Abstract
Epidemiologic evidence suggests an inverse association between sun exposure and follicular lymphoma risk.
We conducted an Australian population-based family case–control study based on 666 cases and 459 controls (288 related, 171 unrelated). Participants completed a lifetime residence and work calendar and recalled outdoor hours on weekdays, weekends, and holidays in the warmer and cooler months at ages 10, 20, 30, and 40 years, and clothing types worn in the warmer months. We used a group-based trajectory modeling approach to identify outdoor hour trajectories over time and examined associations with follicular lymphoma risk using logistic regression.
We observed an inverse association between follicular lymphoma risk and several measures of high lifetime sun exposure, particularly intermittent exposure (weekends, holidays). Associations included reduced risk with increasing time outdoors on holidays in the warmer months [highest category OR = 0.56; 95% confidence interval (CI), 0.42–0.76; Ptrend < 0.01], high outdoor hours on weekends in the warmer months (highest category OR = 0.71; 95% CI, 0.52–0.96), and increasing time outdoors in the warmer and cooler months combined (highest category OR = 0.66; 95% CI, 0.50–0.91; Ptrend 0.01). Risk was reduced for high outdoor hour maintainers in the warmer months across the decade years (OR = 0.71; 95% CI, 0.53–0.96).
High total and intermittent sun exposure, particularly in the warmer months, may be protective against the development of follicular lymphoma.
Although sun exposure is not recommended as a cancer control policy, confirming this association may provide insights regarding the future control of this intractable malignancy.
Introduction
Follicular lymphoma is an indolent lymphoid neoplasm arising from germinal center B cells which accounts for about 20% to 25% of all new non–Hodgkin lymphoma (NHL) cases in Western countries (1). Follicular lymphoma occurs at a slightly higher rate in males than females, at a median age of 61 years (2). The etiology of follicular lymphoma is not fully understood, but risk increases with age, family history, occupational exposure to specific pesticides, and smoking, and decreases with atopic disease (3–7). Sun exposure may reduce follicular lymphoma risk through vitamin D. Vitamin D3 generated via exposure to solar ultraviolet radiation (UVR) has anti-inflammatory properties and may regulate B-cell proliferation (8).
Epidemiologic evidence on the relationship between sun exposure and follicular lymphoma risk suggests an inverse association. A meta-analysis of observational studies examining various measures of solar UVR exposure including outdoor hours on working and nonworking days, working outdoor hours only, and satellite-based ambient UVR levels, found an inverse association between the highest level of exposure to each of these measures and follicular lymphoma risk (9). Pooled case–control study analyses show a reduced risk of follicular lymphoma with increasing recreational outdoor hours (3, 10) and total outdoor hours (3). The only study to examine outdoor hours at different age periods found no association with follicular lymphoma risk (11). A single case–control study observed an inverse association between increasing number of sunburns in childhood and follicular lymphoma risk (12), while others found no association with number of sunburns in childhood or adulthood (13–15). No cohort studies have examined the relationship between outdoor hours and follicular lymphoma risk. In addition, no prior studies examined patterns of outdoor hours over a lifetime and their association with follicular lymphoma risk. The group-based trajectory modeling approach has previously been used to identify distinct trajectories or patterns of exposure over the life course and disease associations (16–18), including sunburn trajectories and skin cancer risk (19).
We conducted a population-based family case–control study to further explore these associations in Australia, a setting with relatively high ambient UVR.
Materials and Methods
Cases
Eligible cases were patients notified to the New South Wales (NSW) or Victoria population-based cancer registry with newly diagnosed follicular lymphoma between 2011 and 2016, ages between 20 and 74 years, resident in NSW or Victoria, the two most populous states in Australia (20). Cases were also eligible if they had no history of hematopoietic malignancy and provided written informed consent. We identified a total of 1,791 cases. Of these, 213 cases, enriched for those with low confidence in the diagnosis on the basis of pathology report review of all cases, underwent slide review by an anatomic pathologist with a particular interest in hematopathology (J. Turner) to assess confidence in the diagnosis of follicular lymphoma (21); this identified 13 (6.1%) ineligible cases where the pathologic diagnosis could not be confirmed. Of the remaining 1,778 eligible and contactable cases, 733 (41.2%) declined and 1,045 (58.8%) consented to be approached by the study coordinating center. Of those approached by the study, 77 cases could not be reached, 770 (79.5%) were enrolled, and 198 (20.5%) declined. Of those enrolled, 717 cases (92.5%) completed the sun exposure and lifetime residence questionnaire (Supplementary Fig. S1). As per previous studies (22), we excluded participants of non-European origin (non-Caucasians; n = 46) from our analyses to ensure a relatively homogenous group as cutaneous absorption of solar radiation varies by cutaneous pigmentation.
Controls
Controls were related (siblings) or unrelated (spouse/partner) family members of cases, ages between 20 and 74 years, with no history of hematopoietic malignancy, who were able to give written informed consent. During recruitment, case participants were asked for consent to invite their family members to participate as controls in the study. When a case had multiple siblings, those of the same sex and closest in the age were approached first. Where cases had no siblings or consenting siblings, they nominated their spouse/partner. Of those approached, 65 controls were unreachable for a response. A total of 517 (80.0%) controls were enrolled and 130 (20.0%) declined. The participation rate for sibling and spouse controls were 80.0% and 79.8%, respectively. Of those enrolled, 491 (95.0%) controls completed the sun exposure and lifetime residence questionnaire (Supplementary Fig. S1). As for cases, we excluded participants of non-European origin (n = 32).
Ethics approval for this study was obtained from the NSW Population and Health Services Research Ethics Committee (2011/07/337) and the Cancer Council Victoria Human Research Ethics Committee (approval number 1114). The research was conducted in accordance with the Declaration of Helsinki and the National Health and Medical Research Councils’ National Statement on Ethical Conduct in Human Research.
Exposure assessment
Participants completed a lifetime residence and work calendar to aid their recall. They completed a questionnaire that asked how many hours they would normally have spent outdoors between 9 am and 5 pm on working or school days, nonworking days or weekends, and during holidays in the warmer months and the cooler months of the years they turned 10, 20, 30 and 40 (“decade years”). This method has shown good reproducibility (intraclass coefficient = 0.71) for total lifetime sun exposure (23) and has been used in multiple epidemiologic studies (24–26). Evidence suggests moderate correlation between self-reported outdoor hours and UV dosimeter-measured outdoor hours and standard erythemal doses, with a higher correlation for weekday exposure (27, 28). Participants reported how much time they spent on holidays (weeks) in both the warmer and cooler months at each decade year. The warmer months were defined as October to April in the Southern Hemisphere and March to August in the Northern Hemisphere. Evidence suggests reduced circulating levels of vitamin D during the cooler compared with warmer months (29). Participants reported the style of clothing they normally wore outdoors on the upper body (nothing, bikini top, singlet top, short- or long-sleeved shirt) and lower body (nothing, bikini bottom, shorts or short skirt, long pants or long skirt) during the warmer months in each decade year.
As for previous studies of risk factors for cutaneous melanoma (30) and NHL (31), participants described the skin color on the inside of their upper arm without tanning, the propensity of their skin to burn on exposure to bright sunlight for the first time in summer, and their skin tanning ability on repeated exposure to bright sunlight in summer without any sun protection.
Case clinical data
The managing clinician provided data on stage of disease (Ann Arbor system; I–IV), clinical variables necessary to calculate the Follicular Lymphoma International Prognostic Index (FLIPI/FLIPI-2; ref. 32), and the date and type of first-line treatment (none, radiotherapy and/or immunochemotherapy). We extracted histologic grade (1–3B) from pathology reports.
