Abstract
Purpose
Associations between sun exposure (a primary source of Vitamin D) and risk of ovarian cancer have been inconsistent. Furthermore, studies have not investigated whether sun exposure at different periods in the lifetime of a person results in differences in risk associations, and little is known about differences according to histological subtype.
Methods
Using a population-based case-control study of 1,334 non-Hispanic white women diagnosed with epithelial ovarian cancer in western Washington State between 2002–2009 and 1,679 non-Hispanic white controls, we assessed the relation of epithelial ovarian cancer with constitutional pigmentation characteristics, sun exposure behaviors, and an index of ultraviolet (UV) exposure based on residential history. Information was collected through in-person interviews. Logistic regression was used to compute odds ratios, 95% confidence intervals and trend P values (Ptrend).
Results
We noted no association with residence-based measures of UV exposure or self-reported sun exposure, either over the lifetime or within specific age intervals. Also, we observed little evidence of association between constitutional pigmentation characteristics and risk, save for a suggestion of increased risk among women who reported increased ability to suntan upon prolonged sun exposure (Ptrend= 0.03).
Conclusions
Results from this study suggest that sun exposure has little influence on the risk of epithelial ovarian cancer. Additional studies in populations with a wider gradient of sun exposure may yet be warranted.
Keywords: Ovarian cancer, sun exposure, histology, epidemiology
Introduction
Ecological studies comparing rates of ovarian cancer in higher versus lower latitudes suggest a relationship between increasing sun exposure and decreasing incidence of ovarian cancer [1–3]. It is known that exposure of the skin to solar UV-B radiation produces circulating vitamin D, as measured by the 25-Hydroxyvitamin D (25(OH)D) metabolite in serum [4], and that sun exposure is the largest source of vitamin D for most people [5]. Experimental studies indicate that higher levels of vitamin D decrease cell proliferation and induce apoptosis in ovarian tissue [6, 7]. Despite these observations, the findings of epidemiological studies examining the relation of sun exposure or vitamin D levels in blood with risk of ovarian cancer have been inconsistent [8–13]. A study of 516 cases and 770 matched controls nested within a pool of seven prospective studies reported no overall association between circulating 25(OH)D levels and the risk of epithelial ovarian cancer [13], while a subsequent comparison within the National Health and Nutrition Examination Survey (NHANES) found lower levels of 25(OH)D among women with a history of ovarian cancer (n=28) than among 7,245 women without such a history [8]. No study has yet explored the possible effects of sun exposure early in life as compared to sun exposure close to the diagnosis of ovarian cancer, as has been done in other cancers [14–16] and few studies have assessed differences in risk according to histological subtype [12].
Using data from a large population-based case-control study, we examined associations of constitutional pigmentation characteristics, self-reported sun exposure, and residential-history based measures of potential UV exposure with risk of epithelial ovarian cancer. We further explored the extent to which early-life and close-to-diagnosis measures of sun exposure influence this risk and characterized associations according to histological subtype.
Methods
A population-based case control study was conducted among residents of 13 counties of Western Washington State. Eligible cases were women diagnosed with primary (borderline or invasive) epithelial ovarian cancer between January 1, 2002 and December 31, 2009, and were able to communicate in English. From 2002–2005, women aged 35–74 years were considered eligible and from 2006–2009, women aged 35–69 years were eligible. They were identified through the Cancer Surveillance System (CSS), a population-based cancer registry participating in the Surveillance, Epidemiology, and End Results Program of the National Cancer Institute [17]; of the potentially eligible women, 61 cases were excluded due to a language barrier. Of the 2,025 eligible cases, 1,502 (74.2%) were interviewed and 1,108 had invasive disease and 394 had borderline tumors. Histologic type was collected and coded by the CSS using the International Classification of Diseases for Oncology (ICD-O) morphology codes [18] and grouped according to the guidelines of the WHO [19] as serous, mucinous, endometrioid, clear cell and “other” epithelial tumors (comprised largely of unspecified adenocarcinoma and carcinoma).
Controls were selected from among English-speaking women with no prior history of ovarian cancer and at least one intact ovary who resided in the 13-county area covered by the CSS. Random digit dialing (RDD) methods were used, with stratified sampling in 5-year age categories and two county strata. From 2002–2005, a 2:1 ratio of controls (35–74 years of age) to women with invasive ovarian cancer was selected using the Waksberg-Mitofsky RDD method [20, 21]. From 2006–2009, a 1:1 ratio of controls (35–69 years of age) to women with invasive disease was used; list-assisted RDD [22] was employed during 2006–2007 and Waksberg-Mitofsky methods during 2008–2009. In total, for 19,092 (78.2%) of the 24,400 telephone numbers belonging to residences, we determined whether an eligible woman resided there; of the 2,351 eligible women identified, 1,849 were interviewed (78.6%) (Table 1). They were assigned a randomly selected calendar year between 2002 and 2009, the reference date, and this date was frequency matched to the year of diagnosis for the cases.
Table 1.
