Cigarette smoking is an important cause of preventable death globally, but associations between smoking and mortality vary substantially across country and calendar time. Although methodological biases have been discussed, it is biologically plausible that persistent organic pollutants (POPs) like polychlorinated biphenyls (PCBs) and organochlorine (OC) pesticides can affect this association. This study was performed to evaluate if associations of cigarette smoking with mortality were modified by serum concentrations of PCBs and OC pesticides. We evaluated cigarette smoking in 111 total deaths among 986 men and women aged 70 years in the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) with mean follow-up for 7.7 years. The association between cigarette smoking and total mortality depended on serum concentration of PCBs and OC pesticides (P value for interaction = 0.02). Among participants in the highest tertile of the serum POPs summary score, former and current smokers had 3.7 (95% CI, 1.5–9.3) and 6.4 (95% CI, 2.3–17.7) times higher mortality hazard, respectively, than never smokers. In contrast, the association between cigarette smoking and total mortality among participants in the lowest tertile of the serum POPs summary score was much weaker and statistically non-significant. The strong smoking-mortality association observed among elderly people with high POPs was mainly driven by low risk of mortality among never smokers with high POPs. As smoking is increasing in many low-income and middle-income countries and POPs contamination is a continuing problem in these areas, the interactions between these two important health-related issues should be considered in future research.
Citation: Lee D-H, Lind L, Jacobs DR Jr, Salihovic S, van Bavel B, Lind PM (2014) Does Mortality Risk of Cigarette Smoking Depend on Serum Concentrations of Persistent Organic Pollutants? Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) Study. PLoS ONE 9(5): e95937. doi:10.1371/journal.pone.0095937
Editor: Keitaro Matsuo, Kyushu University Faculty of Medical Science, Japan
Received: December 23, 2013; Accepted: April 2, 2014; Published: May 14, 2014
Copyright: © 2014 Lee et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was supported by the Swedish Research Council (VR) and the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS), Korea Health technology R&D Project, Ministry of Health & Welfare, Republic of Korea (A111716), and BK21 PLUS KNU Biomedical Convergence Program for Creative Talent. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Cigarette smoking, as a major risk factor for cancer, cardiovascular diseases, and pulmonary diseases, is an important cause of preventable death globally . However, the magnitude of associations between cigarette smoking and mortality varies across countries and calendar time , , , . To date, this variation has been attributed mainly to methodological biases such as misclassification of smoking status, cohort effects, or random variation , , , .
In our previous study , we formulated a novel hypothesis that persistent organic pollutants (POPs) modify the risk of cigarette smoking on death. POPs are lipophilic chemicals that accumulate in adipose tissue and are associated with the risk of various chronic diseases , , . The biological plausibility for our hypothesis was that experimental studies in mice reported that pretreatment with some POPs increased toxicity of important chemicals contained in cigarette smoke, like benzopyrene, dimethylnitrosamine, and N-nitrosodiethlyamine , .
Supporting our prior hypothesis, we observed different associations between cigarette smoking and total mortality depending on serum concentrations of POPs among the elderly in the U.S. . In that study, one surprising finding was that cigarette smoking did not increase the risk of mortality in the lowest category of POPs. Despite the possibility that smoking-related diseases risks are lower among the elderly than in younger persons due to selective survival , finding any subgroup with no association between cigarette smoking and mortality was unexpected.
Our earlier study is the only one published on this topic and the number of current smokers was too small . Thus, we studied here whether the finding was replicable in another dataset. We evaluated if there are similar interactions of serum concentrations of POPs with cigarette smoking on the risk of mortality among men and women aged 70 years, living in the community of Uppsala, Sweden (Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study). Because our previous study found that, among various POPs, polychlorinated biphenyls (PCBs) or organochlorine (OC) pesticides showed clear interactions with cigarette smoking and mortality , we used POPs burden calculated based on serum concentrations of both OC pesticides and PCBs in this study.
Materials and Methods
Study subjects at baseline were 1,016 men and women, residents of Uppsala, Sweden and age 70 years at time of examination between April 2001 and June 2004. Among participants at baseline, 986 subjects had valid measurement of PCBs and OC pesticides at baseline. The study was approved by the Ethics Committee of the University of Uppsala. The participants gave written informed consent.
