with differing levels of citrus consumption. All statistical tests were two-tailed, and the significance level was set at P < 0.05.
Cox proportional hazards models were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) of melanoma associated with dietary consumption of total citrus, citrus fruit, citrus juice, and non-citrus fruit and juice. Each participant contributed person-time from the time of the baseline questionnaire until the date of first diagnosis of any cancer, death, or the end of follow-up. Both age-adjusted and multivariable-adjusted models were tested. Multivariable models were adjusted for clinically important confounders including age, body mass index (BMI), education, physical activity, alcohol consumption, history of NMSC, regional solar radiation, skin reaction to sun, average daily time outdoors in summer currently, and sunscreen use. Models that additionally adjusted for total calorie intake, coffee consumption, smoking status, use of postmenopausal hormone replacement therapy, and average daily time outdoors in summer as a child did not generate materially different results and are therefore not presented. Trend tests across categories of dietary consumption were performed by assigning median values for these categories and treating the new variable as a continuous term in the models.
Supplemental analyses were conducted to explore potential differences in associations between citrus and melanoma among subgroups of participants. We performed subgroup analyses stratified by variables that showed evidence of interaction with citrus consumption. In subgroup analyses, multivariable adjusted HRs with 95% CIs were calculated.
Results:
Total citrus intake varied significantly by several demographic and lifestyle characteristics in this population. Participants with higher total citrus consumption tended to be older, had lower BMIs, had a higher total energy intake, had more education, and consumed more alcohol than those with lower total citrus consumption (Table 1). Several sun-related factors also varied by citrus consumption level. Those who consumed the most citrus reported spending more time outdoors in the summer during childhood and currently. Higher citrus consumers were also higher users of sunscreen and frequently reported having a skin type prone to tanning without burning.
Table 1.
Baseline characteristics of participants according to frequency of total citrus consumption (medium serving size)
| Total Citrus Intake |
||||||||
|---|---|---|---|---|---|---|---|---|
| Level |
<2/wk |
2–4/wk |
5–6/wk |
1–1.4/d |
1.5+/d |
Total | trend | |
| N (%) | 15,947 (28.4) | 12,386 (22.0) | 5,744 (10.2) | 13,702 (24.4) | 8,426 (15.0) | 56,205 | p-value | p-value |
| Median intake/day | 0.08767 | 0.5000 | 0.77808 | 1.07671 | 1.72329 | |||
| Citrus Fruit, daily medium servings (mean, SD) | 0.063 (0.076) | 0.214 (0.182) | 0.381 (0.261) | 0.400 (0.373) | 0.808 (0.690) | 0.323 (0.424) | <0.001 | <0.001 |
| Citrus Juice, daily medium servings (mean, SD) | 0.048 (0.068) | 0.276 (0.191) | 0.396 (0.268) | 0.701 (0.377) | 1.163 (0.660) | 0.460 (0.511) | <0.001 | <0.001 |
| Non-Citrus Fruit and Juice, daily medium servings (mean, SD) | 1.655 (1.344) | 1.817 (1.233) | 1.997 (1.290) | 2.053 (1.347) | 2.588 (1.885) | 1.963 (1.444) | <0.001 | <0.001 |
| Age (yrs) | 62.3 (7.1) | 63.3 (7.1) | 63.4 (7.2) | 64.3 (7.1) | 64.0 (7.3) | 63.4 (7.2) | <0.001 | <0.001 |
| NMSC at baseline (n, %) | 1956 (12.3) | 1516 (12.2) | 742 (12.9) | 1861 (13.6) | 1123 (13.3) | 7198 (12.8) | ||
| Regional solar radiation Langleys (g-cal/cm2) (n, %) | <0.001 | |||||||
| 300–325 | 4580 (28.7) | 4059 (32.8) | 1923 (33.5) | 4914 (35.9) | 2933 (34.8) | 18409 (32.8) | ||
| 350 | 3110 (19.5) | 2483 (20.1) | 1172 (20.4) | 3031 (22.1) | 1999 (23.7) | 11795 (21.0) | ||
| 375–380 | 1776 (11.1) | 1400 (11.3) | 602 (10.5) | 1500 (11.0) | 895 (10.6) | 6173 (11.0) | ||
