ABSTRACT
Background
In prior studies, higher citrus consumption was associated with higher risk of cutaneous malignant melanoma, squamous cell carcinoma (SCC), and basal cell carcinoma (BCC). Furocoumarins, compounds with phototoxicity and photocarcinogenicity in citrus, may be responsible for the association.
Objectives
The objective of the study was to investigate the association between furocoumarin intake and skin cancer risk.
Methods
A total of 47,453 men from the Health Professionals Follow-Up Study (HPFS) and 75,291 women from the Nurses’ Health Study (NHS) with diet data collected every 2–4 y in the 2 prospective cohort studies were included. A furocoumarin food composition database for 7 common furocoumarins [bergaptol, psoralen, 8-methoxypsoralen, bergapten, 6',7'-dihydroxybergamottin (6'7'-DHB), epoxybergamottin, and bergamottin] was developed and used to calculate participants’ cumulative average and energy-adjusted furocoumarin intake. Multivariate HRs and 95% CIs of the associations between furocoumarin intake and skin cancer risk were estimated using Cox proportional hazards models. Analyses were performed separately in each cohort as well as pooled using a fixed-effects model.
Results
Throughout follow-up (1984–2012 in the NHS and 1986–2012 in the HPFS), we identified 1593 melanoma, 4066 SCC, and 28,630 BCC cases. Higher intake of total furocoumarins was associated with an increased risk of BCC; the pooled HR comparing the top with the bottom quintile was 1.16 (95% CI: 1.11, 1.21; P-trend = 0.002). Higher intakes of bergaptol, bergapten, 6'7'-DHB, and bergamottin were also significantly associated with increased BCC risk. No significant associations were found between intake of total furocoumarins and the risks of SCC or melanoma.
Conclusions
Intakes of total furocoumarins as well as some individual furocoumarins were associated with an increased risk of skin cancer, especially BCC, in 2 cohorts of US health professionals.
Keywords: epidemiology, cohort study, citrus fruit, furocoumarin, melanoma, skin cancer
Introduction
Skin cancer including melanoma, squamous cell carcinoma (SCC), and basal cell carcinoma (BCC) is the most common type of cancer and associated with high morbidity and mortality across the world (1–3). Hence, identification of risk factors for skin cancer is crucial for the prevention of the disease. In recent analyses in the Nurses’ Health Study (NHS) and Health Professionals Follow-up Study (HPFS), 2 large prospective cohort studies of women and men, respectively, higher citrus consumption was associated with an increased risk of melanoma, BCC, and SCC (4, 5).
Citrus fruits are major dietary sources of furocoumarins (6–9). As phytoalexins, dozens of congeners of furocoumarins show photoactivity upon UV radiation as a defense against micro-organism infection (10). Several experimental studies have demonstrated the phototoxicity and photocarcinogenicity of furocoumarins (11–14); furocoumarins are photoactive, and can intercalate DNA and induce mutation (6, 13, 15–19). Furthermore, epidemiological studies have observed that psoriasis patients treated with a combination therapy of UVA and oral psoralen (PUVA therapy), a type of furocoumarins, had increased risk of malignant melanoma, BCC, and SCC (20–22).
Therefore, we hypothesized that the increased risk of skin cancer among individuals with higher amounts of citrus intake was due to furocoumarins in citrus. Using a recently developed furocoumarin food composition database (9), we examined the association between furocoumarin intake and the risk of skin cancer among participants in the NHS and HPFS.
Methods
Study population
Established in 1976, the NHS consists of 121,700 US female registered nurses whose ages ranged from 30 to 55 y at enrollment. The HPFS was initiated in 1986, and 51,529 US male health professionals between the ages of 40 and 75 y were enrolled at baseline. Participants received biennial questionnaires, which collected information on medical history and lifestyle factors. Participants completed FFQs every 2–4 y to assess dietary intake. Individuals with no information on diet or a history of any cancer at baseline were excluded from our analysis. Owing to the small proportion and lower risk of skin cancer among nonwhite participants, we limited our analysis to white participants. Although the NHS assessed diet in 1980, we started the follow-up from 1984 to use the dietary data from 1984, which were derived from an expanded FFQ with several food items contributing to furocoumarin intake additional to the 1980 FFQ. We used dietary intake from FFQs in 1984, 1986, 1990, 1994, and 1998 in the NHS and 1986, 1990, 1994, and 1998 in the HPFS. Later FFQs were not used because grapefruit and grapefruit juice were asked together as 1 food item. This study was approved by the Institutional Review Boards of the Brigham and Women's Hospital and Harvard TH Chan School of Public Health, and those of participating registries as required. The completion and return of the self-reported questionnaires by the study participants was considered as informed consent.
Development of the furocoumarin food composition database and assessment of furocoumarin intake
For each food item on the FFQ, participants indicated their frequency of intake on average over the previous year (choosing from 9 categories ranging from “never” to “more than 6 servings per day”). Selection of the food items in the FFQ was based on popularity of foods among the participants in the specific time period. Pilot studies have been conducted to assess changes in food choices and marketplace changes repeatedly over the years. Among the food items in the FFQ, 10 food items were known to contain furocoumarins, including orange, regular orange juice, fortified orange juice with calcium or vitamin D, grapefruit, grapefruit juice, raw carrots, cooked carrots, carrot juice, celery, and lemonade. These food items were matched to a furocoumarin database of popularly consumed foods. Briefly, to construct the database, 3 samples for each food item were purchased from 17 grocery stores in central and eastern Connecticut (9). The concentrations of 7 common furocoumarins [bergaptol, psoralen, 8-methoxypsoralen (8-MOP; methoxsalen), bergapten (5-methoxypsoralen), 6',7'-dihydroxybergamottin (6'7'-DHB), epoxybergamottin, and bergamottin] that are known to be rich in grapefruit were selected for analysis in these food items, and their concentrations were measured in each food item using ultra-performance liquid chromatography with tandem mass spectrometry (UPLC-MS/MS). The mean concentration of each furocoumarin among the 3 samples of each food item was calculated and this mean value was reported in the database and used as the basis for our furocoumarin exposure estimation.
The intake of each furocoumarin for each participant in the NHS and HPFS was calculated as the product of the consumption frequency of a serving of each furocoumarin-containing food item and the furocoumarin content per serving. Based on previous results (23), bergamottin, bergapten, bergaptol, and 6'7'-DHB contributed most to total furocoumarin intake, whereas the intake levels of other furocoumarins including psoralen, 8-MOP, and epoxybergamottin from diet were low. Therefore, we examined both total furocoumarins as well as the individual furocoumarins that contribute most to total furocoumarin intake (i.e., bergamottin, bergapten, 6'7'-DHB, and bergaptol). We used the regression-residual method to adjust furocoumarin intakes for total energy intake (24).
Because dietary intake may affect carcinogenesis over an extended period of time, we calculated cumulative average intakes of furocoumarins to take advantage of repeated dietary intake information and to best estimate long-term intake. For example, in the NHS, 1984 intake was used for the 1984–1986 follow-up period, and the average of 1984 and 1986 intake was used for the 1986–1990 follow-up and so on to conduct a prospective analysis (25). Thus, average furocoumarin intake of the previous time period was evaluated with disease outcomes.
Ascertainment of skin cancer cases
Participants reported incident melanoma, SCC, and BCC during follow-up. After obtaining consent from participants who reported new diagnosis of melanoma or SCC, medical and pathological records were collected and reviewed by physicians who were blinded to the patients’ exposure information. Although we did not obtain medical records for self-reported BCC cases, previous studies of validation by pathological records showed high validity of self-reported BCC in both men and women (26, 27). According to medical records, melanoma and SCC were further classified as in situ or invasive. Only invasive cases were included in the analyses. Melanoma and SCC were further classified into 2 subgroups based on tumor location: tumors that occurred on the more intensively sun-exposed body sites [head, neck, upper extremity (from arm to elbow), leg, ankle, and knee] and those that occurred on the less sun-exposed body sites (trunk, shoulder, hip, back, abdomen, thigh, buttock, anus, vulva, and chest).
Assessment of covariates
Information on the following potential covariates was obtained and updated whenever available through the biennial questionnaires: age (continuous), number of lifetime blistering sunburns (none, 1–2, 3–5, or ≥6), number of arm moles (0, 1–2, 3–5, or ≥6), natural hair color (red, blonde, light brown, or dark brown or black), sunburn susceptibility as child or adolescent (none or some redness, burn, or painful burn or blisters), family history of melanoma (yes/no), alcohol intake (0, 0.1–4.9, 5.0–9.9, 10.0–19.9, or ≥20.0 g/d), total energy intake (quintiles), routine physical examination (yes/no), UV exposure at residence (quintiles), and physical activity (quintiles).