All subjects were investigated in the morning after an overnight fast, with no medication or smoking allowed after midnight. The participants were asked about their health behaviors, medical history, and regular medication. Body mass index (BMI) was derived from measured height and weight (kg/m2). Serum cholesterol and triglyceride concentrations were determined in an enzymatic assay (Abbott, Abbott Park, IL, USA).
POPs were measured in stored plasma samples collected at baseline. Analyses of POPs were performed using a Micromass Autospec Ultima (Waters, Mildford, MA, USA) high resolution chromatography coupled to high resolution mass spectrometry (HRGC/HRMS) system based on the method by Sandau et al  with some modifications. All details of POPs analyses were provided elsewhere . Among 16 PCB congeners and 5 OC pesticides, 2 OC pesticides (trans-chlordane and cis-chlordane) with detection rate <10% were not included in the final analyses. An established summation formula based on serum cholesterol and serum triglyceride concentrations was used to calculate the total amount of lipid in each plasma sample . Thereafter the wet-weight concentrations of the POPs were divided by this lipid estimate to obtain lipid-normalized concentrations. As models based on wet-weight concentrations adjusted for serum cholesterol and triglyceride as covariates showed similar results, we presented the results based on lipid-normalized concentrations for the consistency of analytic strategy with the previous study .
The fact of death was ascertained through linkage to the Swedish Register of Death Causes at the National Board of Health and Welfare. Causes of death data were not available. Follow-up time for each person was calculated as the difference between the first examination date and the last known date alive or censored. Persons who survived the entire follow-up period were censored on Jan 1, 2012. Median follow-up time was 7.7 years (range 0.3–9.8 years) and we documented 111deaths.
For the analysis, smoking status was expressed as never, former and current. First, we calculated the summary measure of 16 PCB congeners and 5 OC pesticides by summing the rank orders of the individual POPs for subjects with detectable values of each POP, assigning rank 0 to not detectable values. Three summary measures were made based on both PCBs and OC pesticides, PCBs only, and OC pesticides only.
Next, we checked if the associations between cigarette smoking and mortality differed by tertiles of the summary measures of POPs in predicting total mortality using Cox proportional hazard models. P values for interaction were calculated based on three categories of each of POPs and cigarette smoking. Adjusting covariates were gender, physical activity (none, moderate, and vigorous), BMI (kg/m2), and alcohol consumption (g/day). We further considered medication for diabetes or hypertension and history of myocardial infarction or stroke as possible confounders.
In addition to stratified analyses, adjusted HRs with the common reference group of never smokers within the 1st tertile of POPs were presented. Also, the same analyses were applied to the U.S. elderly within our previous study  because only results stratified analyses by POPs levels were previously reported in that study. Methodologic details have been presented . All statistical analyses were performed with PC-SAS version 9.1.
Baseline characteristics and history of chronic diseases according to cigarette smoking status were shown in Table 1. Compared to never or current smokers, former smokers tended to be men, more obese, alcohol drinkers, and more under diabetes medication. Current smokers were less obese and more physically inactive than never or former smokers. There were no statistical significant differences across smoking categories for prevalence of hypertension medication or history of myocardial infarction or stroke.
Table 2 indicates mean concentrations of individual POPs depending on cigarette smoking status. Among 16 PCBs, PCB074, PCB105, and PCB118 were statistically significant or marginally significantly lower in current vs. never smokers. These three PCBs have weak dioxin activity, however, other PCBs with dioxin activity like PCB126, PCB156, and PCB169 did not show any trend. On the contrary, among 3 OC pesticides, p,p′-DDE had higher levels among current than never smokers.
When POPs were not considered in analyses, adjusted hazard ratios (HRs) for all-cause mortality were 1.3 for former smokers and 2.1 (95% CI: 1.2–3.7) for current smokers, compared with never smokers. However, the associations were substantially different depending on summary measures of POPs (Table 3). Compared with 70 year old people within the 1st or 2nd tertiles of summary measures of PCBs and OC pesticides, those within the 3rd tertile showed strong associations between cigarette smoking and total mortality. Adjusted HRs were 3.7 (1.5–9.3) for former smokers and 6.4 (2.3–17.7) for current smokers. Further adjustment for medication for diabetes or hypertension and history of myocardial infarction or stroke did not change the result. Summary measures calculated from PCBs or OC pesticides separately showed similar associations (Table S1).