Statistical analysis
Recalled outdoor hours
We analyzed total outdoor hours on weekdays, weekends, and holidays in the warmer and cooler months in each decade year of age. We calculated the total outdoor hours on weekdays during the decade years in the cooler and warmer months by multiplying the hours reported per day by 5 and then by the total numbers of weeks (excluding holidays). Likewise, we calculated the total outdoor hours on weekends by multiplying the hours reported per day by 2 and then by the total numbers of weeks (excluding holidays). We generated the total outdoor hours during holidays by multiplying the hours reported by 7 and then by the total number of weeks. We summed these totals to obtain the total cumulative outdoor hours, and generated tertiles for each measure using the control distribution. We considered outdoor hours of sun exposure on weekends and holidays as markers of intermittent exposure (33).
Trajectories of outdoor hours
We analyzed the life course trajectories of outdoor hours separately for weekdays, weekends, and holidays by applying the group-based trajectory modeling approach (34). This approach involves the use of maximum likelihood method to identify underlying longitudinal trajectories. We used the STATA “Traj” plugin to identify groups that shared similar outdoor hour trajectories across the decade years based on the control distribution (35). We excluded 6 participants with four missing data points. We used the Bayesian information criterion (BIC) to determine the optimal number of groups and outdoor hours trajectories using a two-stage approach (36). The first stage was to determine the number of groups using the quadratic form for all trajectory groups. We started with 2-group and compared the BIC with that of 3- and 4-groups. Once we identified that the 2-group models best fitted our data, we then determined in the second stage the order of the polynomial function specifying the shape of each trajectory. We compared the BIC of the 2-group models, testing the zero-order, linear, quadratic, and cubic function. We observed the 2-group models with up to quadratic order terms 1 2 and 2 2 best fitted our data (Supplementary Table S1a). We assessed the average posterior probability of assignment in a group for each model and obtained values ranging from 0.82 to 0.98, exceeding the recommended minimum 0.70 (Supplementary Table S1b; ref. 36). We examined the correlation between each group's estimated probability and the proportion of participants classified to that group according to the maximum posterior probability assignment rule and obtained values ranging from 10.1 to 89.9, exceeding the recommended minimum of 5 (Supplementary Table S1c; ref. 36). We then described the trajectories of total outdoor hours, and separately for weekdays, weekends, and holidays, based on their visual pattern (Fig. 1).
Estimated lifetime trajectories of outdoor hours in solid lines and group percentages with 95% pointwise CIs: outdoor hours on weekdays only (A), weekends only (B), holidays only (C), outdoor hours in warmer months (D), outdoor hours in cooler months (E), total outdoor hours in warmer and cooler months (F).
Estimated lifetime trajectories of outdoor hours in solid lines and group percentages with 95% pointwise CIs: outdoor hours on weekdays only (A), weekends only (B), holidays only (C), outdoor hours in warmer months (D), outdoor hours in cooler months (E), total outdoor hours in warmer and cooler months (F).
Clothing type
For each clothing type worn outdoors in the warmer months of each decade year, we assigned a measure of the relative body surface area potentially exposed to sunlight. The whole body surface area was assigned a value of 100%; top half uncovered (50.0%), bikini top (37.5%), singlet top (25.0%), short sleeve top (12.5%), long sleeve top (0%), bottom half uncovered (50.0%), bikini bottom (33.3%), shorts or short skirt (16.7%), long pants or long skirt (0%). We summed the proportions for the top and bottom half to estimate the total proportion of skin exposed to sunlight for each participant. We then multiplied this value by the hours outdoors in the warmer months. We categorized the weighted values into tertiles based on their distribution in controls.
Follicular lymphoma risk
In our primary analyses, we examined the association between outdoor hours and follicular lymphoma risk using unconditional logistic regression and estimated ORs with 95% confidence intervals (CI). We combined siblings and spouse controls using the maximum likelihood approach and applied robust SEs to account for correlation of exposure among cases and sibling controls (37). We used directed acyclic graphs (38) to guide inclusion of covariates in our multivariable models, and adjusted for age (years), sex (male, female), state (NSW, Victoria), smoking (never, former, current), and propensity to burn as this pigmentary characteristic might be associated with sun avoidance (ref. 24; Supplementary Fig. S2). We used the lowest tertile category as the reference group in all analyses. We performed two sensitivity analyses: (i) we excluded cases and controls with missing data; (ii) we stratified by control source—excluding cases without sibling controls in the related controls conditional logistic regression model, and including all cases in the unrelated controls unconditional logistic regression model.
