Summary
This prospective study evaluated citrus consumption in relation to risk of cutaneous basal cell carcinoma and squamous cell carcinoma using data from two large cohorts of women and men, and found positive associations between consumption and skin cancer risk that may suggest photocarcinogenic properties of dietary furocoumarins.
Abstract
Animal experiments have demonstrated the photocarcinogenic properties of furocoumarins, a group of naturally occurring chemicals that are rich in citrus products. We conducted a prospective study for citrus consumption and risk of basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) of the skin based on data from 41530 men in the Health Professionals Follow-up Study (1986–2010) and 63759 women in the Nurses’ Health Study (1984–2010) who were free of cancers at baseline. Over 24–26 years of follow-up, we documented 20840 incident BCCs and 3544 incident SCCs. Compared to those who consumed citrus products less than twice per week, the pooled multivariable-adjusted hazard ratios were 1.03 [95% confidence interval (95% CI): 0.99–1.08] for BCC and 1.14 (95% CI: 1.00–1.30) for SCC for those who consumed two to four times per week, 1.06 (95% CI: 1.01–1.11) for BCC and 1.15 (95% CI: 1.02–1.28) for SCC for five to six times per week, 1.11 (95% CI: 1.06–1.16) for BCC and 1.22 (95% CI: 1.08–1.37) for SCC for once to 1.4 times per day and 1.16 (95% CI: 1.09–1.23) for BCC and 1.21 (95% Cl: 1.06–1.38) for SCC for 1.5 times per day or more (P trend = 0.001 for BCC and 0.04 for SCC). In contrast, consumption of non-citrus fruit and juice appeared to be inversely associated with risk of BCC and SCC. Our findings support positive associations between citrus consumption and risk of cutaneous BCC and SCC in two cohorts of men and women, and call for further investigations to better understand the potential photocarcinogenesis associated with dietary intakes.
Introduction
Basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) of the skin are the most frequently diagnosed malignancies in the population (1,2). They are more common than all other cancers combined and can cause substantial morbidity and rising cost to the health-care system (3,4). Knowledge on the modifiable risk factors of these skin cancers is important for the prevention of cancer incidence. Solar ultraviolet (UV) radiation has been recognized as the major environmental factor for skin cancer. However, previous experimental studies suggest that certain photoactive agents, such as furocoumarins, may also increase the risk of skin cancer in the presence of UV radiation (5–11). Furocoumarins (furanocoumarins) are a group of naturally occurring chemicals that are rich in certain plants, including citrus products (12–15). Furocoumarins have high UV absorbance and mutagenic properties (16,17). Oral application of psoralens (a group of furocoumarin derivatives, e.g. 8-methoxypsoralen and 5-methoxypsoralen/bergapten) and UVA radiation (PUVA) has been used as an effective therapy for severe psoriasis and other cutaneous problems (18,19). Interestingly, epidemiologic studies have demonstrated an increased risk of BCC and SCC among patients receiving PUVA treatment (8–10). Commonly consumed citrus products, such as grapefruit and orange, may contain varying amounts of psoralens/furocoumarins (12,13). However, whether dietary consumption of furocoumarin-rich foods may increase the risk of skin cancer is unknown.
Our previous investigation based on data from two large ongoing cohort studies, the Health Professionals Follow-up Study (HPFS, 1986–2010) and the Nurses’ Health Study (NHS, 1984–2010), identified an increased risk of cutaneous malignant melanoma associated with citrus consumption (20). In the present study, we further examined the association between citrus consumption and risk of two other major forms of skin cancer (i.e. BCC and SCC) in the HPFS and NHS.
Methods
Study population
The study population consisted of participants from two ongoing longitudinal cohort studies: HPFS and NHS. The HPFS consisted of 51529 male health professionals who were aged 40–75 and completed their initial questionnaire in 1986. Information on medical history and lifestyle factors was collected biennially via mailed questionnaires in the two cohorts. The NHS was established in 1976 when 121701 married, registered, female nurses who were aged 30–55 years and residing in the USA at the time of enrollment responded to an initial questionnaire regarding their medical history and lifestyle risk factors. The present study was approved by the Institutional Review Boards of Brigham and Women’s Hospital and Harvard School of Public Health. We consider the participants’ completion and return of the self-administered questionnaire as informed consent.
We followed participants for incident BCC and SCC starting from 1986 in the HPFS and 1976 in the NHS. At baseline, 49617 HPFS men and 81685 NHS women completed the dietary questionnaires. Participants who had a history of cancer at baseline were excluded. Owing to small number and low risk of skin cancer in non-white participants (1), the present study only included participants of Caucasian ancestries. After exclusions, 41530 men and 63759 women (total n = 105289) remained in the present study.
Assessment of dietary consumption and other skin cancer risk factors
The present study used a food frequency questionnaire (FFQ) to collect dietary information every 4 years since 1986 in the HPFS and NHS, with an additional dietary questionnaire completed in 1984 in the NHS. Participants responded to the questions regarding how often on average (in nine categories ranging from never to six+ servings per day) during the previous year they had consumed grapefruit (‘a half’), oranges (‘one’), grapefruit juice and orange juice (‘one small glass’ of six fluid ounces) and other food items. Grapefruit and grapefruit juice were asked as a single item in the 2002 and 2006 FFQs. Overall citrus consumption was calculated as the sum of the reported frequencies of these individual citrus products. Dietary intake collected using the FFQ has been demonstrated to be a valid estimator of relative food intake when compared with multiple diet records (21,22). The correlation coefficients ranged from 0.75 to 0.84 for the correlations between intakes of individual citrus products assessed on the baseline FFQ and intakes assessed on two 1-week dietary records (21). Information on other dietary factors, including intakes of total energy, alcohol, coffee, other fruits and juices (13 items) and vegetables (26 items), was also collected by the FFQs.
