To the Editor
Whereas smoking is associated with increased risk of most cancers, observational studies have suggested a lower risk with malignant melanoma (MM) and are inconsistent for non-melanoma skin cancer (NMSC).1–4 Using a large, prospective cohort of the Women’s Health Initiative observational study, we evaluated the relationships of smoking to MM and NMSC. Incidence of MM and NMSC in current smokers was compared with incidence in never smokers and past smokers.
A total of 514 MM cases and 9046 NMSC cases were identified during a mean follow-up period of 11 years (Table I; available at http://www.jaad.org). Approximately 51% of women were never smokers, 43.5% reported being past smokers, and 5.6% were current smokers. Compared with never smokers, current smokers had lower odds of developing MM (hazard ratio 0.55, 95% confidence interval [CI] 0.32–0.94, Pglobal = .053) and NMSC (odds ratio 0.86, 95% CI 0.77–0.96, Pglobal = .033) in fully adjusted models; past smokers, however, did not differ with respect to MM (hazard ratio 1.06, 95% CI 0.88–1.27) or NMSC (odds ratio 0.99, 95% CI 0.95–1.04) risk (Table II).
Table I.
Baseline characteristics and outcomes by smoking status, no. (%)
| Variable | Smoking status
|
Total | χ2 P value | ||
|---|---|---|---|---|---|
| Never | Past | Current | |||
| Total | 28,664 (51) | 24,453 (43) | 3144 (5.6) | 56,261 | |
| Age group | <.0001 | ||||
| <50–59 y | 8763 (30) | 7899 (32) | 1290 (41) | 17,952 | |
| 60–69 y | 12,661 (44) | 11,272 (46) | 1384 (44) | 25,317 | |
| 70–≥79 | 7240 (25) | 5282 (22) | 470 (15) | 12,992 | |
| BMI category | <.0001 | ||||
| <25 | 12,666 (44.19) | 10,383 (42.46) | 1603 (50.99) | 24,652 | |
| 25–<30 | 9585 (33.44) | 8400 (34.35) | 987 (31.39) | 18,972 | |
| ≥30 | 6413 (22.37) | 5670 (23.19) | 554 (17.62) | 12,637 | |
| Ambient sun exposure, Langley | <.0001 | ||||
| 300–325 | 9343 (32.59) | 7991 (32.68) | 930 (29.58) | 18,264 | |
| 350 | 5989 (20.89) | 5246 (21.45) | 679 (21.60) | 11,914 | |
| 375–380 | 3198 (11.16) | 2501 (10.23) | 343 (10.91) | 6042 | |
| 400–430 | 4869 (16.99) | 3740 (15.29) | 466 (14.82) | 9075 | |
| 475–500 | 5265 (18.37) | 4975 (20.35) | 726 (23.09) | 10,966 | |
| Skin reaction to the sun | <.0001 | ||||
| Does not burn | 10,633 (37.10) | 8053 (32.93) | 1224 (38.93) | 19,910 | |
| Burns and tans | 6718 (23.44) | 6883 (28.15) | 789 (25.10) | 14,390 | |
| Burns and tans minimally | 7770 (27.11) | 6864 (28.07) | 789 (25.10) | 15,423 | |
| Burns and does not tan | 3543 (12.36) | 2653 (10.85) | 342 (10.88) | 6538 | |
| Education | <.0001 | ||||
| High school | 5846 (20.39) | 4101 (16.77) | 693 (22.04) | 10,640 | |
| Some college | 9934 (34.66) | 9095 (37.19) | 1329 (42.27) | 20,358 | |
| College | 12,884 (44.95) | 11,257 (46.04) | 1122 (35.69) | 25,263 | |
| History of NMSC | .49 | ||||
| No | 25,939 (90.49) | 22,100 (90.38) | 2862 (91.03) | 50,901 | |
| Yes | 2725 (9.51) | 2353 (9.62) | 282 (8.97) | 5360 | |
| History of melanoma | .61 | ||||
| No | 28,248 (98.55) | 24,114 (98.61) | 3094 (98.41) | 55,456 | |
| Yes | 416 (1.45) | 339 (1.39) | 50 (1.59) | 805 | |
| Childhood exposure to summer sun | <.0001 | ||||
| <30 min | 706 (2.46) | 480 (1.96) | 60 (1.91) | 1246 | |
| 30 min to 2 h | 7734 (26.98) | 6065 (24.80) | 704 (22.39) | 14,503 | |
| >2 h | 20,224 (70.56) | 17,908 (73.23) | 2380 (75.70) | 40,512 | |
| Current exposure to summer sun | <.0001 | ||||