Data availability
Data are available from the authors upon reasonable request with the permission of all relevant human research ethics committees.
Results
Table 1 shows the characteristics of study participants. The median age of cases and controls was 61.0 [interquartile range (IQR), 52.8–67.2] years and 61.1 (IQR, 53.6–66.7) years, respectively. Approximately 48% of cases and 59% of controls were females. Of the 1,125 participants, 8 (0.7%) and 41 (3.6%) were below the age of 30 and 40, respectively. The proportion of participants with missing outdoor hours data was: age 10 (6.5%), 20 (8.4%), 30 (6.1%), and 40 (7.5%). The distribution of outdoor hours for each decade year is shown in Supplementary Fig. S3.
Characteristics of follicular lymphoma cases and controls.
. | . | Controls . | |
---|---|---|---|
Characteristics . | Cases n (%) . | Related n (%) . | Unrelated n (%) . |
Total | 666 (59.2) | 288 (26.6) | 171 (15.2) |
Sex | |||
Male | 349 (52.4) | 116 (40.3) | 70 (40.9) |
Female | 317 (47.6) | 172 (59.7) | 101 (59.1) |
Age (years) | |||
20–29 | 5 (0.8) | 2 (0.7) | 1 (0.6) |
30–39 | 21 (3.2) | 10 (3.5) | 2 (1.2) |
40–49 | 92 (13.8) | 45 (15.6) | 17 (9.9) |
50–59 | 192 (28.8) | 100 (34.7) | 42 (24.6) |
60–69 | 256 (38.4) | 103 (35.8) | 85 (49.7) |
70–74 | 100 (15.0) | 28 (9.7) | 24 (14.0) |
Smoking | |||
Never | 345 (51.8) | 165 (57.3) | 108 (63.2) |
Current | 61 (9.2) | 21 (7.3) | 13 (7.6) |
Former | 260 (39.0) | 102 (35.4) | 50 (29.2) |
Skin color | |||
Very fair | 94 (14.1) | 35 (12.1) | 26 (15.2) |
Fair | 364 (54.7) | 148 (51.4) | 91 (53.2) |
Light olive | 182 (27.3) | 99 (34.4) | 52 (30.4) |
Dark olive | 25 (3.8) | 6 (2.1) | 2 (1.2) |
Missing | 1 (0.1) | — | — |
Skin reaction to sun at first exposure in summer (propensity to burn) | |||
Brown without sunburn | 36 (5.4) | 13 (4.5) | 14 (8.2) |
Painful sunburn then peel | 427 (64.1) | 179 (62.2) | 107 (62.6) |
Severe sunburn with blisters | 201 (30.2) | 96 (33.3) | 50 (29.2) |
Missing | 2 (0.3) | — | — |
Skin reaction to sun on repeated exposure in summer (ability to tan) | |||
Very brown and deeply tanned | 160 (24.0) | 71 (24.7) | 35 (20.5) |
Moderately tanned | 299 (44.9) | 130 (45.1) | 73 (42.7) |
Mildly or occasionally tanned | 184 (27.6) | 82 (28.5) | 61 (35.6) |
No suntan at all or only get freckled | 21 (3.2) | 5 (1.7) | 2 (1.2) |
Missing | 2 (0.3) | — | — |
Follicular lymphoma stage at diagnosisa | |||
I–II | 168 (25.2) | ||
III–IV | 327 (49.1) | ||
Missing | 171 (25.7) | ||
Follicular lymphoma histologic gradea | |||
1–2 | 459 (68.9) | ||
3A–3Bb | 140 (21.0) | ||
Missing | 26 (3.9) | ||
Composite FL/DLBCLc | 41 (6.2) | ||
FLIPI scorea | |||
Low (0–1) | 168 (25.2) | ||
Intermediate (2) | 112 (16.8) | ||
High (3–4) | 135 (20.3) | ||
Missing | 251 (37.7) | ||
Treatmenta | |||
None | 156 (23.4) | ||
Immunochemotherapy | 272 (40.8) | ||
Radiotherapy | 43 (6.5) | ||
Immunochemotherapy/radiotherapy | 27 (4.1) | ||
Missing | 168 (25.2) |
. | . | Controls . | |
---|---|---|---|
Characteristics . | Cases n (%) . | Related n (%) . | Unrelated n (%) . |
Total | 666 (59.2) | 288 (26.6) | 171 (15.2) |
Sex | |||
Male | 349 (52.4) | 116 (40.3) | 70 (40.9) |
Female | 317 (47.6) | 172 (59.7) | 101 (59.1) |
Age (years) | |||
20–29 | 5 (0.8) | 2 (0.7) | 1 (0.6) |
30–39 | 21 (3.2) | 10 (3.5) | 2 (1.2) |
40–49 | 92 (13.8) | 45 (15.6) | 17 (9.9) |
50–59 | 192 (28.8) | 100 (34.7) | 42 (24.6) |
60–69 | 256 (38.4) | 103 (35.8) | 85 (49.7) |
70–74 | 100 (15.0) | 28 (9.7) | 24 (14.0) |
Smoking | |||
Never | 345 (51.8) | 165 (57.3) | 108 (63.2) |
Current | 61 (9.2) | 21 (7.3) | 13 (7.6) |
Former | 260 (39.0) | 102 (35.4) | 50 (29.2) |
Skin color | |||
Very fair | 94 (14.1) | 35 (12.1) | 26 (15.2) |
Fair | 364 (54.7) | 148 (51.4) | 91 (53.2) |
Light olive | 182 (27.3) | 99 (34.4) | 52 (30.4) |
Dark olive | 25 (3.8) | 6 (2.1) | 2 (1.2) |
Missing | 1 (0.1) | — | — |
Skin reaction to sun at first exposure in summer (propensity to burn) | |||
Brown without sunburn | 36 (5.4) | 13 (4.5) | 14 (8.2) |
Painful sunburn then peel | 427 (64.1) | 179 (62.2) | 107 (62.6) |
Severe sunburn with blisters | 201 (30.2) | 96 (33.3) | 50 (29.2) |
Missing | 2 (0.3) | — | — |
Skin reaction to sun on repeated exposure in summer (ability to tan) | |||
Very brown and deeply tanned | 160 (24.0) | 71 (24.7) | 35 (20.5) |
Moderately tanned | 299 (44.9) | 130 (45.1) | 73 (42.7) |
Mildly or occasionally tanned | 184 (27.6) | 82 (28.5) | 61 (35.6) |
No suntan at all or only get freckled | 21 (3.2) | 5 (1.7) | 2 (1.2) |
Missing | 2 (0.3) | — | — |
Follicular lymphoma stage at diagnosisa | |||
I–II | 168 (25.2) | ||
III–IV | 327 (49.1) | ||
Missing | 171 (25.7) | ||
Follicular lymphoma histologic gradea | |||
1–2 | 459 (68.9) | ||
3A–3Bb | 140 (21.0) | ||
Missing | 26 (3.9) | ||
Composite FL/DLBCLc | 41 (6.2) | ||
FLIPI scorea | |||
Low (0–1) | 168 (25.2) | ||
Intermediate (2) | 112 (16.8) | ||
High (3–4) | 135 (20.3) | ||
Missing | 251 (37.7) | ||
Treatmenta | |||
None | 156 (23.4) | ||
Immunochemotherapy | 272 (40.8) | ||
Radiotherapy | 43 (6.5) | ||
Immunochemotherapy/radiotherapy | 27 (4.1) | ||
Missing | 168 (25.2) |
Abbreviation: FLIPI, Follicular Lymphoma International Prognostic Index.
aCases only.
bGrade 3B = 41 cases.
cFL/DLBCL = Follicular lymphoma and diffuse large B-cell lymphoma.
Follicular lymphoma risk
Weekday outdoor hours