| 400–430 | 3051 (19.1) | 2027 (16.4) | 917 (16.0) | 1859 (13.6) | 1113 (13.2) | 8967 (16.0) | ||
| 475–500 | 3430 (21.5) | 2417 (19.5) | 1130 (19.7) | 2398 (17.5) | 1486 (17.6) | 10861 (19.3) | ||
| Average daily time outdoors in summer as a child (n, %) | <0.001 | |||||||
| <30 minutes | 375 (2.4) | 270 (2.2) | 119 (2.1) | 281 (2.1) | 178 (2.1) | 1223 (2.2) | ||
| 30 min to 2 hours | 4132 (25.9) | 3362 (27.1) | 1431 (24.9) | 3614 (26.4) | 2032 (24.1) | 14571 (25.9) | ||
| 2+ hours | 11440 (71.7) | 8754 (70.7) | 4194 (73.0) | 9807 (71.6) | 6216 (73.8) | 40411 (71.9) | ||
| Average daily time outdoors in summer currently (n, %) | <0.001 | |||||||
| <30 minutes | 5302 (33.3) | 3680 (29.7) | 1580 (27.5) | 3944 (28.8) | 2367 (28.1) | 16873 (30.0) | ||
| 30 min to 2 hours | 7741 (48.5) | 6344 (51.2) | 3020 (52.6) | 7104 (51.9) | 4154 (49.3) | 28363 (50.5) | ||
| 2+ hours | 2904 (18.2) | 2362 (19.1) | 1144 (19.9) | 2654 (19.4) | 1905 (22.6) | 10969 (19.5) | ||
| Sunscreen SPF (n, %) | ||||||||
| None | 8156 (51.1) | 5618 (45.4) | 2548 (44.4) | 6055 (44.2) | 3666 (43.5) | 26043 (46.3) | ||
| Something but don’t know | 230 (1.4) | 173 (1.4) | 78 (1.4) | 189 (1.4) | 115 (1.4) | 785 (1.4) | ||
| 2–14 | 738 (4.6) | 657 (5.3) | 311 (5.4) | 684 (5.0) | 414 (4.9) | 2804 (5.0) | ||
| 15–24 | 4301 (27.0) | 3752 (30.3) | 1812 (31.6) | 4389 (32.0) | 2686 (31.9) | 16940 (30.1) | ||
| 25+ | 2522 (15.8) | 2186 (17.7) | 995 (17.3) | 2385 (17.4) | 1545 (18.3) | 9633 (17.1) | ||
| Skin reaction to sun (n, %) | 0.010 | |||||||
| No change | 955 (6.0) | 730 (5.9) | 342 (6.0) | 779 (5.7) | 461 (5.5) | 3267 (5.8) | ||
| Tans, does not burn | 4649 (29.2) | 3707 (29.9) | 1709 (29.8) | 4063 (29.7) | 2609 (31.0) | 16737 (29.8) | ||
| Burns, then tans | 4045 (25.4) | 3157 (25.5) | 1501 (26.1) | 3591 (26.2) | 2173 (25.8) | 14467 (25.7) | ||
| Burns, then tans minimally | 4344 (27.2) | 3423 (27.6) | 1563 (27.2) | 3787 (27.6) | 2235 (26.5) | 15352 (27.3) | ||
| Burns, does not tan | 1954 (12.3) | 1369 (11.1) | 629 (11.0) | 1482 (10.8) | 948 (11.3) | 6382 (11.4) | ||
| Education (n, %) | <0.001 | |||||||
| < high school graduate | 590 (3.7) | 352 (2.8) | 131 (2.3) | 282 (2.1) | 166 (2.0) | 1521 (2.7) | ||
| high school graduate | 2978 (18.7) | 1986 (16.0) | 899 (15.7) | 1977 (14.4) | 1116 (13.2) | 8956 (15.9) | ||
| some college | 5988 (37.6) | 4602 (37.2) | 1939 (33.8) | 4682 (34.2) | 2869 (34.1) | 20080 (35.7) | ||
| college graduate | 6262 (39.3) | 5368 (43.3) | 2743 (47.8) | 6676 (48.7) | 4217 (50.1) | 25266 (45.0) | ||
| BMI, kg/m2 (mean, SD) | 27.3 (6.0) | 27.0 (5.8) | 27.0 (5.7) | 26.6 (5.6) | 26.6 (5.8) | 26.9 (5.8) | <0.001 | <0.001 |
| Physical activity (n, %) | <0.001 | |||||||
| Inactive (0–1.7) | 3482 (22.1) | 1976 (16.1) | 839 (14.7) | 2020 (14.9) | 1091 (13.1) | 9408 (16.9) | ||
| Low Activity (1.8–8.3) | 4336 (27.5) | 3275 (26.7) | 1460 (25.7) | 3322 (24.5) | 1935 (23.1) | 14328 (25.7) | ||
| Mod Activity (8.4–20) | 4487 (28.4) | 3880 (31.6) | 1784 (31.3) | 4481 (33.0) | 2665 (31.9) | 17297 (31.1) | ||
| High Activity (>20) | 3476 (22.0) | 3156 (25.7) | 1609 (28.3) | 3740 (27.6) | 2670 (31.9) | 14651 (26.3) | ||
| Cigarette smoking status (n, %) | <0.001 | |||||||
| Never | 7295 (46.3) | 6202 (50.6) | 2918 (51.4) | 7030 (52.0) | 4401 (52.8) | 27846 (50.2) | ||
| Past | 7216 (45.8) | 5420 (44.3) | 2517 (44.3) | 5974 (44.2) | 3578 (43.0) | 24705 (44.5) | ||
| Current | 1232 (7.8) | 627 (5.1) | 247 (4.4) | 521 (3.9) | 351 (4.2) | 2978 (5.4) | ||
| Alcohol consumption, g/day (mean, SD) | 5.77 (11.68) | 6.23 (11.26) | 6.57 (12.07) | 6.74 (11.38) | 6.50 (11.41) | 6.30 (11.52) | <0.001 | <0.001 |
| Coffee consumption, cups/day (mean, SD) | 2.26 (1.83) | 2.22 (1.73) | 2.24 (1.71) | 2.14 (1.62) | 2.13 (1.70) | 2.20 (1.73) | <0.001 | <0.001 |
| Hormone Therapy status (n, %) | <0.001 | |||||||