In the biennial follow-up questionnaires, we inquired and updated information on body weight and height, physical activity, cigarette smoking and menopausal status and post-menopausal hormone use among women. Data on the following skin cancer risk related variables were also collected through the questionnaires (23,24): natural hair color at an early age; number of moles on arms; family history of melanoma in first-degree relatives; skin reaction to sun exposure for 2h or more as a child/adolescent; number of lifetime blistering sunburns; average time spent in direct sunlight since high school; cumulative UV flux at residence since baseline and use of sunscreen.
Assessment of BCC and SCC cases
Biennial questionnaires mailed to all study participants included questions on diagnoses of BCC and SCC during the previous 2 years. We obtained permission from participants who reported new diagnoses of SCC to review their medical and pathological reports. Study physicians who were blinded of the exposure status reviewed the records to validate the diagnoses and retrieve information on tumor stage and location if available. SCCs were further classified into the following two subgroups according to tumor location: tumors occurred on the body sites with higher continuous sun exposure (head, neck and extremities), and tumors occurred on the body sites with lower continuous sun exposure (truncal sites including shoulder, back, hip, genitals, abdomen and chest). Although medical records were not obtained for self-reported BCC, previous validation studies have demonstrated high validity of self-reported BCC in the two cohorts, with 96% women and 84% men confirmed by pathological records (25,26). Over 2 million person-years of follow-up, we documented a total of 20840 incident BCCs (9033 in men and 11807 in women) and 3544 incident SCCs (1540 in men and 2004 in women). SCCs include 2329 invasive cases and 1215 in situ cases, among which 2758 occurred on the body sites with higher continuous sun exposure (head, neck and extremities) and 483 occurred on the body sites with lower continuous sun exposure (truncal sites).
Statistical analysis
To create the best estimates of long-term intake and to minimize within-person variation, each dietary intake was calculated as the average of all reported intakes up to that time prior to every 2-year follow-up interval. Because grapefruit and grapefruit juice were asked as a single item in the 2002 and 2006 FFQs, analyses for separate grapefruit and grapefruit juice used cumulative average intakes up to 1998 for the subsequent follow-up. We created a new intake variable for combined grapefruit and grapefruit juice for sensitivity analyses. Each participant contributed person-time from the return month of the baseline questionnaire to the date of the first report of any cancer, date of death, or the end of follow-up (1 January 2010 for men; 1 June 2010 for women), whichever came first. We used SAS software version 9.2 (SAS Institute, Cary, NC) for all statistical analyses. All statistical tests were two-tailed, and the significance level was set at P < 0.05.
Cox proportional hazards models were used to compute the hazard ratios (HR) with 95% confidence intervals (CI) of BCC and SCC associated with dietary intakes. Multivariable analyses were performed with adjustment for other skin cancer risk factors and potential lifestyle and dietary confounders. We used the most recent information for time-varying variables (e.g. body mass index) prior to each follow-up interval to take into account potential changes over the follow-up. Missing data during any follow-up period were coded as a missing indicator category for categorical variables (e.g. smoking status) and with carried-forward values for continuous variables (e.g. body mass index). Trend tests for a given citrus product were performed by assigning median values for citrus intake categories and treating the new variable as a continuous term in the models. The analyses were performed among men and women separately and then pooled using a random-effects model. We also performed analyses for SCC subtypes divided by tumor stage and site. The two highest consumption categories for each citrus variable were combined to maintain the statistical power in subtype analyses.
Several sensitivity analyses were performed to test the specificity and consistency of the reported associations. To examine whether the positive association with risk of BCC and SCC was specific to citrus products, we computed the HRs of BCC and SCC in association with consumption of non-citrus fruits and juices and vegetables. To address the concern about potential reverse causality between dietary assessment and cancer diagnosis, we performed lag analyses by adding a 2-year interval between dietary intake and cohort follow-up (e.g. we used citrus consumption from the 1984 questionnaire for the follow-up period from 1986 to 1988 in the NHS).
Results
Table 1 shows the baseline characteristics of the study population. There was no appreciable difference in known skin cancer risk factors including phenotypic traits (e.g. red/blonde hair color and skin reaction to sun as a child/adolescent) and sun exposure related variables (i.e. number of lifetime blistering sunburns, average time spent in direct sunlight since high school and annual UV flux at residence) over the citrus intake categories, suggesting homogeneous characteristics of the study participants in terms of host risk profile and sun exposure. Consumption levels of citrus products remained relatively constant over the follow-up, and orange juice was the major contributor of overall citrus consumption (Supplementary Table 1, available at Carcinogenesis Online).
Table 1.