Statistical analysis
Participants contributed person-time from the return date of the 1984 (NHS) or 1986 (HPFS) questionnaire until the date of melanoma, SCC, or BCC diagnosis; diagnosis of any other cancer; death; or the end of follow-up (June 2012 for the NHS or January 2012 for the HPFS), whichever came first. After excluding ineligible individuals, 47,453 men and 75,291 women remained in the analysis for melanoma (Supplemental Figure 1). The numbers were slightly different for BCC and SCC owing to the exclusion of prevalent cases at baseline.
Furocoumarin intake was categorized into quintiles based on the overall distribution of person-time (Supplemental Table 1). For some of the furocoumarins, the bottom quintile consisted of those with “0” intake. Spearman correlation coefficients (ρ) were calculated between intakes of total and individual furocoumarins and citrus intake. To estimate the association between cumulative average intake of furocoumarins and risk of skin cancer, Cox proportional hazards models were used to estimate the HRs and 95% CIs. Initial models adjusted for age and calendar time, whereas multivariable-adjusted models further adjusted for potential dietary and lifestyle confounders and known skin cancer risk factors (number of lifetime blistering sunburns, number of arm moles, natural hair color, sunburn susceptibility as child or adolescent, family history of melanoma, alcohol intake, total energy intake, routine physical examination, UV exposure at residence, and physical activity). Tests for trend across categories of furocoumarin intake were performed by assigning the median value of each quintile of intake and modeling this variable as a continuous term. In a secondary analysis for melanoma and SCC, we evaluated the cases according to their occurrence on body sites, comparing high- with low-exposed sites. We also stratified the analyses by other skin cancer risk factors including number of lifetime sunburns that blistered (<6 or ≥6 times), UV exposure at residence (≥176 mW/m2 or <176 mW/m2), hair color (red/blonde compared with dark/light brown or black color), and tanning ability in the NHS (practically none, light tan, average tan, or deep tan). Analyses were conducted separately in the NHS and HPFS, and pooled using a fixed-effects model. SAS software version 9.4 (SAS Institute) was used for all statistical analyses. All statistical tests were performed as 2-tailed, and a P value < 0.05 was considered to be significant.
Results
During >2 million person-years of follow-up, 1593 melanoma cases (776 in women, 817 in men), 4066 SCC cases (2222 in women, 1844 in men), and 28,630 BCC cases (17,544 in women, 11,086 in men) were identified. Table 1 shows baseline characteristics of study participants by total furocoumarin intake. Participants with higher furocoumarin intake were more likely to be older, have higher citrus intake, and show higher physical activity in both men and women, and less likely to get frequent sunburn in men. No appreciable difference was observed in terms of other skin cancer risk factors, such as hair color, burn reaction during childhood or adolescence, UV exposure at residence, and number of lifetime sunburns. Citrus fruit intake was correlated with total furocoumarin intake (ρ = 0.49 in the NHS and 0.52 in the HPFS; Supplemental Table 2). Total furocoumarin intake was highly correlated with intakes of individual furocoumarins (ρ > 0.9) except for bergapten (ρ = 0.56 in the NHS and 0.55 in the HPFS). Major food sources of total furocoumarin intake included grapefruit (70% in the NHS and 74% in the HPFS) and grapefruit juice (29% in the NHS and 25% in the HPFS), which led to a skewed intake distribution of total furocoumarin intake. Other foods such as carrot, celery, orange juice, and coleslaw contributed <1% of the intake.
TABLE 1.
Baseline age-standardized characteristics of study participants by quintiles of energy-adjusted total furocoumarin intake in the NHS and HPFS1
| Intake quintiles | |||||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | Q5 | |
| Women (NHS, 1984) | |||||
| Participants, n | 15,108 | 14,966 | 15,115 | 15,051 | 15,051 |
| Age, y | 49 ± 7 | 50 ± 7 | 50 ± 7 | 51 ± 7 | 52 ± 10 |
| Furocoumarin intake, μg/d | 0.00 ± 0.00 | 22.0 ± 36.0 | 183 ± 45.0 | 419 ± 112 | 1750 ± 1570 |
| Total citrus intake, serving/d | 0.30 ± 0.40 | 0.90 ± 0.70 | 0.90 ± 0.60 | 0.90 ± 0.60 | 1.40 ± 0.90 |
| Total energy intake, kcal/d | 1680 ± 588 | 1890 ± 575 | 1950 ± 388 | 1660 ± 493 | 1540 ± 470 |
| UV exposure at residence, mW/m2 | 194 ± 28 | 191 ± 27 | 192 ± 26 | 192 ± 28 | 191 ± 28 |
| Family history of melanoma | 2.6 | 2.9 | 2.8 | 2.7 | 2.7 |
| Red/blonde hair | 16 | 16 | 16 | 15 | 16 |
| Painful burn or blistering reaction to sun exposure as a child/adolescent | 36 | 36 | 35 | 33 | 34 |
| ≥6 lifetime sunburns | 8 | 7 | 8 | 7 | 7 |
| ≥6 moles on the arms | 5 | 5 | 4 | 5 | 5 |
| Physical exam | 68 | 72 | 72 | 72 | 73 |
| BMI, kg/m2 | 25 ± 5 | 25 ± 5 | 25 ± 5 | 25 ± 5 | 25 ± 4 |
| Physical activity, MET h/wk | 12 ± 18 | 14 ± 20 | 15 ± 21 | 17 ± 22 | 19 ± 25 |
| Alcohol intake, g/d | 6.50 ± 12.2 | 7.00 ± 11.9 | 7.40 ± 11.5 | 6.90 ± 10.6 | 6.90 ± 10.3 |
| Current smoking | 33 | 26 | 22 | 21 | 20 |
| Men (HPFS, 1986) | |||||
| Participants, n | 8364 | 10,599 | 9498 | 9497 | 9495 |
| Age, y | 52 ± 10 | 54 ± 10 | 53 ± 10 | 55 ± 10 | 57 ± 10 |
| Furocoumarin intake, μg/d | 0.00 ± 0.00 | 28.0 ± 48.0 | 241 ± 60.0 | 588 ± 169 | 2400 ± 1980 |
| Total citrus intake, serving/d | 0.20 ± 0.30 | 1.00 ± 0.80 | 0.90 ± 0.70 | 1.00 ± 0.70 | 1.60 ± 1.10 |
| Total energy intake, kcal/d | 1840 ± 603 | 2110 ± 591 | 2140 ± 426 | 1860 ± 567 | 1760 ± 496 |
| UV exposure at residence, mW/m2 | 194 ± 28 | 191 ± 27 | 192 ± 28 | 192 ± 28 | 191 ± 28 |
| Family history of melanoma | 3.0 | 3.1 | 3.5 | 3.3 | 2.4 |
| Red/blonde hair | 13.8 | 14.5 | 15.0 | 13.4 | 13.0 |
| Painful burn or blistering reaction to sun exposure as a child/adolescent | 54 | 55 | 57 | 55 | 53 |
| ≥6 lifetime sunburns | 36 | 36 | 36 | 35 | 33 |
| ≥6 moles on the arms | 5.1 | 5.0 | 5.4 | 6.1 | 5.4 |
| Physical exam | 73 | 76 | 77 | 77 | 80 |
| BMI, kg/m2 | 25 ± 5 | 25 ± 5 | 25 ± 5 | 25 ± 5 | 25 ± 5 |
| Physical activity level, MET h/wk | 16 ± 26 | 20 ± 29 | 21 ± 27 | 22 ± 29 | 24 ± 35 |
| Alcohol intake, g/d | 12.0 ± 18.0 | 13.0 ± 17.0 | 12.0 ± 16.0 | 11.0 ± 15.0 | 11.0 ± 14.0 |
| Current smoking | 15 | 11 | 9 | 8 | 7 |
Values are means ± SDs or percentages unless otherwise indicated, and standardized to the age distribution of the study population except for age and number of participants. HPFS, Health Professionals Follow-Up Study; MET, metabolic-equivalent; NHS, Nurses’ Health Study.
