Associations of organochlorine pesticides and polychlorinated biphenyls with total, cardiovascular, and cancer mortality in elders with differing fat mass.

Environmental Research 138 (2015) 1–7

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

Associations of organochlorine pesticides and polychlorinated biphenyls with total, cardiovascular, and cancer mortality in elders with differing fat mass Se-A Kim a,b,c, Ki-Su Kim a, Yu-Mi Lee a, David R. Jacobs d, Duk-Hee Lee a,b,c,n a

Department of Preventive Medicine, School of Medicine, Kyungpook National University, Daegu, Republic of Korea Department of Biomedical Science, Kyungpook National University, Daegu, Republic of Korea c BK21 Plus KNU Biomedical Convergence Program, Department of Biomedical Science, Kyungpook National University, Republic of Korea d Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, MN, USA b

art ic l e i nf o

a b s t r a c t

Article history: Received 5 September 2014 Received in revised form 26 January 2015 Accepted 28 January 2015

Background: We investigated if certain persistent organic pollutants (POPs), namely polychlorinated biphenyls (PCBs) and organochlorine (OC) pesticides, predicted total, cardiovascular disease (CVD), and cancer mortality among the elderly, with the hypothesis that associations differ by the amount of fat mass. Methods: We studied serum concentrations of 11 PCBs in 633 elders (ageZ 70 years) and of 5 OC pesticides in 675 elders within the National Health and Nutrition Examination Survey (NHANES) 1999–2004. Mean follow-up was 4.1-years. Results: Neither PCBs nor OC pesticides were associated with total mortality when fat mass was not considered in analyses. However, associations of PCBs and OC pesticides with total mortality depended on fat mass (Pinteraction o 0.01 and 0.06, respectively). PCBs associated inversely with total mortality for high fat mass, but not for lower fat mass. On the contrary, OC pesticides associated positively with total mortality for low fat mass and this association weakened at higher fat mass. The interaction was also observed with CVD mortality. In elders with low fat mass, higher PCBs associated with 2–3 fold higher risk of CVD mortality, while this association was absent in elders with more fat mass (Pinteraction ¼ 0.03). The positive association between OC pesticides and CVD mortality was also observed only among elderly with low fat mass (Pinteraction ¼ 0.03). Conclusions: The possibility of interaction between POPs and the amount of fat mass on risk of mortality from chronic diseases is clinically important in modern societies with an obesity epidemic and requires confirmation in other studies with larger sample size. & 2015 Elsevier Inc. All rights reserved.

Keywords: Persistent organic pollutants Polychlorinated biphenyls Organochlorine pesticides Fat mass Obesity Cardiovascular mortality Cancer mortality

1. Introduction Background exposure to persistent organic pollutants (POPs) has recently been linked to a variety of chronic diseases. However, few studies have focused on associations between POPs and mortality in general populations. Exposure to a mixture of dioxinlike chemicals was associated with an increased total mortality risk among U.S. adults (Lin et al., 2012). POPs without dioxin activity, such as nondioxin-like polychlorinated biphenyls (PCBs) or organochlorine (OC) pesticides, have not been evaluated in relation to mortality in general populations yet. n Correspondence to: Department of Preventive Medicine, School of Medicine, Kyungpook University, 101 Dongin-dong, Jung-gu, Daegu 700-422, Republic of Korea. Fax: þ82 53 425 2447. E-mail address: [email protected] (D.-H. Lee).

http://dx.doi.org/10.1016/j.envres.2015.01.021 0013-9351/& 2015 Elsevier Inc. All rights reserved.