Figure 1 shows adjusted HRs when never smokers within the 1st tertile of PCBs and OC pesticides were used as the common reference group. Never smokers within the 3rd tertile of PCBs and OC pesticides showed a statistically significantly lower risk of mortality with adjusted HR of 0.3 (0.1–0.9) while current smokers with the same levels of POPs concentrations had a statistically significantly higher risk of mortality with adjusted HR of 2.3 (1.0–5.0). When we applied the same analyses to the U.S. elderly who were included in our previous study, patterns were similar to those in the current study, although no individual HR reached statistical significance (Figure S1). In that study, the summary measures for PCBs and for OC pesticides were separate because PCB vs OC pesticide measurements were performed in different participants.
Hazard ratios (HRs) were estimated using a common reference group (never smokers & T1), adjusted for gender, BMI, exercise, and alcohol consumption. T1, first tertile; T2, second tertile; T3, third tertile, Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study.
To better understand possible confounding, Table 4 compared baseline characteristics of 4 subgroups (never smokers & low POPs, current smokers & high POPs, never smokers & low POPs, and current smokers & high POPs), in particular focusing on never smokers with high POPs who had the lowest mortality. Compared to other subgroups, 70 year old participants within the never smokers & high POPs group were the most physically active and had the lowest prevalence of history of myocardial infarction.
Generally, the present study replicated our report in a U.S. elderly population . The association between cigarette smoking and total mortality depended on serum concentration of summary scores reflecting the background mixture of PCBs and/or OC pesticides. When POPs were not considered in the analyses, the risk of mortality among current-smokers was about two times higher than never smokers. However, among elderly with relatively high POPs, former or current smokers had about 4 to 7 times higher mortality than never smokers while the association between cigarette smoking and total mortality much weaker and statistically non-significant among elderly with relatively low serum concentrations of POPs. In addition to the main results, a subsidiary finding of lower concentrations of dioxin-like PCBs among smokers than never smokers was also similar to finding in the previous study .
These findings suggested that different concentrations of POPs among populations may partly explain variability in smoking-related total mortality association across previous epidemiological studies , , , . More importantly, the similar weak association between cigarette smoking and mortality among the elderly people in the lowest category of POPs from these two studies could indicate that the presence of certain levels POPs is a necessary factor for cigarette smoking to increase the risk of death in the elderly.
At first, we had expected the interaction between these two factors to be driven by the high risk of death among current smokers with high POPs. However, these two studies showed that the strong association between cigarette smoking and mortality among the elderly with high POPs had a strong component of low mortality in never smokers as well as of high mortality in former or current smokers.
The very low mortality among the 150 never smokers in PIVUS with high POPs is a provocative finding. A similar pattern of reduced mortality was apparent in the National Health and Nutrition Examination Survey (NHANES) (Figure S1), suggesting that this finding is not simply bias or chance. An ecologic finding in the cohort study of male British doctors may be viewed as concordant. In that study, the excess mortality associated with smoking was greater during the second half of follow-up (1971∼1991) than the first half (1951∼1971) , largely because the mortality rate among non-smokers had decreased substantially over time while the mortality rate among smokers had remained about constant . Historically, PCBs and OC pesticides were widely used after World War II and periods with the highest body burden of POPs were 1960s and 1970s . Therefore, any effect due to high POPs may be more strongly reflected in the latter part of cohort study.
It is difficult to explain very low mortality among never smokers with high POPs. One possible explanation is survival bias. However, for survival bias to explain this finding, never smokers with high POPs must have had a higher death rate before reaching age 70 than other subgroups like active smokers with high POP, leaving healthier survivors to participate in the PIVUS and NHANES studies. This would be an odd pattern. Also, compared to other subgroups, the elderly never smokers with high POPs were the most physically active and had the lowest prevalence of history of myocardial infarction, both of which might predispose to reduced mortality rate. Therefore, this type of survival bias would seem to be unlikely even though we cannot totally exclude it.