Higher intakes of total furocoumarins as well as individual furocoumarins including bergaptol, bergapten, bergamottin, and 6'7'-DHB were significantly associated with a modest increase in risk of BCC (Table 2). The pooled multivariable-adjusted HR of BCC comparing the top with the bottom quintile was 1.16 (95% CI: 1.11, 1.21; P-trend = 0.002) for total furocoumarins. For bergaptol, bergapten, bergamottin, and 6'7'-DHB, the pooled multivariable-adjusted HRs were similar to that of total furocoumarins. The positive associations were consistent across the cohorts. Because we evaluated furocoumarin intake as a potentially responsible component of citrus fruit in relation to BCC risk, we examined the association between citrus intake and BCC risk by adjusting for furocoumarin intake. The pooled multivariable-adjusted HR of BCC comparing the top with the bottom intake category of citrus (<2 servings/wk compared with ≥1.6 servings/d) changed from 1.21 (95% CI: 1.14, 1.28; P-trend < 0.001) to 1.15 (95% CI: 1.08, 1.22; P-trend < 0.001) after adjusting for total furocoumarin intake.
TABLE 2.
HRs and 95% CIs of basal cell carcinoma according to quintiles of cumulative updated furocoumarin intake in the NHS and HPFS1
| Intake quintiles | |||||||
|---|---|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | Q5 | P-trend | ||
| Total furocoumarins | |||||||
| NHS | Median intake, μg/d | 0.00 | 2.00 | 182 | 398 | 1270 | |
| Cases, n | 2942 | 3400 | 3515 | 3679 | 4008 | ||
| Age-adjusted HR | 1.00 (referent) | 1.11 (1.06, 1.17) | 1.14 (1.08, 1.19) | 1.16 (1.10, 1.22) | 1.21 (1.16, 1.27) | <0.0001 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.08 (1.03, 1.13) | 1.10 (1.05, 1.16) | 1.12 (1.07, 1.18) | 1.18 (1.12, 1.24) | <0.0001 | |
| HPFS | Median intake, μg/d | 0.00 | 2.00 | 237 | 557 | 1800 | |
| Cases, n | 1843 | 2216 | 2255 | 2329 | 2443 | ||
| Age-adjusted HR | 1.00 (referent) | 1.12 (1.05, 1.19) | 1.16 (1.09, 1.24) | 1.17 (1.10, 1.24) | 1.18 (1.11, 1.25) | 0.0004 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.05 (0.99, 1.12) | 1.10 (1.03, 1.17) | 1.10 (1.04, 1.17) | 1.13 (1.06, 1.20) | 0.0006 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.07 (1.03, 1.11) | 1.10 (1.06, 1.14) | 1.12 (1.07, 1.16) | 1.16 (1.11, 1.21) | 0.002 |
| Bergaptol | |||||||
| NHS | Median intake, μg/d | 0.00 | 0.80 | 1.50 | 12.2 | 57.9 | |
| Cases, n | 3267 | 3197 | 3791 | 3583 | 3706 | ||
| Age-adjusted HR | 1.00 (referent) | 1.11 (1.05, 1.16) | 1.14 (1.09, 1.20) | 1.14 (1.09, 1.19) | 1.14 (1.09, 1.20) | 0.001 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.09 (1.04, 1.15) | 1.12 (1.07, 1.18) | 1.10 (1.05, 1.15) | 1.10 (1.05, 1.16) | 0.061 | |
| HPFS | Median intake, μg/d | 0.00 | 0.90 | 1.90 | 17.0 | 64.6 | |
| Cases, n | 2193 | 1913 | 2366 | 2362 | 2252 | ||
| Age-adjusted HR | 1.00 (referent) | 1.13 (1.06, 1.20) | 1.17 (1.10, 1.24) | 1.17 (1.10, 1.24) | 1.13 (1.06, 1.20) | 0.15 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.08 (1.02, 1.15) | 1.11 (1.05, 1.18) | 1.10 (1.04, 1.17) | 1.09 (1.03, 1.16) | 0.20 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.09 (1.05, 1.13) | 1.12 (1.08, 1.16) | 1.10 (1.06, 1.14) | 1.10 (1.06, 1.14) | 0.02 |
| Bergapten | |||||||
| NHS | Median intake, μg/d | 0.02 | 0.18 | 0.40 | 0.82 | 1.90 | |
| Cases, n | 3110 | 3546 | 3424 | 3657 | 3807 | ||
| Age-adjusted HR | 1.00 (referent) | 1.08 (1.03, 1.13) | 1.03 (0.98, 1.08) | 1.07 (1.02, 1.13) | 1.10 (1.05, 1.16) | 0.0006 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.06 (1.01, 1.11) | 1.00 (0.95, 1.04) | 1.04 (0.99, 1.09) | 1.06 (1.01, 1.11) | 0.07 | |
| HPFS | Median intake, μg/d | 0.08 | 0.25 | 0.43 | 0.84 | 1.87 | |
| Cases, n | 2019 | 2143 | 2254 | 2355 | 2315 | ||
| Age-adjusted HR | 1.00 (referent) | 1.02 (0.96, 1.08) | 1.05 (0.99, 1.11) | 1.09 (1.03, 1.16) | 1.06 (1.00, 1.12) | 0.07 | |
| Multivariable-adjusted HR | 1.00 (referent) | 0.99 (0.93, 1.05) | 1.01 (0.95, 1.07) | 1.05 (0.99, 1.11) | 1.03 (0.97, 1.09) | 0.11 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.03 (0.96, 1.10) | 1.00 (0.96, 1.04) | 1.04 (1.00, 1.08) | 1.05 (1.01, 1.09) | 0.02 |
| 6',7'-Dihydroxybergamottin | |||||||
| NHS | Median intake, μg/d | 0.10 | 0.70 | 105 | 210 | 649 | |
| Cases, n | 2972 | 3349 | 3487 | 3716 | 4020 | ||
| Age-adjusted HR | 1.00 (referent) | 1.11 (1.06, 1.17) | 1.14 (1.08, 1.19) | 1.18 (1.12, 1.23) | 1.22 (1.16, 1.28) | <0.0001 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.09 (1.03, 1.14) | 1.10 (1.05, 1.16) | 1.14 (1.09, 1.20) | 1.18 (1.13, 1.24) | <0.0001 | |
| HPFS | Median intake, μg/d | 0.10 | 0.80 | 131 | 279 | 909 | |
| Cases, n | 1887 | 2186 | 2213 | 2317 | 2483 | ||
| Age-adjusted HR | 1.00 (referent) | 1.14 (1.08, 1.22) | 1.15 (1.08, 1.22) | 1.18 (1.11, 1.25) | 1.20 (1.13, 1.27) | <0.0001 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.08 (1.02, 1.15) | 1.09 (1.03, 1.16) | 1.12 (1.05, 1.19) | 1.15 (1.08, 1.22) | <0.0001 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.09 (1.04, 1.13) | 1.10 (1.06, 1.14) | 1.13 (1.09, 1.18) | 1.17 (1.12, 1.21) | 0.0003 |
| Bergamottin | |||||||
| NHS | Median intake, μg/d | 0.00 | 0.10 | 76.6 | 163 | 524 | |
| Cases, n | 2872 | 3468 | 3499 | 3710 | 3995 | ||
| Age-adjusted HR | 1.00 (referent) | 1.11 (1.05, 1.16) | 1.13 (1.08, 1.19) | 1.17 (1.11, 1.23) | 1.21 (1.15, 1.27) | <0.0001 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.08 (1.02, 1.13) | 1.10 (1.05, 1.16) | 1.13 (1.08, 1.19) | 1.17 (1.12, 1.23) | <0.0001 | |
| HPFS | Median intake, μg/d | 0.10 | 0.30 | 98.4 | 220 | 719 | |
| Cases, n | 1934 | 2166 | 2220 | 2327 | 2439 | ||
| Age-adjusted HR | 1.00 (referent) | 1.16 (1.09, 1.24) | 1.15 (1.08, 1.22) | 1.18 (1.11, 1.25) | 1.18 (1.11, 1.26) | 0.0007 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.10 (1.03, 1.17) | 1.09 (1.02, 1.16) | 1.12 (1.05, 1.19) | 1.13 (1.07, 1.20) | 0.001 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.08 (1.04, 1.13) | 1.10 (1.05, 1.14) | 1.13 (1.08, 1.17) | 1.16 (1.11, 1.20) | 0.003 |
Multivariable analyses were further adjusted for number of lifetime blistering sunburns (none, 1–2, 3–5, or ≥6), number of arm moles (0, 1–2, 3–5, or ≥6), natural hair color (red, blonde, light brown, or dark brown or black), sunburn susceptibility as child or adolescent (none or some redness, burn, or painful burn or blisters), family history of melanoma (yes or no), alcohol intake (0, 0.1–4.9, 5.0–9.9, 10.0–19.9, or ≥20.0 g/d), UV exposure at residence (quintiles), physical activity (quintiles), routine physical examination (yes or no), intake of total energy (quintiles), and history of squamous cell carcinoma and melanoma. HPFS, Health Professionals Follow-Up Study; NHS, Nurses’ Health Study.