| Never | 3934 (25.1) | 3210 (26.4) | 1499 (26.7) | 3734 (27.8) | 2401 (29.1) | 14778 (26.8) | ||
| Past | 3049 (19.4) | 2299 (18.9) | 1070 (19.0) | 2632 (19.6) | 1557 (18.8) | 10607 (19.2) | ||
| Current | 8704 (55.5) | 6675 (54.8) | 3051 (54.3) | 7064 (52.6) | 4307 (52.1) | 29801 (54.0) | ||
During an average follow-up of 15.7 years, 956 melanoma cases were documented. Total citrus consumption was not significantly associated with incident melanoma in both age- and multivariable-adjusted models (Table 2). The age-adjusted HR among the participants in the highest citrus consumption category (1.5+ servings per day) compared to the lowest consumers (<2 servings/week) was 1.24 (95% CI: 1.01, 1.51) (p-trend = 0.05), and the multivariable-adjusted HR was 1.12 (95% CI: 0.91, 1.37). Citrus juice consumption showed a positive association with incident melanoma risk in the age-adjusted model, but not in the multivariable-adjusted model. In the age-adjusted model, the HR for incident melanoma among the highest citrus juice consumers (1+ servings/day), using the lowest consumers (<1 serving/week) as a reference, was 1.21 (95% CI: 1.03, 1.41) (p-trend = 0.01). After full adjustment, the HR among the highest citrus juice consumers, using the lowest consumers as a reference, was 1.13 (95% CI: 0.96, 1.32) (p-trend = 0.11). Citrus fruit consumption was not associated with melanoma risk, nor was consumption of non-citrus fruits and juices.
Table 2.
Hazard ratios for incident melanoma according to frequency of citrus consumption
| Serving Category | P for trend | |||||
|---|---|---|---|---|---|---|
| Total Citrus | <2/wk | 2–4/wk | 5–6/wk | 1–1.4/d | 1.5+/d | |
| No. of person-years | 248,899 | 193,710 | 90,868 | 216,254 | 132,152 | |
| No. of cases | 245 | 224 | 85 | 243 | 159 | |
| Age-adjusted HR (95% CI) | (ref) | 1.18 (0.99, 1.42) | 0.96 (0.75, 1.22) | 1.16 (0.97, 1.38) | 1.24 (1.01, 1.51) | 0.05 |
| Multivariable-adjusted HR (95% CI) | (ref) | 1.14 (0.95, 1.37) | 0.90 (0.70, 1.16) | 1.06 (0.88, 1.28) | 1.12 (0.91, 1.37) | 0.45 |
| Citrus Juice | <1/wk | 1/wk | 2–4/wk | 5–6/wk | 1+/d | |
| No. of person-years | 382,946 | 78,667 | 148,035 | 53,658 | 218,576 | |
| No. of cases | 402 | 72 | 145 | 62 | 275 | |
| Age-adjusted HR (95% CI) | (ref) | 0.87 (0.68, 1.12) | 0.95 (0.78, 1.14) | 1.10 (0.85, 1.44) | 1.21 (1.03, 1.41) | 0.01 |
| Multivariable-adjusted HR (95% CI) | (ref) | 0.86 (0.67, 1.11) | 0.95 (0.78, 1.15) | 1.08 (0.82, 1.42) | 1.13 (0.96, 1.32) | 0.11 |
| Citrus Fruit | Never | <1/wk | 1/wk | 2–4/wk | 5+/wk | |
| No. of person-years | 146,291 | 338,313 | 114,702 | 124,254 | 158,323 | |
| No. of cases | 143 | 384 | 129 | 133 | 167 | |
| Age-adjusted HR (95% CI) | (ref) | 1.15 (0.95, 1.40) | 1.15 (0.90, 1.46) | 1.10 (0.87, 1.39) | 1.09 (0.87, 1.36) | 0.87 |
| Multivariable-adjusted HR (95% CI) | (ref) | 1.05 (0.86, 1.28) | 1.03 (0.81, 1.31) | 0.97 (0.76, 1.23) | 0.95 (0.76, 1.20) | 0.44 |
| Non-Citrus Fruit/Juice | <0.75/d | 0.75–1.2/d | 1.3–1.9/d | 2.0–2.9/d | 3+/d | |
| No. of person-years | 143,368 | 180,183 | 212,029 | 192,722 | 153,580 | |
| No. of cases | 155 | 187 | 229 | 215 | 170 | |
| Age-adjusted HR (95% CI) | (ref) | 0.96 (0.78, 1.19) | 1.00 (0.82, 1.23) | 1.04 (0.85, 1.28) | 1.04 (0.83, 1.29) | 0.52 |
| Multivariable-adjusted HR (95% CI) | (ref) | 0.92 (0.73, 1.14) | 0.90 (0.73, 1.11) | 0.92 (0.74, 1.14) | 0.92 (0.73, 1.15) | 0.63 |
Multivariable-adjusted HRs were adjusted for age (continuous), BMI (<25.0, 25.0–29.9, 30.0–34.9, ≥35.0 kg/m2), education (less than high school graduate, high school graduate, some college, college graduate), physical activity (inactive, low activity, moderate activity, high activity), alcohol consumption (continuous), history of NMSC (yes, no), regional solar radiation (300–325, 350, 375–380, 400–430, 475–500 Langleys), skin reaction to sun (no burns, with or without tanning; burns, with or without tanning), average daily time outdoors currently (<30min, 30min - 2hrs, 2+ hrs), and sunscreen SPF use (none, something but don’t know, SPF 2–14, SPF 15–24, SPF 25+)