| <30 min | 8669 (30.24) | 7541 (30.84) | 1015 (32.28) | 17,225 | |
| 30 min to 2 h | 14,658 (51.14) | 12,164 (49.74) | 1433 (45.58) | 28,255 | |
| >2 h | 5337 (18.62) | 4748 (19.42) | 696 (22.14) | 10,781 | |
| Usually use sunscreen outside | <.0001 | ||||
| No | 13,599 (47.44) | 10,641 (43.52) | 1773 (56.39) | 26,013 | |
| Yes | 15,065 (52.56) | 13,812 (56.48) | 1371 (43.61) | 30,248 | |
| Vitamin D from supplements, diet | <.0001 | ||||
| <400 IU | 14,249 (49.71) | 12,090 (49.44) | 1852 (58.91) | 28,191 | |
| ≥400 IU | 14,415 (50.29) | 12,363 (50.56) | 1292 (41.09) | 28,070 | |
| Passive smoking exposure (any) | <.0001 | ||||
| None | 2496 (8.71) | 506 (2.07) | 38 (1.21) | 3040 | |
| Exposure as child, adult at home, adult at work | 26,168 (91.29) | 23,947 (97.93) | 3106 (98.79) | 53,221 | |
| Passive smoking exposure category | <.0001 | ||||
| None | 2496 (8.71) | 506 (2.07) | 38 (1.21) | 3040 | |
| Child only | 1754 (6.12) | 403 (1.65) | 37 (1.18) | 2194 | |
| Adult home only | 1768 (6.17) | 1227 (5.02) | 87 (2.77) | 3082 | |
| Adult work only | 3440 (12.00) | 1228 (5.02) | 108 (3.44) | 4776 | |
| >1 | 19,206 (67.00) | 21,089 (86.24) | 2874 (91.41) | 43,169 | |
| Pack years smoked | <.0001 | ||||
| ≤20 | 28,664 (100.00) | 15,371 (62.86) | 1141 (36.29) | 45,176 | |
| >20 | 0 (0.00) | 9082 (37.14) | 2003 (63.71) | 11,085 | |
| Alcohol consumption | <.0001 | ||||
| Nondrinker | 4333 (15.12) | 498 (2.04) | 82 (2.61) | 4913 | |
| Past drinker | 4527 (15.79) | 4120 (16.85) | 534 (16.98) | 9181 | |
| <1 drink/mo | 3700 (12.91) | 2323 (9.50) | 434 (13.80) | 6457 | |
| <1 drink/wk | 6283 (21.92) | 4746 (19.41) | 573 (18.23) | 11,602 | |
| 1–<7 drinks/wk | 7256 (25.31) | 7968 (32.58) | 809 (25.73) | 16,033 | |
| ≥7 drinks/wk | 2565 (8.95) | 4798 (19.62) | 712 (22.65) | 8075 | |
| Annual family income | <.0001 | ||||
| <$20,000 | 3662 (12.78) | 2723 (11.14) | 589 (18.73) | 6974 | |
| $20,000–$35,000 | 6862 (23.94) | 5315 (21.74) | 804 (25.57) | 12,981 | |
| $35,000–$50,000 | 6004 (20.95) | 5071 (20.74) | 651 (20.71) | 11,726 | |
| $50,000–$75,000 | 5962 (20.80) | 5402 (22.09) | 606 (19.27) | 11,970 | |
| ≥$75,000 | 6174 (21.54) | 5942 (24.30) | 494 (15.71) | 12,610 | |
| Any insurance | <.0001 | ||||
| No | 609 (2.12) | 534 (2.18) | 200 (6.36) | 1343 | |
| Yes | 28,055 (97.88) | 23,919 (97.82) | 2944 (93.64) | 54,918 | |
| Last medical visit within 1 y | <.0001 | ||||
| No | 4450 (15.52) | 3617 (14.79) | 664 (21.12) | 8731 | |
| Yes | 24,214 (84.48) | 20,836 (85.21) | 2480 (78.88) | 47,530 | |
| Physical activity (metabolic equivalents min/wk) | <.0001 | ||||
| 0–100 | 4969 (17.34) | 3839 (15.70) | 876 (27.86) | 9684 | |
| 100–500 | 7595 (26.50) | 5785 (23.66) | 975 (31.01) | 14,355 | |
| 500–1200 | 8922 (31.13) | 7861 (32.15) | 785 (24.97) | 17,568 | |
| ≥1200 | 7178 (25.04) | 6968 (28.50) | 508 (16.16) | 14,654 | |
| Melanoma during follow-up | .002 | ||||
| No | 28,417 (99.14) | 24,200 (98.97) | 3130 (99.55) | 55,747 | |
| Yes | 247 (0.86) | 253 (1.03) | 14 (0.45) | 514 | |
| NMSC during follow-up | <.0001 | ||||
| No | 24,085 (84.03) | 20,405 (83.45) | 2725 (86.67) | 47,215 | |
| Yes | 4579 (15.97) | 404816.55 | 419 (13.33) | 9046 | |
| Death during follow-up | <.0001 | ||||
| No | 26,754 (93.34) | 22,339 (91.35) | 2686 (85.43) | 51,779 | |
| Yes | 1910 (6.66) | 2114 (8.65) | 458 (14.57) | 4482 | |
BMI, Body mass index; NMSC, nonmelanoma skin cancer.