Pooled multivariate analysis was defined as combining the data of the 2 cohorts (NHS and HPFS) using a fixed-effects model.
Higher total furocoumarin intake was similarly but nonsignificantly associated with SCC risk (Table 3). The results were similar across the cohorts, although only statistically significant in the NHS. In terms of individual furocoumarins, higher intakes of bergamottin and 6'7'-DHB were positively associated with risk of SCC.
TABLE 3.
HRs and 95% CIs of squamous cell carcinoma according to quintiles of cumulative updated furocoumarin intake in the NHS and HPFS1
| Intake quintiles | |||||||
|---|---|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | Q5 | P-trend | ||
| Total furocoumarins | |||||||
| NHS | Cases, n | 366 | 440 | 448 | 437 | 531 | |
| Age-adjusted HR | 1.00 (referent) | 1.16 (1.01, 1.34) | 1.17 (1.01, 1.34) | 1.10 (0.96, 1.27) | 1.28 (1.12, 1.47) | 0.002 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.11 (0.97, 1.28) | 1.10 (0.95, 1.26) | 1.03 (0.89, 1.18) | 1.18 (1.03, 1.36) | 0.04 | |
| HPFS | Cases, n | 278 | 393 | 376 | 411 | 386 | |
| Age-adjusted HR | 1.00 (referent) | 1.34 (1.15, 1.56) | 1.28 (1.10, 1.50) | 1.35 (1.16, 1.57) | 1.19 (1.02, 1.40) | 0.70 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.19 (1.02, 1.39) | 1.16 (0.99, 1.36) | 1.23 (1.05, 1.43) | 1.13 (0.97, 1.33) | 0.54 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.14 (1.03, 1.27) | 1.12 (1.01, 1.25) | 1.12 (0.94, 1.33) | 1.16 (1.05, 1.29) | 0.15 |
| Bergaptol | |||||||
| NHS | Cases, n | 395 | 447 | 443 | 454 | 483 | |
| Age-adjusted HR | 1.00 (referent) | 1.22 (1.06, 1.40) | 1.10 (0.96, 1.26) | 1.15 (1.00, 1.31) | 1.20 (1.05, 1.38) | 0.10 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.18 (1.03, 1.36) | 1.04 (0.91, 1.20) | 1.08 (0.95, 1.24) | 1.12 (0.98, 1.28) | 0.49 | |
| HPFS | Cases, n | 318 | 371 | 389 | 410 | 356 | |
| Age-adjusted HR | 1.00 (referent) | 1.28 (1.10, 1.49) | 1.21 (1.04, 1.41) | 1.29 (1.11, 1.49) | 1.13 (0.97, 1.32) | 0.71 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.18 (1.01, 1.37) | 1.09 (0.93, 1.26) | 1.17 (1.01, 1.36) | 1.08 (0.92, 1.25) | 0.85 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.18 (1.07, 1.31) | 1.06 (0.96, 1.18) | 1.12 (1.02, 1.24) | 1.10 (0.99, 1.21) | 0.72 |
| Bergapten | |||||||
| NHS | Cases, n | 394 | 434 | 432 | 471 | 491 | |
| Age-adjusted HR | 1.00 (referent) | 1.04 (0.91, 1.20) | 1.02 (0.89, 1.17) | 1.09 (0.95, 1.24) | 1.11 (0.97, 1.27) | 0.09 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.01 (0.88, 1.16) | 0.98 (0.86, 1.13) | 1.03 (0.90, 1.18) | 1.03 (0.90, 1.18) | 0.56 | |
| HPFS | Cases, n | 352 | 337 | 410 | 389 | 356 | |
| Age-adjusted HR | 1.00 (referent) | 0.91 (0.78, 1.05) | 1.07 (0.93, 1.24) | 1.01 (0.87, 1.17) | 0.91 (0.78, 1.05) | 0.24 | |
| Multivariable-adjusted HR | 1.00 (referent) | 0.87 (0.75, 1.01) | 1.01 (0.88, 1.17) | 0.96 (0.83, 1.11) | 0.89 (0.77, 1.03) | 0.24 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 0.94 (0.85, 1.05) | 1.00 (0.90, 1.10) | 1.00 (0.91, 1.10) | 0.97 (0.87, 1.07) | 0.76 |
| 6',7'-Dihydroxybergamottin | |||||||
| NHS | Cases, n | 368 | 426 | 449 | 457 | 522 | |
| Age-adjusted HR | 1.00 (referent) | 1.14 (0.99, 1.31) | 1.17 (1.02, 1.35) | 1.16 (1.01, 1.33) | 1.26 (1.10, 1.44) | 0.006 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.09 (0.95, 1.26) | 1.11 (0.96, 1.27) | 1.08 (0.94, 1.24) | 1.16 (1.01, 1.33) | 0.081 | |
| HPFS | Cases, n | 287 | 390 | 370 | 401 | 396 | |
| Age-adjusted HR | 1.00 (referent) | 1.33 (1.14, 1.55) | 1.25 (1.07, 1.46) | 1.31 (1.12, 1.52) | 1.20 (1.03, 1.40) | 0.56 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.18 (1.02, 1.38) | 1.14 (0.97, 1.33) | 1.19 (1.03, 1.39) | 1.13 (0.97, 1.32) | 0.53 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.13 (1.02, 1.26) | 1.12 (1.01, 1.24) | 1.13 (1.02, 1.25) | 1.15 (1.04, 1.27) | 0.12 |
| Bergamottin | |||||||
| NHS | Cases, n | 360 | 441 | 455 | 430 | 536 | |
| Age-adjusted HR | 1.00 (referent) | 1.16 (1.01, 1.34) | 1.19 (1.04, 1.37) | 1.09 (0.95, 1.26) | 1.31 (1.14, 1.50) | 0.001 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.11 (0.96, 1.28) | 1.13 (0.98, 1.29) | 1.02 (0.89, 1.17) | 1.20 (1.05, 1.38) | 0.03 | |
| HPFS | Cases, n | 290 | 385 | 372 | 407 | 390 | |
| Age-adjusted HR | 1.00 (referent) | 1.33 (1.14, 1.55) | 1.25 (1.07, 1.46) | 1.32 (1.14, 1.54) | 1.19 (1.02, 1.39) | 0.59 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.18 (1.01, 1.38) | 1.14 (0.97, 1.33) | 1.20 (1.03, 1.40) | 1.13 (0.97, 1.32) | 0.49 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.14 (1.03, 1.26) | 1.13 (1.02, 1.25) | 1.10 (0.94, 1.30) | 1.17 (1.06, 1.29) | 0.12 |
Multivariable analyses were further adjusted for number of lifetime blistering sunburns (none, 1–2, 3–5, or ≥6), number of arm moles (0, 1–2, 3–5, or ≥6), natural hair color (red, blonde, light brown, or dark brown or black), sunburn susceptibility as child or adolescent (none or some redness, burn, or painful burn or blisters), family history of melanoma (yes or no), alcohol intake (0, 0.1–4.9, 5.0–9.9, 10.0–19.9, or ≥20.0 g/d), UV exposure at residence (quintiles), physical activity (quintiles), routine physical examination (yes or no), intake of total energy (quintiles), and history of basal cell carcinoma and melanoma. HPFS, Health Professionals Follow-Up Study; NHS, Nurses’ Health Study.
Pooled multivariate analysis was defined as combining the data of the 2 cohorts (NHS and HPFS) using a fixed-effects model.
The direction and magnitude of the associations with risk of melanoma for either total furocoumarin intake or intake of any of the individual furocoumarins were similar to those of BCC or SCC (Table 4). However, possibly due to much smaller number of cases, none of the associations were significant.
TABLE 4.