In subgroup analyses of women stratified by skin type, regional solar radiation, time outdoors in summer currently, sunscreen use, and history of NMSC, total citrus consumption was not significantly associated with risk of incident melanoma in any subgroups (Table 3). However, citrus juice consumption was significantly associated with increased melanoma risk among women who spent more than 30 minutes per day outdoors in the summer as adults (Table 4). Among these women, HRs for those consuming 1 serving/wk, 2–4 servings/wk, 5–6 servings/wk, and 1+ servings/day of citrus juice, using those consuming <1 serving/wk as a reference, were 0.90 (0.67, 1.20), 0.94 (0.75, 1.18), 1.04 (0.76, 1.44), and 1.22 (1.02, 1.46), respectively (p-trend = 0.03). No statistically significant trends in HRs were noted for citrus juice consumption among subgroups of skin reaction to the sun, regional solar radiation level, history of NMSC, or sunscreen use.
Table 3.
Multivariable hazard ratios for incident melanoma according to frequency of total citrus consumption in subgroups of potential confounders
| Total Citrus Serving Category | P for trend | |||||
|---|---|---|---|---|---|---|
| <2/wk | 2–4/wk | 5–6/wk | 1–1.4/d | 1.5+/d | ||
| Time Outdoors in Summer Currently | ||||||
| < 30 minutes/day | ||||||
| Number of person-years | 80,406 | 55,199 | 23,766 | 59,672 | 35,652 | |
| Number of cases | 70 | 64 | 20 | 59 | 31 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.38 (0.98, 1.96) | 0.92 (0.55, 1.54) | 1.08 (0.75, 1.55) | 1.00 (0.64, 1.54) | 0.76 |
| > 30 minutes/day | ||||||
| Number of person-years | 168,494 | 138,511 | 67,102 | 156,581 | 96,499 | |
| Number of cases | 175 | 160 | 65 | 184 | 128 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.06 (0.85, 1.33) | 0.89 (067, 1.19) | 1.05 (0.85, 1.30) | 1.15 (0.91, 1.46) | 0.29 |
| Skin reaction to sun | ||||||
| No burn, with or without tanning | ||||||
| Number of person-years | 84,832 | 67,067 | 31,553 | 74,619 | 46,685 | |
| Number of cases | 64 | 45 | 20 | 52 | 33 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.81 (0.54, 1.20) | 0.75 (0.45, 1.26) | 0.86 (0.59, 1.27) | 0.85 (0.55, 1.31) | 0.51 |
| Burn, with or without tanning | ||||||
| Number of person-years | 164,067 | 126,643 | 59,315 | 141,634 | 85,466 | |
| Number of cases | 181 | 179 | 65 | 191 | 126 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.25 (1.01, 1.55) | 0.95 (0.71, 1.27) | 1.14 (0.92, 1.40) | 1.21 (0.95, 1.53) | 0.24 |
| Regional solar radiation | ||||||
| 300–350 | ||||||
| Number of person-years | 121,819 | 103,326 | 49,609 | 125,681 | 78,120 | |
| Number of cases | 109 | 120 | 46 | 148 | 93 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.24 (0.95, 1.62) | 0.97 (0.68, 1.37) | 1.19 (0.92, 1.54) | 1.18 (0.89, 1.57) | 0.34 |
| 375–500 | ||||||
| Number of person-years | 127,080 | 90,384 | 41,259 | 90,572 | 54,032 | |
| Number of cases | 136 | 104 | 39 | 95 | 66 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.06 (0.82, 1.38) | 0.86 (0.60, 1.23) | 0.93 (0.71, 1.22) | 1.08 (0.80, 1.46) | 0.96 |
| History of NMSC | ||||||
| No history of NMSC | ||||||
| Number of person-years | 219,598 | 170,986 | 79,718 | 187,597 | 114,993 | |
| Number of cases | 180 | 168 | 68 | 174 | 121 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.19 (0.96, 1.48) | 1.01 (0.76, 1.34) | 1.09 (0.88, 1.36) | 1.19 (0.94, 1.51) | 0.28 |
| History of NMSC | ||||||
| Number of person-years | 29,302 | 22,724 | 11,150 | 28,657 | 17,159 | |
| Number of cases | 65 | 56 | 17 | 69 | 38 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.99 (0.68, 1.44) | 0.59 (0.34, 1.04) | 1.00 (0.70, 1.42) | 0.91 (0.60, 1.38) | 0.70 |
| Sunscreen Use | ||||||
| No sunscreen use | ||||||