Baseline characteristics of study participants according to frequency of overall citrus consumption in the HPFS and NHS
| <2 per week | 2–4 per week | 5–6 per week | 1–1.4 per day | ≥1.5 per day | |
|---|---|---|---|---|---|
| Men (1986) | |||||
| No. of participants | 10617 | 6236 | 6921 | 9496 | 8260 |
| Age (year)a | 52.0 (9.2) | 52.2 (9.4) | 53.3 (9.6) | 54.3 (9.8) | 54.7 (9.8) |
| Red/blonde hair (%) | 13.3 | 14.2 | 13.0 | 13.8 | 12.6 |
| Arm with moles (%) | 31.4 | 32.2 | 32.7 | 31.9 | 32.0 |
| Painful burn/blisters skin reaction as a child/adolescent (%) | 23.7 | 23.7 | 23.9 | 24.0 | 22.7 |
| Family history of melanoma (%) | 4.2 | 3.6 | 3.9 | 4.3 | 3.7 |
| No. of lifetime blistering sunburns | 13.0 (12.0) | 13.1 (12.1) | 13.1 (12.1) | 12.6 (12.0) | 12.6 (12.2) |
| Average time spent in direct sunlight since high school (h/week) | 9.2 (5.4) | 9.1 (5.3) | 9.2 (5.4) | 9.1 (5.5) | 9.0 (5.6) |
| Annual UV flux at residence (×10–4 RB count) | 132.0 (27.5) | 130.1 (27.1) | 129.1 (26.7) | 127.4 (26.2) | 126.3 (26.2) |
| Body mass index (kg/m2) | 25.1 (5.2) | 25.1 (5.3) | 25.1 (5.1) | 24.9 (4.6) | 24.8 (5.0) |
| Physical activity level (metabolic equivalents h/week) | 17.1 (27.0) | 19.7 (29.3) | 21.1 (30.7) | 22.0 (28.9) | 25.8 (32.3) |
| Current smoker (%) | 14.1 | 10.1 | 8.4 | 7.7 | 6.3 |
| Alcohol intake (g/day) | 12.6 (17.9) | 11.8 (16.0) | 11.8 (15.8) | 11.8 (15.2) | 10.6 (14.6) |
| Coffee (cup/day) | 1.6 (1.8) | 1.4 (1.6) | 1.3 (1.5) | 1.2 (1.5) | 1.1 (1.4) |
| Women (1984) | |||||
| No. of participants | 17660 | 10217 | 9771 | 15269 | 10842 |
| Age (year)a | 49.2 (7.1) | 49.5 (7.2) | 50.0 (7.2) | 50.7 (7.2) | 51.3 (7.0) |
| Red/blonde hair (%) | 15.5 | 16.1 | 15.4 | 15.4 | 16.6 |
| Arm with moles (%) | 36.5 | 37.6 | 36.5 | 37.5 | 39.2 |
| Painful burn/blisters skin reaction as a child/adolescent (%) | 15.6 | 14.8 | 14.0 | 13.8 | 14.6 |
| Family history of melanoma (%) | 6.7 | 6.8 | 7.1 | 7.3 | 7.3 |
| No. of lifetime blistering sunburns | 8.6 (7.0) | 8.8 (7.0) | 8.8 (6.9) | 8.7 (6.9) | 8.6 (7.0) |
| Average time spent in direct sunlight since high school (h/week) | 4.8 (2.7) | 4.9 (2.7) | 5.0 (2.7) | 4.9 (2.7) | 5.0 (2.7) |
| Annual UV flux at residence (×10–4 Robertson-Berger count) | 124.4 (25.9) | 123.0 (25.1) | 122.0 (24.3) | 119.7 (23.0) | 118.8 (22.1) |
| Body mass index (kg/m2) | 25.0 (4.9) | 25.3 (4.8) | 25.2 (4.7) | 24.9 (4.7) | 25.0 (4.8) |
| Physical activity level (metabolic-equivalents h/week) | 12.1 (19.7) | 13.1 (18.8) | 14.1 (19.1) | 14.2 (19.1) | 17.3 (24.6) |
| Current smoker (%) | 32.1 | 25.2 | 21.0 | 20.0 | 18.6 |
| Menopausal status (%) | 46.5 | 46.1 | 45.8 | 45.9 | 46.3 |
| Postmenopausal hormone use (%)b | 24.5 | 24.0 | 25.0 | 24.9 | 24.8 |
| Alcohol intake (g/day) | 7.2 (12.7) | 6.7 (11.4) | 6.9 (11.0) | 7.3 (11.3) | 7.1 (11.0) |
| Coffee (cup/day) | 1.9 (1.9) | 1.8 (1.8) | 1.7 (1.7) | 1.6 (1.6) | 1.6 (1.7) |
Values are means (SD) or percentages and have been standardized to the age distribution of the study population.
aValues are not age adjusted.
bPercentages among postmenopausal women.
No significant heterogeneity was detected between sex-specific results for BCC and SCC (Supplementary Tables 2 and Table 3, available at Carcinogenesis Online, all P for heterogeneity > 0.05), and therefore the pooled risk estimates were presented as the main results. Overall citrus consumption was significantly associated with increased risk of BCC in the pooled analyses (Table 2). Among the individual citrus products, grapefruit and orange juice showed significant positive associations with risk of BCC. Neither grapefruit juice nor oranges was significantly associated with risk of BCC. Similarly, overall citrus consumption was also positively associated with risk of SCC (Table 3). Among the individual citrus products, grapefruit showed the most apparent association with risk of SCC, followed by orange juice, and neither grapefruit juice nor oranges was significantly associated with risk of SCC.
Table 2.