On the other hand, mortality studies among workers who were exposed to high levels of PCBs or OC pesticides in occupational settings have reported lower total and cardiovascular diseases (CVD) mortality among workers compared to general populations and associations with site-specific cancer mortality differ across studies (Brown, 1992; Kimbrough and Krouskas, 2003; Kimbrough et al., 2014; Prince et al., 2006; Ruder et al., 2006; Swaen et al., 2002). The decreased risk of mortality among workers has commonly been interpreted as a bias due to healthy worker effect (Brown, 1992; Ruder et al., 2006). Interestingly, one prospective study among the elderly reported that obesity was differently associated with total mortality depending on serum concentrations of POPs including nondioxinlike PCBs or OC pesticides (Hong et al., 2012). In that study, mortality was not analyzed and interpreted from the viewpoint of POPs because it was performed to evaluate if the obesity paradox,

2

Se-A Kim et al. / Environmental Research 138 (2015) 1–7

better survival among obese patients or elders than among lean persons, can be explained by POPs. Although obesity is commonly regarded as harmful to human health, from the viewpoint of lipophilic chemicals like POPs with very long half-lives, adipose tissue may provide a relatively safe harbor for POPs in humans, protecting other critical organs from POP insults (La Merrill et al., 2013; Lee et al., 2014). Experimental studies also suggested that adipose tissue can protect against harm from POPs. Coplanar polychlorinated biphenyls (PCBs) impaired glucose homeostasis in lean mice but not in obese mice and obese mice developed impaired glucose homeostasis after weight loss (Baker et al., 2013). In rats, diet-induced obesity also significantly increased survival time following exposure to a potentially lethal dose of dioxin (Tuomisto et al., 1999). Therefore, this study was performed with two purposes. First, we evaluated associations of serum concentrations of PCBs, including both dioxin-like and nondioxin-like PCBs, and OC pesticides with total, CVD, and cancer mortality. Second, we further tested if POPs were differently associated with mortality by the amount of fat mass. As the findings were more clearly observed with aging, we focused on men and women aged 70 or over.

2. Methods 2.1. Study subjects The continuous NHANES, conducted annually since 1999 by the Centers for Disease Control and Prevention (CDC), is an ongoing survey designed to measure the health and nutritional status of the civilian noninstitutionalized U.S. population. In the NHANES 1999–2002, all POPs were measured in the same subsample while in the NHANES 2003–2004, OC pesticides were measured in a different subsample from the subsample in which other POPs were measured. Thus, the final partially overlapping sample sizes were 633 for PCBs and 675 for OC pesticides, including subjects aged 70 years or older with information on fat mass and followed for mortality through the end of 2006. The study protocol was reviewed and approved by the institutional review board of Center for Disease Control in the U.S. Also, informed written consent was obtained from all subjects before they took part in the study. 2.2. Measurement Detailed sample collection and laboratory analysis was reported in the NHANES Laboratory/Medical Technologists Procedures Manual (CDC, 1999–2000; CDC, 2001–2002; CDC, 2003– 2004). Briefly, venous blood samples were collected and shipped weekly at 20 °C. PCBs and OC pesticides were all measured as individual chemicals by high-resolution gas chromatography/ high-resolution mass spectrometry using isotope dilution for quantification. We selected 11 PCBs (PCB074, PCB099, PCB118, PCB126, PCB146, PCB153, PCB156, PCB169, PCB170, PCB180, and PCB187) and 5 OC pesticides (p,p’-DDE, trans-nonachlor, oxychlordane, heptachlor epoxide, and β-hexachlorocyclohexane) for which at least 80% of study subjects had concentrations more than the limit of detection. Fat mass was measured by dual X-ray absorptiometry (DEXA) using Hologic QDR 4500A fan-beam densitometers (Hologic, Bedford, MA, USA). 2.3. Mortality follow-up Probabilistic matching was used to link NHANES participants with the National Death Index to ascertain vital status. Matching was based on 12 identifiers for each participant (eg, Social Security number, sex, and date of birth). The cause of death was