Outcomes of telephone screening for control recruitment, western Washington State, 2002–2009.
| Total telephone numbers called | 84,462 | ||||
| Ineligible telephone number | 53,653 | ||||
| Business | 11,357 | ||||
| Non-working | 36,282 | ||||
| Non-business (institutions, group quarters, data-lines, cell phones) | 6,014 | ||||
| Unknown if residential telephone number | 6,409 | ||||
| Known residential telephone number | 24,400 | ||||
| Unknown if individual eligible | 5,308 | ||||
| Answering machine on all attempts | 2,012 | ||||
| Refused age/county questions | 2,297 | ||||
| Age/county eligible, ovarian status unknown | 488 | ||||
| Other (language/communication barrier) | 511 | ||||
| Respondent screened | 19,092 | ||||
| Not eligible | 16,741 | ||||
| Did not fit age/county eligibility or frequency matching criteria | 16,087 | ||||
| No ovaries | 624 | ||||
| Prior ovarian cancer | 6 | ||||
| Selected woman had communication barrier | 24 | ||||
| Eligible | 2,351 | ||||
| Not interviewed | 502 | ||||
| Interviewed | 1,849 |
All women provided signed, informed consent before participating in the study, which was approved by the Institutional Review Board of the Fred Hutchinson Cancer Research Center. In-person interviews were conducted in relation to events that occurred before a woman’s diagnosis/reference date and included demographic and lifestyle factors, family history of cancer and reproductive history. Multiple measures related to pigmentation characteristics and sun exposure were collected, including tendency to sunburn on first exposure to summer sun and ability to suntan after prolonged sun exposure, as well as measures of sun-related behaviors. Information on sun-exposure behaviors was collected for each 10-year age period, starting in the teens. Specifically, subjects were questioned about the time spent in the mid-day sun during the summer months, the months per year with a tan and the use of sunscreen in the mid-day sun. A weekly weighted average of daily (summer) sun exposure within each 10-year age period was computed based on questions in which women separately reported time spent in the mid-day sun on weekends and weekdays.
Place of residence (as city, within state or country) was obtained for locations where participants had lived for at least a year after age 25 as well as the age intervals lived at each location. These residential data were linked to latitude and longitudes, which were then linked to a 5-year averaged daily summary measure from NASA’s Total Ozone Mapping Spectrometer (TOMS) satellite from 1987–1992 (http://ozoneaq.gsfc.nasa.gov/TOMSUVExposure.md) to compute an index of erythemal exposure (EE). EE is a measure of the potential for biologic damage due to solar radiation, and is computed for every 1° latitude by 1.25° longitude; factors taken into account include the distance from the sun to the earth, the biological action spectrum for erythemal damage, cloud cover, total column ozone, solar zenith angle at a particular time, and the spectral irradiance at the earth’s surface under clear skies [23]. As examples, the EE value for Seattle, WA, USA is 1897; low EE areas include Vancouver, BC, Canada (EE value, 1599) and Anchorage, AK, USA (983); high EE areas include Los Angeles, CA, USA (3548) and Miami, FL, USA (4036) [23]. For analysis, we computed the mean EE based on residential history from age 25 to one year before the diagnosis or comparable reference date, as well as the mean EE in the 10 years prior to this latter time point.
Given that production of vitamin D from sunlight is influenced by skin pigmentation, which is correlated with race, analyses were restricted to non-Hispanic whites (1,334 cases (88.8% of those interviewed) and 1,679 controls (90.8% of those interviewed)). Unconditional logistic regression was used to compute odds ratios (OR) and 95% confidence intervals (CI). ORs were adjusted by the matching variables (age at diagnosis/reference date categorized in 5-year intervals, year of diagnosis/reference date and county of residence) as well as number of full term pregnancies and duration of use of hormonal contraceptives. Adjustment for additional potential confounders including body mass index (BMI), tendency to burn upon first seasonal (summer) sun exposure, ability to tan after prolonged sun exposure and mean EE levels did not substantially change the ORs we report, and thus were not included in the final models. Trend p-values were obtained as the p-value associated with the corresponding continuous variable in the logistic model. For the EE related variables, the trend p-values were computed considering the category with the lower level as the reference category. For examination of risk within histologic subtypes of invasive cancers, we excluded women with mucinous cancers as there were too few for meaningful analysis. All analyses were done using the STATA software package (version 10.1, STATA Corporation, College Station, TX) for Macintosh.
Results
Cases were less likely to have had a full term pregnancy, to have taken hormonal contraceptives, or to have undergone tubal ligation (Table 2). In addition, cases tended to report a higher BMI.
Table 2.
Characteristics of non-Hispanic white women with invasive or borderline epithelial ovarian cancer and controls, western Washington State, 2002–2009.
| Characteristics | Cases (N=1,334) n (%) |
Controls (N=1,679) n (%) |
|---|---|---|
| Age (years) | ||
| 35–44 | 163 (12.2) | 177 (10.5) |
| 45–54 | 443 (33.2) | 482 (28.7) |
| 55–64 | 483 (36.2) | 626 (37.3) |
| 65–74 | 245 (18.4) | 394 (23.5) |
| BMI at age 30 (Kg/m2) | ||
| < 25.0 | 1,038 (78.3) | 1,386 (83.3) |
| 25.0–29.9 | 168 (12.7) | 128 (11.3) |
| ≥30 | 120 ( 9.0) | 61 ( 5.4) |
| Missing | 8 | 16 |
| Education | ||
| High school or less | 325 (24.4) | 360 (21.5) |
| Some college or technical school | 504 (37.8) | 612 (36.5) |
| College graduate | 303 (22.7) | 396 (23.6) |
| Post-college studies | 200 (15.0) | 310 (18.5) |
| Missing | 2 | 1 |
| Full term pregnancies | ||
| 0 | 329 (24.7) | 258 (15.4) |
| 1 | 219 (16.4) | 226 (13.5) |
| 2 | 403 (30.2) | 570 (33.9) |
| ≥3 | 383 (28.7) | 625 (37.2) |
| Hormonal contraceptive use (years) | ||
| Never | 293 (22.0) | 283 (16.9) |
| Less than 6 months | 117 ( 8.8) | 108 ( 6.4) |
| 6 months to less than 5 years | 488 (36.6) | 609 (36.3) |
| 5 years to less than 10 years | 247 (18.5) | 356 (21.2) |
| 10 years or more | 187 (14.0) | 323 (19.2) |
| Missing | 2 | 0 |
| Tubal ligation | ||
| No | 1,098 (82.3) | 1,294 (77.1) |
| Yes | 236 (17.7) | 385 (22.9) |
| Family history of breast and ovarian cancer | ||
| Not known | 739 (57.8) | 975 (59.8) |
| Breast cancer only | 385 (30.1) | 540 (33.1) |
| Ovarian cancer only | 85 ( 6.7) | 65 ( 4.0) |
| Breast and ovarian cancer | 69 ( 5.4) | 50 ( 3.1) |
| Missing | 56 | 49 |
In general, there were no clear differences in pigmentation characteristics, including natural skin color, natural hair color and natural eye color, between cases and controls, either overall or when borderline and invasive tumors were considered separately (Table 3). Relative to women who reported no ability to tan after prolonged sun exposure, women who reported moderate or deep degrees of tanning after such exposure were at a modestly increased risk of ovarian cancer (Ptrend=0.03). Average EE levels (based on place(s) of residence) from age 25 to the diagnosis/reference date minus 1 year ranged from 1,109 to 4,158 among cases and from 1,127 to 4,321 among controls, with median value 1,897 for both groups. Average EE levels in the decade before the diagnosis/reference date minus 1 year ranged from 1,249 to 3,771 among cases and from 1,123 to 3,769 among controls, with median value 1,897 for both groups. Neither measure was associated with ovarian cancer risk (Table 3). These results changed only slightly when additionally adjusted for tendency to burn, ability to tan, and/or BMI.