In addition, low mortality in relation to high POPs is biologically plausible under some conditions, a case in point relating to associations between POPs and telomeres. Telomere length has been proposed as a marker of mitotic cell age and as a general index of human aging . In our previous cross-sectional study among apparently healthy Koreans , telomere length was increasing across serum concentrations of POPs within the lower range of POPs. In that study, the interpretation was that low dose POPs may act as a tumor promoter in carcinogenesis based on experimental findings on arsenic. For example, low concentrations of arsenic relevant to current human exposure elongated telomeres in vitro and increased myc and ras oncogenes while high concentrations of arsenic decreased telomere length . As myc oncogene can activate telomerase , the authors interpreted these experimental findings on arsenic as a role of tumor promoter of low dose arsenic in human.
However, an opposite interpretation of how POPs relate to telomere length is also possible. It is well-known that shorter telomere length is associated with higher risk of early death . There is a higher mortality rate, especially from heart disease and infectious disease, among elderly people who have shorter telomeres in blood DNA . Therefore, in PIVUS and NHANES persons with relatively high serum POPs concentrations within background exposure levels may have had a longer survival than persons with lower serum POPs concentrations. As cigarette smoking is reported to decrease telomere length , longer survival with higher POPs could be observed only in never smokers, as we observed in these two studies.
Although molecular mechanisms for low dose POPs to increase telomere length have not been studied, one speculation is that certain levels of POPs might excite production of cytoprotective and restorative proteins including growth factors, phase II and antioxidant enzymes, and protein chaperones, known as hormetic effects . Increased telomerase activity or slow-down of age-dependent telomere shortening has been regarded as one marker of activation of cytoprotective and restorative proteins . Some previous human and experimental studies support possible beneficial effects of certain levels of POPs. For example, decreased risk of soft tissue sarcoma was reported with increased concentrations of dioxins or PCBs with dioxin activity in the general population. Also, in some animal studies of low doses of TCDD or DDT, there was a tendency towards fewer tumors or altered hepatic foci than in controls, indicating an anti-carcinogenic process , . Furthermore, decreased lipid peroxidation level in rats treated with low dose DDT was reported, compared to control rats .
Studies of the levels of POPs in the global environment show that emission sources of a number of POPs in the last 20 years have shifted from industrialized countries of the Northern Hemisphere to less developing countries in tropical and sub-tropical regions . This is due to a late production ban on OC pesticides: some OC pesticides are still being used in agriculture and for the control of diseases, such as malaria . Also, there is another active exposure source of PCBs through e-waste recycling in these countries . Although the burden of cigarette smoking use is currently greatest in high-income countries, rates of smoking are increasing in many low-income and middle-income countries. Therefore, the interaction between these two important health-related issues, POPs and cigarette smoking, should be further studied from a variety of viewpoints including molecular mechanisms.
The strengths of our study include the homogenous age and community-based sampling of study subjects. There were some limitations to our study. Due to a limited sample size, we could not consider more detailed information on cigarette smoking like total pack-years or duration of smoking cessation in analyses. In addition, analyses focusing on cause-specific mortality were not possible as cause of death information was not available. Finally, even though the consistency of findings across two studies lends credence to our findings and seems to reduce the likelihood that our finding is explainable by chance or bias, our findings still require replication by other cohort studies.
This study has confirmed a strong association between cigarette smoking and mortality among elderly people with relatively high POPs, but a much weaker and not statistically significant smoking-mortality association among those with relatively low POPs. However, the observation that the lowest mortality was seen among never smokers with high POPs is provocative and requires further studies on the role of POPs in longevity. In addition more prospective studies in human, in-vitro and in-vivo experimental studies would help to elucidate potential molecular mechanisms.