determined using the underlying cause listed on death certificates, and was coded using the International Classification of Diseases, 10th Revision (ICD-10). Cause-specific mortality was ascertained for cardiovascular disease (codes: I00-I78), and cancer (codes: C00-C97). Persons who survived the entire follow-up period were censored on December 31, 2006. Follow-up time for each person was calculated as the difference between the NHANES examination date and the last known date alive or censored. More detailed information on the matching process and calibration study can be found elsewhere (CDC, 2014). 2.4. Statistical analysis First, we checked if there were associations between serum concentrations of each of 11 PCBs and 5 OC pesticides (tertiles) and total, CVD, and cancer mortality using Cox proportional hazard models with linear continuous POP terms. Tertile cutpoints of individual POPs were presented in Supplementary Table 1. In addition to serum concentrations of each compound, we used summary measures of PCBs and of OC pesticides, again assessing pvalues for continuous trends in proportion al hazards models. The summary measurements were estimated by summing the rank orders of the individual compounds belonging to PCBs or OC pesticides from rank 1 to 635 (PCBs) or 675 (OC pesticides). Concentrations lower than the LOD were assigned rank 1, and the remaining subjects were ranked according to increasing concentration. The ranks for individual compounds were summed, and the summed values were categorized into tertile groups. We did not use summary measurements formed by adding the absolute concentration of each compound because they were strongly influenced only by several compounds with high concentrations. Second, we evaluated if there were statistically significant interactions of PCBs or OC pesticides (tertiles) with fat mass (quartiles) in predicting total, CVD, and cancer mortality. As most outcomes showed at least marginally statistically significant interactions with PCBs or OC pesticides, we presented results stratified by fat mass. For stratified analyses, we regrouped the quartiles of fat mass into o25th, 25th o75th, and Z75th percentile of fat mass for statistical stability of the middle category. Adjusting covariates were age, gender, race-ethnicity, smoking status (current, ex, and never), physical activity (vigorous, moderate, and sedentary), and body mass index. Estimates of main results were calculated accounting for NHANES stratification and clustering (Korn and Graubard, 1991), adjusting for age, gender, race-ethnicity, smoking status, and physical activity instead of using sample weights; this adjustment has been regarded as a good compromise between efficiency and bias (Graubard and Korn, 1999; Korn and Graubard, 1991). As results were very similar with SAS 9.1 and SUDAAN 9.0, we present the results based on SAS 9.1.

3. Results Among 633 elders with information on PCBs, 143 subjects (underlying causes: 50 cardiovascular, 34 cancer, 17 respiratory disease, and 42 others) died during the mean follow-up time of 4.1 years. Among 675 elders with information on OC pesticides, the numbers of all causes, CVD, and cancer deaths were 146, 56, and 32, respectively. Table 1 shows characteristics of study subjects. Among participants with information on PCBs, mean age was 77.6 years and proportions of men, white race, obesity, current smoker, and physically inactive were 52.6%, 69.5%, 23.9%, 6.5%, and 55.3%, respectively. Prevalence of physician-diagnosed heart diseases, cancer, diabetes, and hypertension were 24.5%, 23.4%, 15.6%, and 49.0%. Similar distributions were observed among 675


Table 1 Characteristics of study subjects (633 elders for analyses on polychlorinated biphenyls (PCBs) or 675 elders for analyses on organochlorine (OC) pesticides).

Agea, (years) Men, (%) White, (%) Body mass index, (kg/m2) Body mass index Z 30 kg/m2, (%) Fat mass, (kg) Current smoker, (%) Physically inactive, (%) Physician-diagnosed heart diseases, (%) Physician-diagnosed cancer, (%) Physician-diagnosed diabetes, (%) Physician-diagnosed hypertension, (%)

Dataset for PCBs (n ¼633)

Dataset for OC pesticides (n¼675)

77.6 75.1 52.6 69.5 27.3 7 4.6 23.9% 26.9 78.6 6.5 55.3 24.5 23.4 15.6 49.0

77.4 7 5.1 51.3 67.7 27.3 74.9 23.6% 27.17 9.1 6.3 58.3 24.5 24.5 15.8 47.2

a Individuals aged 85 and over (13.6%) were recoded by the National Health and Nutrition Examination Survey to 85 years of age.