HRs and 95% CIs of melanoma according to quintiles of cumulative updated furocoumarin intake in the NHS and HPFS1
| Intake quintiles | |||||||
|---|---|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | Q5 | P-trend | ||
| Total furocoumarins | |||||||
| NHS | Cases, n | 128 | 157 | 160 | 183 | 148 | |
| Age-adjusted HR | 1.00 (referent) | 1.21 (0.96, 1.52) | 1.23 (0.97, 1.55) | 1.39 (1.11, 1.75) | 1.11 (0.87, 1.41) | 0.82 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.15 (0.91, 1.46) | 1.16 (0.92, 1.47) | 1.32 (1.05, 1.66) | 1.06 (0.83, 1.34) | 0.997 | |
| HPFS | Cases, n | 129 | 165 | 178 | 172 | 173 | |
| Age-adjusted HR | 1.00 (referent) | 1.20 (0.95, 1.51) | 1.31 (1.05, 1.65) | 1.23 (0.98, 1.54) | 1.19 (0.94, 1.50) | 0.56 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.15 (0.91, 1.45) | 1.25 (1.00, 1.57) | 1.17 (0.93, 1.47) | 1.16 (0.92, 1.46) | 0.57 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.15 (0.98, 1.36) | 1.21 (1.02, 1.42) | 1.24 (1.06, 1.46) | 1.11 (0.94, 1.31) | 0.64 |
| Bergaptol | |||||||
| NHS | Cases, n | 150 | 132 | 179 | 167 | 148 | |
| Age-adjusted HR | 1.00 (referent) | 1.11 (0.88, 1.41) | 1.31 (1.05, 1.62) | 1.23 (0.98, 1.53) | 1.09 (0.86, 1.36) | 0.62 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.08 (0.85, 1.37) | 1.24 (1.00, 1.55) | 1.16 (0.93, 1.45) | 1.03 (0.82, 1.30) | 0.44 | |
| HPFS | Cases, n | 151 | 139 | 177 | 189 | 161 | |
| Age-adjusted HR | 1.00 (referent) | 1.09 (0.87, 1.38) | 1.22 (0.98, 1.51) | 1.31 (1.05, 1.62) | 1.13 (0.91, 1.41) | 0.70 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.05 (0.83, 1.33) | 1.16 (0.93, 1.45) | 1.25 (1.00, 1.55) | 1.12 (0.89, 1.40) | 0.59 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.07 (0.90, 1.26) | 1.20 (1.03, 1.40) | 1.20 (1.03, 1.41) | 1.08 (0.92, 1.26) | 0.93 |
| Bergapten | |||||||
| NHS | Cases, n | 138 | 175 | 167 | 153 | 143 | |
| Age-adjusted HR | 1.00 (referent) | 1.23 (0.99, 1.54) | 1.18 (0.94, 1.48) | 1.07 (0.85, 1.35) | 1.00 (0.79, 1.26) | 0.31 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.19 (0.95, 1.49) | 1.13 (0.90, 1.41) | 1.02 (0.81, 1.29) | 0.94 (0.74, 1.19) | 0.15 | |
| HPFS | Cases, n | 148 | 151 | 168 | 171 | 179 | |
| Age-adjusted HR | 1.00 (referent) | 0.97 (0.78, 1.22) | 1.08 (0.86, 1.34) | 1.08 (0.87, 1.35) | 1.13 (0.91, 1.40) | 0.17 | |
| Multivariable-adjusted HR | 1.00 (referent) | 0.94 (0.75, 1.18) | 1.03 (0.82, 1.28) | 1.05 (0.84, 1.30) | 1.11 (0.89, 1.39) | 0.15 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.06 (0.90, 1.24) | 1.07 (0.92, 1.26) | 1.03 (0.88, 1.21) | 1.03 (0.89, 1.21) | 0.91 |
| 6',7'-Dihydroxybergamottin | |||||||
| NHS | Cases, n | 140 | 147 | 156 | 178 | 155 | |
| Age-adjusted HR | 1.00 (referent) | 1.06 (0.84, 1.34) | 1.11 (0.88, 1.40) | 1.26 (1.01, 1.57) | 1.08 (0.85, 1.36) | 0.55 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.01 (0.80, 1.28) | 1.05 (0.83, 1.32) | 1.19 (0.95, 1.49) | 1.02 (0.81, 1.29) | 0.76 | |
| HPFS | Cases, n | 137 | 157 | 173 | 166 | 184 | |
| Age-adjusted HR | 1.00 (referent) | 1.13 (0.90, 1.42) | 1.24 (0.99, 1.55) | 1.15 (0.91, 1.44) | 1.21 (0.97, 1.52) | 0.20 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.08 (0.86, 1.36) | 1.18 (0.94, 1.48) | 1.10 (0.87, 1.38) | 1.18 (0.94, 1.48) | 0.22 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.05 (0.89, 1.23) | 1.11 (0.95, 1.31) | 1.14 (0.97, 1.34) | 1.10 (0.94, 1.29) | 0.24 |
| Bergamottin | |||||||
| NHS | Cases, n | 129 | 159 | 164 | 166 | 158 | |
| Age-adjusted HR | 1.00 (referent) | 1.17 (0.92, 1.47) | 1.23 (0.97, 1.55) | 1.23 (0.97, 1.55) | 1.15 (0.91, 1.46) | 0.49 | |
| Multivariable-adjusted HR | 1.00 (referent) | 1.12 (0.88, 1.41) | 1.16 (0.92, 1.47) | 1.16 (0.92, 1.47) | 1.09 (0.86, 1.38) | 0.70 | |
| HPFS | Cases, n | 146 | 150 | 173 | 175 | 173 | |
| Age-adjusted HR | 1.00 (referent) | 1.04 (0.83, 1.31) | 1.18 (0.94, 1.47) | 1.15 (0.92, 1.43) | 1.10 (0.88, 1.37) | 0.54 | |
| Multivariable-adjusted HR | 1.00 (referent) | 0.99 (0.79, 1.25) | 1.12 (0.90, 1.40) | 1.09 (0.88, 1.37) | 1.07 (0.85, 1.33) | 0.55 | |
| Pooled2 | Multivariable-adjusted HR | 1.00 (referent) | 1.05 (0.89, 1.24) | 1.14 (0.97, 1.34) | 1.13 (0.96, 1.32) | 1.08 (0.92, 1.27) | 0.48 |
Multivariable analyses were further adjusted for number of lifetime blistering sunburns (none, 1–2, 3–5, or ≥6), number of arm moles (0, 1–2, 3–5, or ≥6), natural hair color (red, blonde, light brown, or dark brown or black), sunburn susceptibility as child or adolescent (none or some redness, burn, or painful burn or blisters), family history of melanoma (yes or no), alcohol intake (0, 0.1–4.9, 5.0–9.9, 10.0–19.9, or ≥20.0 g/d), UV exposure at residence (quintiles), physical activity (quintiles), routine physical examination (yes or no), intake of total energy (quintiles), and history of basal cell carcinoma and squamous cell carcinoma. HPFS, Health Professionals Follow-Up Study; NHS, Nurses’ Health Study.
Pooled multivariate analysis was defined as combining the data of the 2 cohorts (NHS and HPFS) using a fixed-effects model.
We also explored absolute cutoffs of total furocoumarin intake using those with 0 μg/d intake as the reference; the pooled multivariable-adjusted HRs for the top intake category of >600 μg/d were 1.18 (95% CI: 1.12, 1.24) for BCC, 1.28 (95% CI: 1.08, 1.52) for SCC, and 1.31 (95% CI: 1.02, 1.67) for melanoma.
Additional analysis for total furocoumarin intake was conducted according to melanoma subtypes based on body location (Supplemental Table 3). However, the results were not consistent by cohort. Total furocoumarin intake was associated with risk of melanomas of the head, neck, and extremities in women in the NHS and those of the trunk, shoulder, and thigh in men in the HPFS. For SCC, the positive association between total furocoumarin intake and SCC risk was significant with SCCs of the head, neck, and extremities, but not with those of the trunk in the pooled analyses (Supplemental Table 4).
We also performed stratified analyses to determine if the intake of total furocoumarins was differentially associated with the risk of melanoma according to melanoma risk factors including UV exposure at residence, lifetime number of sunburns, hair color, and tanning ability (Supplemental Table 5). No significant interactions were observed for the risk of melanoma by the selected melanoma risk factors. However, the association between furocoumarin intake and melanoma was significant among those with low UV exposure and higher number of sunburns. In similar stratified analyses, no interaction was found for SCC (Supplemental Table 6) and BCC (Supplemental Table 7), except with number of sunburns. For SCC, the association between furocoumarin intake and SCC was significant among those with high UV exposure, higher number of sunburns (P-interaction = 0.05), both red/blonde and dark/light brown or black hair color, and those getting deep tan after prolonged sun exposure (Supplemental Table 6). For BCC, most of the associations in the stratified analyses were significant by the melanoma risk factors (Supplemental Table 7). However, the P value for interaction was only significant for number of sunburns (P = 0.007).
Discussion
In the 2 prospective cohort studies, we determined the associations between individual and total furocoumarin intakes and the risk of skin cancer by taking advantage of a newly constructed food composition database for furocoumarins. Intakes of total and some individual furocoumarins (bergaptol, bergapten, 6'7'-DHB, and bergamottin) were modestly and significantly associated with an increased risk of BCC. The associations with SCC or melanoma were, although of similar magnitude, nonsignificant.