Interactions of cigarette smoking with summary measure of polychlorinated biphenyls (PCB) or organochlorine (OC) pesticides predicting mortality in the U.S. general population (reanalyses of published results using the same National Health and Nutrition Examination Survey (NHANES) datasets (Lee, 2013). Number of study subjects and deaths were 610 and 142 for PCB analyses and 702 and 157 for OC pesticide analyses. Hazard ratios (HRs) were estimated using a common reference group (never smokers & T1), adjusted for gender, BMI, exercise, and alcohol consumption. T1, first tertile; T2, second tertile; T3, third tertile. Although none of the HRs was statistically significantly different from the reference group, as previously reported, interaction p-values were 0.008 for PCBs and 0.024 for OC pesticides.
Conceived and designed the experiments: DHL LL PML. Performed the experiments: DHL LL PML. Analyzed the data: DHL. Contributed reagents/materials/analysis tools: SS BB. Wrote the paper: DHL LL DRJ SS BB PML.
- 1. Murray CJ, Lopez AD (1997) Mortality by cause for eight regions of the world: Global Burden of Disease Study. Lancet 349: 1269–1276.
- 2. Doll R, Peto R, Wheatley K, Gray R, Sutherland I (1994) Mortality in relation to smoking: 40 years' observations on male British doctors. Bmj 309: 901–911.
- 3. Hunt D, Blakely T, Woodward A, Wilson N (2005) The smoking-mortality association varies over time and by ethnicity in New Zealand. Int J Epidemiol 34: 1020–1028.
- 4. Jacobs DR Jr, Adachi H, Mulder I, Kromhout D, Menotti A, et al. (1999) Cigarette smoking and mortality risk: twenty-five-year follow-up of the Seven Countries Study. Arch Intern Med 159: 733–740.
- 5. van de Mheen PJ, Gunning-Schepers LJ (1996) Differences between studies in reported relative risks associated with smoking: an overview. Public Health Rep 111: 420–426 discussion 427.
- 6. Lee YM, Bae SG, Lee SH, Jacobs DR, Lee DH (2013) Associations between cigarette smoking and total mortality differ depending on serum concentrations of persistent organic pollutants among the elderly. J Korean Med Sci 28: 1122–1128.
- 7. Carpenter DO (2011) Health effects of persistent organic pollutants: the challenge for the Pacific Basin and for the world. Rev Environ Health 26: 61–69.
- 8. Ha MH, Lee DH, Jacobs DR (2007) Association between serum concentrations of persistent organic pollutants and self-reported cardiovascular disease prevalence: results from the National Health and Nutrition Examination Survey, 1999–2002. Environ Health Perspect 115: 1204–1209.
- 9. Lee DH, Lee IK, Song K, Steffes M, Toscano W, et al. (2006) A strong dose-response relation between serum concentrations of persistent organic pollutants and diabetes: results from the National Health and Examination Survey 1999–2002. Diabetes Care 29: 1638–1644.
- 10. Diwan BA, Ward JM, Kurata Y, Rice JM (1994) Dissimilar frequency of hepatoblastomas and hepatic cystadenomas and adenocarcinomas arising in hepatocellular neoplasms of D2B6F1 mice initiated with N-nitrosodiethylamine and subsequently given Aroclor-1254, dichlorodiphenyltrichloroethane, or phenobarbital. Toxicol Pathol 22: 430–439.
- 11. Hutton JJ, Meier J, Hackney C (1979) Comparison of the in vitro mutagenicity and metabolism of dimethylnitrosamine and benzo[a]pyrene in tissues from inbred mice treated with phenobarbital, 3-methylcholanthrene or polychlorinated biphenyls. Mutat Res 66: 75–94.
- 12. Psaty BM, Koepsell TD, Manolio TA, Longstreth WT Jr, Wagner EH, et al. (1990) Risk ratios and risk differences in estimating the effect of risk factors for cardiovascular disease in the elderly. J Clin Epidemiol 43: 961–970.
- 13. Sandau CD, Sjodin A, Davis MD, Barr JR, Maggio VL, et al. (2003) Comprehensive solid-phase extraction method for persistent organic pollutants. Validation and application to the analysis of persistent chlorinated pesticides. Anal Chem 75: 71–77.
- 14. Salihovic S, Lampa E, Lindstrom G, Lind L, Lind PM, et al. (2012) Circulating levels of persistent organic pollutants (POPs) among elderly men and women from Sweden: results from the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS). Environ Int 44: 59–67.