We found no significant association between follicular lymphoma risk and high or increasing outdoor hours on weekdays. This included for the warmer months, the warmer months weighted by the proportion of skin on body exposed, the cooler months, and the warmer and cooler months combined (Table 2).
ORs and 95% CIs for follicular lymphoma risk in relation to total lifetime outdoor hours during decade years.
Exposure measure . | Cases . | Related controls . | Unrelated controls . | OR (95% CI)a . | P . | Ptrend . |
---|---|---|---|---|---|---|
Weekdays | ||||||
Outdoor hours in warmer months | ||||||
Tertile 1 (<790.0) | 253 | 95 | 62 | Ref. | 0.12 | 0.18 |
Tertile 2 (790.0–1220.0) | 196 | 93 | 56 | 0.78 (0.58–1.05) | ||
Tertile 3 (>1220.0) | 212 | 99 | 53 | 0.80 (0.58–1.07) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<148.4) | 223 | 85 | 62 | Ref. | 0.84 | 0.88 |
Tertile 2 (148.4–296.0) | 211 | 97 | 52 | 1.06 (0.77–1.46) | ||
Tertile 3 (>296.0) | 201 | 93 | 54 | 0.98 (0.71–1.34) | ||
Outdoor hours in cooler months | ||||||
Tertile 1 (<585.0) | 218 | 99 | 54 | Ref. | 0.59 | 0.80 |
Tertile 2 (585.0–1017.0) | 218 | 89 | 63 | 0.96 (0.70–1.30) | ||
Tertile 3 (>1017.0) | 224 | 99 | 53 | 0.92 (0.68–1.25) | ||
Outdoor hours in warmer and cooler months | ||||||
Tertile 1 (<1400.0) | 240 | 98 | 58 | Ref. | 0.32 | 0.14 |
Tertile 2 (1400.0–2258.0) | 207 | 90 | 60 | 0.86 (0.65–1.18) | ||
Tertile 3 (>2258.0) | 214 | 99 | 53 | 0.79 (0.58–1.08) | ||
Weekends | ||||||
Outdoor hours in warmer months | ||||||
Tertile 1 (<640.0) | 229 | 94 | 57 | Ref. | 0.03 | 0.08 |
Tertile 2 (640.0–888.0) | 216 | 103 | 51 | 0.87 (0.64–1.18) | ||
Tertile 3 (>888.0) | 216 | 90 | 63 | 0.71 (0.52–0.96) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<180.8) | 225 | 83 | 64 | Ref. | 0.59 | 0.61 |
Tertile 2 (180.8–279.3) | 196 | 101 | 47 | 0.85 (0.64–1.14) | ||
Tertile 3 (>279.3) | 214 | 91 | 57 | 0.92 (0.69–1.24) | ||
Outdoor hours in cooler months | ||||||
Tertile 1 (<464.0) | 222 | 91 | 63 | Ref. | 0.18 | 0.34 |
Tertile 2 (464.0–722.0) | 213 | 102 | 49 | 0.94 (0.68–1.28) | ||
Tertile 3 (>722.0) | 225 | 94 | 48 | 0.80 (0.58–1.10) | ||
Outdoor hours in warmer and cooler months | ||||||
Tertile 1 (<1136.0) | 227 | 98 | 57 | Ref. | 0.14 | 0.07 |
Tertile 2 (1136.0–1594.0) | 214 | 98 | 53 | 0.93 (0.69–1.27) | ||
Tertile 3 (>1594.0) | 220 | 91 | 61 | 0.74 (0.54–1.02) | ||
Holidays | ||||||
Outdoor hours in warmer months | ||||||
Tertile 1 (<504.0) | 283 | 96 | 64 | Ref. | <0.01 | <0.01 |
Tertile 2 (504.0–756.0) | 211 | 90 | 61 | 0.73 (0.54–0.99) | ||
Tertile 3 (>756.0) | 156 | 100 | 46 | 0.56 (0.42–0.76) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<145.1) | 267 | 85 | 62 | Ref. | <0.01 | <0.01 |
Tertile 2 (145.1–257.2) | 195 | 90 | 58 | 0.73 (0.53–1.01) | ||
Tertile 3 (>257.2) | 162 | 99 | 48 | 0.60 (0.45–0.81) | ||
Outdoor hours in cooler months | ||||||
Tertile 1 (<154.0) | 242 | 97 | 57 | Ref. | 0.10 | 0.07 |
Tertile 2 (154.0–294.0) | 231 | 82 | 65 | 0.99 (0.73–1.34) | ||
Tertile 3 (>294.0) | 167 | 102 | 46 | 0.67 (0.50–1.02) | ||
Outdoor hours in warmer and cooler months | ||||||
Tertile 1 (<672.0) | 259 | 95 | 58 | Ref. | <0.01 | <0.01 |
Tertile 2 (672.0–1029.0) | 228 | 85 | 67 | 0.78 (0.58–1.06) | ||
Tertile 3 (>1029.0) | 163 | 106 | 46 | 0.59 (0.44–0.80) | ||
Total (weekdays, weekends and holidays combined) | ||||||
Outdoor hours in warmer monthsb | ||||||
Tertile 1 (<2096.0) | 260 | 97 | 57 | Ref. | 0.01 | 0.01 |
Tertile 2 (2096.0–2802.0) | 206 | 94 | 58 | 0.73 (0.54–1.00) | ||
Tertile 3 (>2802.0) | 195 | 96 | 56 | 0.64 (0.47–0.86) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<576.4) | 265 | 90 | 61 | Ref. | 0.18 | 0.10 |
Tertile 2 (576.4–875.1) | 218 | 94 | 58 | 0.80 (0.60–1.07) | ||
Tertile 3 (>875.1) | 177 | 101 | 51 | 0.65 (0.48–1.08) | ||
Outdoor hours in cooler monthsb | ||||||
Tertile 1 (<1388.0) | 248 | 96 | 57 | Ref. | 0.10 | 0.07 |
Tertile 2 (1388.0–2042.0) | 195 | 90 | 62 | 0.81 (0.60–1.10) | ||
Tertile 3 (>2042.0) | 217 | 101 | 51 | 0.73 (0.55–1.02) | ||
Outdoor hours in warmer and cooler monthsb | ||||||
Tertile 1 (<3444.0) | 248 | 94 | 56 | Ref. | 0.03 | 0.01 |
Tertile 2 (3444.0–4816.0) | 210 | 92 | 64 | 0.75 (0.55–1.01) | ||
Tertile 3 (>4816.0) | 203 | 101 | 51 | 0.66 (0.50–0.91) |
Exposure measure . | Cases . | Related controls . | Unrelated controls . | OR (95% CI)a . | P . | Ptrend . |
---|---|---|---|---|---|---|
Weekdays | ||||||
Outdoor hours in warmer months | ||||||
Tertile 1 (<790.0) | 253 | 95 | 62 | Ref. | 0.12 | 0.18 |
Tertile 2 (790.0–1220.0) | 196 | 93 | 56 | 0.78 (0.58–1.05) | ||
Tertile 3 (>1220.0) | 212 | 99 | 53 | 0.80 (0.58–1.07) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<148.4) | 223 | 85 | 62 | Ref. | 0.84 | 0.88 |
Tertile 2 (148.4–296.0) | 211 | 97 | 52 | 1.06 (0.77–1.46) | ||
Tertile 3 (>296.0) | 201 | 93 | 54 | 0.98 (0.71–1.34) | ||
Outdoor hours in cooler months | ||||||
Tertile 1 (<585.0) | 218 | 99 | 54 | Ref. | 0.59 | 0.80 |
Tertile 2 (585.0–1017.0) | 218 | 89 | 63 | 0.96 (0.70–1.30) | ||
Tertile 3 (>1017.0) | 224 | 99 | 53 | 0.92 (0.68–1.25) | ||
Outdoor hours in warmer and cooler months | ||||||
Tertile 1 (<1400.0) | 240 | 98 | 58 | Ref. | 0.32 | 0.14 |
Tertile 2 (1400.0–2258.0) | 207 | 90 | 60 | 0.86 (0.65–1.18) | ||
Tertile 3 (>2258.0) | 214 | 99 | 53 | 0.79 (0.58–1.08) | ||
Weekends | ||||||
Outdoor hours in warmer months | ||||||
Tertile 1 (<640.0) | 229 | 94 | 57 | Ref. | 0.03 | 0.08 |