Studies by our group and others have shown that orally ingested furocoumarins can be rapidly absorbed into the bloodstream of humans (28–30). These studies have demonstrated that furocoumarins in grapefruit or grapefruit juice are detected in plasma of healthy adults as early as 15 min after consumption (29, 30). Once absorbed into the bloodstream, hydrophobic furocoumarins bind to proteins, particularly albumin, to be carried in the blood and distributed into multiple tissue types (31). A tissue distribution study in humans found that skin furocoumarin concentrations reach their peak 1 h after oral consumption and decline over the next 2 h (32). Another study showed that after oral administration, 8-MOP reached peak concentrations in the skin in 1–4 h and was detected in the skin for ≥7 h after the administration (33). A case report describes a healthy woman who consumed ∼450 g celeriac (containing ∼45 mg furocoumarins) and visited a suntan parlor 1 h later, then developed a severe phototoxic reaction over the next 48 h, indicating the photoactivation of furocoumarins in her skin (34).
Epidemiological studies have observed an increased skin cancer risk among patients receiving PUVA therapies to treat psoriasis or vitiligo (20, 22). 8-MOP is the most common furocoumarin used in PUVA therapy, typically ingested orally rather than topically (23, 35). Animal studies have also reported carcinogenicity of 8-MOP either orally or topically administered in combination with UV light (12, 36). An animal study revealed induction of melanocytic tumors under PUVA therapy with topical application of 8-MOP (37), and more apparent outgrowth of melanoma cells was observed with PUVA than with UVA or psoralen alone (11, 13, 38). The actions of furocoumarins (ingested orally or administered topically) in the skin include their ability to form DNA adducts after UV irradiation (18, 39–41). Photocarcinogenic properties of furocoumarins have been demonstrated in experimental studies which date back to as early as 1958 (11–13, 36, 37, 42). Some furocoumarins such as bergamot oil or bergapten were used as tanning activators and sunscreens until 1996. However, these products elevated the risk of melanoma (42–44). As a result, strict regulations were imposed on furocoumarin-containing tanning lotions and other cosmetic products (43). The International Agency for Research on Cancer (IARC) classified PUVA as a Group 1 human carcinogen (i.e., evidence for carcinogenicity to humans is sufficient) (45), although it is still clinically used with caution. The IARC also categorized bergapten with UV as a Group 2A carcinogen [i.e., probably carcinogenic to humans (46)].
Typical doses of 8-MOP administered in PUVA therapy range from ∼10–70 mg depending on the patient's weight (47), which is >50 times higher than consumed in the top quintile of our population and another study based on the NHANES (48). However, habitual consumption of low-level intake from diet may still have some impact on skin cancer risk. Intake of citrus, 1 of the major food sources of furocoumarins, was associated with BCC, SCC, and melanoma with some suggestion of stronger association with higher UV exposure in the NHS and HPFS (4, 5). Those studies motivated us to construct the food composition database for dietary furocoumarins so that we could calculate furocoumarin intake in the cohorts. Our study was the first prospective study evaluating the intake of furocoumarins in relation to skin cancer because food composition data for furocoumarins have not previously been available (9). We evaluated total and some individual furocoumarins. Because 8-MOP exists in small amounts in a limited number of foods, we were not able to evaluate 8-MOP intake separately.
Our findings support the hypothesis that dietary furocoumarins are associated with skin cancer risk. When we evaluated the associations by sex, the association of total furocoumarin intake and BCC was significant in both men and women. For SCC, the association was only significant in women, probably due to the smaller sample size in men than in women. The associations were significant with BCC and SCC, but not with melanoma potentially owing to the much smaller number of cases, although the direction and magnitude of the associations were similar across the different skin cancer types. Our findings on melanoma were consistent with a recent cross-sectional study based on the NHANES, where higher furocoumarin intake was associated with ≤75% higher prevalence of melanoma (48).
There were some suggestions of interaction between furocoumarin intake and sun exposure–related factors such as UV exposure at residence and number of sunburns in relation to skin cancer. It is of interest that the interactions were manifested more with SCC than with other skin cancers. The association between furocoumarin intake and SCC risk was stronger among those with high UV exposure, higher number of sunburns, light hair color, and those getting a deep tan after prolonged sun exposure. The association with furocoumarin intake was also largely limited to SCCs of sun-exposed sites, especially in men. This is consistent with the fact that SCC is more strongly related to UV exposure than are other skin cancers and further supports the photocarcinogenicity of furocoumarin intake (49).
Total furocoumarin intake was associated with risk of melanomas of the head, neck, and extremities in women and those of truncal sites in men. This discrepancy might be due to a different sun exposure pattern in women than in men, in that men tend to get more sun exposure on the truncal site. It may also be related to the fact that intermittent sun exposure is more relevant to melanoma than to other skin cancers (49).
We have created the food composition database based on 7 furocoumarins, many of which are rich in grapefruits. PUVA therapy generally uses furocoumarins with linear structures including 8-MOP and bergapten (50), whereas there is difficulty in distinguishing between furocoumarin isomers reported in studies for furocoumarin identification or measurement (8, 51, 52), hence it is possible that the observed association can be partly explained by the relation between biological effect and chemical structure of the compound (53). Future studies on the effect of structure-specific furocoumarins from multiple food items on the risk of skin cancer might be warranted.
In our FFQs, we had 10 food items containing furocoumarins, including orange, regular orange juice, fortified orange juice with calcium or vitamin D, grapefruit, grapefruit juice, raw carrots, cooked carrots, carrot juice, celery, and lemonade. The ρ values between citrus fruit intake and total furocoumarin intake were ∼0.5 in the cohorts, supportive of food items other than citrus contributing to furocoumarin intake. Still, most FFQs include a rather small number of food items with high furocoumarin content. Furocoumarins are rich in herbs, lemons, limes, parsnips, and figs, yet these items are not typically included in FFQs (9, 54). However, these food items are typically consumed in relatively small amounts also. Even for the citrus fruits which are usually assessed in FFQs, the furocoumarin contents in whole fruits compared with juice are quite different. We were fortunate to have grapefruits and grapefruit juice as well as oranges and orange juice as separate food items in the FFQs we used. Other FFQs often do not include them as separate food items, leading to potential misclassification of furocoumarin intake. Furthermore, because we selected furocoumarins largely based on those in grapefruits, we might have missed some other furocoumarins which commonly exist in foods other than citrus and have similar biological effects related to skin cancer. Evaluation of furocoumarin consumption in NHANES 2003–2012, which used two 24-h dietary recalls instead of FFQs and thus had a wider range of foods contributing to furocoumarins than our FFQs, found that grapefruit and grapefruit juice accounted for ∼76% of total furocoumarin consumption (48). Other than those food items, lime juice, parsley, lemon juice, and celery contributed most significantly to furocoumarin intake in the study. In our FFQ, we did not collect information on lime juice, parsley, and lemon juice.
Our study had several strengths. To our knowledge, this was the first prospective study that examined the relation between dietary furocoumarin intake and skin cancer risk. With >2 million person-years of follow-up and thousands of skin cancer cases included in the study, we had sufficient power to detect modest associations, especially for BCC. The repetitive assessment of diet likely reduces any misclassification of furocoumarin intake. A wide range of skin cancer risk factors including UV exposure and constitutional factors were taken into consideration in statistical analyses.
However, several limitations of this study should be noted. First, furocoumarin intake was estimated based on the FFQ, which is a self-reported measurement and thus subject to misclassification. However, by using cumulative average intake of multiple FFQs, measurement errors were minimized. Second, furocoumarin intake estimated by the FFQ has not been validated. However, the food items contributing to furocoumarin intake in the FFQ were validated with two 1-wk diet records (55). Also, the UPLC-MS/MS methods used to measure furocoumarin concentrations for this database were developed and validated by our group (29). The database was constructed using 3 varieties of each food sample. Third, the ongoing 2 large cohorts mostly consist of educated white health care professionals, which may limit the generalizability of these findings. Nevertheless, restricting the study population to only health professionals may limit confounding by socioeconomic status. Also, skin cancer is not common in nonwhites. Sources of misclassification regarding furocoumarin intake estimation should be noted as well. The concentrations of furocoumarins may be subject to change based on seasons, cooking method, storage, or even their geographical origins (56, 57). For example, cooked carrot had much less furocoumarin content than raw carrot based on our database (9). Because vegetables are often consumed cooked, there could be more misclassification of furocoumarin intake from vegetables than from fruits. Fourth, we have evaluated multiple furocoumarins and 3 different types of skin cancer, which might lead to an issue of multiple testing. Therefore, any significant associations may need to be interpreted with caution and need to be replicated in future studies.