Table 3.
Pigmentation characteristics, tendency to burn, ability to tan and erythemal exposure (EE)‡ and risk of epithelial ovarian cancer by tumor type, western Washington State, 2002–2009.
| All tumors | Borderline tumors | Invasive tumors | |||||
|---|---|---|---|---|---|---|---|
| Controls (N=1,679) |
Cases (N=1,334) |
Cases (N=351) |
Cases (N=983) |
||||
| n (%) | n (%) | OR* (95% CI) | n (%) | OR* (95% CI) | n (%) | OR* (95% CI) | |
| Natural skin color | |||||||
| Light | 1,194 (71.1) | 962 (72.3) | 1.00 (ref.) | 258 (73.7) | 1.00 (ref.) | 704 (71.8) | 1.00 (ref.) |
| Medium | 480 (28.6) | 363 (27.3) | 1.07 (0.90–1.26) | 91 (26.0) | 1.03 (0.78–1.35) | 272 (27.7) | 1.06 (0.88–1.28) |
| Dark | 5 ( 0.3) | 6 ( 0.5) | 1.72 (0.51–5.82) | 1 ( 0.3) | 0.90 (0.09–8.51) | 5 ( 0.5) | 2.15 (0.60–7.74) |
| Ptrend | 0.36 | 0.87 | 0.38 | ||||
| Natural hair color (as a teenager) | |||||||
| Blonde or red | 524 (31.2) | 438 (32.9) | 1.00 (ref.) | 122 (34.8) | 1.00 (ref.) | 316 (32.2) | 1.00 (ref.) |
| Light brown | 348 (20.7) | 239 (17.9) | 0.81 (0.65–1.00) | 57 (16.2) | 0.65 (0.46–0.93) | 182 (18.5) | 0.86 (0.68–1.09) |
| Medium or dark brown | 774 (46.1) | 642 (48.2) | 0.96 (0.81–1.14) | 167 (47.6) | 0.89 (0.68–1.17) | 475 (48.4) | 0.98 (0.81–1.18) |
| Black | 33 ( 2.0) | 14 ( 1.1) | 0.52 (0.27–1.02) | 5 ( 1.4) | 0.79 (0.28–2.18) | 9 ( 0.9) | 0.43 (0.20–0.93) |
| Ptrend | 0.46 | 0.51 | 0.54 | ||||
| Natural eye color | |||||||
| Blue or Grey | 655 (39.0) | 529 (39.7) | 1.00 (ref.) | 135 (38.5) | 1.00 (ref.) | 394 (40.1) | 1.00 (ref.) |
| Green | 232 (13.8) | 185 (13.9) | 0.98 (0.77–1.24) | 47 (13.4) | 0.87 (0.59–1.27) | 138 (14.0) | 1.03 (0.80–1.33) |
| Hazel | 388 (23.1) | 281 (21.1) | 0.88 (0.72–1.07) | 70 (19.9) | 0.88 (0.64–1.23) | 211 (21.5) | 0.88 (0.71–1.10) |
| Brown | 404 (24.1) | 339 (25.4) | 1.01 (0.83–1.23) | 99 (28.2) | 1.10 (0.81–1.49) | 240 (24.4) | 0.97 (0.78–1.20) |
| Ptrend | 0.77 | 0.70 | 0.52 | ||||
| Tendency to burn§ | |||||||
| Painful burn with blistering or peeling | 651 (39.1) | 530 (39.9) | 1.00 (ref.) | 131 (37.5) | 1.00 (ref.) | 399 (40.8) | 1.00 (ref.) |
| Mild burn then tan | 809 (48.5) | 619 (46.6) | 0.90 (0.76–1.06) | 166 (47.6) | 0.93 (0.71–1.21) | 453 (46.3) | 0.87 (0.73–1.04) |
| Tan without burn | 207 (12.4) | 178 (13.4) | 1.07 (0.84–1.36) | 52 (14.9) | 1.35 (0.92–1.98) | 126 (12.9) | 0.97 (0.74–1.26) |
| Ptrend | 0.92 | 0.33 | 0.42 | ||||
| Ability to tan¶ | |||||||
| Freckle or no tan | 203 (12.1) | 150 (11.3) | 1.00 (ref.) | 36 (10.3) | 1.00 (ref.) | 114 (11.6) | 1.00 (ref.) |
| Mild tan | 389 (23.2) | 277 (20.8) | 0.92 (0.70–1.21) | 73 (20.9) | 0.92 (0.58–1.45) | 204 (20.8) | 0.92 (0.68–1.24) |
| Moderate tan | 773 (46.1) | 638 (47.9) | 1.11 (0.87–1.42) | 166 (47.4) | 1.05 (0.70–1.58) | 472 (48.1) | 1.13 (0.87–1.48) |
| Deep tan | 310 (18.5) | 267 (20.0) | 1.22 (0.92–1.61) | 75 (21.4) | 1.24 (0.79–1.96) | 192 (19.6) | 1.21 (0.89–1.64) |
| Ptrend | 0.03 | 0.18 | 0.05 | ||||
| Mean EE level from age 25 to diagnosis/reference date minus 1 year£ | |||||||
| < 1897 | 415 (24.7) | 337 (25.3) | 0.89 (0.70–1.13) | 97 (27.6) | 1.08 (0.74–1.59) | 240 (24.4) | 0.83 (0.64–1.07) |
| 1897 | 499 (29.8) | 426 (31.9) | 1.00 (ref.) | 107 (30.5) | 1.00 (ref.) | 319 (32.5) | 1.00 (ref.) |
| > 1897 and < 2042 | 343 (20.5) | 218 (16.3) | 0.78 (0.62–0.97) | 54 (15.4) | 0.87 (0.60–1.27) | 164 (16.7) | 0.75 (0.59–0.96) |
| ≥2042 | 420 (25.0) | 353 (26.5) | 0.97 (0.79–1.19) | 93 (26.5) | 1.10 (0.79–1.53) | 260 (26.4) | 0.94 (0.75–1.18) |
| Ptrend | 0.98 | 0.96 | 0.86 | ||||
| Mean EE level in the 10 years prior to diagnosis/reference date minus 1 year£ | |||||||
| < 1897 | 495 (29.5) | 397 (29.8) | 0.92 (0.71–1.20) | 110 (31.3) | 1.14 (0.75–1.74) | 287 (29.2) | 0.84 (0.63–1.12) |
| 1897 | 924 (57.4) | 748 (56.1) | 1.00 (ref.) | 182 (51.9) | 1.00 (ref.) | 566 (57.6) | 1.00 (ref.) |
| > 1897 | 220 (13.1) | 189 (14.2) | 1.05 (0.71–1.20) | 59 (16.8) | 1.21 (0.84–1.73) | 130 (13.2) | 1.00 (0.76–1.30) |
| Ptrend | 0.37 | 0.68 | 0.30 | ||||