Tertile 2 (640.0–888.0) | 216 | 103 | 51 | 0.87 (0.64–1.18) | ||
Tertile 3 (>888.0) | 216 | 90 | 63 | 0.71 (0.52–0.96) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<180.8) | 225 | 83 | 64 | Ref. | 0.59 | 0.61 |
Tertile 2 (180.8–279.3) | 196 | 101 | 47 | 0.85 (0.64–1.14) | ||
Tertile 3 (>279.3) | 214 | 91 | 57 | 0.92 (0.69–1.24) | ||
Outdoor hours in cooler months | ||||||
Tertile 1 (<464.0) | 222 | 91 | 63 | Ref. | 0.18 | 0.34 |
Tertile 2 (464.0–722.0) | 213 | 102 | 49 | 0.94 (0.68–1.28) | ||
Tertile 3 (>722.0) | 225 | 94 | 48 | 0.80 (0.58–1.10) | ||
Outdoor hours in warmer and cooler months | ||||||
Tertile 1 (<1136.0) | 227 | 98 | 57 | Ref. | 0.14 | 0.07 |
Tertile 2 (1136.0–1594.0) | 214 | 98 | 53 | 0.93 (0.69–1.27) | ||
Tertile 3 (>1594.0) | 220 | 91 | 61 | 0.74 (0.54–1.02) | ||
Holidays | ||||||
Outdoor hours in warmer months | ||||||
Tertile 1 (<504.0) | 283 | 96 | 64 | Ref. | <0.01 | <0.01 |
Tertile 2 (504.0–756.0) | 211 | 90 | 61 | 0.73 (0.54–0.99) | ||
Tertile 3 (>756.0) | 156 | 100 | 46 | 0.56 (0.42–0.76) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<145.1) | 267 | 85 | 62 | Ref. | <0.01 | <0.01 |
Tertile 2 (145.1–257.2) | 195 | 90 | 58 | 0.73 (0.53–1.01) | ||
Tertile 3 (>257.2) | 162 | 99 | 48 | 0.60 (0.45–0.81) | ||
Outdoor hours in cooler months | ||||||
Tertile 1 (<154.0) | 242 | 97 | 57 | Ref. | 0.10 | 0.07 |
Tertile 2 (154.0–294.0) | 231 | 82 | 65 | 0.99 (0.73–1.34) | ||
Tertile 3 (>294.0) | 167 | 102 | 46 | 0.67 (0.50–1.02) | ||
Outdoor hours in warmer and cooler months | ||||||
Tertile 1 (<672.0) | 259 | 95 | 58 | Ref. | <0.01 | <0.01 |
Tertile 2 (672.0–1029.0) | 228 | 85 | 67 | 0.78 (0.58–1.06) | ||
Tertile 3 (>1029.0) | 163 | 106 | 46 | 0.59 (0.44–0.80) | ||
Total (weekdays, weekends and holidays combined) | ||||||
Outdoor hours in warmer monthsb | ||||||
Tertile 1 (<2096.0) | 260 | 97 | 57 | Ref. | 0.01 | 0.01 |
Tertile 2 (2096.0–2802.0) | 206 | 94 | 58 | 0.73 (0.54–1.00) | ||
Tertile 3 (>2802.0) | 195 | 96 | 56 | 0.64 (0.47–0.86) | ||
Outdoor hours in warmer months weighted by proportion of skin on body exposedb,c | ||||||
Tertile 1 (<576.4) | 265 | 90 | 61 | Ref. | 0.18 | 0.10 |
Tertile 2 (576.4–875.1) | 218 | 94 | 58 | 0.80 (0.60–1.07) | ||
Tertile 3 (>875.1) | 177 | 101 | 51 | 0.65 (0.48–1.08) | ||
Outdoor hours in cooler monthsb | ||||||
Tertile 1 (<1388.0) | 248 | 96 | 57 | Ref. | 0.10 | 0.07 |
Tertile 2 (1388.0–2042.0) | 195 | 90 | 62 | 0.81 (0.60–1.10) | ||
Tertile 3 (>2042.0) | 217 | 101 | 51 | 0.73 (0.55–1.02) | ||
Outdoor hours in warmer and cooler monthsb | ||||||
Tertile 1 (<3444.0) | 248 | 94 | 56 | Ref. | 0.03 | 0.01 |
Tertile 2 (3444.0–4816.0) | 210 | 92 | 64 | 0.75 (0.55–1.01) | ||
Tertile 3 (>4816.0) | 203 | 101 | 51 | 0.66 (0.50–0.91) |
aMultivariable model—adjusted for age, sex, state, smoking, and propensity to burn.
bWeekdays, weekends, and holidays combined over decade years.
cImputations (number of participants with missing values): total outdoor hours in warmer months weighted by proportion of skin on body exposed (5).
Weekend outdoor hours
We observed an inverse association between follicular lymphoma risk and high outdoor hours on weekends in the warmer months (highest category OR = 0.71; 95% CI, 0.52–0.96; Table 2). We found no association with weekend outdoor hours in the warmer months weighted by the proportion of skin on body exposed, the cooler months, or the warmer and cooler months combined (Table 2).
Holiday outdoor hours
We found an inverse association between follicular lymphoma risk and high and increasing outdoor hours on holidays. This association was observed in the warmer months (Ptrend < 0.01), the warmer months weighted by the proportion of skin on body exposed (Ptrend < 0.01), and the warmer and cooler months combined (Ptrend 0.01; Table 2).
Total outdoor hours (weekday, weekend, and holiday combined)
We found an inverse association between follicular lymphoma risk and high and increasing total outdoor hours in the warmer months (Ptrend 0.01) and the warmer and cooler months combined (Ptrend 0.01; Table 2).
Outdoor hours in individual decade years
In terms of individual decade years, only high outdoor holiday hours at age 20 (Ptrend < 0.01) showed an inverse association with follicular lymphoma risk (Table 3). Nevertheless, the point estimates for the highest categories of outdoor hours were consistently below 1 for each measure and each decade year.
ORs and 95% CIs for follicular lymphoma risk in relation to outdoor hours on weekdays, weekends, and holidays at each decade year.
Exposure measure . | Cases . | Related controls . | Unrelated controls . | OR (95% CI)a . | P . | Ptrend . |
---|---|---|---|---|---|---|
Outdoor hours on weekdaysb | ||||||
Age 10 | ||||||
Tertile 1 (<420.0) | 240 | 109 | 51 | Ref. | 0.49 | 0.36 |
Tertile 2 (420.0–630.0) | 216 | 85 | 71 | 0.86 (0.63–1.15) | ||
Tertile 3 (>630.0) | 200 | 90 | 48 | 0.87 (0.65–1.18) | ||
Age 20 | ||||||
Tertile 1 (<245.0) | 245 | 98 | 58 | Ref. | 0.30 | 0.18 |
Tertile 2 (245.0–480.0) | 184 | 94 | 53 | 0.80 (0.60–1.08) | ||
Tertile 3 (>480.0) | 226 | 95 | 56 | 0.82 (0.61–1.11) | ||
Age 30 | ||||||
Tertile 1 (<260.0) | 223 | 96 | 58 | Ref. | 0.93 | 0.79 |
Tertile 2 (260.0–590.0) | 212 | 90 | 59 | 0.97 (0.72–1.30) | ||
Tertile 3 (>590.0) | 214 | 98 | 50 | 0.96 (0.71–1.30) | ||
Age 40 | ||||||
Tertile 1 (<245.0) | 242 | 90 | 61 | Ref. | 0.28 | 0.18 |