In conclusion, based on 2 large prospective cohorts of US health professionals, our results demonstrated that total furocoumarin intake was modestly associated with increased risk of BCC. The associations were consistent in women and men for BCC. For SCC, the association was only significant in women, probably due to a larger sample size in women than in men. Among individual furocoumarins, bergaptol, bergapten, 6',7'-DHB, and bergamottin were significantly associated with BCC risk. More in-depth pathological or pharmacological studies are needed to reveal the site-specific relation between furocoumarin intake and skin cancers. Also, evaluation of biological biomarkers of furocoumarin intake may overcome the limitations of dietary assessment.
Supplementary Material
Acknowledgments
We 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, AR, AZ, CA, CO, CT, DE, FL, GA, IA, ID, IL, IN, KY, LA, MA, MD, ME, MI, NC, ND, NE, NH, NJ, NY, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, and WY. We assume full responsibility for the analyses and interpretation of these data. The authors’ responsibilities were as follows—AAQ and EC: designed the research; WS, MKP, OKC, and MMM: conducted the research; WS and MKP: analyzed the data; EC: had primary responsibility for the final content; and all authors: wrote the paper and read and approved the final manuscript.
Notes
Supported by NIH grants CA186107, CA167552 (to WCW), and CA198216 (to EC) and the Dermatology Foundation Research Career Development Award (to WL).
Author disclosures: The authors report no conflicts of interest.
Supplemental Figure 1 and Supplemental Tables 1–7 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.
Abbreviations used: BCC, basal cell carcinoma; HPFS, Health Professionals Follow-Up Study; IARC, International Agency for Research on Cancer; NHS, Nurses’ Health Study; PUVA, psoralen and UVA; SCC, squamous cell carcinoma; UPLC-MS/MS, ultraperformance liquid chromatography with tandem mass spectrometry; 6'7'-DHB, 6',7'-dihydroxybergamottin; 8-MOP, 8-methoxypsoralen (methoxsalen).
References
- 1. Jemal A, Saraiya M, Patel P, Cherala SS, Barnholtz-Sloan J, Kim J, Wiggins CL, Wingo PA. Recent trends in cutaneous melanoma incidence and death rates in the United States, 1992–2006. J Am Acad Dermatol. 2011;65(5 Suppl 1):S17–25..e1–3. [DOI] [PubMed] [Google Scholar]
- 2. Lomas A, Leonardi-Bee J, Bath-Hextall F. A systematic review of worldwide incidence of nonmelanoma skin cancer. Br J Dermatol. 2012;166(5):1069–80. [DOI] [PubMed] [Google Scholar]
- 3. Wehner MR, Shive ML, Chren MM, Han J, Qureshi AA, Linos E. Indoor tanning and non-melanoma skin cancer: systematic review and meta-analysis. BMJ. 2012;345:e5909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Wu S, Cho E, Feskanich D, Li WQ, Sun Q, Han J, Qureshi AA. Citrus consumption and risk of basal cell carcinoma and squamous cell carcinoma of the skin. Carcinogenesis. 2015;36(10):1162–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Wu S, Han J, Feskanich D, Cho E, Stampfer MJ, Willett WC, Qureshi AA. Citrus consumption and risk of cutaneous malignant melanoma. J Clin Oncol. 2015;33(23):2500–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. De Castro WV, Mertens-Talcott S, Rubner A, Butterweck V, Derendorf H. Variation of flavonoids and furanocoumarins in grapefruit juices: a potential source of variability in grapefruit juice−drug interaction studies. J Agric Food Chem. 2006;54(1):249–55. [DOI] [PubMed] [Google Scholar]
- 7. Manthey JA, Buslig BS. Distribution of furanocoumarins in grapefruit juice fractions. J Agric Food Chem. 2005;53(13):5158–63. [DOI] [PubMed] [Google Scholar]
- 8. Frerot E, Decorzant E. Quantification of total furocoumarins in citrus oils by HPLC coupled with UV, fluorescence, and mass detection. J Agric Food Chem. 2004;52(23):6879–86. [DOI] [PubMed] [Google Scholar]
- 9. Melough MM, Lee SG, Cho E, Kim K, Provatas AA, Perkins C, Park MK, Qureshi A, 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. 2017;65(24):5049–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Ostertag E, Becker T, Ammon J, Bauer-Aymanns H, Schrenk D. Effects of storage conditions on furocoumarin levels in intact, chopped, or homogenized parsnips. J Agric Food Chem. 2002;50(9):2565–70. [DOI] [PubMed] [Google Scholar]
- 11. Cartwright LE, Walter JF. Psoralen-containing sunscreen is tumorigenic in hairless mice. J Am Acad Dermatol. 1983;8(6):830–6. [DOI] [PubMed] [Google Scholar]
- 12. Mullen MP, Pathak MA, West JD, Harrist TJ, Dall'Acqua F. Carcinogenic effects of monofunctional and bifunctional furocoumarins. Natl Cancer Inst Monogr. 1984;66:205–10. [PubMed] [Google Scholar]
- 13. Zajdela F, Bisagni E. 5-methoxypsoralen, the melanogenic additive in sun-tan preparations, is tumorigenic in mice exposed to 365 nm u.v. radiation. Carcinogenesis. 1981;2(2):121–7. [DOI] [PubMed] [Google Scholar]
- 14. Kornhauser A, Wamer WG, Giles AL Jr. Psoralen phototoxicity: correlation with serum and epidermal 8-methoxypsoralen and 5-methoxypsoralen in the guinea pig. Science. 1982;217(4561):733–5. [DOI] [PubMed] [Google Scholar]
- 15. Bissonnette L, Arnason JT, Smith ML. Real-time fluorescence-based detection of furanocoumarin photoadducts of DNA. Phytochem Anal. 2008;19(4):342–7. [DOI] [PubMed] [Google Scholar]
- 16. Messer A, Nieborowski A, Strasser C, Lohr C, Schrenk D. Major furocoumarins in grapefruit juice I: levels and urinary metabolite(s). Food Chem Toxicol. 2011;49(12):3224–31. [DOI] [PubMed] [Google Scholar]
- 17. Parrish JA, Fitzpatrick TB, Tanenbaum L, Pathak MA. Photochemotherapy of psoriasis with oral methoxsalen and longwave ultraviolet light. N Engl J Med. 1974;291(23):1207–11. [DOI] [PubMed] [Google Scholar]
- 18. Vos JM, Hanawalt PC. Processing of psoralen adducts in an active human gene: repair and replication of DNA containing monoadducts and interstrand cross-links. Cell. 1987;50(5):789–99. [DOI] [PubMed] [Google Scholar]
- 19. Walter JF, Gange RW, Mendelson IR. Psoralen-containing sunscreen induces phototoxicity and epidermal ornithine decarboxylase activity. J Am Acad Dermatol. 1982;6(6):1022–7. [DOI] [PubMed] [Google Scholar]
- 20. Stern RS. The risk of melanoma in association with long-term exposure to PUVA. J Am Acad Dermatol. 2001;44(5):755–61. [DOI] [PubMed] [Google Scholar]
- 21. 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. 2012;66(4):553–62. [DOI] [PubMed] [Google Scholar]
- 22. Stern RS, Nichols KT, Väkevä LH; for The PUVA Follow-up Study. Malignant melanoma in patients treated for psoriasis with methoxsalen (psoralen) and ultraviolet A radiation (PUVA). N Engl J Med. 1997;336(15):1041–5. [DOI] [PubMed] [Google Scholar]
- 23. Archier E, Devaux S, Castela E, Gallini A, Aubin F, Le Maître M, Aractingi S, Bachelez H, Cribier B, Joly P et al.. Carcinogenic risks of psoralen UV-A therapy and narrowband UV-B therapy in chronic plaque psoriasis: a systematic literature review. J Eur Acad Dermatol Venereol. 2012;26(Suppl 3):22–31. [DOI] [PubMed] [Google Scholar]
- 24. Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol. 1986;124(1):17–27. [DOI] [PubMed] [Google Scholar]
- 25. Rivera A, Nan H, Li T, Qureshi A, Cho E. Alcohol intake and risk of incident melanoma: a pooled analysis of three prospective studies in the United States. Cancer Epidemiol Biomarkers Prev. 2016;25(12):1550–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Colditz GA, Martin P, Stampfer MJ, Willett WC, Sampson L, Rosner B, Hennekens CH, Speizer FE. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol. 1986;123(5):894–900. [DOI] [PubMed] [Google Scholar]
