Skip to main content
NCBI home page
American Journal of Epidemiology logoLink to American Journal of Epidemiology
. 2022 Jun 7;191(10):1742–1752. doi: 10.1093/aje/kwac101

Cumulative Erythemal Ultraviolet Radiation and Risk of Cancer in 3 Large US Prospective Cohorts

Michael S Chang, Rebecca I Hartman, Nicole Trepanowski, Edward L Giovannucci, Hongmei Nan, Xin Li ✉
PMCID: PMC9991893  PMID: 35671977

Abstract

Ultraviolet radiation (UVR) exposure is the major risk factor for melanoma. However, epidemiologic studies on UVR and noncutaneous cancers have reported inconsistent results, with some suggesting an inverse relationship potentially mediated by vitamin D. To address this, we examined 3 US prospective cohorts, the Health Professionals Follow-up Study (HPFS) (1986) and Nurses’ Health Study (NHS) I and II (1976 and 1989), for associations between cumulative erythemal UVR and incident cancer risk, excluding nonmelanoma skin cancer. We used a validated spatiotemporal model to calculate erythemal UVR. Participants (47,714 men; 212,449 women) were stratified into quintiles by cumulative average erythemal UVR, using the first quintile as referent, for Cox proportional hazards regression analysis. In the multivariable-adjusted meta-analysis of all cohorts, compared with the lowest quintile, risk of any cancer was slightly increased across all other quintiles (highest quintile hazard ratio (HR) = 1.04, 95% confidence interval (CI): 1.01, 1.07; P for heterogeneity = 0.41). All UVR quintiles were associated with similarly increased risk of any cancer excluding melanoma. As expected, erythemal UVR was positively associated with risk of melanoma (highest quintile HR = 1.17, 95% CI: 1.04, 1.31; P for heterogeneity = 0.83). These findings suggest that elevated UVR is associated with increased risk of both melanoma and noncutaneous cancers.

Keywords: cancer, carcinogenesis, erythemal exposure, melanoma, noncutaneous cancers, risk, ultraviolet radiation, spatiotemporal modeling

Abbreviations

CI

confidence interval

HPFS

Health Professionals Follow-up Study

HR

hazard ratio

NHS

Nurses’ Health Study

UV

ultraviolet

UVR

ultraviolet radiation

Ultraviolet radiation (UVR) is a major risk factor for skin cancer. Ultraviolet (UV)-B (290–320 nanometers in wavelength) substantially increases risk for melanoma, particularly in fair-skinned individuals. Substantial epidemiologic evidence demonstrates that behaviors associated with increased UV exposure, including indoor tanning and time spent outdoors, increase risk of nonmelanoma skin cancer and melanoma (1, 2). The underlying carcinogenic etiology of UVR is thought to be related to its ability to cause direct and indirect DNA damage through disrupting molecular bonds and causing structural aberrations, specifically the formation of pyrimidine dimers (3).

However, the association between UVR and noncutaneous cancers is less well-characterized. Multiple studies have suggested a protective, inverse relationship may exist between cumulative UVR exposure and cancer incidence, specifically with hepatocellular carcinoma, colorectal cancer, and non-Hodgkin lymphoma (4–6). Mixed evidence demonstrates that early life UVR exposure may be protective for certain types of breast cancer, while other data demonstrate no association between ambient, cumulative UVR exposure in adulthood and breast cancer risk (7). Thus, these discrepancies necessitate further investigation to characterize the relationship between UVR and noncutaneous cancers.

Limitations in prior epidemiologic studies suggest that these inconsistencies between UVR and cancer risk may be due to confounding and coarse UVR measures. To more granularly investigate the potential association between UVR and cancer risk, we examined 3 ongoing prospective cohort studies in the United States: the Health Professionals Follow-up Study (HPFS) and Nurses’ Health Study (NHS) I and II. Using a validated spatiotemporal model (8), we sought to assess the relationship between cumulative erythemal UVR and risk of cancer development.

METHODS

Study population

The HPFS began in 1986 with 51,529 male enrollees from multiple health professions (29,683 dentists, 4,185 pharmacists, 3,745 optometrists, 2,220 osteopath physicians, 1,600 podiatrists, and 10,098 veterinarians) (9). The Nurses’ Health Study (NHS) I and II began in 1976 and 1989 with 121,700 (aged 30–55 years) and 116,429 (aged 25–42 years) female nurse enrollees, respectively (10). Participants from HPFS, NHS I, and NHS II receive health-related questionnaires every 2 years, encompassing demographic, medical, and geographic data. Response rates per survey cycle typically exceeded 90%.

UVR exposure was first available in the NHS I cohort in 1980 and has been subsequently updated every 2 years. We therefore set 1980 as baseline year for the NHS I. In the present analysis, we excluded participants with a known diagnosis of any cancer, excluding nonmelanoma skin cancer (basal cell and squamous cell carcinoma), at baseline. We excluded deaths before baseline (n = 481 in NHS I, n = 1 in NHS II, and n = 9 in HPFS), as well as participants who did not have the documented UVR exposure of interest (n = 1,110 in NHS I, n = 164 in NHS II, and n = 1,030 in HPFS). For all 3 cohorts, individuals with missing birth years were also excluded (n = 224).

The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T. H. Chan School of Public Health, and those of participating registries as required.

Exposure assessment

Using previously established methods (8), cumulative erythemal UVR was calculated as average July noon-time erythemal UVR spanning the contiguous United States with spatial resolution of 1 km2 and annual temporal resolution for each cohort’s respective study period. For each participant, UVR was estimated as a time-varying cumulative average and subsequently stratified by quintiles. To develop this model, residential addresses within the United States provided by participants in the HPFS, NHS I, and NHS II were updated biennially. These addresses were geocoded to standardized zip codes to develop a spatiotemporal model of erythemal UVR. Both UV-A and UV-B wavelengths were weighted based on their relative impact on development of erythema in white skin (11), with shorter UV-B wavelengths represented more substantially. Stratified area-to-point residual kriging was applied to downscale gridded remote sensing images from the National Aeronautics and Space Administration Total Ozone Mapping Spectrometer and Ozone Monitoring Instrument satellite sensors. Additional environmental factors affecting UVR, including aerosol optical depth, cloud cover, elevation, ozone, and latitude, were accounted for in this model (7, 12). Prior cross-validation of this spatiotemporal model demonstrated high predictive performance (7, 8). As a secondary exposure, we also evaluated self-reported history of severe sunburns, defined with responses of never, 1–2 times, 3–5 times, or ≥6 times to “How often have you had a severe sunburn?” Self-reported history of severe sunburns was defined over the respondent’s lifetime in NHS I and HPFS, and between the ages of 15 and 20 years in NHS II.