Pooled risk of basal cell carcinoma according to frequency of citrus consumption in the HPFS (1986–2010) and NHS (1984–2010)
| Serving category | P for trend | |||||
|---|---|---|---|---|---|---|
| Overall citrus | <2 per week | 2–4 per week | 5–6 per week | 1–1.4 per day | ≥1.5 per day | |
| No. of person-years | 400456 | 371958 | 465228 | 418628 | 319669 | |
| No. of cases | 3505 | 3665 | 5022 | 4787 | 3861 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.07 (0.99–1.16) | 1.11 (1.06–1.16) | 1.17 (1.12–1.22) | 1.20 (1.15–1.26) | <0.001 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.03 (0.99–1.08) | 1.06 (1.01–1.11) | 1.11(1.06–1.16) | 1.16 (1.09–1.23) | 0.001 |
| Grapefruit | Never | <1 per week | 1 per week | 2 per week | ≥3 per week | |
| No. of person-years | 509252 | 639008 | 349832 | 221228 | 256619 | |
| No. of cases | 4507 | 6460 | 3827 | 2795 | 3251 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.08 (1.04–1.13) | 1.15 (1.10–1.20) | 1.21 (1.15–1.27) | 1.19 (1.14–1.25) | <0.001 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.03 (0.99–1.08) | 1.09 (1.02–1.17) | 1.13 (1.07–1.19) | 1.13 (1.05–1.21) | <0.001 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 1.03 (0.99–1.08) | 1.09 (1.01–1.18) | 1.13 (1.06–1.20) | 1.14 (1.04–1.24) | 0.002 |
| Grapefruit juice | Never | <1 per week | 1 per week | 2 per week | ≥3 per week | |
| No. of person-years | 1092966 | 476005 | 165397 | 110093 | 131479 | |
| No. of cases | 11062 | 5256 | 1768 | 1278 | 1476 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.05 (1.00–1.11) | 1.03 (0.98–1.08) | 1.07 (1.01–1.13) | 1.08 (1.02–1.15) | 0.01 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.00 (0.96–1.04) | 0.99 (0.94–1.04) | 1.02 (0.96–1.08) | 1.06 (1.00–1.12) | 0.08 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 0.98 (0.93–1.04) | 0.96 (0.91–1.01) | 0.98 (0.92–1.04) | 1.02 (0.97–1.08) | 0.80 |
| Oranges | Never | <1 per week | 1 per week | 2 per week | ≥3 per week | |
| No. of person-years | 228853 | 518773 | 401333 | 338726 | 488254 | |
| No. of cases | 2213 | 5157 | 4075 | 3940 | 5455 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.00 (0.95–1.06) | 1.03 (0.97–1.08) | 1.07 (0.93–1.25) | 1.03 (0.94–1.12) | 0.37 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 0.96 (0.91–1.01) | 0.95 (0.90–1.01) | 0.97 (0.86–1.10) | 0.94 (0.86–1.03) | 0.59 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 0.95 (0.90–1.00) | 0.94 (0.89–1.00) | 0.96 (0.84–1.10) | 0.93 (0.84–1.04) | 0.56 |
| Orange juice | <1 per week | 1–2 per week | 3–4 per week | 5–6 per week | ≥1 per day | |
| No. of person-years | 499507 | 451276 | 395566 | 310206 | 319385 | |
| No. of cases | 4723 | 4567 | 4223 | 3706 | 3621 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.05 (1.01–1.09) | 1.09 (1.04–1.13) | 1.15 (1.10–1.20) | 1.17 (1.12–1.22) | <0.001 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.02 (0.97–1.06) | 1.06 (1.01–1.10) | 1.09 (1.04–1.14) | 1.13 (1.05–1.21) | <0.001 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 1.01 (0.97–1.05) | 1.05 (1.01–1.10) | 1.08 (1.03–1.14) | 1.12 (1.05–1.20) | <0.001 |
Multivariable hazard ratios were adjusted for age, natural hair color (red, blonde, light brown, dark brown, black), number of arm moles (0, 1–2, 3–9, ≥10), sunburn susceptibility as a child/adolescent (none/some redness, burn, painful burn/blisters), family history of melanoma (yes, no), number of lifetime blistering sunburns (0, 1–4, 5–9, ≥10), cumulative UV flux since baseline (quintiles), average time spent in direct sunlight since high school (<2, 2–5, 6–10, ≥11h/week), sunscreen use (yes, no), body mass index (<25.0, 25.0–29.9, 30.0–34.9, ≥35.0kg/m2), physical activity (quintiles), smoking status (never, past, current with 1–14, 15–24, or ≥25 cigarettes/day), alcohol intake (0, 0.1–4.9, 5.0–9.9, 10.0–19.9, ≥20.0g/day), coffee intake (0, <1, 1, 2, ≥3 cup/day), and consumption of total other fruit and juice (except citrus products, <0.75, 0.75–1.2, 1.3–1.9, 2.0–2.9, ≥3.0 per day) and total vegetable (<2.0, 2.0–2.9, 3.0–3.9, 4.0–4.9, ≥5.0 per day). Analyses for women were also adjusted for menopausal status and postmenopausal hormone use (premenopausal, postmenopausal never, past or current use). The second multivariable hazard ratios were additionally adjusted for consumption of the other individual citrus products listed in the tables. Results in the HPFS and NHS were pooled using the random-effects model.
Table 3.