3

in the lowest tertile of PCBs after adjusting for age, gender, race, cigarette smoking, physical activity, and BMI (P trend ¼0.02). However, PCBs demonstrated an inverse (though not statistically significant) trend with cancer mortality. OC pesticides showed patterns of associations similar to but weaker than those with PCBs. Tables 3 and 4 show results for PCBs and OC pesticides in analyses stratified by fat mass. Results for individual PCBs and OC pesticides were presented in Supplement Table 3 to 5. In general, the associations of PCBs with total, CVD, and cancer mortality differed by the amount of fat mass (Table 3). There was no clear association between PCBs and total mortality among the elderly with small fat mass while high PCBs showed a lower total mortality as fat mass increased (P for interaction o0.01). In addition, the patterns differed depending on cause of death. Among the elderly with fat mass o75th percentile, those with high PCBs showed 2–3 times higher risk of CVD mortality than those with low PCBs, but this association was not observed among elders with higher fat mass (P for interaction¼ 0.03). On the other hand, the

Table 2 Adjusteda associations of summary measuresb of polychlorinated biphenyls (PCBs) or organochlorine (OC) pesticides with all cause, cardiovascular disease (CVD), or cancer mortality. Tertiles of summary measure of PCBs

Outcome: all cause death

Outcome: CVD death

Outcome: cancer

Ptrend

T1

T2

T3

Death/no. of subjects Crude hazard ratios Adjusted hazard ratios

56/211 Reference Reference

41/211 0.71 (0.48–1.07) 0.64 (0.43–0.97)

46/211 0.82 (0.56–1.22) 0.77 (0.51–1.16)

0.31 0.19



13/211 1.15 (0.51–2.57) 1.06 (0.47–2.40)

26/211 2.36 (1.16–4.77) 2.37 (1.10–5.09)

0.01 0.02



11/211 0.78 (0.35–1.73) 0.71 (0.32–1.57)

9/211 0.65 (0.28–1.51) 0.58 (0.24–1.38)

0.31 0.21

Tertiles of summary measure of OC pesticides

Outcome: all cause death

Outcome: CVD death

Outcome: cancer

Ptrend

T1

T2

T3



48/225 1.05 (0.69–1.59) 1.15 (0.75–1.78)

57/225 1.49 (1.00–2.22) 1.27 (0.83–1.94)

0.05 0.28



18/225 1.27 (0.62–2.60) 1.34 (0.64–2.81)

25/225 2.03 (1.04–3.98) 1.65 (0.81–3.34)

0.03 0.17



9/225 0.62 (0.27–2.75) 0.86 (0.35–2.14)

10/225 0.84 (0.37–1.91) 0.88 (0.36–2.13)

0.23 0.77

a

Adjusted for age, sex, race, cigarette smoking, physical activity, and body mass index. Detectable values of PCBs or OC pesticides were individually ranked and the rank orders of the individual POPs in each subclass were summed to arrive at the subclass value. All not detectable values were ranked as 1. b

participants with information on OC pesticides. Table 2 shows associations of tertiles of the summary measure of PCBs or of OC pesticides with all causes, CVD, and cancer mortality. Results for individual PCBs and OC pesticides were presented in Supplement Table 2. The summary measures of both PCBs and OC pesticides were not clearly associated with total mortality. However, PCBs showed opposite associations according to cause of death. Elderly men and women in the highest tertile of PCBs had about 2.4 times higher CVD mortality compared to those

interaction between PCBs and fat mass with cancer mortality was mainly driven by a strong inverse association among the elderly with higher fat mass (P for interaction¼ 0.09). This inverse association was not seen at lower fat mass. OC pesticides also showed at least marginally statistically significant interactions with fat mass on the prediction of total and CVD mortality (Table 4). Similar to the results for PCBs, the associations between OC pesticides and total mortality depended on fat mass (P for interaction ¼0.06). Compared to the results for

4


Table 3 Adjusteda associations of summary measuresb of polychlorinated biphenyls (PCBs) with all cause, cardiovascular disease (CVD), or cancer mortality stratified by fat massc. Tertiles of summary measure of PCBs