Outcomes and covariates

Participants in the HPFS, NHS I, and NHS II reported new diagnoses of cancer biennially. Patient permission was obtained prior to contacting the participant’s physician or care site to request access for medical records, pathology reports, and other relevant information. These reports were then reviewed to confirm the self-reported diagnosis. Only pathologically confirmed cancer cases were included. The outcomes of interest were risk of any cancer excluding nonmelanoma skin cancer, any cancer excluding both melanoma and nonmelanoma skin cancer, and melanoma only. For patients with multiple cancer diagnoses, we considered only their initial cancer diagnosis for the risk analysis. Specifically, for individuals with an initial diagnosis of melanoma, we included them exclusively in the analysis of risk of melanoma. We controlled for multiple covariates, including age, body mass index, pack-years of smoking, alcohol consumption, coffee consumption, physical activity (metabolic equivalents, hours per week), Alternative Healthy Eating Index score, hair color, mole count, skin reaction to the sun, and menopause status/hormone use (female NHS participants only).

Statistical analysis

Participants were included in the analysis from the initial start of 1980 for NHS I, 1986 for HPFS, and 1989 for NHS II, until either cancer diagnosis (excluding nonmelanoma skin cancer), death, or end of follow-up (June 2016 for NHS I, June 2017 for NHS II, and June 2016 for the HPFS), whichever occurred first. Within each cohort, we used age-adjusted and multivariable-adjusted Cox proportional hazards regression models to evaluate potential associations between cumulative average erythemal UVR, as well as severe sunburn history, and risk of any cancer, any cancer excluding melanoma, and melanoma alone. Cumulative average erythemal UVR was stratified by quintiles, using the lowest as the referent for modeling. We also performed trend testing by using continuous cumulative average erythemal UVR in the regression models. We assessed heterogeneity across studies using Cochran’s Q2 statistic (13). Results from all 3 cohort studies were combined using a random-effects model. We considered 2-sided P < 0.05 to be statistically significant. Patients with missing cumulative UVR data were excluded from analyses. For missing covariates (other than age), we treated missing data as a separate category and adjusted the dummy variable in the Cox models using the missing indicator approach. Statistical analyses were performed using SAS, version 9.4 for UNIX (SAS Institute, Inc., Cary, North Carolina).

Of note, additional analysis examining cancer risk by specific cancer subtype was conducted. In addition, sensitivity analyses were conducted to exclude breast cancer from NHS I and II and prostate cancer from HPFS.

RESULTS

A total of 278,932 participants were included in this study, including 48,744 men from HPFS in addition to 114,983 and 115,205 women in NHS I and NHS II, respectively. During a follow-up of 7,357,547 person-years across all 3 cohorts, we documented a total of 57,269 incident cancer cases (31,029 from NHS I, 12,137 from NHS II, and 14,103 from HPFS). Age-standardized characteristics of participants according to quintiles of baseline erythemal UVR are presented in Table 1. Overall, mean baseline characteristics for body mass index, daily alcohol intake, smoking pack-years, and hourly physical activity per week were consistent across UVR quintiles in each cohort. In addition, the percentage of natural red or blonde hair was greater in the fifth quintile in NHS I and NHS II, relative to other quintiles within cohorts. Mole count on the extremities was comparable across all UVR levels. In the NHS, proportion of women in their premenopause stage was lower in the highest quintile compared with other quintiles.

Table 1.

Age-Standardized Characteristics of Participants in the Nurses’ Health Study I (1980), Nurses’ Health Study II (1989), and Health Professionals Follow-up Study (1986), United States

NHS I (n = 112,507) NHS II (n = 99,942) HPFS (n = 47,714)
Variable  a Mean (SD) No. % Mean (SD) No. % Mean (SD) No. %
Ageb, years 46.7 (7.2) 34.6 (4.7) 54.5 (9.8)
BMIc 24.1 (4.2) 24 (5) 25.5 (3.4)
Smoking pack-years 1.3 (1.5) 0.6 (0.9) 1.4 (1.6)
Alcohol intake, g/day 6.3 (10.5) 3.1 (6) 11.3 (15.4)
Caffeine intake, mg/day 397 (274.7) 243.1 (223.1) 239.2 (250.4)
Physical activity level, metabolic-equivalent-hours/week 11.9 (13.2) 25.2 (37.1) 20.9 (29.7)
AHEI score 43.3 (10.2) 43.3 (10.3) 47 (10.9)
Red/blonde hair, % 17,439 15.5 19,888 19.9 6,489 13.6
Arm with moles, % 41,178 36.6 50,471 50.5 15,363 32.2
Menopausal status/PMH statusd 0
Premenopause, % 55,691 49.5 97,144 97.2
 Postmenopause, never use, % 16,314 14.5 100 0.1
 Postmenopause, current use, % 6,413 5.7 1,899 1.9
 Postmenopause, past use, % 8,551 7.6 100 0.1
 Missing, % 25,652 22.8 600 0.6
Cumulative erythemal UVR
 Quintile 1 156.6 (5.2) 127.9 (19.9) 163.0 (4.8)
 Quintile 2 171.0 (3.0) 149.9 (5.0) 172.3 (2.7)
 Quintile 3 178.5 (1.9) 168.9 (4.2) 183.2 (3.2)
 Quintile 4 186.6 (4.1) 182.0 (4.6) 203.2 (8.7)
 Quintile 5 223.0 (20.2) 232.5 (22.4) 238.1 (13.7)
Cancer type
 Breast 11,956 10.6 5,056 5.1
 Prostate 6,397 13.4
 Lung 3,119 2.8 278 0.3 962 2.0
 Colorectal 2,619 2.3 473 0.5 1,276 2.7
 Hematologice 2,261 2.0 540 0.5 1,306 2.7
 Endometrial 2,131 1.9 706 0.7
 Other gastrointestinalf 1,897 1.7 311 0.3 1,104 2.3
 Melanoma 1,474 1.3 742 0.7 964 2.0
 Genitourinaryg 1,226 1.1 235 0.2 1,204 2.5
 Ovarian 1,151 1.0 396 0.4
 Otherh 3,195 2.8 3,400 3.4 890 1.9
Open in a new tab