EE: mean index of erythemal exposure, computed based upon place of residence (see text). Numbers might not add to total due to missing values.
Adjusted for age, county of residence, calendar year, number of full term pregnancies, and duration of hormonal contraceptives.
Tendency to burn based on first exposure to summer sun (1 hour or more).
Ability to tan upon prolonged exposure to summer sun.
The EE value for Seattle area, WA, US, is 1897.
Table 4 shows the association between behavioral sun exposure factors and risk of invasive ovarian cancer separately for two different age intervals. While we observed age-related differences in self-reported sun exposure (e.g., only 10.1% of cases and 9.5% of controls reported having less than 1 hour/day of daily summer sun exposure during the teen age years, while 45.5% of cases and 42.6% of controls reported that degree of exposure in the decade prior to the diagnosis/reference date; 15.6% of cases and 17.3% of controls reported not having a sun tan during the teen age years, versus 30.4% of cases and 28.8% of controls reported no tan in the decade prior to the diagnosis/reference date), and in the use of sun screen (83.6% of cases and 84.1% of controls reported not using sun screen during the teen age years, compared to 31.4% of cases and 29.0% of controls in the decade prior to the diagnosis/reference date), no associations were observed between these factors and risk of invasive ovarian cancer in either period. When we combined the weekly average of sun exposure in summer months with the use of sunscreen, we again did not observe any associations with the risk of ovarian cancer, although the numbers in the individual exposure categories were relatively small for this analysis (data not shown).
Table 4.
Behavioral sun exposure factors and risk of invasive epithelial ovarian cancer in two different age intervals, western Washington State, 2002–2009.
| Teen age years | Decade prior to diagnosis/reference date | |||||
|---|---|---|---|---|---|---|
| Controls (N=1,679) |
Cases (N=983) |
Controls (N=1,679) |
Cases (N=983) |
|||
| n (%) | n (%) | OR* (95% CI) | n (%) | n (%) | OR* (95% CI) | |
| Weekly average of daily sun exposure in summer months | ||||||
| < 1 hour/day | 159 ( 9.5) | 99 (10.1) | 1.00 (ref.) | 716 (42.6) | 446 (45.5) | 1.00 (ref.) |
| 1 to < 2 hours/day | 304 (18.1) | 164 (16.7) | 0.90 (0.65–1.25) | 504 (30.0) | 287 (29.3) | 0.92 (0.75–1.11) |
| 2 to < 3 hours/day | 273 (16.3) | 160 (16.3) | 0.93 (0.67–1.30) | 224 (13.3) | 124 (12.6) | 0.85 (0.65–1.10) |
| 3 to < 4 hours/day | 345 (20.6) | 222 (22.6) | 1.09 (0.79–1.49) | 145 ( 8.6) | 71 ( 7.2) | 0.88 (0.64–1.22) |
| 4 to < 5 hours/day | 224 (13.3) | 132 (13.5) | 0.97 (0.69–1.38) | 53 ( 3.2) | 28 ( 2.9) | 0.79 (0.48–1.30) |
| ≥5 hours/day | 373 (22.2) | 204 (20.8) | 0.88 (0.64–1.22) | 37 ( 2.2) | 25 ( 2.5) | 1.08 (0.63–1.87) |
| Ptrend | 0.80 | 0.32 | ||||
| Months with sun tan per year | ||||||
| Never had a sun tan | 290 (17.3) | 153 (15.6) | 1.00 (ref.) | 484 (28.8) | 298 (30.4) | 1.00 (ref.) |
| 1–3 months | 791 (47.1) | 479 (48.9) | 1.17 (0.92–1.48) | 889 (53.0) | 499 (50.9) | 0.95 (0.79–1.15) |
| 4–6 months | 426 (25.4) | 260 (26.5) | 1.12 (0.87–1.46) | 234 (13.9) | 136 (13.9) | 0.94 (0.72–1.23) |
| 7–9 months | 105 ( 6.3) | 46 ( 4.7) | 0.86 (0.57–1.30) | 40 ( 2.4) | 24 ( 2.4) | 0.96 (0.56–1.66) |
| 10–12 months | 67 ( 4.0) | 42 ( 4.3) | 1.15 (0.73–1.80) | 31 ( 1.8) | 23 ( 2.3) | 1.33 (0.74–2.40) |
| Ptrend | 0.95 | 0.87 | ||||
| Use of sun screen | ||||||
| Never or rarely | 1412 (84.1) | 821 (83.6) | 1.00 (ref.) | 487 (29.0) | 309 (31.4) | 1.00 (ref.) |
| Less than half the time | 120 ( 7.1) | 66 ( 6.7) | 0.97 (0.70–1.35) | 187 (11.1) | 111 (11.3) | 0.89 (0.67–1.19) |
| About half the time | 73 ( 4.3) | 53 ( 5.4) | 1.05 (0.72–1.55) | 234 (13.9) | 154 (15.7) | 1.08 (0.83–1.40) |
| More than half the time | 32 ( 1.9) | 17 ( 1.7) | 0.87 (0.47–1.62) | 247 (14.7) | 112 (11.4) | 0.71 (0.53–0.94) |