Tertile 2 (245.0–486.0) | 180 | 88 | 55 | 0.81 (0.60–1.11) | ||
Tertile 3 (>486.0) | 202 | 95 | 50 | 0.82 (0.60–1.11) | ||
Outdoor hours on weekendsb | ||||||
Age 10 | ||||||
Tertile 1 (<320.0) | 235 | 101 | 56 | Ref. | 0.12 | 0.06 |
Tertile 2 (320.0–452.0) | 216 | 85 | 62 | 0.92 (0.68–1.24) | ||
Tertile 3 (>452.0) | 203 | 98 | 53 | 0.75 (0.55–1.01) | ||
Age 20 | ||||||
Tertile 1 (<260.0) | 213 | 93 | 62 | Ref. | 0.13 | 0.18 |
Tertile 2 (260.0–419.0) | 232 | 97 | 52 | 1.11 (0.80–1.53) | ||
Tertile 3 (>419.0) | 210 | 96 | 55 | 0.79 (0.58–1.09) | ||
Age 30 | ||||||
Tertile 1 (<256.0) | 207 | 98 | 54 | Ref. | 0.26 | 0.21 |
Tertile 2 (256.0–414.0) | 232 | 98 | 52 | 1.10 (0.81–1.49) | ||
Tertile 3 (>414.0) | 211 | 87 | 62 | 0.83 (0.61–1.13) | ||
Age 40 | ||||||
Tertile 1 (<240.0) | 204 | 86 | 60 | Ref. | 0.14 | 0.11 |
Tertile 2 (240.0–396.0) | 224 | 97 | 49 | 1.00 (0.73–1.38) | ||
Tertile 3 (>396.0) | 198 | 89 | 57 | 0.76 (0.55–1.06) | ||
Outdoor hours during holidaysb | ||||||
Age 10 | ||||||
Tertile 1 (<322.0) | 217 | 87 | 62 | Ref. | 0.40 | 0.22 |
Tertile 2 (322.0–469.0) | 220 | 81 | 59 | 0.93 (0.70–1.26) | ||
Tertile 3 (>469.0) | 182 | 101 | 43 | 0.82 (0.61–1.12) | ||
Age 20 | ||||||
Tertile 1 (<140.0) | 282 | 98 | 60 | Ref. | <0.01 | <0.01 |
Tertile 2 (140.0–224.0) | 179 | 79 | 50 | 0.72 (0.54–1.26) | ||
Tertile 3 (>224.0) | 141 | 93 | 49 | 0.54 (0.40–0.73) | ||
Age 30 | ||||||
Tertile 1 (<112.0) | 226 | 87 | 51 | Ref. | 0.13 | 0.18 |
Tertile 2 (112.0–196.0) | 186 | 73 | 64 | 0.89 (0.66–1.20) | ||
Tertile 3 (>196.0) | 157 | 101 | 40 | 0.62 (0.46–1.08) | ||
Age 40 | ||||||
Tertile 1 (<105.0) | 233 | 81 | 52 | Ref. | 0.08 | 0.07 |
Tertile 2 (105.0–168.0) | 175 | 73 | 62 | 0.74 (0.50–1.09) | ||
Tertile 3 (>168.0) | 164 | 98 | 41 | 0.60 (0.43–1.03) |
Exposure measure . | Cases . | Related controls . | Unrelated controls . | OR (95% CI)a . | P . | Ptrend . |
---|---|---|---|---|---|---|
Outdoor hours on weekdaysb | ||||||
Age 10 | ||||||
Tertile 1 (<420.0) | 240 | 109 | 51 | Ref. | 0.49 | 0.36 |
Tertile 2 (420.0–630.0) | 216 | 85 | 71 | 0.86 (0.63–1.15) | ||
Tertile 3 (>630.0) | 200 | 90 | 48 | 0.87 (0.65–1.18) | ||
Age 20 | ||||||
Tertile 1 (<245.0) | 245 | 98 | 58 | Ref. | 0.30 | 0.18 |
Tertile 2 (245.0–480.0) | 184 | 94 | 53 | 0.80 (0.60–1.08) | ||
Tertile 3 (>480.0) | 226 | 95 | 56 | 0.82 (0.61–1.11) | ||
Age 30 | ||||||
Tertile 1 (<260.0) | 223 | 96 | 58 | Ref. | 0.93 | 0.79 |
Tertile 2 (260.0–590.0) | 212 | 90 | 59 | 0.97 (0.72–1.30) | ||
Tertile 3 (>590.0) | 214 | 98 | 50 | 0.96 (0.71–1.30) | ||
Age 40 | ||||||
Tertile 1 (<245.0) | 242 | 90 | 61 | Ref. | 0.28 | 0.18 |
Tertile 2 (245.0–486.0) | 180 | 88 | 55 | 0.81 (0.60–1.11) | ||
Tertile 3 (>486.0) | 202 | 95 | 50 | 0.82 (0.60–1.11) | ||
Outdoor hours on weekendsb | ||||||
Age 10 | ||||||
Tertile 1 (<320.0) | 235 | 101 | 56 | Ref. | 0.12 | 0.06 |
Tertile 2 (320.0–452.0) | 216 | 85 | 62 | 0.92 (0.68–1.24) | ||
Tertile 3 (>452.0) | 203 | 98 | 53 | 0.75 (0.55–1.01) | ||
Age 20 | ||||||
Tertile 1 (<260.0) | 213 | 93 | 62 | Ref. | 0.13 | 0.18 |
Tertile 2 (260.0–419.0) | 232 | 97 | 52 | 1.11 (0.80–1.53) | ||
Tertile 3 (>419.0) | 210 | 96 | 55 | 0.79 (0.58–1.09) | ||
Age 30 | ||||||
Tertile 1 (<256.0) | 207 | 98 | 54 | Ref. | 0.26 | 0.21 |
Tertile 2 (256.0–414.0) | 232 | 98 | 52 | 1.10 (0.81–1.49) | ||
Tertile 3 (>414.0) | 211 | 87 | 62 | 0.83 (0.61–1.13) | ||
Age 40 | ||||||
Tertile 1 (<240.0) | 204 | 86 | 60 | Ref. | 0.14 | 0.11 |
Tertile 2 (240.0–396.0) | 224 | 97 | 49 | 1.00 (0.73–1.38) | ||
Tertile 3 (>396.0) | 198 | 89 | 57 | 0.76 (0.55–1.06) | ||
Outdoor hours during holidaysb | ||||||
Age 10 | ||||||
Tertile 1 (<322.0) | 217 | 87 | 62 | Ref. | 0.40 | 0.22 |
Tertile 2 (322.0–469.0) | 220 | 81 | 59 | 0.93 (0.70–1.26) | ||
Tertile 3 (>469.0) | 182 | 101 | 43 | 0.82 (0.61–1.12) | ||
Age 20 | ||||||
Tertile 1 (<140.0) | 282 | 98 | 60 | Ref. | <0.01 | <0.01 |
Tertile 2 (140.0–224.0) | 179 | 79 | 50 | 0.72 (0.54–1.26) | ||
Tertile 3 (>224.0) | 141 | 93 | 49 | 0.54 (0.40–0.73) | ||
Age 30 | ||||||
Tertile 1 (<112.0) | 226 | 87 | 51 | Ref. | 0.13 | 0.18 |
Tertile 2 (112.0–196.0) | 186 | 73 | 64 | 0.89 (0.66–1.20) | ||
Tertile 3 (>196.0) | 157 | 101 | 40 | 0.62 (0.46–1.08) | ||
Age 40 | ||||||
Tertile 1 (<105.0) | 233 | 81 | 52 | Ref. | 0.08 | 0.07 |
Tertile 2 (105.0–168.0) | 175 | 73 | 62 | 0.74 (0.50–1.09) | ||
Tertile 3 (>168.0) | 164 | 98 | 41 | 0.60 (0.43–1.03) |
aMultivariable model—adjusted for age, sex, state, smoking, and propensity to burn.
Trajectories of outdoor hours across the decade years
We first considered the pattern of sun exposure and were able to identify two different trajectories for each of the weekday, weekend, and holiday exposures over time. Compared with those with a trajectory from moderate to stable low total outdoor hours on weekdays, those who maintained high outdoor hours showed no significant association with follicular lymphoma risk (Table 4; Fig. 1). Compared with the low outdoor hour maintainers on weekends, those who maintained moderate outdoor hours showed no significant association with follicular lymphoma risk (Table 4; Fig. 1). For holidays, those with a trajectory from moderate to low outdoor hours showed no association with follicular lymphoma risk, compared with those that maintained low outdoor hours (Table 4; Fig. 1).
ORs and 95% CIs for follicular lymphoma risk in relation to trajectories of outdoor hours across the decade years.