Abbreviations: AHEI, Alternative Healthy Eating Index; BMI, body mass index; HPFS, Health Professionals Follow-up Study; NHS, Nurses’ Health Study; PMH, postmenopausal hormone; SD, standard deviation; UVR, ultraviolet radiation.

a Values of polytomous variables may not sum to 100% due to rounding. Participants missing UVR measurements at baseline are not included in this table; therefore, the total numbers of participants in each study shown here is lower than the total numbers included in the regression analysis.

b Value is not age-adjusted.

c BMI is calculated as weight (kg)/height (m)2.

d Available only for female participants in NHS I and II.

e Hematologic cancers include lymphosarcoma, Hodgkin disease, non-Hodgkin lymphoma, multiple myeloma, lymphatic leukemia, myeloid leukemia, and unspecified leukemia.

f Other gastrointestinal cancers include oral, esophageal, stomach, small intestine, liver, gallbladder, pancreatic, and peritoneal cancers.

g Genitourinary cancers include bladder, kidney, and ureter cancers.

h Other cancers include eye cancer, nose cancer, larynx cancer, pharynx cancer, brain cancer, cervical cancer, nervous system cancer, thyroid cancer, bone cancer, cancer of connective tissue, and cancer of unknown site.

In the meta-analysis of all 3 cohorts, we observed elevated cancer risk comparing higher quintiles of UVR exposure with the lowest quintile (Table 2). Although the trend test was not significant, risk of any incident cancer was significantly increased among individuals in the highest UVR quintile compared with those in the lowest quintile (multivariable-adjusted hazard ratio (HR) = 1.04, 95% confidence interval (CI): 1.01, 1.07; P for heterogeneity = 0.41). For individual cohorts, participants in NHS I were found to have a significant, positive association with risk of any cancer across nearly all UVR quintiles in both the age-adjusted and multivariable-adjusted models (fifth quintile HR = 1.05, 95% CI: 1.01, 1.09). In addition, a similar association was found in the HPFS cohort for the second highest quintile in both the age-adjusted (fourth quintile HR = 1.07, 95% CI: 1.01, 1.12) and multivariable-adjusted (fourth quintile HR = 1.06, 95% CI: 1.01, 1.12). No association was found in the NHS II for either model. Additionally, we repeated the analysis by excluding melanoma skin cancer from the outcome of interest and obtained similar results (Table 2). As expected, the risk of melanoma alone was most strongly positively associated with UVR. In the meta-analysis, cumulative UV was significantly associated with increased risk of melanoma (multivariable-adjusted HR comparing highest quintile vs. lowest quintile, 1.17, 95% CI: 1.04, 1.31; P for heterogeneity = 0.83; P for trend = 0.02) (Table 2).

Table 2.

Association Between Cumulative Erythemal Ultraviolet Radiation and Cancer Risk in the Nurses’ Health Study I (1980–2016), Nurses’ Health Study II (1989–2017), and Health Professionals Follow-up Study (1986–2016), United States