| Always or nearly always | 42 ( 2.5) | 25 ( 2.5) | 0.95 (0.56–1.61) | 523 (31.2) | 297 (30.2) | 0.91 (0.74–1.13) |
| Ptrend | 0.82 | 0.23 | ||||
Numbers might not add to total due to missing values.
Adjusted for age, county of residence, calendar year, number of full term pregnancies, and duration of hormonal contraceptives.
We observed little evidence that associations with pigmentation characteristics or EE levels varied markedly by histologic subtype of invasive disease (Table 5). Relative to women who reported no ability to tan after prolonged sun exposure, increased risks were noted in women who reported a greater ability to tan for serous cancers, while little evidence of similar risk increases was noted for the combined group of endometrioid/clear cell cancers or for “other” epithelial cancers. Although risk of serous cancers was weakly increased among women with increasing EE (Ptrend=0.06), the direction of this effect was opposite to what had been hypothesized, assuming increased EE indicates increased vitamin D synthesis from ultraviolet exposure. In additional stratified analyses, we noted no differences in associations of ovarian cancer risk with behavioral or residence-based measures of sun exposure when we restricted our analyses to groups of women who reported no or mild ability to tan versus women who tanned moderately or deeply (data not shown).
Table 5.
Pigmentation characteristics, tendency to burn, ability to tan, and erythemal exposure (EE)‡ and risk of invasive ovarian cancer by histological subtype, western Washington State, 2002–2009.
| Serous invasive tumors | Endometroid/clear cell invasive tumors |
Other epithelial (non- mucinous) invasive tumors |
|||||
|---|---|---|---|---|---|---|---|
| Controls (N=1,679) |
Cases (N=577) |
Cases (N=234) |
Cases (N=144) |
||||
| n (%) | n (%) | OR* (95% CI) | n (%) | OR* (95% CI) | n (%) | OR* (95% CI) | |
| Natural skin color | |||||||
| Light | 1,194 (71.1) | 404 (70.1) | 1.00 (ref.) | 182 (77.8) | 1.00 (ref.) | 99 (69.2) | 1.00 (ref.) |
| Medium | 480 (28.6) | 170 (29.5) | 1.12 (0.90–1.39) | 49 (20.9) | 0.80 (0.56–1.15) | 44 (30.8) | 1.18 (0.81–1.74) |
| Dark | 5 ( 0.3) | 2 ( 0.3) | 2.02 (0.38–10.88) | 3 ( 1.3) | 6.71 (1.40–32.15) | 0 ( 0.0) | -- |
| Ptrend | 0.26 | 0.61 | 0.45 | ||||
| Natural hair color (as a teenager) | |||||||
| Blonde or red | 524 (31.2) | 188 (32.6) | 1.00 (ref.) | 76 (32.5) | 1.00 (ref.) | 40 (28.0) | 1.00 (ref.) |
| Light brown | 348 (20.7) | 94 (16.3) | 0.73 (0.55–0.98) | 54 (23.1) | 1.00 (0.67–1.49) | 28 (19.6) | 1.03 (0.61–1.71) |
| Medium or dark brown | 774 (46.1) | 289 (50.1) | 0.99 (0.79–1.24) | 101 (43.2) | 0.84 (0.59–1.18) | 75 (52.4) | 1.25 (0.83–1.88) |
| Black | 33 ( 2.0) | 6 ( 1.0) | 0.47 (0.19–1.18) | 3 ( 1.3) | 0.80 (0.22–2.87) | 0 ( 0.0) | -- |
| Ptrend | 0.80 | 0.27 | 0.57 | ||||
| Natural eye color | |||||||
| Blue or Grey | 655 (39.0) | 236 (40.9) | 1.00 (ref.) | 103 (44.0) | 1.00 (ref.) | 44 (30.6) | 1.00 (ref.) |
| Green | 232 (13.8) | 79 (13.7) | 0.99 (0.72–1.34) | 24 (10.3) | 0.63 (0.38–1.04) | 31 (21.5) | 2.10 (1.28–3.44) |
| Hazel | 388 (23.1) | 122 (21.1) | 0.86 (0.66–1.11) | 52 (22.2) | 0.85 (0.58–1.24) | 29 (20.1) | 1.13 (0.69–1.84) |
| Brown | 404 (24.1) | 140 (24.3) | 1.00 (0.77–1.28) | 55 (23.5) | 0.81 (0.56–1.18) | 40 (27.8) | 1.42 (0.90–2.25) |
| Ptrend | 0.68 | 0.29 | 0.29 | ||||
| Tendency to burn§ | |||||||
| Painful burn with blistering or peeling | 651 (39.1) | 232 (40.4) | 1.00 (ref.) | 90 (38.6) | 1.00 (ref.) | 68 (47.6) | 1.00 (ref.) |
| Mild burn then tan | 809 (48.5) | 264 (46.0) | 0.89 (0.72–1.10) | 120 (51.5) | 0.88 (0.64–1.21) | 53 (37.1) | 0.61 (0.41–0.89) |
| Tan without burn | 207 (12.4) | 78 (13.6) | 1.00 (0.73–1.37) | 23 ( 9.9) | 0.80 (0.48–1.34) | 22 (15.4) | 0.99 (0.59–1.67) |
| Ptrend | 0.66 | 0.32 | 0.28 | ||||
| Ability to tan¶ | |||||||
| Freckle or no tan | 203 (12.1) | 63 (10.9) | 1.00 (ref.) | 30 (12.8) | 1.00 (ref.) | 19 (13.2) | 1.00 (ref.) |