Outcome: all cause deaths Fat masso 25% Fat mass 25 o75% Fat massZ 75%

T1

T2

T3

Death/no. of subjects Adjusted hazard ratios Death/no. of subjects Adjusted hazard ratios Death/no. of subjects Adjusted hazard ratios

12/40 Reference 28/109 Reference 16/62 Reference

13/52 0.75 (0.33–1.71) 20/106 0.68 (0.38–1.22) 8/53 0.49 (0.20–1.21)

21/66 1.12 (0.50–2.49) 21/102 0.72 (0.39–1.32) 4/43 0.32 (0.10–1.05)



3/52 0.62 (0.12–3.22) 6/106 1.48 (0.41–5.27) 4/53 1.29 (0.30–5.48)

12/66 2.33 (0.59–9.27) 13/102 3.42 (1.02–11.4) 1/43 0.51 (0.05–5.38)



3/52 0.93 (0.14–6.15) 6/106 0.85 (0.28–2.56) 2/53 0.29 (0.05–1.83)

4/66 1.18 (0.18–7.61) 4/102 0.59 (0.16–2.19) 1/43 0.15 (0.01–1.58)

Ptrend

Pinteractionb

0.67

o 0.01

0.25 0.04

Outcome: CVD deaths Fat masso 25% Fat mass 25 o75% Fat massZ 75%

0.10

0.03

0.04 0.72

Outcome: cancer deaths Fat masso 25% Fat mass 25 o75% Fat massZ 75%

0.83

0.09

0.43 0.09

a

Adjusted for age, sex, race, cigarette smoking, physical activity, and body mass index. Detectable values of PCBs were individually ranked and the rank orders of the individual compound were summed to arrive at the subclass value. All not detectable values were ranked as 1. c The cutoff point of fat mass 25% and 75% were 21.0 kg and 31.5 kg, respectively. b

Table 4 Adjusteda associations of summary measuresbof organochlorine (OC) pesticides with all cause, cardiovascular disease (CVD), or cancer mortality stratified by fat massc. Tertiles of summary measure of OC pesticides

Outcome: all cause deaths Fat masso 25% Fat mass 25 o75% Fat massZ 75%

T1

T2

T3



15/54 1.64 (0.70–3.83) 27/114 1.26 (0.69–2.30) 6/57 0.44 (0.15–1.27)

23/55 2.86 (1.31–6.25) 19/106 0.85 (0.44–1.64) 15/64 0.88 (0.36–2.20)



1/54 0.26 (0.03–2.37) 14/114 2.42 (0.84–7.01) 3/57 0.67 (0.14–3.31)

13/55 4.54 (1.38–15.0) 7/106 1.13 (0.34–3.71) 5/64 0.96 (0.22–4.17)



3/54 1.08 (0.22–5.25) 4/114 1.00 (0.25–4.06) 2/57 0.37 (0.05–2.62)

5/55 1.44 (0.37–5.63) 3/106 0.79 (0.17–3.67) 2/64 0.33 (0.04–2.70)

Ptrend

Pinteractionb

o0.01

0.06

0.61 0.99

Outcome: CVD deaths Fat masso 25% Fat mass 25 o75% Fat massZ 75%

o0.01

0.03

0.99 1.00

Outcome: cancer deaths Fat masso 25% Fat mass 25 o75% Fat massZ 75%

a

0.61

0.13

0.77 0.29

Adjusted for age, sex, race, cigarette smoking, physical activity, and body mass index. Detectable values of OC pesticides were individually ranked and the rank orders of the individual compound were summed to arrive at the subclass value. All not detectable values were ranked as 1. c The cutoff point of fat mass 25% and 75% were 20.7 kg and 31.4 kg, respectively. b


5

Fig. 1. Interactions of polychlorinated biphenyls or organochlorine pesticides with fat mass on all cause, cardiovascular disease (CVD), or cancer mortality. Adjusted for age, sex, race, cigarette smoking, physical activity, and body mass index. T1, first tertile; T2, second tertile; T3, third tertile.