Pooled risk of squamous cell carcinoma according to frequency of citrus consumption in the HPFS (1986–2010) and NHS (1984–2010)
| Serving category | P for trend | |||||
|---|---|---|---|---|---|---|
| Overall citrus | <2 per week | 2–4 per week | 5–6 per week | 1–1.4 per day | ≥1.5 per day | |
| No. of person-years | 400456 | 371958 | 465228 | 418628 | 319669 | |
| No. of cases | 530 | 672 | 913 | 825 | 604 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.19 (1.02–1.39) | 1.20 (1.07–1.33) | 1.26 (1.13–1.41) | 1.20 (1.07–1.35) | 0.001 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.14 (1.00–1.30) | 1.15 (1.02–1.28) | 1.22 (1.08–1.37) | 1.21 (1.06–1.38) | 0.04 |
| Grapefruit | Never | <1 per week | 1 per week | 2 per week | ≥3 per week | |
| No. of person-years | 509252 | 639008 | 349832 | 221228 | 256619 | |
| No. of cases | 635 | 1182 | 641 | 530 | 556 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.26 (1.14–1.39) | 1.24 (1.09–1.42) | 1.34 (1.19–1.51) | 1.33 (1.18–1.49) | <0.001 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.16 (1.05–1.28) | 1.15 (0.93–1.41) | 1.23 (1.05–1.45) | 1.30 (1.14–1.48) | <0.001 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 1.16 (1.04–1.28) | 1.15 (0.94–1.41) | 1.25 (1.07–1.47) | 1.31 (1.15–1.49) | <0.001 |
| Grapefruit juice | Never | <1 per week | 1 per week | 2 per week | ≥3 per week | |
| No. of person-years | 1092966 | 476005 | 165397 | 110093 | 131479 | |
| No. of cases | 1776 | 988 | 301 | 212 | 267 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.10 (1.02–1.19) | 1.00 (0.87–1.14) | 0.95 (0.82–1.09) | 1.17 (0.97–1.40) | 0.37 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.03 (0.95–1.11) | 0.96 (0.84–1.11) | 0.90 (0.78–1.04) | 1.17 (1.02–1.36) | 0.42 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 0.99 (0.91–1.07) | 0.91 (0.80–1.04) | 0.84 (0.73–0.98) | 1.11 (0.97–1.27) | 0.85 |
| Oranges | Never | <1 per week | 1 per week | 2 per week | ≥3 per week | |
| No. of person-years | 228853 | 518773 | 401333 | 338726 | 488254 | |
| No. of cases | 295 | 892 | 747 | 752 | 858 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.11 (0.97–1.27) | 1.20 (1.05–1.38) | 1.16 (1.01–1.33) | 1.02 (0.89–1.17) | 0.45 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.01 (0.88–1.16) | 1.06 (0.92–1.22) | 1.01 (0.87–1.17) | 0.93 (0.80–1.08) | 0.12 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 0.99 (0.86–1.13) | 1.04 (0.90–1.20) | 0.99 (0.85–1.15) | 0.91 (0.78–1.06) | 0.12 |
| Orange juice | <1 per week | 1–2 per week | 3–4 per week | 5–6 per week | ≥1 per day | |
| No. of person-years | 499507 | 451276 | 395566 | 310206 | 319385 | |
| No. of cases | 723 | 816 | 768 | 662 | 575 | |
| Age-adjusted HR (95% CI) | 1.00 | 1.10 (0.95–1.26) | 1.17 (1.00–1.36) | 1.15 (1.03–1.28) | 1.22 (1.10–1.37) | <0.001 |
| Multivariable-adjusted HR1 (95% CI) | 1.00 | 1.04 (0.94–1.15) | 1.12 (1.01–1.24) | 1.06 (0.95–1.18) | 1.18 (1.06–1.32) | 0.009 |
| Multivariable-adjusted HR2 (95% CI) | 1.00 | 1.03 (0.93–1.14) | 1.11 (0.99–1.23) | 1.05 (0.94–1.17) | 1.17 (1.05–1.31) | 0.01 |
Multivariable hazard ratios were further adjusted for the covariates listed in the Table 2. Results in the HPFS and NHS were pooled using the random-effects model.
Associations between citrus consumption and risk of BCC and SCC remained essentially unchanged when we added a 2-year lag between dietary assessment and cohort follow-up (data not shown). Subtype analyses for SCC showed similar associations with citrus consumption for invasive and in situ cases (data not shown). Interestingly, the positive association of SCC with citrus consumption appeared to be more apparent for tumors occurred on the body sites with higher continuous sun exposure (head, neck and extremities) than for tumors occurred on the body sites with lower continuous sun exposure (truncal sites) (Table 4). There were significant trends towards higher risk for SCC on head, neck and extremities (all P trend < 0.05) but not for SCC on truncal sites (all P trend > 0.50) in association with consumption of overall citrus, grapefruit and orange juice.
Table 4.
Pooled risk of site-specific squamous cell carcinoma according to frequency of citrus consumption in the HPFS (1986–2010) and NHS (1984–2010)
| Serving category | P for trend | ||||
|---|---|---|---|---|---|
| Overall citrus | <2 per week | 2–4 per week | 5–6 per week | ≥1 per day | |
| No. of person-years | 400456 | 371958 | 465228 | 738297 | |
| Risk of SCC on the body sites with higher continuous sun exposure (head, neck and extremities) | |||||
| No. of cases | 406 | 526 | 722 | 1104 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 1.15 (1.01–1.32) | 1.17 (1.02–1.33) | 1.23 (1.08–1.39) | 0.004 |
| Risk of SCC on the body sites with lower continuous sun exposure (truncal sites) | |||||
| No. of cases | 78 | 91 | 126 | 188 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 1.01 (0.74–1.38) | 1.07 (0.63–1.81) | 1.07 (0.76–1.50) | 0.70 |
| Grapefruit | Never | <1 per week | 1 per week | ≥2 per week | |
| No. of person-years | 509252 | 639008 | 349832 | 477847 | |
| Risk of overall SCC on the body sites with higher continuous sun exposure (head, neck and extremities) | |||||
| No. of cases | 471 | 935 | 496 | 856 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 1.21 (1.08–1.36) | 1.18 (0.94–1.47) | 1.33 (1.15–1.53) | 0.01 |
| Risk of overall SCC on the body sites with lower continuous sun exposure (truncal sites) | |||||