| Mild tan | 389 (23.2) | 122 (21.2) | 1.05 (0.73–1.50) | 47 (20.1) | 0.66 (0.39–1.12) | 32 (22.2) | 0.86 (0.47–1.58) |
| Moderate tan | 773 (46.1) | 282 (49.0) | 1.26 (0.91–1.75) | 117 (50.0) | 0.90 (0.57–1.43) | 58 (40.3) | 0.76 (0.44–1.33) |
| Deep tan | 310 (18.5) | 109 (18.9) | 1.29 (0.89–1.87) | 40 (17.1) | 0.85 (0.50–1.46) | 35 (24.3) | 1.26 (0.69–2.30) |
| Ptrend | 0.07 | 0.84 | 0.50 | ||||
| Mean EE level from age 25 to diagnosis/reference date minus 1 year£ | |||||||
| <1897 | 415 (24.7) | 141 (24.4) | 0.86 (0.62–1.15) | 59 (25.2) | 0.91 (0.57–1.45) | 34 (23.6) | 0.63 (0.36–1.10) |
| 1897 | 499 (29.8) | 176 (30.5) | 1.00 (ref.) | 81 (34.6) | 1.00 (ref.) | 50 (34.7) | 1.00 (ref.) |
| >1897 and <2042 | 343 (20.5) | 105 (18.2) | 0.86 (0.62–1.17) | 34 (14.5) | 0.69 (0.44–1.09) | 22 (15.3) | 0.60 (0.35–1.02) |
| ≥2042 | 420 (25.0) | 155 (26.9) | 1.01 (0.77–1.33) | 60 (25.6) | 0.95 (0.64–1.41) | 38 (26.4) | 0.80 (0.50–1.29) |
| Ptrend | 0.47 | 0.78 | 0.90 | ||||
| Mean EE level in the 10 years prior to diagnosis/reference date minus 1 year£ | |||||||
| <1897 | 495 (29.5) | 166 (28.8) | 0.82 (0.58–1.16) | 63 (26.9) | 0.70 (0.40–1.21) | 52 (36.1) | 1.36 (0.77–2.40) |
| 1897 | 964 (57.4) | 328 (56.8) | 1.00 (ref.) | 145 (62.0) | 1.00 (ref.) | 75 (52.1) | 1.00 (ref.) |
| > 1897 | 220 (13.1) | 83 (14.4) | 1.16 (0.85–1.59) | 26 (11.1) | 0.69 (0.42–1.13) | 17 (11.8) | 1.01 (0.57–1.81) |
| Ptrend | 0.06 | 0.83 | 0.39 | ||||
EE: mean index of erythemal exposure, computed based upon place of residence (see text). Numbers might not add to total due to missing values.
Adjusted for age, county of residence, calendar year, number of full term pregnancies, and duration of hormonal contraceptives.
Tendency to burn based on first exposure to summer sun (1 hour or more).
Ability to tan upon prolonged exposure to summer sun.
The EE value for Seattle area, WA, US, is 1897.
Discussion
This study assessed the relationship between sun exposure and ovarian cancer, using sun exposure as a surrogate of the availability of endogenous vitamin D. For the majority of people, vitamin D is principally synthesized from sun exposure [25], and so sun-related behaviors as well as phenotypic characteristics that are associated with UV-light absorption likely correlate with endogenous levels of vitamin D [26]. Little evidence was found of an association of ovarian cancer risk with either the different measures of pigmentation or the residence-based and behavioral measures of sun exposure in our study.
However, our study may be subject to several limitations. Not all eligible cases or controls could be enrolled; our risk estimates could be influenced if those who chose not to participate differed from the study subjects in factors related to sun exposure. Errors in reporting pigmentation characteristics, residential history and other sun-related behaviors could also affect our results. While there is little reason to expect that cases and controls would differ in the accuracy of self-report of these exposures (given the limited extent to which they have been addressed in prior ovarian cancer studies), such errors in reporting could nevertheless result in nondifferential misclassification. Some evidence suggests that, although sun exposure behaviors may seem difficult to recall throughout the lifetime of a person, information obtained using standardized interview instruments can provide a valid measure of cumulative sun exposure [24]. Another potential source of bias could be a lack of sun exposure among cases in the months immediately preceding diagnosis due to ill health attributable to the presence of the cancer, which could lead to apparent risk increases associated with reduced sun exposure, particularly for the behavioral measures. We therefore attempted to diminish any influence of such bias by considering only exposure up to one year before the diagnosis/reference date when this was possible. Also, it is possible that unknown or unmeasured confounders could have biased our results.