- 15. Phillips DL, Pirkle JL, Burse VW, Bernert JT Jr, Henderson LO, et al. (1989) Chlorinated hydrocarbon levels in human serum: effects of fasting and feeding. Arch Environ Contam Toxicol 18: 495–500.
- 16. Solomon GM, Weiss PM (2002) Chemical contaminants in breast milk: time trends and regional variability. Environ Health Perspect 110: A339–347.
- 17. Blackburn EH (2005) Telomeres and telomerase: their mechanisms of action and the effects of altering their functions. FEBS Lett 579: 859–862.
- 18. Shin JY, Choi YY, Jeon HS, Hwang JH, Kim SA, et al. (2010) Low-dose persistent organic pollutants increased telomere length in peripheral leukocytes of healthy Koreans. Mutagenesis 25: 511–516.
- 19. Zhang TC, Schmitt MT, Mumford JL (2003) Effects of arsenic on telomerase and telomeres in relation to cell proliferation and apoptosis in human keratinocytes and leukemia cells in vitro. Carcinogenesis 24: 1811–1817.
- 20. Wu KJ, Grandori C, Amacker M, Simon-Vermot N, Polack A, et al. (1999) Direct activation of TERT transcription by c-MYC. Nat Genet 21: 220–224.
- 21. Bakaysa SL, Mucci LA, Slagboom PE, Boomsma DI, McClearn GE, et al. (2007) Telomere length predicts survival independent of genetic influences. Aging Cell 6: 769–774.
- 22. Cawthon RM, Smith KR, O'Brien E, Sivatchenko A, Kerber RA (2003) Association between telomere length in blood and mortality in people aged 60 years or older. Lancet 361: 393–395.
- 23. Valdes AM, Andrew T, Gardner JP, Kimura M, Oelsner E, et al. (2005) Obesity, cigarette smoking, and telomere length in women. Lancet 366: 662–664.
- 24. Mattson MP (2008) Hormesis defined. Ageing Res Rev 7: 1–7.
- 25. Yokoo S, Furumoto K, Hiyama E, Miwa N (2004) Slow-down of age-dependent telomere shortening is executed in human skin keratinocytes by hormesis-like-effects of trace hydrogen peroxide or by anti-oxidative effects of pro-vitamin C in common concurrently with reduction of intracellular oxidative stress. J Cell Biochem 93: 588–597.
- 26. Tuomisto J, Pekkanen J, Kiviranta H, Tukiainen E, Vartiainen T, et al. (2005) Dioxin cancer risk—example of hormesis? Dose Response 3: 332–341.
- 27. Sukata T, Uwagawa S, Ozaki K, Ogawa M, Nishikawa T, et al. (2002) Detailed low-dose study of 1,1-bis(p-chlorophenyl)-2,2,2- trichloroethane carcinogenesis suggests the possibility of a hormetic effect. Int J Cancer 99: 112–118.
- 28. Viluksela M, Bager Y, Tuomisto JT, Scheu G, Unkila M, et al. (2000) Liver tumor-promoting activity of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in TCDD-sensitive and TCDD-resistant rat strains. Cancer Res 60: 6911–6920.
- 29. Shutoh Y, Takeda M, Ohtsuka R, Haishima A, Yamaguchi S, et al. (2009) Low dose effects of dichlorodiphenyltrichloroethane (DDT) on gene transcription and DNA methylation in the hypothalamus of young male rats: implication of hormesis-like effects. J Toxicol Sci 34: 469–482.
- 30. Tanabe S, Minh TB (2010) Dioxins and organohalogen contaminants in the Asia-Pacific region. Ecotoxicology 19: 463–478.
- 31. Wong MH, Leung AO, Chan JK, Choi MP (2005) A review on the usage of POP pesticides in China, with emphasis on DDT loadings in human milk. Chemosphere 60: 740–752.
- 32. Someya M, Ohtake M, Kunisue T, Subramanian A, Takahashi S, et al. (2010) Persistent organic pollutants in breast milk of mothers residing around an open dumping site in Kolkata, India: specific dioxin-like PCB levels and fish as a potential source. Environ Int 36: 27–35.