- 27. Van Dam RM, Huang Z, Giovannucci E, Rimm EB, Hunter DJ, Colditz GA, Stampfer MJ, Willett WC. Diet and basal cell carcinoma of the skin in a prospective cohort of men. Am J Clin Nutr. 2000;71(1):135–41. [DOI] [PubMed] [Google Scholar]
- 28. Busch U, Schmid J, Koss FW, Zipp H, Zimmer A. Pharmacokinetics and metabolite-pattern of 8-methoxypsoralen in man following oral administration as compared to the pharmacokinetics in rat and dog. Arch Dermatol Res. 1978;262(3):255–65. [DOI] [PubMed] [Google Scholar]
- 29. Lee SG, Kim K, Vance TM, Perkins C, Provatas A, Wu S, Qureshi A, Cho E, 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. 2016;67(8):881–7. [DOI] [PubMed] [Google Scholar]
- 30. Melough MM, Vance TM, Lee SG, Provatas AA, Perkins C, Qureshi A, Cho E, Chun OK. Furocoumarin kinetics in plasma and urine of healthy adults following consumption of grapefruit (Citrus paradisi Macf.) and grapefruit juice. J Agric Food Chem. 2017;65(14):3006–12. [DOI] [PubMed] [Google Scholar]
- 31. The Nordic Working Group on Food Toxicology and Risk Evaluation. Furocoumarins in plant food- exposure, biological properties, risk assessment and recommendations. TemaNord 1996:600 Copenhagen: Nordic Council of Ministers; 1996. [Google Scholar]
- 32. Lauharanta J, Juvakoski T, Kanerva L, Lassus A. Pharmacokinetics of 8-methoxypsoralen in serum and suction blister fluid. Arch Dermatol Res. 1982;273(1–2):111–4. [DOI] [PubMed] [Google Scholar]
- 33. Tegeder I, Brautigam L, Podda M, Meier S, Kaufmann R, Geisslinger G, Grundmann-Kollmann M. Time course of 8-methoxypsoralen concentrations in skin and plasma after topical (bath and cream) and oral administration of 8-methoxypsoralen. Clin Pharmacol Ther. 2002;71(3):153–61. [DOI] [PubMed] [Google Scholar]
- 34. Ljunggren B. Severe phototoxic burn following celery ingestion. Arch Dermatol. 1990;126(10):1334–6. [PubMed] [Google Scholar]
- 35. Sahin S, Hindioglu U, Karaduman A. PUVA treatment of vitiligo: a retrospective study of Turkish patients. Int J Dermatol. 1999;38(7):542–5. [DOI] [PubMed] [Google Scholar]
- 36. Griffin AC, Hakim RE, Knox J. The wave length effect upon erythemal and carcinogenic response in psoralen treated mice. J Invest Dermatol. 1958;31(5):289–95. [PubMed] [Google Scholar]
- 37. Alcalay J, Bucana C, Kripke ML. Cutaneous pigmented melanocytic tumor in a mouse treated with psoralen plus ultraviolet A radiation. Photodermatol Photoimmunol Photomed. 1990;7(1):28–31. [PubMed] [Google Scholar]
- 38. Aubin F, Donawho CK, Kripke ML. Effect of psoralen plus ultraviolet A radiation on in vivo growth of melanoma cells. Cancer Res. 1991;51(21):5893–7. [PubMed] [Google Scholar]
- 39. Stern RS. Psoralen and ultraviolet A light therapy for psoriasis. N Engl J Med. 2007;357(7):682–90. [DOI] [PubMed] [Google Scholar]
- 40. Dall'Acqua F, Marciani S, Ciavatta L, Rodighiero G. Formation of inter-strand cross-linkings in the photoreactions between furocoumarins and DNA. Z Naturforsch B. 1971;26(6):561–9. [DOI] [PubMed] [Google Scholar]
- 41. Cleaver JE, Killpack S, Gruenert DC. Formation and repair of psoralen-DNA adducts and pyrimidine dimers in human DNA and chromatin. Environ Health Perspect. 1985;62:127–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Ashwood-Smith MJ, Poulton GA, Barker M, Mildenberger M. 5-methoxypsoralen, an ingredient in several suntan preparations, has lethal, mutagenic and clastogenic properties. Nature. 1980;285(5764):407–9. [DOI] [PubMed] [Google Scholar]
- 43. Autier P, Dore JF, Cesarini JP, Boyle P; for the Epidemiology and Prevention Subgroup, EORTC Melanoma Cooperative Group; EORTC Prevention Research Division. Should subjects who used psoralen suntan activators be screened for melanoma?. Ann Oncol. 1997;8(5):435–7. [DOI] [PubMed] [Google Scholar]
- 44. Autier P, Doré JF, Schifflers E, Cesarini JP, Bollaerts A, Koelmel KF, Gefeller O, Liabeuf A, Lejeune F, Lienard D et al.. Melanoma and use of sunscreens: an EORTC case-control study in Germany, Belgium and France. Int J Cancer. 1995;61(6):749–55. [DOI] [PubMed] [Google Scholar]
- 45. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Biological agents. Volume 100 B. A review of human carcinogens. IARC Monogr Eval Carcinog Risks Hum. 2012;100(Pt B):1–441. [PMC free article] [PubMed] [Google Scholar]
- 46. International Agency for Research on Cancer (IARC). Supplement 7: 5-methoxypsoralen (group 2A). [Internet], Lyon (France): International Agency for Research on Cancer; 1987; [updated 1998 Feb 11; cited 2019 Jun 2]. Available from: http://www.inchem.org/documents/iarc/suppl7/methoxypsoralen-5.html. [Google Scholar]
- 47. IARC Working Group on the Evaluation of Carcinogenic Risk to Humans. Methoxsalen plus ultraviolet A radiation. [Internet] In: Pharmaceuticals. Lyon (France): International Agency for Research on Cancer; 2012[cited 2019 Jun 2]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK304315/. [Google Scholar]
- 48. Melough MM, Kim K, Cho E, Chun OK. Relationship between furocoumarin intake and melanoma history among US adults in the National Health and Nutrition Examination Survey 2003–2012. Nutr Cancer. 2020;72(1):24–32. [DOI] [PubMed] [Google Scholar]
- 49. Wu S, Han J, Laden F, Qureshi AA. Long-term ultraviolet flux, other potential risk factors, and skin cancer risk: a cohort study. Cancer Epidemiol Biomarkers Prev. 2014;23(6):1080–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Viola G, Salvador A, Vedaldi D, Dall'Acqua F, Bianchi N, Zuccato C, Borgatti M, Lampronti I, Gambari R. Differentiation and apoptosis in UVA-irradiated cells treated with furocoumarin derivatives. Ann N Y Acad Sci. 2009;1171:334–44. [DOI] [PubMed] [Google Scholar]
- 51. Dugrand A, Olry A, Duval T, Hehn A, Froelicher Y, Bourgaud F. Coumarin and furanocoumarin quantitation in citrus peel via ultraperformance liquid chromatography coupled with mass spectrometry (UPLC-MS). J Agric Food Chem. 2013;61(45):10677–84. [DOI] [PubMed] [Google Scholar]
- 52. Dugo P, Piperno A, Romeo R, Cambria M, Russo M, Carnovale C, Mondello L. Determination of oxygen heterocyclic components in citrus products by HPLC with UV detection. J Agric Food Chem. 2009;57(15):6543–51. [DOI] [PubMed] [Google Scholar]
- 53. Van Rossum JM. The relation between chemical structure and biological activity: discussion of possibilities, pitfalls and limitations. J Pharm Pharmacol. 1963;15(1):285–316. [DOI] [PubMed] [Google Scholar]
- 54. Pathak MA, Daniels F Jr, Fitzpatrick TB. The presently known distribution of furocoumarins (psoralens) in plants. J Invest Dermatol. 1962;39:225–39. [DOI] [PubMed] [Google Scholar]
- 55. Feskanich D, Rimm EB, Giovannucci EL, Colditz GA, Stampfer MJ, Litin LB, Willett WC. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J Am Diet Assoc. 1993;93(7):790–6. [DOI] [PubMed] [Google Scholar]
- 56. Chaudhary PR, Jayaprakasha GK, Porat R, Patil BS. Low temperature conditioning reduces chilling injury while maintaining quality and certain bioactive compounds of ‘star ruby’ grapefruit. Food Chem. 2014;153:243–9. [DOI] [PubMed] [Google Scholar]
- 57. Uesawa Y, Mohri K. The use of heat treatment to eliminate drug interactions due to grapefruit juice. Biol Pharm Bull. 2006;29(11):2274–8. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.