A large proportion of both cases and controls, particularly in the decade preceding the diagnosis or reference date, had average EE levels equivalent to the EE for the Seattle area, WA, the population center for our study. While little evidence of differences in risk were noted for women who resided either in areas of greater or lesser EE, it is possible that associations with residence-based measures of UV exposure may be more clearly linked with risk in other regions or populations, particularly if study populations with a greater extent of variation in EE levels can be observed.
The present study differs in several aspects from previous analytic epidemiologic studies of ovarian cancer risk associated with measures of vitamin D. Most of the previous studies were based on circulating levels of vitamin D (see review [27]) based on a single measurement [8, 13]. Although circulating levels of 25(OH)D have been observed to be stable for about 5 years [28], vitamin D levels are nevertheless expected to vary over the lifetime, given that sun exposure behaviors are very different during the lifetime of a person, as observed in our Table 4. Since the relevant period of exposure influencing the risk of ovarian cancer is unknown, studies based on circulating levels of vitamin D are limited by looking at the exposure at a single point in time. Our study, in contrast, was able to look at behavioral measures of sun exposure over two different periods, and at residential history throughout adulthood. In particular, we were able to look at sun-related behaviors such as time spent in the summer sun early in life, which has been observed to be relevant in other cancers [29]. Studies based on the intake of vitamin D might suffer by not having an accurate assessment of circulating vitamin D levels (as well as differences in this exposure over the lifetime), given that for most people dietary vitamin D represents only a small source of circulating levels of 25(OH)D [5]. The current study further differs from previous ovarian cancer studies of sun exposure in the use of a sun exposure index computed through linking an individual’s residential history, assessed at the city level, with objective measures of UV irradiance measured at each 1° latitude and 1.25° longitude). This finer level of detail contrasts with a case-control study on ovarian cancer mortality in which state-level data on place of residence was more broadly categorized as low/medium/high annual mean daily solar radiation [1].
Despite the examination of several measures thought to be correlated with vitamin D synthesis through sun exposure, we found little evidence of association with ovarian cancer risk. Relative to women who reported that they did not tan with prolonged sun exposure, women who were more likely to tan were at increased ovarian cancer risk. Due to the number of tests performed, this association could be due to chance; if not, this increased risk may reflect a lesser extent of vitamin D synthesis among women in whom more melanin pigment is produced with sun exposure. Risk estimates for variables that examined behavioral measures of sun exposure were similarly null whether exposures were assessed in the teen years or in the recent past, and a lifetime (after age 25) summary variable of EE also failed to detect risk reductions with increased potential sun exposure based on place of residence. While some evidence of a trend in risk associated with EE was noted among women with serous invasive cancer, this association was in the direction opposite to what was hypothesized, and may be due to chance. Our results add to the lack of consistent evidence of the relationship between sun exposure and incidence of ovarian cancer.
Acknowledgements
The study was supported by grants R01 CA112523 and R01 CA87538 from the National Cancer Institute. CB was partially supported by grant number T32 CA009168 from the National Institutes of Health.
Footnotes
Conflict of interest: None.
References
- 1.Freedman DM, Dosemeci M, McGlynn K. Sunlight and mortality from breast, ovarian, colon, prostate, and non-melanoma skin cancer: a composite death certificate based case-control study. Occup Environ Med. 2002;59(4):257–262. doi: 10.1136/oem.59.4.257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Garland CF, Mohr SB, Gorham ED, Grant WB, Garland FC. Role of ultraviolet B irradiance and vitamin D in prevention of ovarian cancer. Am J Prev Med. 2006;31(6):512–514. doi: 10.1016/j.amepre.2006.08.018. [DOI] [PubMed] [Google Scholar]
- 3.Lefkowitz ES, Garland CF. Sunlight, vitamin D, ovarian cancer mortality rates in US women. Int J Epidemiol. 1994;23(6):1133–1136. doi: 10.1093/ije/23.6.1133. [DOI] [PubMed] [Google Scholar]
- 4.Giovannucci E. The epidemiology of vitamin D and cancer incidence and mortality: a review (United States) Cancer Causes Control. 2005;16(2):83–95. doi: 10.1007/s10552-004-1661-4. [DOI] [PubMed] [Google Scholar]
- 5.Holick MF. Vitamin D: importance in the prevention of cancers, type 1 diabetes, heart disease, and osteoporosis. Am J Clin Nutr. 2004;79(3):362–371. doi: 10.1093/ajcn/79.3.362. [DOI] [PubMed] [Google Scholar]
- 6.Ahonen MH, Zhuang YH, Aine R, Ylikomi T, Tuohimaa P. Androgen receptor and vitamin D receptor in human ovarian cancer: growth stimulation and inhibition by ligands. Int J Cancer. 2000;86(1):40–46. doi: 10.1002/(sici)1097-0215(20000401)86:1<40::aid-ijc6>3.0.co;2-e. [DOI] [PubMed] [Google Scholar]