Exposure measure . | Cases . | Related controls . | Unrelated controls . | OR (95% CI)a . | P . |
---|---|---|---|---|---|
Outdoor hours trajectory on weekdays | |||||
Moderate to stable low | 495 | 220 | 133 | Ref. | 0.87 |
Maintained high | 166 | 67 | 38 | 0.98 (0.72–1.32) | |
Outdoor hours trajectory on weekends | |||||
Maintained low | 464 | 205 | 117 | Ref. | 0.10 |
Maintained moderate | 197 | 82 | 54 | 0.79 (0.59–1.05) | |
Outdoor hours trajectory on holidays | |||||
Maintained low | 600 | 234 | 143 | Ref. | 0.21 |
Moderate to low | 50 | 44 | 28 | 0.51 (0.34–1.06) | |
Total (weekdays, weekends, and holidays combined) | |||||
Outdoor hours trajectory in warmer months | |||||
Moderate to low | 505 | 203 | 129 | Ref. | 0.03 |
Maintained high | 156 | 84 | 42 | 0.71 (0.53–0.96) | |
Outdoor hours trajectory in cooler months | |||||
Maintained low | 500 | 215 | 133 | Ref. | 0.30 |
Maintained moderate | 160 | 72 | 37 | 0.85 (0.63–1.15) | |
Outdoor hours trajectory in warmer and cooler months | |||||
Moderate to low | 457 | 192 | 122 | Ref. | 0.22 |
Maintained high | 204 | 95 | 49 | 0.84 (0.64–1.11) |
Exposure measure . | Cases . | Related controls . | Unrelated controls . | OR (95% CI)a . | P . |
---|---|---|---|---|---|
Outdoor hours trajectory on weekdays | |||||
Moderate to stable low | 495 | 220 | 133 | Ref. | 0.87 |
Maintained high | 166 | 67 | 38 | 0.98 (0.72–1.32) | |
Outdoor hours trajectory on weekends | |||||
Maintained low | 464 | 205 | 117 | Ref. | 0.10 |
Maintained moderate | 197 | 82 | 54 | 0.79 (0.59–1.05) | |
Outdoor hours trajectory on holidays | |||||
Maintained low | 600 | 234 | 143 | Ref. | 0.21 |
Moderate to low | 50 | 44 | 28 | 0.51 (0.34–1.06) | |
Total (weekdays, weekends, and holidays combined) | |||||
Outdoor hours trajectory in warmer months | |||||
Moderate to low | 505 | 203 | 129 | Ref. | 0.03 |
Maintained high | 156 | 84 | 42 | 0.71 (0.53–0.96) | |
Outdoor hours trajectory in cooler months | |||||
Maintained low | 500 | 215 | 133 | Ref. | 0.30 |
Maintained moderate | 160 | 72 | 37 | 0.85 (0.63–1.15) | |
Outdoor hours trajectory in warmer and cooler months | |||||
Moderate to low | 457 | 192 | 122 | Ref. | 0.22 |
Maintained high | 204 | 95 | 49 | 0.84 (0.64–1.11) |
aMultivariable model—adjusted for age, sex, state, smoking, and propensity to burn.
We then considered total sun exposure, summing the hours across weekdays, weekends, and holidays, and stratifying by the time of year. Nonoverlapping trajectories were again able to be identified for each of the warmer months, cooler months, and warmer and cooler months combined. Compared with those with a trajectory from moderate to low total outdoor hours in the warmer months, those who maintained high outdoor hours were at reduced risk of follicular lymphoma (OR = 0.71; 95% CI, 0.53–0.96; Table 4; Fig. 1). Higher total outdoor hour trajectories during both the cooler months alone and the warmer and cooler months combined showed no association with follicular lymphoma risk (Table 4; Fig. 1).
Sensitivity analyses
Results from sensitivity analyses were consistent with the primary analyses (Supplementary Tables S2a–S3d; Supplementary Figs. S4 and S5), including the analyses of cases and controls with no missing data.
Discussion
Our population-based family case–control study provides additional evidence of a reduced risk of follicular lymphoma with high sun exposure. Several measures of high lifetime sun exposure, both intermittent and total, were associated with follicular lymphoma risk. The pattern was most consistent for sun exposure in the warmer months of the year. Furthermore, risk was reduced for those who maintained a trajectory of high total outdoor hours in the warmer months compared with those with a trajectory of moderate to low total outdoor hours.
Restricting measurement to the first four decade years of life is a valid measure of total lifetime sun exposure (23), and this approach has been used in several epidemiologic studies (24–26). Our findings broadly align with a recent meta-analysis of five case–control studies that found an inverse association between follicular lymphoma risk and high outdoor hours on working and nonworking days combined (meta-RR = 0.81; 95% CI, 0.67–0.99; ref. 9). Similarly, results from two pooled case–control analyses including previous Australian data consistently showed a reduced risk of follicular lymphoma with increasing recreational outdoor hours (Ptrend < 0.01; refs. 3, 10) and the highest category of recreational and nonrecreational outdoor hours combined (OR = 0.82; 95% CI, 0.69–0.99; quartile 4 vs. quartile 1; ref. 3). These results add to the mounting evidence of a protective effect of sun exposure on some cancers, including NHL and renal (39), colorectal, and breast cancer (40).
Our findings of an inverse association with outdoor hours in the warmer months on weekends and holidays suggests an intermittent sun exposure pattern, when solar UVR levels are highest, may play a protective role against follicular lymphoma. No previous study examined outdoor hours and follicular lymphoma risk in relation to the time of the year. A prior Australian case–control study examining outdoor hours in the warmer and cooler months reported an inverse association between NHL risk and increasing outdoor hours on holidays during the decade years (24), but did not stratify by time of year. This is the first evaluation of outdoor hours in the warmer months weighted by the proportion of skin on the body exposed to sunlight and not covered by clothing, and the inverse association was maintained, strengthening the observation.
With respect to sun exposure at specific times of life and follicular lymphoma risk, the only significant association we observed was with high outdoor hours during holidays at age 20. There are limited prior data on the life period of sun exposure. A U.S. hospital-based case–control study (n = 245 cases) found no association between follicular lymphoma risk and self-reported midday hours of sun exposure at birth–12, 13–21, 22–40, and >40 years (11). This study did not separately ascertain time outdoors on weekdays, weekends, and holidays or by time of year.
No cohort study has examined the association between time outdoors and follicular lymphoma risk. A U.S. (41) and Scandinavian (42) cohort study found no association between NHL risk and time outdoors in the summer months at <13, 13–19, or 20–39 years (41), or annual number of sunburns at <10, 10–19, 20–29, 30–39, 40–49 years (42).
Studies based on ambient UVR levels, as a proxy for individual sun exposure, are inconsistent. Two prospective U.S. cohort studies examining residential ambient UVR levels at enrolment found no evidence of an association with follicular lymphoma risk (13, 43). van Leeuwen and colleagues, in an Australian population-based cancer registry study, also reported no association between residential latitude band at diagnosis and follicular lymphoma risk (44). In contrast, a U.S. population-based study using the average annual noon-time cloud-adjusted solar UVB level (satellite-based) between 1982 and 1992 reported an inverse association with the highest quintile of ambient UVB (548–662 mW/m2) and follicular lymphoma risk among Whites, Hispanics, and all ethnic groups combined (45).
We identified two markedly different trajectories over time for each of weekday, weekend, and holiday outdoor hours. We showed a reduced risk of follicular lymphoma for those who maintained high outdoor hours in the warmer months compared with those with a trajectory from moderate to low outdoor hours. This finding suggests high early life sun exposure maintained to adulthood may be protective against follicular lymphoma. This modeling approach appears worthy of further investigation when examining the association between outdoor hours across the life course and cancer risk. In the only prior use of this approach related to sun exposure, a positive association was observed between the stable high, and high to low, sunburn trajectories and skin cancer risk, compared with the stable low trajectory (19).
Our findings stratified by control type were similar to those for combined controls. For related controls, the point estimates for the highest sun exposure measures were consistently less than 1. Findings were somewhat more variable for the smaller group of unrelated controls. For siblings and spouses/partners, we can expect a degree of correlation in early or later life sun exposure, respectively. Cases and siblings tend to have the same childhood environment and may have similar sun exposure behaviors, biasing our risk estimates toward the null (46). Likewise, cases and spouses/partners are likely to live together and may have similar sun exposure (47), again biasing risk estimates toward the null. However, we accounted for this correlation of exposure in our analyses.