| No. of cases | 93 | 167 | 84 | 139 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 1.09 (0.83–1.43) | 1.03 (0.75–1.43) | 1.15 (0.84–1.57) | 0.58 |
| Grapefruit juice | Never | <1 per week | 1 per week | ≥2 per week | |
| No. of person-years | 1092966 | 476005 | 165397 | 241572 | |
| Risk of overall SCC on the body sites with higher continuous sun exposure (head, neck and extremities) | |||||
| No. of cases | 1364 | 765 | 236 | 393 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 0.97 (0.88–1.06) | 0.91 (0.79–1.05) | 1.02 (0.91–1.15) | 0.89 |
| Risk of overall SCC on the body sites with lower continuous sun exposure (truncal sites) | |||||
| No. of cases | 247 | 140 | 45 | 51 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 0.99 (0.80–1.24) | 0.98 (0.65–1.47) | 0.74 (0.54–1.02) | 0.18 |
| Oranges | Never | <1 per week | 1 per week | ≥2 per week | |
| No. of person-years | 228853 | 518773 | 401333 | 826980 | |
| Risk of overall SCC on the body sites with higher continuous sun exposure (head, neck and extremities) | |||||
| No. of cases | 210 | 699 | 603 | 1246 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 1.03 (0.88–1.21) | 1.10 (0.93–1.30) | 0.96 (0.81–1.13) | 0.41 |
| Risk of overall SCC on the body sites with lower continuous sun exposure (truncal sites) | |||||
| No. of cases | 43 | 117 | 92 | 231 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 0.82 (0.57–1.18) | 0.80 (0.54–1.18) | 0.88 (0.60–1.29) | 0.97 |
| Orange juice | <1 per week | 1–2 per week | 3–4 per week | ≥5 per week | |
| No. of person-years | 499507 | 451276 | 395566 | 629591 | |
| Risk of overall SCC on the body sites with higher continuous sun exposure (head, neck and extremities) | |||||
| No. of cases | 542 | 656 | 587 | 973 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 1.06 (0.93–1.21) | 1.10 (0.96–1.25) | 1.13 (1.01–1.26) | 0.03 |
| Risk of overall SCC on the body sites with lower continuous sun exposure (truncal sites) | |||||
| No. of cases | 102 | 107 | 111 | 163 | |
| Multivariable-adjusted HR (95% CI) | 1.00 | 0.92 (0.70–1.21) | 1.13 (0.75–1.72) | 1.02 (0.64–1.64) | 0.73 |
Multivariable hazard ratios were adjusted for the covariates listed in the Table 2. Results in the HPFS and NHS were pooled using the random-effects model.
We did not find any significant positive association between consumption of other non-citrus fruits, juices and vegetables and risk of BCC or SCC. Instead, we found primarily inverse associations between these food items and disease risk. For example, the fully adjusted HRs comparing the extreme consumption categories of total non-citrus fruit and juice (13 items, ≥3.0 per day versus <0.75 per day) were 0.90 (95% CI: 0.84–0.96, P trend = 0.049) for BCC and 0.87 (95% CI: 0.74–1.03, P trend = 0.06) for SCC (Supplementary Table 4, available at Carcinogenesis Online).
We conducted interaction tests to evaluate whether the association between citrus consumption and risk of SCC and BCC varied by other potential confounders, and found that there was no significant interaction between citrus consumption and other variables adjusted in the analysis (all P interaction > 0.10, data not shown). We further examined the risk of other major non-skin cancers (e.g. breast cancer, prostate cancer, lung cancer and colorectal cancer) in association with citrus consumption, and did not find any similar positive associations (data not shown).
Discussion
In the present study, we performed a detailed analysis for the association between citrus consumption and risk of cutaneous BCC and SCC based on data from two large cohorts of men and women. After adjusting for other known skin cancer risk factors and potential confounders, citrus consumption was positively associated with increased risk of BCC and SCC. Among the individual citrus products, grapefruit and orange juice showed consistent positive associations with risk of BCC and SCC. These findings are generally consistent with our previous investigation for citrus consumption and risk of cutaneous malignant melanoma (20), both of which support a potentially increased risk of skin cancer associated with consumption of furocoumarin-containing foods.
The potential photocarcinogenesis of furocoumarins has well-documented biological plausibility. Furocoumarins/psoralens have been identified as a group of carcinogens for decades (5,7). Previous animal experiments have well demonstrated that furocoumarins are able to induce skin tumors in the presence of UV radiation (5–7,11). Mechanistic investigations have revealed a linear relation between epidermal and serum concentrations of psoralens after oral administration, and the appearance of phototoxicity is associated with the serum concentrations of psoralens (27). Furocoumarins plus UV radiation could induce skin erythema, edema, delayed pigmentation and increased activity of epidermal ornithine decarboxylase, which may serve as a biomarker for cutaneous tumor promotion (28,29). Furocoumarins could also induce lethal, mutagenic and clastogenic effects in mammalian cells and other organisms (16,17). Photoexcited furocoumarins can react with biomolecules, especially with pyrimidine bases in DNA, and form mono- and di-adducts (30,31), and photocycloaddition reactions initiated by furocoumarins play an important role in the formation of DNA adducts (18). Although DNA is generally assumed as the primary site of action for furocoumarins (31,32), they may also bind to other specific and high-affinity sites in mammalian cells which may in turn mediate the furocoumarin-induced phototoxicity in part (33).