| Number of person-years | 125,067 | 86,502 | 39,582 | 93,478 | 56,387 | |
| Number of cases | 92 | 65 | 24 | 79 | 42 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.04 (0.75, 1.43) | 0.77 (0.48,1.21) | 1.07 (0.78, 1.47) | 0.93 (0.64, 1.36) | 0.82 |
| Sunscreen use (any or unknown SPF) | ||||||
| Number of person-years | 123,833 | 107,208 | 51,285 | 122,776 | 75,765 | |
| Number of cases | 153 | 159 | 61 | 164 | 117 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 1.20 (0.95, 1.50) | 0.98 (0.72, 1.32) | 1.06 (0.84, 1.33) | 1.21 (0.94, 1.55) | 0.30 |
Multivariable-adjusted HRs were adjusted for age (continuous), BMI (<25.0, 25.0–29.9, 30.0–34.9, ≥35.0 kg/m2), education (less than high school graduate, high school graduate, some college, college graduate), physical activity (inactive, low activity, moderate activity, high activity), alcohol consumption (continuous), history of NMSC (yes, no), regional solar radiation (300–325, 350, 375–380, 400–430, 475–500 Langleys), skin reaction to sun (no burns, with or without tanning; burns, with or without tanning), average daily time outdoors currently (<30min, 30min - 2hrs, 2+ hrs), and sunscreen SPF use (none, something but don’t know, SPF 2–14, SPF 15–24, SPF 25+). For each stratified analysis, stratifying variable was omitted from models.
Table 4.
Multivariable hazard ratios for incident melanoma according to frequency of citrus juice consumption in subgroups of potential confounders
| Citrus Juice Serving Category | P for trend | |||||
|---|---|---|---|---|---|---|
| <1/wk | 1/wk | 2–4/wk | 5–6/wk | 1+/d | ||
| Time Outdoors in Summer Currently | ||||||
| < 30 minutes/day | ||||||
| Number of person-years | 115,599 | 21,897 | 41,364 | 14,778 | 61,058 | |
| Number of cases | 116 | 17 | 39 | 17 | 55 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.77 (0.46, 1.30) | 0.97 (0.67, 1.40) | 1.21 (0.72, 2.01) | 0.87 (0.63, 1.21) | 0.60 |
| > 30 minutes/day | ||||||
| Number of person-years | 267,348 | 56,770 | 106,671 | 38,880 | 157,518 | |
| Number of cases | 286 | 55 | 106 | 45 | 220 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.90 (0.67, 1.20) | 0.94 (0.75, 1.18) | 1.04 (0.76, 1.44) | 1.22 (1.02, 1.46) | 0.03 |
| Skin reaction to sun | ||||||
| No burn, with or without tanning | ||||||
| Number of person-years | 132,179 | 26,558 | 53,320 | 18,171 | 74,530 | |
| Number of cases | 93 | 13 | 36 | 13 | 59 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.69 (0.38, 1.27) | 1.05 (0.71, 1.56) | 1.00 (0.54, 1.83) | 1.13 (0.81, 1.60) | 0.41 |
| Burn, with or without tanning | ||||||
| Number of person-years | 250,768 | 52,109 | 94,715 | 35,488 | 144,045 | |
| Number of cases | 309 | 59 | 109 | 49 | 216 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.90 (0.68, 1.20) | 0.91 (0.73, 1.14) | 1.10 (0.81, 1.50) | 1.13 (0.95, 1.35) | 0.15 |
| Regional solar radiation | ||||||
| 300–350 | ||||||
| Number of person-years | 190,606 | 43,109 | 84,557 | 29,262 | 131,020 | |
| Number of cases | 190 | 38 | 86 | 28 | 174 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.85 (0.60, 1.21) | 1.01 (0.78, 1.31) | 0.94 (0.63, 1.39) | 1.21 (0.98, 1.49) | 0.08 |
| 375–500 | ||||||
| Number of person-years | 192,340 | 35,557 | 63,478 | 24,396 | 87,556 | |
| Number of cases | 212 | 34 | 59 | 34 | 101 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.87 (0.60, 1.26) | 0.87 (0.65, 1.17) | 1.25 (0.86, 1.82) | 1.01 (0.80, 1.29) | 0.74 |
| History of NMSC | ||||||
| No history of NMSC | ||||||
| Number of person-years | 336,355 | 69,816 | 129,904 | 47,389 | 189,429 | |
| Number of cases | 297 | 54 | 112 | 46 | 202 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.88 (0.66, 1.18) | 0.99 (0.80, 1.24) | 1.08 (0.79, 1.48) | 1.14 (0.95, 1.37) | 0.13 |
| History of NMSC | ||||||
| Number of person-years | 46,591 | 8,851 | 18,131 | 6,270 | 29,147 | |