This is the first study to use a population-based family case–control design to investigate the relationship between sun exposure and follicular lymphoma risk. We achieved robust control participation as family members have stronger motivation to participate as controls, hence reducing potential nonparticipation bias associated with traditional case–control studies (46). All cases were histologically confirmed and recruited via population-based cancer registries.
The precise mechanism through which sun exposure might reduce follicular lymphoma risk is unknown. Plausible biological pathways are through the anti-inflammatory activities of vitamin D and the activation of signaling pathways that regulate B-cell proliferation (8). Solar UVB radiation is absorbed by 7-dehydrocholesterol when the skin is directly exposed to the sun, transforming it to pre-vitamin D3 (48). This is closely regulated and only about 10%–15% of the total cutaneous 7-dehydrocholesterol is converted to pre-vitamin D3 (49). The conversion of pre-vitamin D3 to its biologically active form, 1,25-dihydroxyvitamin D3 (vitamin D3), is mediated by the vitamin D receptor and regulated by the enzyme cytochrome P450 monooxygenase 25(OH)D 1α hydroxylase in the kidney (50), and other extrarenal sites including the skin and lymph nodes and activated macrophages (51, 52). Vitamin D3 levels are positively correlated with the proportion of human body surface area exposed to solar UVB radiation (53). Vitamin D3 inhibits the nuclear factor kappa B (NFκB) pathway which in turn inhibits B-cell proliferation (54) and downregulation of downstream mammalian target of rapamycin (mTOR), a protein activated in follicular lymphoma (55, 56). Inhibition of the mTOR signaling pathway has been explored for the treatment of follicular lymphoma (57, 58).
Exposure to solar UVR is immunosuppressive and induces DNA damage (59, 60). In a cross-sectional study, a correlation was observed between increasing solar UVR exposure and increasing DNA damage in human lymphocytes (61). Of direct relevance to our findings however, Narbutt and colleagues, reported increased apoptosis of human peripheral blood mononuclear cells after whole body exposure to low-dose UVB, equivalent to about 35 minutes of midday exposure in summer, over 10 days (62). Intriguingly, this UVB exposure decreased the expression of B-cell lymphoma-2 (BCL-2), the protein overexpressed in most follicular lymphoma (63, 64). Furthermore, Nair-Shalliker and colleagues observed reduced DNA methylation in circulating human lymphocytes with increasing exposure to solar UVR (65). Somatic mutations in epigenetic regulators, transcription factors, and elements of the mTOR signaling pathway predominate in follicular lymphoma and are believed to follow t(14;18) translocation (66–68). Finally, although multiple common germline variants have been associated with follicular lymphoma risk (69, 70), particularly in the human leukocyte antigen class regions, none seems to implicate a pathway involving UVR, although the immune modifying effects of UVR are not well understood.
Melanoma and nonmelanoma skin cancers are characterized by “UV signatures,” unique germline mutations (predominantly C>T; ref. 71). A UV signature has also been detected in cutaneous T-cell lymphoma (72), but not in B-cell NHLs (73). A UV signature would not be anticipated for cancers exhibiting a protective association with UVR.
Our study was restricted to participants of European origin (Caucasians) living in a high UVR setting, and controls were family members, limiting generalizability. We also acknowledge not all those who were eligible agreed to participate, and the nonparticipation may have biased our results. Furthermore, we cannot rule out measurement error and recall bias given participants self-reported their outdoor hours in the distant past. The bias is likely to be nondifferential because sun exposure is not a well-known or established risk factor for follicular lymphoma, and it is unlikely that cases would link their cancer to time outdoors. This nondifferential misclassification may have biased our risk estimates toward the null. On the other hand, we ascertained recalled outdoor hours using an established approach, including a lifetime calendar to aid recall, and we modeled exposure over the life course. Nevertheless, we did not utilize objective sun exposure measures, and as a result, caution should be taken in the interpretation of our results. We acknowledge the multiplicity of our analyses may have resulted in chance findings. Furthermore, we did not collect data on the use of sunscreen or hats, and this may have resulted in misclassification of exposure in our analyses of outdoor hours weighted by the proportion of skin exposed. Also, we did not collect data on sunburn history or account for actual ambient UVR levels. Finally, we did not ascertain exposure to artificial UVR including use of tanning beds or arc welding.
In summary, findings from this population-based case–control study are consistent with prior evidence indicating a reduced risk of follicular lymphoma with high sun exposure, particularly intermittent sun exposure. Our assessment of outdoor hour trajectories indicates that consistently high sun exposure during the warmer months over a life course may reduce follicular lymphoma risk. Despite these findings, sun exposure is not recommended as a cancer control policy given that solar UVR is an established carcinogen for skin cancer (74). Prospective studies are needed to affirm our findings.
Authors' Disclosures
M.T. van Leeuwen reports grants from National Health and Medical Research Council during the conduct of the study. J. Trotman reports other support from BeiGene, Roche, Janssen, Cellectar, and BMS outside the submitted work. E. Verner reports grants from Janssen and personal fees from BeiGene outside the submitted work. G.G. Giles reports grants from National health and medical research council (Australia) during the conduct of the study. C.M. Vajdic reports grants from Australian National Health and Medical Research Council and Cancer Institute New South Wales, and other support from University of New South Wales International Postgraduate Award during the conduct of the study. No disclosures were reported by the other authors.
Authors' Contributions
M.K. Odutola: Data curation, formal analysis, investigation, visualization, methodology, writing–original draft. M.T. van Leeuwen: Conceptualization, resources, supervision, funding acquisition, validation, investigation, methodology, writing–review and editing. F. Bruinsma: Conceptualization, resources, investigation, writing–review and editing. J. Turner: Conceptualization, resources, formal analysis, funding acquisition, investigation, methodology, writing–review and editing. M. Hertzberg: Resources, data curation, writing–review and editing. J.F. Seymour: Conceptualization, resources, data curation, funding acquisition, writing–review and editing. H.M. Prince: Conceptualization, resources, data curation, funding acquisition, writing–review and editing. J. Trotman: Resources, data curation, writing–review and editing. E. Verner: Resources, data curation, writing–review and editing. F. Roncolato: Resources, data curation, writing–review and editing. S. Opat: Resources, data curation, writing–review and editing. R. Lindeman: Resources, data curation, writing–review and editing. C. Tiley: Resources, data curation, writing–review and editing. S.T. Milliken: Conceptualization, resources, data curation, funding acquisition, writing–review and editing. C.R. Underhill: Resources, data curation, writing–review and editing. G. Benke: Conceptualization, resources, data curation, funding acquisition, writing–review and editing. G.G. Giles: Conceptualization, resources, funding acquisition, methodology, writing–review and editing. C.M. Vajdic: Conceptualization, resources, supervision, funding acquisition, validation, investigation, methodology, writing–review and editing.
Acknowledgments
The authors thank the study participants and the participating clinical sites and clinicians including Dr Michael Harvey, Dr Duncan Carradice, Dr Cecily Forsyth, Dr Pauline Warburton, and Dr William Stevenson. We acknowledge the Victorian Cancer Registry and NSW Cancer Registry for supporting patient recruitment. We also acknowledge the assistance of the AIHW Data Linkage Unit for undertaking the data linkage to the National Death Index.
This work was supported by the National Health and Medical Research Council of Australia (ID 1006707). The National Health and Medical Research Council also supported M.T. van Leeuwen (ID 1012141). M.K. Odutola is supported by a University of New South Wales International Postgraduate Award Scholarship and a Cancer Institute New South Wales Translational Cancer Research Network PhD Scholarship Top-up award.
The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.
Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).