Citrus products are known to contain furocoumarins (12–15), and orange and grapefruit are the two mostly consumed citrus fruits in the population over the past several decades, accounting for over 90% of citrus market shares (34). Although grapefruit and orange juice were significantly and positively associated with risk of BCC and SCC, grapefruit juice and oranges did not showed appreciable positive associations with these outcome diseases. Several reasons may help explain the different associations. Grapefruit generally contains higher levels of furocoumarins than oranges (12,13). The null association of grapefruit juice with skin cancer risk may be explained by its much lower consumption levels (Supplementary Table 1, available at Carcinogenesis Online) and a much larger number of non-consumers (Table 2) when compared with the other individual citrus products. In contrast, orange juice contributed to more than 50% of the overall citrus consumption, and the significant association of orange juice with skin cancer risk may be explained by its much higher consumption levels when compared with the other individual citrus products (Supplementary Table 1, available at Carcinogenesis Online).
In addition, we found that the positive association between citrus consumption and SCC risk appeared to be more apparent for tumors occurred on the body sites with higher continuous sun exposure (head, neck and extremities) than for tumors occurred on the body sites with lower continuous sun exposure (truncal sites). Interestingly, animal experiments have demonstrated that exposure to furocoumarins or UVA alone is not tumorigenic in mice whereas exposure to furocoumarins plus UVA substantially increases the number of mice with skin tumors (6,11). Therefore, our findings may suggest a potential synergistic effect between citrus consumption and UV radiation.
Our study has several strengths, including the prospective design, the large sample size and large number of skin cancer cases, the long-term follow-up over 24–26 years, the repeated assessment of dietary and lifestyle factors, and the ability to include a number of potential confounders. Nevertheless, our study also has several limitations. Our study populations consisted of well-educated Caucasian health professionals and may not be representative of the general population. Nevertheless, restricting the sample to health professionals also reduces potential residual confounding from socioeconomic status. Our results need to be replicated in future studies with sufficient power to detect similar associations among other populations. In addition, the diagnosis of BCC was assessed based on self-reports without pathological validation. However, the health care background of our study participants suggest that their reports were likely to be highly accurate, as proven in previous validation studies (25,26). The positive association between citrus consumption and BCC risk as reported herein is also consistent between the two study cohorts. These data suggest that the bias due to self-reported BCC is likely to be minimal and unlikely to affect the study results materially. Furthermore, we do not have data on exposure to arsenic, which has been linked to the risk of skin cancer (35). However, our study participants have better health awareness than the general population and are therefore less likely to be exposed to arsenic, either occupationally or non-occupationally.
In conclusion, our study based on two large cohorts of men and women demonstrated that citrus consumption was associated with an increased risk of cutaneous BCC and SCC, the most commonly diagnosed malignancies in the population. These findings together with our previous investigation on citrus consumption and risk of cutaneous malignant melanoma (20) provide evidence for the potential photocarcinogenic effect of commonly consumed foods. However, our study findings need to be confirmed in future studies before a clear causal inference could be obtained. Nevertheless, given the high prevalence of citrus consumption as well as skin cancers in the population, our findings hold general public health significance and may serve as the first effort to initiate future research in this area.
Supplementary material
Supplementary Tables 1-4. can be found at http://carcin.oxfordjournals.org/
Funding
National Institute of Health (UM1 CA186107, P01 CA87969, UM1 CA167552 and R01 CA137165). Dr. Shaowei Wu also received an Amgen Medical Dermatology Fellowship support from the National Psoriasis Foundation.
Supplementary Material
Acknowledgements
We would like to thank the participants and staff of the Nurses’ Health Study and Health Professionals Follow-up Study for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for analyses and interpretation of these data.
Conflict of Interest Statement: None declared.
Glossary
Abbreviations
- BCC
basal cell carcinoma
- CI
confidence interval
- FFQ
food frequency questionnaire
- HR
hazardous ratio
- HPFS
Health Professionals Follow-up Study
- NHS
Nurses’ Health Study
- SCC
squamous cell carcinoma
- UV
ultraviolet
References
- 1. American Cancer Society. (2014) Cancer Facts & Figures 2014. American Cancer Society, Atlanta, GA. [Google Scholar]
- 2. Wehner M.R., et al. (2012) Indoor tanning and non-melanoma skin cancer: systematic review and meta-analysis. BMJ, 345, e5909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Rogers H.W., et al. (2010) Incidence estimate of nonmelanoma skin cancer in the United States, 2006. Arch. Dermatol., 146, 283–287. [DOI] [PubMed] [Google Scholar]
- 4. Guy G.P., et al. (2011) Years of potential life lost and indirect costs of melanoma and non-melanoma skin cancer: a systematic review of the literature. Pharmacoeconomics, 29, 863–874. [DOI] [PubMed] [Google Scholar]