| Number of cases | 105 | 18 | 33 | 16 | 73 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.80 (0.47, 1.35) | 0.81 (0.54, 1.21) | 1.03 (0.59, 1.81) | 1.09 (0.80, 1.49) | 0.55 |
| Sunscreen Use | ||||||
| No sunscreen use | ||||||
| Number of person-years | 182,539 | 34,102 | 66,289 | 23,824 | 94,262 | |
| Number of cases | 133 | 21 | 47 | 22 | 79 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.89 (0.56, 1.41) | 1.00 (0.71, 1.40) | 1.20 (0.75, 1.90) | 1.08 (0.81, 1.44) | 0.49 |
| Sunscreen use (any or unknown SPF) | ||||||
| Number of person-years | 200,408 | 44,564 | 81,746 | 29,835 | 124,313 | |
| Number of cases | 269 | 51 | 98 | 40 | 196 | |
| Multivariable Adjusted HR (95% CI) | (ref) | 0.85 (0.63, 1.15) | 0.92 (0.73, 1.17) | 1.02 (0.73, 1.43) | 1.14 (0.95, 1.38) | 0.15 |
Multivariable-adjusted HRs were adjusted for age (continuous), BMI (<25.0, 25.0–29.9, 30.0–34.9, ≥35.0 kg/m2), education (less than high school graduate, high school graduate, some college, college graduate), physical activity (inactive, low activity, moderate activity, high activity), alcohol consumption (continuous), history of NMSC (yes, no), regional solar radiation (300–325, 350, 375–380, 400–430, 475–500 Langleys), skin reaction to sun (no burns, with or without tanning; burns, with or without tanning), average daily time outdoors currently (<30min, 30min - 2hrs, 2+ hrs), and sunscreen SPF use (none, something but don’t know, SPF 2–14, SPF 15–24, SPF 25+). For each stratified analysis, stratifying variable was omitted from models.
Discussion:
In this prospective cohort study of postmenopausal women there was no clear association between citrus consumption and risk of incident melanoma. The data suggested that citrus fruit juice consumption may be more closely associated with melanoma risk than total citrus consumption, yet this trend was not statistically significant when adjusted for important confounders. Subgroup analyses indicated that citrus juice consumption was associated with an increased melanoma risk among women who spent the most time outdoors in summer as adults.
These results somewhat differ from the findings of our previous work documenting a positive association between total citrus consumption and melanoma risk in the HPFS and NHS (9). Discrepancies between these studies could be related to various factors. If a relationship between citrus consumption and melanoma risk exists, and if the relationship is related to furocoumarin exposure, the use of total citrus as a crude indicator of furocoumarin consumption may reduce the ability to detect significant relationships in this study. Although citrus products are collectively believed to be the major source of furocoumarins in the American diet, there is wide variability in furocoumarin concentrations between individual citrus products. In our previous work we documented that the total concentration of seven furocoumarin compounds varied from 21858 ng/g in grapefruit, to 9151 ng/g in limes, to 0.5 ng/g in oranges (15). Furthermore, individual citrus products vary widely in the concentrations of individual furocoumarin compounds. For example, grapefruits are particularly rich in bergamottin and 6′7′-dihydroxybergamottin (6′7′-DHB), but limes exceed grapefruits in their concentrations of bergaptol and bergapten (15). While the NHS and HPFS asked separate questions on grapefruits, grapefruit juice, oranges, and orange juice, WHI participants reported consumption of oranges, grapefruits, and tangerines together, and orange juice and grapefruit juice together, which did not allow us to evaluate these items separately. Furthermore, other major citrus products that are rich in furocoumarins, such as lemons and limes, were not assessed in the FFQ used in the WHI. Therefore, assessment of total citrus consumption using the WHI FFQ is an imperfect approach for examining a potential relationship between furocoumarin consumption and melanoma risk.