Table II.
Association of current smoking on melanoma or nonmelanoma skin cancer risk in age-adjusted and fully adjusted models
| Age-adjusted (CI) | Fully adjusted (CI) | |
|---|---|---|
| Melanoma (HR) | Pglobal = .004 | Pglobal = .053 |
| Current vs never smoker | 0.53 (0.31, 0.91) | 0.55 (0.32, 0.94) |
| Past vs never smoker | 1.20 (1.00, 1.42) | 1.06 (0.88, 1.27) |
| NMSC (OR) | Pglobal = .0002 | Pglobal = .033 |
| Current vs never smoker | 0.85 (0.76, 0.95) | 0.86 (0.77, 0.96) |
| Past vs never smoker | 1.05 (1.01, 1.1) | 0.99 (0.95, 1.04) |
Adjusted model is adjusted for age group, body mass index category, ethnicity/race, skin reaction to the sun, Langley exposure, education, physical activity, history of NMSC, history of malignant melanoma, summer sun exposure as a child, summer sun exposure currently, vitamin D, alcohol use, last medical visit within 1 year, any insurance, and income.
For melanoma, Cox proportional HR was used as the exact date of diagnosis was ascertained from medical records. For NMSC, subjects self-reported the diagnosis and thus we used logistic regression (OR).
CI, Confidence intervals; HR, hazard ratio; NMSC, nonmelanoma skin cancer; OR, odds ratio.
When stratified by duration and quantity of smoking, according to pack years for current and past smokers (N = 27,597), there was no significant trend overall between increasing pack years and risk of incident NMSC or MM. However, each 10-year increase in pack years smoked was associated with a small, but significant, 2% decrease in risk on NMSC (P =.010).
In this large cohort of older Caucasian women, we observed lower odds of both melanoma and NMSC among current smokers compared with lifelong nonsmokers. These results corroborate previous research suggesting an inverse relationship3 between cigarette smoking and melanoma while differing from null findings of past studies.4
An inverse relationship of smoking to skin cancer seems counterintuitive, given smoking’s positive association with many other cancers5; however, this relationship persisted despite controlling for age, education, alcohol use, sun-exposure variables, and other potential confounders. One explanation for the lower odds of MM and NMSC in current smokers could be that ‘‘cancer-prone’’ individuals may succumb to other cancers at an earlier age than required for developing MM and NMSC and this fact may mask or skew the true rate of skin cancers in smokers (if all would have survived). We hypothesize that the observed lower skin cancer risk in smokers may be caused by detection bias related to lower screening for skin cancer in smokers. In fact, current smokers in the Women’s Health Initiative are less likely to have medical insurance and are less likely to see physicians for cancer screening. Fewer current smokers ever had a mammogram (78% smokers vs 93% never smokers, respectively) or colonoscopy (46% of current smokers vs 56% of never smokers) (P value < .0001), similar to past studies on colorectal screening rates.5 Current smokers had lower rates of screening compared with never and past smokers despite having a current care provider and having insurance. In conclusion, smoking does not increase risk of skin cancer. Current smokers are associated with a lower risk of MM and NMSC possibly as a result of detection bias related to lower screening for skin cancer in smokers.
Acknowledgments
Funding sources: None.
Footnotes
Conflicts of interest: None declared.
References
- 1.Hertog SD, Wensveen C, Bastiaens M, et al. Relation between smoking and skin cancer. J Clin Oncol. 2001;19:231–238. doi: 10.1200/JCO.2001.19.1.231. [DOI] [PubMed] [Google Scholar]
- 2.Song F, Qureshi AA, Gao X, Li T, Han J. Smoking and risk of skin cancer: a prospective analysis and a meta-analysis. Int J Epidemiol. 2012;41:1694–1705. doi: 10.1093/ije/dys146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.DeLancey JO, Hannan LM, Gapstur SM, Thun MJ. Cigarette smoking and the risk of incident and fatal melanoma in a large prospective cohort study. Cancer Causes Control. 2011;22:937–942. doi: 10.1007/s10552-011-9766-z. [DOI] [PubMed] [Google Scholar]
- 4.Westerdahl J, Olsson H, Masback A, Ingvar C, Jonsson N. Risk of malignant melanoma in relation to drug intake, alcohol, smoking and hormonal factors. Br J Cancer. 1996;73:1126–1131. doi: 10.1038/bjc.1996.216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Meissner HI, Breen N, Klabunde CN, Vernon SW. Patterns of colorectal cancer screening uptake among men and women in the United States. Cancer Epidemiol Biomarkers Prev. 2006;15:389–394. doi: 10.1158/1055-9965.EPI-05-0678. [DOI] [PubMed] [Google Scholar]