Cohort and UVR Exposure Quintile No. of Cases Person-Years Age-Adjusted Model Multivariable-Adjusted Model  a
HR 95% CI HR 95% CI
Any Cancer
NHS I, quintile
 1 5,968 674,848 1.00 Referent 1.00 Referent
 2 6,274 680,678 1.05b 1.01, 1.09 1.05b 1.01, 1.08
 3 6,301 677,651 1.06b 1.02, 1.10 1.05b 1.01, 1.08
 4 6,161 668,277 1.04 1.00, 1.07 1.02 0.99, 1.06
 5 6,325 650,097 1.04b 1.00, 1.08 1.05b 1.01, 1.09
  P for trend 0.35 0.06
NHS II, quintile
 1 2,448 595,695 1.00 Referent 1.00 Referent
 2 2,398 595,805 0.98 0.93, 1.04 0.99 0.94, 1.05
 3 2,438 599,456 1.00 0.95, 1.06 1.00 0.94, 1.06
 4 2,358 597,899 0.98 0.93, 1.04 0.99 0.94, 1.05
 5 2,495 594,828 0.98 0.92, 1.03 1.00 0.95, 1.06
  P for trend 0.29 0.99
HPFS, quintile
 1 2,753 204,303 1.00 Referent 1.00 Referent
 2 2,879 205,047 1.05 0.99, 1.10 1.05 1.00, 1.11
 3 2,797 207,434 1.02 0.97, 1.07 1.02 0.96, 1.07
 4 2,867 203,028 1.07b 1.01, 1.12 1.06b 1.01, 1.12
 5 2,807 202,501 1.03 0.98, 1.09 1.05 0.99, 1.11
  P for trend 0.59 0.31
Meta-analysisc
 1 11,169 1,474,846 1.00 Referent 1.00 Referent
 2 11,551 1,481,530 1.03 0.99, 1.07 1.03 1.00, 1.07
 3 11,536 1,484,541 1.03 1.00, 1.07 1.03b 1.00, 1.06
 4 11,386 1,469,204 1.03 0.98, 1.07 1.03 0.99, 1.06
 5 11,627 1,447,426 1.02 0.98, 1.06 1.04b 1.01, 1.07
  P for trend 0.64 0.06
Any Cancer (Excluding Melanoma)
NHS I
 1 5,710 674,557 1.00 Referent 1.00 Referent
 2 5,970 680,309 1.04b 1.01, 1.08 1.04b 1.00, 1.08
 3 6,011 677,311 1.06b 1.02, 1.10 1.04b 1.01, 1.08
 4 5,837 667,915 1.03 0.99, 1.06 1.01 0.98, 1.05
 5 6,027 649,741 1.03 1.00, 1.07 1.04b 1.01, 1.08
  P for trend 0.57 0.11
NHS II
 1 2,309 595,532 1.00 Referent 1.00 Referent
 2 2,261 595,655 0.98 0.93, 1.04 0.99 0.93, 1.05
 3 2,296 599,297 1.00 0.94, 1.06 1.00 0.94, 1.06
 4 2,193 597,708 0.97 0.91, 1.02 0.98 0.93, 1.04
 5 2,336 594,646 0.97 0.91, 1.03 0.99 0.94, 1.06
  P for trend 0.12 0.65
HPFS
 1 2,574 204,101 1.00 Referent 1.00 Referent
 2 2,678 204,839 1.04 0.98, 1.10 1.04 0.99, 1.10
 3 2,606 207,228 1.02 0.96, 1.07 1.01 0.96, 1.07
 4 2,696 202,834 1.07b 1.02, 1.13 1.07b 1.01, 1.13
 5 2,585 202,260 1.02 0.96, 1.08 1.04 0.98, 1.09
  P for trend 0.89 0.45
Meta-analysisc
 1 10,593 1,474,190 1.00 Referent 1.00 Referent
 2 10,909 1,480,803 1.03 0.99, 1.06 1.03 1.00, 1.06
 3 10,913 1,483,836 1.03 0.99, 1.07 1.03 1.00, 1.05
 4 10,726 1,468,457 1.02 0.97, 1.08 1.02 0.98, 1.07
 5 10,948 1,446,647 1.01 0.97, 1.05 1.03b 1.00, 1.06
  P for trend 0.75 0.18
Melanoma Only
NHS I
 1 258 668,448 1.00 Referent 1.00 Referent
 2 304 673,975 1.18 1.00, 1.39 1.19b 1.01, 1.40
 3 290 671,008 1.11 0.94, 1.32 1.11 0.94, 1.32
 4 324 661,673 1.25b 1.06, 1.48 1.23b 1.04, 1.45
 5 298 643,396 1.18 1.00, 1.40 1.17 0.98, 1.38
  P for trend 0.08 0.13
NHS II
 1 139 593,108 1.00 Referent 1.00 Referent
 2 137 593,192 1.00 0.79, 1.26 1.01 0.79, 1.27
 3 142 596,788 1.04 0.82, 1.31 1.03 0.82, 1.30
 4 165 595,324 1.21 0.96, 1.52 1.21 0.97, 1.52
 5 159 592,146 1.14 0.90, 1.43 1.11 0.88, 1.40
  P for trend 0.05 0.10
HPFS
 1 179 201,530 1.00 Referent 1.00 Referent
 2 201 202,186 1.13 0.93, 1.39 1.16 0.95, 1.42
 3 191 204,646 1.06 0.86, 1.30 1.06 0.86, 1.30
 4 171 200,135 0.98 0.80, 1.22 0.96 0.77, 1.18
 5 222 199,688 1.26b 1.03, 1.54 1.22 0.99, 1.49
  P for trend 0.11 0.36
Meta-analysisc
 1 576 1,463,086 1.00 Referent 1.00 Referent
 2 642 1,469,353 1.12 1.00, 1.25 1.14b 1.01, 1.27
 3 623 1,472,442 1.08 0.96, 1.21 1.07 0.96, 1.20
 4 660 1,457,132 1.15 0.99, 1.33 1.13 0.97, 1.32
 5 679 1,435,230 1.19b 1.07, 1.33 1.17b 1.04, 1.31
  P for trend 0.003b 0.02b
Open in a new tab

Abbreviations: CI, confidence interval; HPFS, Health Professionals Follow-up Study; HR, hazard ratio; NHS, Nurses’ Health Study; UVR, ultraviolet radiation.

a Adjusted for age, body mass index, pack-years of smoking, alcohol consumption, coffee consumption, physical activity (metabolic equivalents, hours per week), Alternative Healthy Eating Index score, hair color, mole count, skin reaction to the sun, and menopause status/hormone use (in female Nurses’ Health Studies I and II cohorts only).

b Denotes significance (P < 0.05).

c No significant heterogeneity was found among results from the 3 cohorts (P for heterogeneity > 0.05).

A history of severe sunburn ≥6 times was significantly associated with increased risk of any cancer (HR = 1.03, 95% CI: 1.00, 1.07; P for heterogeneity = 0.39), but not of any cancer excluding melanoma (HR = 1.00, 95% CI: 0.97, 1.04; P for heterogeneity = 0.38), in the multivariable-adjusted meta-analysis. As expected, the combined results from the 3 cohorts demonstrated that history of severe sunburn was associated with increased melanoma risk, greatest in magnitude for those with ≥6 sunburns in both the age-adjusted (HR = 2.30, 95% CI: 2.03, 2.61; P for heterogeneity = 0.98) and multivariable-adjusted models (HR = 1.69, 95% CI: 1.48, 1.93; P for heterogeneity = 0.77) (Table 3).

Table 3.

Association Between Severe Sunburn History and Cancer Risk in the Nurses’ Health Study I (1980–2016), Nurses’ Health Study II (1989–2017), and Health Professionals Follow-up Study (1986–2016), United States