Nonetheless, the results of the present study suggested that higher consumption of citrus juices may be associated with incident melanoma among women spending at least 30 minutes outdoors daily during the summer as adults. This study included several sub-analyses examining multiple citrus products and participant subgroups, which could result in false positive findings due to chance. However, the association observed in women spending more time outdoors suggests a potential synergistic interaction between furocoumarins and UV radiation, which is consistent with our hypothesis based on previous data. Interestingly, we found no association between consumption of whole fruits and melanoma risk. This may be related in part to the differing furocoumarin profiles and concentrations between whole citrus fruits and their juices (15, 21, 22). Furocoumarins that are normally found primarily in the peel of citrus fruits may be distributed into the processed fruit juice through the squeezing process (22). Given that various furocoumarin compounds differ in their cytotoxic and carcinogenic potentials (23, 24), the different furocoumarin profiles of citrus fruit compared to juice may therefore contribute to some differences in associated risk observed in this study. However, further research is needed to ascertain which specific furocoumarin compounds may impact skin cancer risk in humans, and whether the amounts consumed in typical diets can meaningfully affect risk.
While citrus consumption was not found to be associated with incident melanoma risk in this study, our data, along with findings of previous human and animal studies, suggest a need for further investigation of a potential relationship between citrus and skin cancer risk, especially among women who have high sun exposure. Human studies including both men and women, and using long-term follow-up and detailed dietary data collection methods will be essential for understanding any public health risk associated with citrus products or dietary furocoumarins. Additional cellular and animal studies are needed to clarify whether certain furocoumarin compounds found in citrus products may meaningfully impact skin cancer risk in humans and to further clarify the underlying mechanisms.
Acknowledgments
Financial Support: This study was supported by the University of Connecticut USDA Hatch-Multistate Competitive Capacity Grant Program (CONS01012, PI: Dr. Ock K. Chun) and the National Cancer Institute (CA198216, PI: Dr. Eunyoung Cho). The Women’s Health Initiative is funded by the National Heart, Lung, and Blood Institute, NIH, U.S. Department of Health and Human Services (grant numbers N01WH22110, 24152, 32100–2, 32105–6, 32108–9, 32111–13, 32115, 32118–32119, 32122, 42107–26, 42129–32, and 44221).
Footnotes
Disclosure Statement: The authors have no conflicts of interest to declare.
References
- 1.Linos E, Swetter SM, Cockburn MG, Colditz GA and Clarke CA: Increasing Burden of Melanoma in the United States. J Invest Dermatol 129: 1666–1674, 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Gandini S, Sera F, Cattaruzza MS, Pasquini P, Zanetti R, Masini C, Boyle P and Melchi CF: Meta-analysis of risk factors for cutaneous melanoma: III. Family history, actinic damage and phenotypic factors. Eur J Cancer 41: 2040–2059, 2005. [DOI] [PubMed] [Google Scholar]
- 3.Bataille V: Sun exposure, sunbeds and sunscreens and melanoma. What are the controversies? Curr Oncol Rep 15: 526–532, 2013. [DOI] [PubMed] [Google Scholar]
- 4.Sample A and He YY: Mechanisms and prevention of UV-induced melanoma. Photodermatol Photoimmunol Photomed: 1–12, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wu H, Reeves KW, Qian J and Sturgeon SR: Coffee, tea, and melanoma risk among postmenopausal women. Eur J Cancer Prev 24: 347–352, 2015. [DOI] [PubMed] [Google Scholar]
- 6.de Waure C, Quaranta G, Gualano MR, Cadeddu C, Jovic-Vranes A, Djikanovic B, La Torre G and Ricciardi W: Systematic review of studies investigating the association between dietary habits and cutaneous malignant melanoma. Public Health 129: 1099–1113, 2015. [DOI] [PubMed] [Google Scholar]