- 5. GRIFFIN A.C., et al. (1958) The wave length effect upon erythemal and carcinogenic response in psoralen treated mice. J. Invest. Dermatol., 31, 289–295. [PubMed] [Google Scholar]
- 6. Cartwright L.E., et al. (1983) Psoralen-containing sunscreen is tumorigenic in hairless mice. J. Am. Acad. Dermatol., 8, 830–836. [DOI] [PubMed] [Google Scholar]
- 7. Mullen M.P., et al. (1984) Carcinogenic effects of monofunctional and bifunctional furocoumarins. Natl. Cancer Inst. Monogr., 66, 205–210. [PubMed] [Google Scholar]
- 8. Stern R.S., et al. (1984) Cutaneous squamous-cell carcinoma in patients treated with PUVA. N. Engl. J. Med., 310, 1156–1161. [DOI] [PubMed] [Google Scholar]
- 9. Stern R.S., et al. (1998) Oral psoralen and ultraviolet-A light (PUVA) treatment of psoriasis and persistent risk of nonmelanoma skin cancer. PUVA Follow-up Study. J. Natl. Cancer Inst., 90, 1278–1284. [DOI] [PubMed] [Google Scholar]
- 10. Stern R.S. (2012) 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, 553–62. [DOI] [PubMed] [Google Scholar]
- 11. Zajdela F., et al. (1981) 5-Methoxypsoralen, the melanogenic additive in sun-tan preparations, is tumorigenic in mice exposed to 365nm u.v. radiation. Carcinogenesis, 2, 121–127. [DOI] [PubMed] [Google Scholar]
- 12. Dugo P., et al. (2000) LC-MS for the identification of oxygen heterocyclic compounds in citrus essential oils. J. Pharm. Biomed. Anal., 24, 147–154. [DOI] [PubMed] [Google Scholar]
- 13. Dugo P., et al. (2009) Determination of oxygen heterocyclic components in citrus products by HPLC with UV detection. J. Agric. Food Chem., 57, 6543–6551. [DOI] [PubMed] [Google Scholar]
- 14. Frérot E., et al. (2004) Quantification of total furocoumarins in citrus oils by HPLC coupled with UV, fluorescence, and mass detection. J. Agric. Food Chem., 52, 6879–6886. [DOI] [PubMed] [Google Scholar]
- 15. Lin Y.K., et al. (2009) Development of a reversed-phase high-performance liquid chromatographic method for analyzing furanocoumarin components in citrus fruit juices and Chinese herbal medicines. J. Chromatogr. Sci., 47, 211–215. [DOI] [PubMed] [Google Scholar]
- 16. Ashwood-Smith M.J., et al. (1980) 5-Methoxypsoralen, an ingredient in several suntan preparations, has lethal, mutagenic and clastogenic properties. Nature, 285, 407–409. [DOI] [PubMed] [Google Scholar]
- 17. Scott B.R., et al. (1976) Molecular and genetic basis of furocoumarin reactions. Mutat. Res., 39, 29–74. [DOI] [PubMed] [Google Scholar]
- 18. Kitamura N., et al. (2005) Molecular aspects of furocoumarin reactions: Photophysics, photochemistry, photobiology, and structural analysis. J. Photochem. Photobiol. C Photochem. Rev., 6, 168–185. [Google Scholar]
- 19. Parrish J.A., et al. (1974) Photochemotherapy of psoriasis with oral methoxsalen and longwave ultraviolet light. N. Engl. J. Med., 291, 1207–1211. [DOI] [PubMed] [Google Scholar]
- 20. Wu S., et al. (2015) Citrus consumption and risk of cutaneous malignant melanoma. J. Clin. Oncol., pii: JCO.2014.57.4111, epub ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Feskanich D., et al. (1993) Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J. Am. Diet. Assoc., 93, 790–796. [DOI] [PubMed] [Google Scholar]
- 22. Salvini S., et al. (1989) Food-based validation of a dietary questionnaire: the effects of week-to-week variation in food consumption. Int. J. Epidemiol., 18, 858–867. [DOI] [PubMed] [Google Scholar]
- 23. Wu S., et al. (2013) Basal-cell carcinoma incidence and associated risk factors in U.S. women and men. Am. J. Epidemiol., 178, 890–897. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wu, S. et al. (2014) Cumulative ultraviolet radiation flux in adulthood and risk of incident skin cancers in women. Brit. J. Cancer, 110, 1855–1861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Colditz G.A., et al. (1986) Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am. J. Epidemiol., 123, 894–900. [DOI] [PubMed] [Google Scholar]
- 26. van Dam R.M., et al. (2000) Diet and basal cell carcinoma of the skin in a prospective cohort of men. Am. J. Clin. Nutr., 71, 135–141. [DOI] [PubMed] [Google Scholar]
- 27. Kornhauser A., et al. (1982) Psoralen phototoxicity: correlation with serum and epidermal 8-methoxypsoralen and 5-methoxypsoralen in the guinea pig. Science, 217, 733–735. [DOI] [PubMed] [Google Scholar]
- 28. Lowe N.J. (1980) Epidermal ornithine decarboxylase, polyamines, cell proliferation, and tumor promotion. Arch. Dermatol., 116, 822–825. [PubMed] [Google Scholar]
- 29. Walter J.F., et al. (1982) Psoralen-containing sunscreen induces phototoxicity and epidermal ornithine decarboxylase activity. J. Am. Acad. Dermatol., 6, 1022–1027. [DOI] [PubMed] [Google Scholar]
- 30. Tessman J.W., et al. (1985) Photochemistry of the furan-side 8-methoxypsoralen-thymidine monoadduct inside the DNA helix. Conversion to diadduct and to pyrone-side monoadduct. Biochemistry, 24, 1669–1676. [DOI] [PubMed] [Google Scholar]
- 31. Vos J.M., et al. (1987) Processing of psoralen adducts in an active human gene: repair and replication of DNA containing monoadducts and interstrand cross-links. Cell, 50, 789–799. [DOI] [PubMed] [Google Scholar]
- 32. Dall’Acqua F., et al. (1971) Formation of inter-strand cross-linkings in the photoreactions between furocoumarins and DNA. Z. Naturforsch. B., 26, 561–569. [DOI] [PubMed] [Google Scholar]
- 33. Laskin J.D., et al. (1985) A possible mechanism of psoralen phototoxicity not involving direct interaction with DNA. Proc. Natl. Acad. Sci. USA, 82, 6158–6162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. US Department of Agriculture Economic Research Service (2014) Fruit and tree nut data. http://www.ers.usda.gov/data-products/fruit-and-tree-nut-data/yearbook-tables.aspx. (31 January 2015, date last accessed). [Google Scholar]
- 35. Surdu S. (2014) Non-melanoma skin cancer: occupational risk from UV light and arsenic exposure. Rev. Environ. Health, 29, 255–264. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.