Cohort and Severe Sunburn History No. of Cases Person-Years Age-Adjusted Model Multivariable-Adjusted Model  a
HR 95% CI HR 95% CI
Any Cancer
NHS I
 Never 14,753 1,521,521 1.00 Referent 1.00 Referent
 1–2 times 4,881 508,998 1.05b 1.01, 1.08 1.02 0.99, 1.06
 3–5 times 1,828 199,074 1.01 0.96, 1.06 0.98 0.93, 1.03
 ≥6 times 1,831 178,898 1.12b 1.07, 1.17 1.06b 1.00, 1.11
NHS II
 Never 4,119 1,058,085 1.00 Referent 1.00 Referent
 1–2 times 4,677 1,155,397 1.05b 1.01, 1.10 1.02 0.97, 1.06
 3–5 times 2,091 501,147 1.07b 1.01, 1.13 1.02 0.96, 1.08
 ≥6 times 1,272 290,061 1.11b 1.04, 1.18 1.04 0.97, 1.11
HPFS
 Never 1,903 145,203 1.00 Referent 1.00 Referent
 1–2 times 2,853 211,089 1.02 0.96, 1.08 1.00 0.94, 1.06
 3–5 times 2,816 208,736 1.05 0.99, 1.11 1.02 0.96, 1.08
 ≥6 times 4,046 312,175 1.05 0.99, 1.11 1.00 0.94, 1.06
Meta-analysisc
 Never 20,775 2,724,809 1.00 Referent 1.00 Referent
 1–2 times 12,411 1,875,484 1.04b 1.02, 1.07 1.02 0.99, 1.04
 3–5 times 6,735 908,957 1.04b 1.00, 1.08 1.00 0.97, 1.03
 ≥6 times 7,149 781,134 1.09b 1.05, 1.14 1.03b 1.00, 1.07
Any Cancer (Excluding Melanoma)
NHS I
 Never 14,198 1,520,893 1.00 Referent 1.00 Referent
 1–2 times 4,613 508,684 1.03 1.00, 1.07 1.01 0.98, 1.05
 3–5 times 1,702 198,919 0.98 0.93, 1.03 0.96 0.91, 1.01
 ≥6 times 1,684 178,744 1.07b 1.02, 1.13 1.02 0.97, 1.08
NHS II
 Never 3,941 1,057,884 1.00 Referent 1.00 Referent
 1–2 times 4,390 1,155,062 1.04 0.99, 1.08 1.01 0.96, 1.05
 3–5 times 1,925 500,961 1.03 0.97, 1.09 0.99 0.94, 1.05
 ≥6 times 1,157 289,934 1.06 0.99, 1.13 1.01 0.94, 1.08
HPFS
 Never 1,821 145,112 1.00 Referent 1.00 Referent
 1–2 times 2,699 210,919 1.01 0.95, 1.07 0.99 0.93, 1.05
 3–5 times 2,614 208,500 1.02 0.96, 1.08 1.00 0.94, 1.06
 ≥6 times 3,663 311,779 0.99 0.94, 1.05 0.97 0.91, 1.03
Meta-analysisc
 Never 19,960 2,723,889 1.00 Referent 1.00 Referent
 1–2 times 11,702 1,874,665 1.03b 1.00, 1.05 1.01 0.98, 1.03
 3–5 times 6,241 908,380 1.01 0.97, 1.04 0.98 0.95, 1.01
 ≥6 times 6,504 780,457 1.04 0.99, 1.09 1.00 0.97, 1.04
Melanoma Only
NHS I
 Never 555 1,505,806 1.00 (ref) 1.00 (ref)
 1–2 times 268 503,909 1.47b 1.27, 1.70 1.32b 1.14, 1.53
 3–5 times 126 197,180 1.75b 1.44, 2.13 1.48b 1.22, 1.81
 ≥6 times 147 177,024 2.30b 1.91, 2.76 1.77b 1.46, 2.16
NHS II
 Never 178 1,053,539 1.00 Referent 1.00 Referent
 1–2 times 287 1,150,314 1.49b 1.24, 1.80 1.25b 1.03, 1.51
 3–5 times 166 498,990 1.99b 1.61, 2.46 1.51b 1.22, 1.89
 ≥6 times 115 288,716 2.34b 1.85, 2.96 1.58b 1.23, 2.03
HPFS
 Never 82 143,218 1.00 Referent 1.00 Referent
 1–2 times 154 208,202 1.30 0.99, 1.70 1.19 0.90, 1.56
 3–5 times 202 205,932 1.75b 1.35, 2.27 1.44b 1.10, 1.87
 ≥6 times 383 308,214 2.27b 1.78, 2.89 1.66b 1.29, 2.14
Meta-analysisc
 Never 815 2,702,563 1.00 Referent 1.00 Referent
 1–2 times 709 1,862,425 1.45b 1.30, 1.61 1.27b 1.14, 1.42
 3–5 times 494 902,102 1.83b 1.61, 2.08 1.48b 1.30, 1.69
 ≥6 times 645 773,954 2.30b 2.03, 2.61 1.69b 1.48, 1.93
Open in a new tab

Abbreviations: CI, confidence interval; HPFS, Health Professionals Follow-up Study; HR, hazard ratio; NHS, Nurses’ Health Study.

a Adjusted for age, body mass index, pack-years of smoking, alcohol consumption, coffee consumption, physical activity (metabolic equivalents, hours per week), Alternative Healthy Eating Index score, hair color, mole count, skin reaction to the sun, and menopause status/hormone use (in female Nurses’ Health Studies I and II cohorts only).

b Denotes significance (P< 0.05).

c No significant heterogeneity was found among results from the 3 cohorts (P for heterogeneity > 0.05).

DISCUSSION

In this analysis of 3 large well-characterized US cohorts, we examined associations of cumulative erythemal UVR, and severe sunburn history as a secondary exposure of interest, with any cancer risk, any cancer (excluding melanoma) risk, and melanoma risk. In the multivariable-adjusted model, we found that cumulative UVR, after adjustment for potential confounders, was associated with a slightly increased risk of any cancer, even when excluding melanoma cases. In addition, in the multivariable-adjusted analysis, history of severe sunburn was associated with increased risk of any cancer but only when including melanoma cases. As expected, both UVR and history of severe sunburn were most strongly associated with melanoma risk alone, providing internal validity to our analyses.

Of note, we separately examined the association of UVR and risk of cancer in 11 different sites, excluding melanoma (Web Table 1, available at https://doi.org/10.1093/aje/kwac101). Breast and prostate cancer, which are the most common cancers in women and men, respectively, in the United States (14), in addition to lung and other gastrointestinal cancers were the only sites with identifiable associations in the multivariable model. We did not appreciate an association in the other cancers, likely due to lack of statistical power, and our results were no longer significant when excluding breast and prostate cancer in the sensitivity analyses (Web Table 2). In the meta-analysis of both NHS cohorts, we found small positive associations between the highest quintiles of cumulative UVR and risk of breast cancer in both the age- and multivariable-adjusted models. In HPFS, we found an elevated risk of prostate cancer associated with the second highest cumulative UVR quintiles.