- 7.Loftfield E, Freedman ND, Graubard BI, Hollenbeck AR, Shebl FM, Mayne ST and Sinha R: Coffee drinking and cutaneous melanoma risk in the NIH-AARP diet and health study. J Natl Cancer Inst 107: 1–9, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Wu S, Cho E, Feskanich D, Li W-Q, Sun Q, Han J and Qureshi AA: Citrus consumption and risk of basal cell carcinoma and squamous cell carcinoma of the skin. Carcinogenesis 36: 1162–1168, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wu S, Han J, Feskanich D, Cho E, Stampfer MJ, Willett WC and Qureshi A a.: Citrus Consumption and Risk of Cutaneous Malignant Melanoma. J Clin Oncol 33: 2500–2508, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Melough MM, Cho E and Chun OK: Furocoumarins: A review of biochemical activities, dietary sources and intake, and potential health risks. Food Chem Toxicol 113: 99–107, 2018. [DOI] [PubMed] [Google Scholar]
- 11.Zarebska Z, Waszkowska E, Caffieri S and Dall’Acqua F: PUVA (psoralen + UVA) photochemotherapy: Processes triggered in the cells. Farmaco 55: 515–520, 2000. [DOI] [PubMed] [Google Scholar]
- 12.International Agency for Research on Cancer: Methoxsalen Plus Ultraviolet A. IARC Monogr Eval Carcinog Risks to Humans 100A, 2012. [PMC free article] [PubMed] [Google Scholar]
- 13.Stern RS: The risk of squamous cell and basal cell cancer associated with psoralen and ultraviolet A therapy: A 30-year prospective study. J Am Acad Dermatol 66: 553–562, 2012. [DOI] [PubMed] [Google Scholar]
- 14.Stern RS: The risk of melanoma in association with long-term exposure to PUVA. J Am Acad Dermatol 44: 755–761, 2001. [DOI] [PubMed] [Google Scholar]
- 15.Melough MM, Lee SG, Cho E, Kim K, Provatas AA, Perkins C, Park MK, Qureshi A and Chun OK: Identification and Quantitation of Furocoumarins in Popularly Consumed Foods in the U.S. Using QuEChERS Extraction Coupled with UPLC-MS/MS Analysis. J Agric Food Chem 65: 5049–5055, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Anderson GL, Manson J, Wallace R, Lund B, Hall D, Davis S, Shumaker S, Wang CY, Stein E and Prentice RL: Implementation of the Women’s Health Initiative study design. Ann Epidemiol 13: 5–17, 2003. [DOI] [PubMed] [Google Scholar]
- 17.Langer RD, White E, Lewis CE, Kotchen JM, Hendrix SL and Trevisan M: The women’s health initiative observational study: Baseline characteristics of participants and reliability of baseline measures. Ann Epidemiol 13, 2003. [DOI] [PubMed] [Google Scholar]
- 18.Patterson RE, Kristal AR, Tinker LF, Carter RA, Bolton MP and Agurs-Collins T: Measurement Characteristics of the Women’s Health Initiative Food Frequency Questionnaire. Ann Epidemiol 9: 178–187, 1999. [DOI] [PubMed] [Google Scholar]
- 19.Curb JD, Mctiernan A, Heckbert SR, Kooperberg C, Stanford J, Nevitt M, Johnson KC, Proulx-Burns L, Pastore L, Criqui M and Daugherty S: Outcomes ascertainment and adjudication methods in the women’s health initiative. Ann Epidemiol 13: S122–S128, 2003. [DOI] [PubMed] [Google Scholar]
- 20.Rosenberg CA, Khandekar J, Greenland P, Rodabough RJ and McTiernan A: Cutaneous melanoma in postmenopausal women after nonmelanoma skin carcinoma: The Women’s Health Initiative Observational Study. Cancer 106: 654–663, 2006. [DOI] [PubMed] [Google Scholar]
- 21.Lee SG, Kim K, Vance TM, Perkins C, Provatas A, Wu S, Qureshi A, Cho E and Chun OK: Development of a comprehensive analytical method for furanocoumarins in grapefruit and their metabolites in plasma and urine using UPLC-MS/MS: a preliminary study. Int J Food Sci Nutr 7486: 1–7, 2016. [DOI] [PubMed] [Google Scholar]
- 22.Wangensteen H, Molden E, Christensen H and Malterud KE: Identification of epoxybergamottin as a CYP3A4 inhibitor in grapefruit peel. Eur J Clin Pharmacol 58: 663–668, 2003. [DOI] [PubMed] [Google Scholar]
- 23.Raquet N and Schrenk D: Application of the equivalency factor concept to the phototoxicity and -genotoxicity of furocoumarin mixtures. Food Chem Toxicol 68: 257–266, 2014. [DOI] [PubMed] [Google Scholar]
- 24.Lohr C, Raquet N and Schrenk D: Application of the concept of relative photomutagenic potencies to selected furocoumarins in V79 cells. Toxicol Vitr 24: 558–566, 2010. [DOI] [PubMed] [Google Scholar]