- 7.Zhang X, Jiang F, Li P, et al. Growth suppression of ovarian cancer xenografts in nude mice by vitamin D analogue EB1089. Clin Cancer Res. 2005;11(1):323–328. [PubMed] [Google Scholar]
- 8.Bakhru A, Mallinger JB, Buckanovich RJ, Griggs JJ. Casting light on 25-hydroxyvitamin D deficiency in ovarian cancer: a study from the NHANES. Gynecol Oncol. 2010;119(2):314–318. doi: 10.1016/j.ygyno.2010.07.006. [DOI] [PubMed] [Google Scholar]
- 9.Clendenen TV, Arslan AA, Koenig KL, et al. Vitamin D receptor polymorphisms and risk of epithelial ovarian cancer. Cancer Lett. 2008;260(1–2):209–215. doi: 10.1016/j.canlet.2007.11.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lurie G, Wilkens LR, Thompson PJ, et al. Vitamin D receptor gene polymorphisms and epithelial ovarian cancer risk. Cancer Epidemiol Biomarkers Prev. 2007;16(12):2566–2571. doi: 10.1158/1055-9965.EPI-07-0753. [DOI] [PubMed] [Google Scholar]
- 11.Tworoger SS, Gate MA, Lee I-M, et al. Polymorphisms in the Vitamin D Receptor and Risk of Ovarian Cancer in Four Studies. Cancer Res. 2009;69(5):1885–1891. doi: 10.1158/0008-5472.CAN-08-3515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Tworoger SS, Lee I-M, Buring JE, Rosner B, Hollis BW, Hankinson SE. Plasma 25-Hydroxyvitamin D and 1,25-Dihydroxyvitamin D and Risk of Incident Ovarian Cancer. Cancer Epidemiol Biomarkers Prev. 2007;16(4):783–788. doi: 10.1158/1055-9965.EPI-06-0981. [DOI] [PubMed] [Google Scholar]
- 13.Zheng W, Danforth KN, Tworoger SS, et al. Circulating 25-Hydroxyvitamin D and Risk of Epithelial Ovarian Cancer: Cohort Consortium Vitamin D Pooling Project of Rarer Cancers. Am J Epidemiol. 2010;172(1):70–80. doi: 10.1093/aje/kwq118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.English DR, Armstrong BK, Kricker A, Winter MG, Heenan PJ, Randell PL. Case-control study of sun exposure and squamous cell carcinoma of the skin. Int J Cancer. 1998;77(3):347–353. doi: 10.1002/(sici)1097-0215(19980729)77:3<347::aid-ijc7>3.0.co;2-o. [DOI] [PubMed] [Google Scholar]
- 15.John EM, Koo J, Schwartz GG. Sun exposure and prostate cancer risk: evidence for a protective effect of early-life exposure. Cancer Epidemiol Biomarkers Prev. 2007;16(6):1283–1286. doi: 10.1158/1055-9965.EPI-06-1053. [DOI] [PubMed] [Google Scholar]
- 16.Knight JA, Lesosky M, Barnett H, Raboud JM, Vieth R. Vitamin D and reduced risk of breast cancer: a population-based case-control study. Cancer Epidemiol Biomarkers Prev. 2007;16(3):422–429. doi: 10.1158/1055-9965.EPI-06-0865. [DOI] [PubMed] [Google Scholar]
- 17.Hankey BF, Ries LA, Edwards BK. The surveillance, epidemiology, and end results program: a national resource. Cancer Epidemiol Biomarkers Prev. 1999;8(12):1117–1121. [PubMed] [Google Scholar]
- 18.Fritz AG. International classification of diseases for oncology: ICD-O. World Health Organization; 2000. [Google Scholar]
- 19.World Health Orgnaization Classification of Tumours. Pathology and Genetcis: Tumours of the Breast and Female Genital Organs. Lyon: IARC Press; 2003. [Google Scholar]
- 20.Waksberg J. Random digit dialing sampling for case-control studies. In: Gail MH, Benichou J, editors. Encyclopedia of epidemiologic methods. New York: John Wiley & Sons; 2000. pp. 749–753. [Google Scholar]
- 21.Waksberg J. Sampling methods for random digit dialing. Journal of the American Statistical Association. 1978;73(371):40–46. [Google Scholar]
- 22.Casady RJ, Lepkowski JM. Stratified telephone survey designs. Survey Methodology. 1993;19(1):103–113. [Google Scholar]
- 23.Solomon CC, White E, Kristal AR, Vaughan T. Melanoma and lifetime UV radiation. Cancer Causes Control. 2004;15(9):893–902. doi: 10.1007/s10552-004-1142-9. [DOI] [PubMed] [Google Scholar]
- 24.Karagas MR, Zens MS, Nelson HH, et al. Measures of cumulative exposure from a standardized sun exposure history questionnaire: a comparison with histologic assessment of solar skin damage. Am J Epidemiol. 2007;165(6):719–726. doi: 10.1093/aje/kwk055. [DOI] [PubMed] [Google Scholar]
- 25.Holick MF. Vitamin D and Sunlight: Strategies for Cancer Prevention and Other Health Benefits. Clinical Journal of the American Society of Nephrology. 2008;3(5):1548–1554. doi: 10.2215/CJN.01350308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Norman AW. Sunlight, season, skin pigmentation, vitamin D, 25-hydroxyvitamin D: integral components of the vitamin D endocrine system. The American Journal of Clinical Nutrition. 1998;67(6):1108–1110. doi: 10.1093/ajcn/67.6.1108. [DOI] [PubMed] [Google Scholar]
- 27.Cook LS, Neilson HK, Lorenzetti DL, Lee RC. A systematic literature review of vitamin D and ovarian cancer. Am J Obstet Gynecol. 2010;203(1):70.e1–70.e8. doi: 10.1016/j.ajog.2010.01.062. [DOI] [PubMed] [Google Scholar]
- 28.Hofmann JN, Yu K, Horst RL, Hayes RB, Purdue MP. Long-term Variation in Serum 25-Hydroxyvitamin D Concentration among Participants in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Cancer Epidemiol Biomarkers Prev. 2010;19(4):927–931. doi: 10.1158/1055-9965.EPI-09-1121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Veierod MB, Weiderpass E, Thorn M, et al. A prospective study of pigmentation, sun exposure, and risk of cutaneous malignant melanoma in women. J Natl Cancer Inst. 2003;95(20):1530–1538. doi: 10.1093/jnci/djg075. [DOI] [PubMed] [Google Scholar]