Previous findings have demonstrated overall increased mortality associated with UVR, including total deaths and cancer, among others (15). Our findings similarly suggest no protective benefit between UVR and risk of any cancer, advancing our understanding of a controversial area of research. Among the mixed evidence supporting decreased risk of noncutaneous cancers and UVR, a prior study of the National Institutes of Health–AARP cohort in 2013 found that erythemal UVR estimated from National Aeronautics and Space Administration Total Ozone Mapping Spectrometer and Ozone Monitoring Instrument satellite imaging and US Census data on subject residence was associated with decreased risk of colon, prostate, squamous cell lung, pleural, kidney, and bladder cancers, as well as non-Hodgkin lymphoma (16). In addition, nonlinear associations were found with thyroid and pancreatic cancer (16). Conversely, other epidemiologic studies have also demonstrated no observed associations between UVR and breast cancer risk (7, 17). Although the aforementioned National Institutes of Health–AARP study utilized a spatiotemporal model with similar data (16), our investigation employed an improved model that was not limited to coarse geographic variables (e.g., city, clinic location, etc.) or specific geographic regions of the United States. In addition, our follow-up time with the HPFS, NHS I, and NHS II cohorts were substantially longer than their 14-year period.

Our findings of slightly increased cancer risk with cumulative erythemal UVR are in contrast with hypotheses of UVR-induced anticancer benefits. First, UV-B irradiation is known to be the major source of 25-hydroxyvitamin D production in the skin. Specifically in breast cancer studies, experimental evidence has suggested that vitamin D may inhibit cellular growth and activate apoptosis in breast cancer cells (18–20). Vitamin D has also demonstrated antihepatocarcinogenic effects in murine models, including regulating bile acid levels and suppressing hepatic stellate cell proliferation, and inhibiting downstream fibrogenic pathways via vitamin D receptors signaling (5, 21–24). UVR has also been found to be associated with improved mood and well-being, which have been attributed to the role of vitamin D in various neural pathways (25, 26). However, results from the Vitamin D and Omega-3 Trial (VITAL) showed no evidence of lower incidence of invasive cancer with vitamin D supplementation, which is in accordance with our large-scale epidemiologic findings of UVR and risk of any cancer (27).

In addition, although cumulative erythemal UVR was associated with an increased risk of cancer both with and without melanoma cases, history of severe sunburn was not. The discrepancy between the association of risk of cancer with cumulative UVR and history of severe sunburn, a well-established marker for acute UVR, suggests that temporality of UVR may have an impact on carcinogenesis (28). We conducted a separate analysis investigating risk of any cancer and timing of UVR exposure (cumulative average erythemal UVR, most recent UVR, and baseline UVR), which demonstrated significant associations between risk of cancer and cumulative average UVR and most recent UVR but not baseline UVR, when the survey was initially administered (1980 for NHS I, 1989 for NHS II, 1986 for HPFS) (Web Table 3). Severe sunburn history and a pattern of intermittent sun exposure have been significantly linked to melanoma risk, whereas continuous, chronic exposure has been more strongly tied to development of nonmelanoma skin cancers (29–32). In addition, sunburns more commonly occur in younger age groups, whereas the regional cumulative UVR measured in this study is a more updated variable that is weighted towards older age exposure (33). Preliminary scientific evidence suggests that acute and chronic pathways of UVR activate separate downstream pathways. For instance, while both acute and chronic UVR result in epidermal hyperplasia and lower epidermal T-cell density, chronic UVR increases peripheral T-cell activity due to higher cytokine signaling causing marked upregulation of Th2/T-regulatory cells (34). T-cells have been demonstrated to be important in the antitumor response, with Th2-type cytokines linked to downregulating antitumor immune activity (35). In addition, T-regulatory cells migrate to sites of inflammation to suppress immune effector cells and, in the context of cancer, are responsible for promoting tumor immune escape mechanisms and decreased immune surveillance (36). In addition, T-regulatory cells are often found near tumors in the body (36). Thus, these potential immunoregulatory factors may help explain discrepancies in risk of cancer between the 2 outcomes of cumulative UVR and severe sunburn history in this study. Although further work is necessary to better characterize the underlying mechanisms for UVR and carcinogenesis, temporality appears to have an impact on cancer risk at both an epidemiologic and biological level.

Likewise, there was a discrepancy in the association between severe sunburn history and risk of any cancer when considering melanoma diagnoses. This may be attributable to both epidemiologic and statistical factors. For instance, individuals who are less likely to engage in skin cancer prevention practices are more likely use alcohol, illicit drugs, and smoking products, among other potentially risky or unhealthy practices (37). Although we controlled for smoking and alcohol in our multivariable model, there may be residual confounding. As such, these associated behaviors may also increase one’s risk for developing noncutaneous cancers as well. In addition, this discrepancy may also be explained by the relatively strong association between risk of melanoma and history of severe sunburn, which may positively skew the association with risk of any cancer overall.

Our study was inherently limited by lack of known sun exposure behaviors (such as percentage of body exposed to UV, outdoor activity, etc.). However, among the various covariates we adjusted for in our multivariable-adjusted analyses, physical activity served as a proxy for time spent outdoors, and suspected risk factors, such as hair color and mole count, were accounted for. Despite these adjustments, we acknowledge that there may still be residual confounding. Moreover, these cohorts were predominantly White health professionals, which may limit generalizability of our findings, particularly given inherent differences in UVR-induced cutaneous synthesis of vitamin D by skin phototype and differences in UV exposure by occupational patterns (38). In terms of our secondary outcome of sunburn history, this variable was collected differently in the NHS II cohort (during ages 15–20 years) compared with NHS I and HPFS (lifetime history). Furthermore, based on our study’s findings, there does not appear to be a dose-dependent relationship between cumulative average erythemal UVR and risk of cancer. Likewise, we are limited in our conclusions given the low magnitude of the calculated HRs. There are also data suggesting that vitamin D confers a benefit for cancer mortality, but not incidence; thus, there may be UVR-related effects on survival that we did not examine in this analysis (39, 40).

The strengths of this study include utilization of one of the largest, geographically representative pooled cohorts assessing impact of erythemal UVR and carcinogenesis across the HPFS, NHS I, and NHS II. In addition, this study implements the highest-resolution, validated spatiotemporal model known to date, allowing for granular assessment of cumulative erythemal UVR across the entire United States by geocoded residential addresses updated biennially (7, 8). Relative to other national prospective studies, our study also benefits from having longer follow-up periods, ranging from 31 to 41 years. When determining cancer risk over a shorter 10-year follow-up period, similar to the National Institutes of Health–AARP cohort study, our findings are no longer significantly different, suggesting that our findings are enhanced by our cohort’s longer follow-up period (Web Table 4).

Our findings provide further evidence for positive association between cumulative UVR and risk of both melanoma and noncutaneous cancers. However, sensitivity analyses demonstrate that this association is driven largely by the substantially greater proportion of breast and prostate cancer cases in our cohorts. Despite emerging evidence for the protective benefits of UVR against cancer, further research is necessary to understand the health effects of sun exposure, vitamin D, and their underlying biological mechanisms.

Supplementary Material

Web_Material_kwac101
Click here for additional data file. (23.8KB, zip)

ACKNOWLEDGMENTS

Author affiliations: Department of Dermatology, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, United States (Michael S. Chang, Rebecca I. Hartman, Nicole Trepanowski); Department of Dermatology, VA Integrated Service Network 1 (VISN-1), Jamaica Plain, Massachusetts, United States (Rebecca I. Hartman); Boston University School of Medicine, Boston, Massachusetts, United States (Nicole Trepanowski); Department of Nutrition, Harvard T. H. Chan School of Public Health, Boston, Massachusetts, United States (Edward L. Giovannucci); Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, United States (Edward L. Giovannucci); Department of Global Health, Richard M. Fairbanks School of Public Health, Indiana University, Indianapolis, Indiana, United States (Hongmei Nan); Department of Epidemiology, Richard M. Fairbanks School of Public Health, Indiana University, Indianapolis, Indiana, United States (Xin Li); and Indiana University Melvin and Bren Simon Cancer Center, Indianapolis, Indiana, United States (Xin Li).

M.S.C. and R.I.H. contributed equally as first authors.

This work was funded by the National Institutes of Health (grants UM1 CA186107, P01 CA87969, U01 CA176726, and U01 CA167552). R.I.H. is supported by the Department of Defense and a VISN-1 Career Development Award.

Further information, including the procedures to obtain and access data from the Nurses’ Health Studies and Health Professionals Follow-up Study, is described at https://www.nurseshealthstudy.org/researchers (contact e-mail: nhsaccess@channing.harvard.edu) and https://sites.sph.harvard.edu/hpfs/for-collaborators/.

We thank the participants and staff of the cohort studies for their dedication to this research. We thank the Channing Division of Network Medicine, Brigham and Women’s Hospital, Harvard Medical School. The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention’s National Program of Cancer Registries (NPCR) and/or the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program. Central registries may also be supported by state agencies, universities, and cancer centers. Participating central cancer registries include the following: Alabama, Alaska, Arizona, Arkansas, California, Delaware, Colorado, Connecticut, Florida, Georgia, Hawaii, Idaho, Indiana, Iowa, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Mississippi, Montana, Nebraska, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Puerto Rico, Rhode Island, Seattle SEER Registry, South Carolina, Tennessee, Texas, Utah, Virginia, West Virginia, and Wyoming.

Presented at the 2021 Society for Investigative Dermatology Virtual Meeting (online), May 3–8, 2021.

The views expressed in this article are those of the authors and do not reflect those of the National Institutes of Health.

Conflict of interest: none declared.

REFERENCES

  • 1. Zhang  M, Qureshi AA, Geller AC, et al.  Use of tanning beds and incidence of skin cancer. J Clin Oncol. 2012;30(14):1588–1593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Bauer  A, Diepgen TL, Schmitt J. Is occupational solar ultraviolet irradiation a relevant risk factor for basal cell carcinoma? A systematic review and meta-analysis of the epidemiological literature. Br J Dermatol. 2011;165(3):612–625. [DOI] [PubMed] [Google Scholar]
  • 3. Amaro-Ortiz  A, Yan B, D’Orazio JA. Ultraviolet radiation, aging and the skin: prevention of damage by topical cAMP manipulation. Molecules. 2014;19(5):6202–6219. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Boscoe  FP, Schymura MJ. Solar ultraviolet-B exposure and cancer incidence and mortality in the United States, 1993–2002. BMC Cancer. 2006;6:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Vopham  T, Bertrand KA, Yuan JM, et al.  Ambient ultraviolet radiation exposure and hepatocellular carcinoma incidence in the United States. Environ Health. 2017;16(1):1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Gonzalez  CA, Riboli E, Overvad K, et al.  Diet and cancer prevention: contributions from the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Eur J Cancer. 2010;46(14):2555–2562. [DOI] [PubMed] [Google Scholar]
  • 7. VoPham  T, Bertrand KA, DuPré NC, et al.  Ultraviolet radiation exposure and breast cancer risk in the Nurses’ Health Study II. Environ Epidemiol. 2019;3(4):e057. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. VoPham  T, Hart JE, Bertrand KA, et al.  Spatiotemporal exposure modeling of ambient erythemal ultraviolet radiation. Environ Health. 2016;15(1):1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Van Dam  RM, Willett WC, Rimm EB, et al.  Dietary fat and meat intake in relation to risk of type 2 diabetes in men. Diabetes Care. 2002;25(3):417–424. [DOI] [PubMed] [Google Scholar]
  • 10. Bao  Y, Bertoia ML, Lenart EB, et al.  Origin, methods, and evolution of the three Nurses’ Health Studies. Am J Public Health. 2016;106(9):1573–1581. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. National Aeronautics and Space Administration (NASA) . Erythemal Exposure Data Product. Greenbelt, MD: NASA Goddard Space Flight Center; 2017. http://ozoneaq.gsfc.nasa.gov/media/docs/erynotes.pdf. Accessed January 27, 2022. [Google Scholar]
  • 12. Kerr  JB, Fioletov VE. Surface ultraviolet radiation. Atmosphere-Ocean. 2008;46(1):159–184. [Google Scholar]
  • 13. Crippa  A, Khudyakov P, Wang M, et al.  A new measure of between-studies heterogeneity in meta-analysis. Stat Med. 2016;35(21):3661–3675. [DOI] [PubMed] [Google Scholar]
  • 14. Siegel  RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin. 2020;70(1):7–30. [DOI] [PubMed] [Google Scholar]
  • 15. Lin  SW, Wheeler DC, Park Y, et al.  Prospective study of ultraviolet radiation exposure and mortality risk in the United States. Am J Epidemiol. 2013;178(4):521–533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Lin  SW, Wheeler DC, Park Y, et al.  Prospective study of ultraviolet radiation exposure and risk of cancer in the United States. Int J Cancer. 2012;131(6):E1015–E1023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Anderson  LN, Cotterchio M, Kirsh VA, et al.  Ultraviolet sunlight exposure during adolescence and adulthood and breast cancer risk: a population-based case-control study among Ontario women. Am J Epidemiol. 2011;174(3):293–304. [DOI] [PubMed] [Google Scholar]
  • 18. Gross  M, Kost SB, Ennis B, et al.  Effect of 1,25-dihydroxyvitamin D3 on mouse mammary tumor (GR) cells: evidence for receptors, cellular uptake, inhibition of growth and alteration in morphology at physiologic concentrations of hormone. J Bone Miner Res. 1986;1(5):457–467. [DOI] [PubMed] [Google Scholar]
  • 19. Shao  Y, Zhong DS. Manual lymphatic drainage for breast cancer-related lymphoedema. Eur J Cancer Care (Engl). 2017;26(5):ecc.12517. [DOI] [PubMed] [Google Scholar]
  • 20. Simboli-Campbell  M, Gagnon AM, Franks DJ, et al.  1,25-Dihydroxyvitamin D3 translocates protein kinase Cβ to nucleus and enhances plasma membrane association of protein kinase Cα in renal epithelial cells. J Biol Chem. 1994;269(5):3257–3264. [PubMed] [Google Scholar]
  • 21. Ogura  M, Nishida S, Ishizawa M, et al.  Vitamin D3 modulates the expression of bile acid regulatory genes and represses inflammation in bile duct-ligated mice. J Pharmacol Exp Ther. 2009;328(2):564–570. [DOI] [PubMed] [Google Scholar]
  • 22. Schmidt  DR, Holmstrom SR, Tacer KF, et al.  Regulation of bile acid synthesis by fat-soluble vitamins A and D. J Biol Chem. 2010;285(19):14486–14494. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Ding  N, Yu RT, Subramaniam N, et al.  A vitamin D receptor/SMAD genomic circuit gates hepatic fibrotic response. Cell. 2013;153(3):601–613. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Abramovitch  S, Dahan-Bachar L, Sharvit E, et al.  Vitamin D inhibits proliferation and profibrotic marker expression in hepatic stellate cells and decreases thioacetamide-induced liver fibrosis in rats. Gut. 2011;60(12):1728–1737. [DOI] [PubMed] [Google Scholar]
  • 25. Veleva  BI, Bezooijen RL, Chel VGM, et al.  Effect of ultraviolet light on mood, depressive disorders and well-being. Photodermatol Photoimmunol Photomed. 2018;34(5):288–297. [DOI] [PubMed] [Google Scholar]
  • 26. Cui  X, Gooch H, Petty A, et al.  Vitamin D and the brain: genomic and non-genomic actions. Mol Cell Endocrinol. 2017;453:131–143. [DOI] [PubMed] [Google Scholar]
  • 27. Manson  JE, Cook NR, Lee I-M, et al.  Vitamin D supplements and prevention of cancer and cardiovascular disease. N Engl J Med. 2019;380(1):33–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Wu  S, Cho E, Li WQ, et al.  History of severe sunburn and risk of skin cancer among women and men in 2 prospective cohort studies. Am J Epidemiol. 2016;183(9):824–833. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Wu  S, Han J, Vleugels RA, et al.  Cumulative ultraviolet radiation flux in adulthood and risk of incident skin cancers in women. Br J Cancer. 2014;110(7):1855–1861. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Gandini  S, Puntoni M, Heckman-Stoddard BM, et al.  Metformin and cancer risk and mortality: a systematic review and meta-analysis taking into account biases and confounders. Cancer Prev Res. 2014;7(9):867–885. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Chang  YM, Barrett JH, Bishop TD, et al.  Sun exposure and melanoma risk at different latitudes: a pooled analysis of 5700 cases and 7216 controls. Int J Epidemiol. 2009;38(3):814–830. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Green  AC, Olsen CM. Cutaneous squamous cell carcinoma: an epidemiological review. Br J Dermatol. 2017;177(2):373–381. [DOI] [PubMed] [Google Scholar]
  • 33. Holman  DM, Ding H, Guy GP, et al.  Prevalence of sun protection use and sunburn and association of demographic and behaviorial characteristics with sunburn among US adults. JAMA. Dermatology. 2018;154(5):561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Weill  FS, Cela EM, Ferrari A, et al.  Skin exposure to chronic but not acute UV radiation affects peripheral T-cell function. J Toxicol Environ Health A. 2011;74(13):838–847. [DOI] [PubMed] [Google Scholar]
  • 35. Ellyard  JI, Simson L, Parish CR. Th2-mediated anti-tumour immunity: friend or foe? Tissue Antigens. 2007;70(1):1–11. [DOI] [PubMed] [Google Scholar]
  • 36. Togashi  Y, Shitara K, Nishikawa H. Regulatory T cells in cancer immunosuppression—implications for anticancer therapy. Nat Rev Clin Oncol. 2019;16(6):356–371. [DOI] [PubMed] [Google Scholar]
  • 37. Merten  JW, King JL, Walsh-Childers K, et al.  Skin cancer risk and other health risk behaviors: a scoping review. Am J Lifestyle Med. 2017;11(2):182–196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Malvy  DJ-M, Guinot C, Preziosi P, et al.  Relationship between vitamin D status and skin Phototype in general adult population. Photochem Photobiol. 2000;71(4):466–469. [DOI] [PubMed] [Google Scholar]
  • 39. Chandler  PD, Chen WY, Ajala ON, et al.  Effect of vitamin D3 supplements on development of advanced cancer: a secondary analysis of the VITAL randomized clinical trial. JAMA Netw Open. 2020;3(11):e2025850. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Keum  N, Lee DH, Greenwood DC, et al.  Vitamin D supplementation and total cancer incidence and mortality: a meta-analysis of randomized controlled trials. Ann Oncol. 2019;30(5):733–743. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Web_Material_kwac101
Click here for additional data file. (23.8KB, zip)

Articles from American Journal of Epidemiology are provided here courtesy of Oxford University Press

RESOURCES