PCBs, a positive association among the elderly with low fat mass was evident, in contrast to the inverse association among the elderly with higher fat mass. The elderly with high OC pesticides showed about 2.9 times higher total mortality compared to those with low OC pesticides (P trend o 0.01). Patterns with CVD mortality and cancer mortality were also similar to those of PCBs. In

the case of CVD mortality, the interactions between OC pesticides and fat mass were mainly influenced by the positive association among the elderly with low fat mass; the elderly with high OC pesticides showed 4.5 times higher risk of CVD mortality if they had low fat mass (P trendo0.01), but this association was not observed among the elderly with large fat mass (P for

6


interaction¼ 0.03). When cancer death was used as the outcome, p value of interaction was not statistically significant. However, similar to the findings for PCBs, the inverse trend between OC pesticides and cancer mortality was observed only among elders with higher fat mass while this pattern was not observed among elders with low fat mass. Fig. 1 shows adjusted HRs when elders with fat mass o25th percentile and the lowest PCBs or OC pesticides were used as the common reference group. For clarity of figures, we did not present results among elders with fat mass in the 25th–75th percentiles. The inverse association between PCBs and total mortality among elders with fat mass Z75th percentile was driven by both the high mortality in the lowest PCBs and the low mortality in the highest PCBs. In the case of OC pesticides, the positive association with total mortality among the elderly with fat mass o25th percentile seemed to be more prominent. However, results varied by cause of death. When CVD mortality was the outcome, both PCBs and OC pesticides showed the highest mortality when they had fat mass o25th percentile and the highest concentrations. Cancer mortality was the highest among elders with fat mass Z75th percentile and the lowest concentrations, especially PCBs. When we interpreted the figure from the viewpoint of fat mass, more fat mass were related to higher risk of CVD or cancer mortality among elders with low PCBs or OC pesticides whereas elders with less fat mass showed a higher risk of CVD or cancer when they had high levels of PCBs or OC pesticides. As it was likely that history of CVD or cancer at baseline could affect both serum concentrations of POPs and fat mass, we performed sensitivity analyses after excluding elders with either condition. Even though statistical power became lower because the sample sizes were reduced from 633 to 367 and from 675 to 392 in the datasets for PCBs and OC pesticides, respectively, the pattern for total mortality was similarly observed (Supplement Table 6), but power in this study is insufficient to fully evaluate this issue. Also, further adjustment for weight change during the most recent 1 year before survey or dropping body mass index from the list of covariates did not materially change the results.

4. Discussion This study demonstrated complicated interactions of PCBs or OC pesticides and fat mass in prediction of mortality. When fat mass was not considered, neither PCBs nor OC pesticides was clearly associated with total mortality. However, in analyses stratified by fat mass, PCBs demonstrated an inverse association with total mortality predominantly among the elderly with higher fat mass while this association was not present as fat mass decreased. On the contrary, OC pesticides increased the risk of total mortality only among elders with low fat mass and this association weakened at higher fat mass. Different patterns of associations depending on causes of death made the findings more complicated. In the case of CVD mortality, both PCBs and OC pesticides increased the risk of CVD mortality only among elders with low fat mass. However, cancer mortality showed a strong inverse association with PCBs among the elderly with higher fat mass. OC pesticides also showed the inverse trend. This inverse association was strongly influenced by the high risk of cancer mortality among the elderly in the lowest tertile of PCBs or OC pesticides. This inverse association was absent in the elderly with lower fat mass. Lower total or CVD mortality and inconsistent associations with cancer mortality among PCBs or OC pesticides workers have mainly been attributed to a healthy worker effect (Brown, 1992; Ruder et al., 2006). However, the current findings from the general population with low dose background exposure to PCBs or OC

pesticides suggests that there may be important issues in evaluation of the health effects of PCBs or OC pesticides, other than a bias such as the healthy worker effect. Regardless of cause of mortality, two biological mechanisms are commonly considered to link POPs to fat mass. First is that POPs may act as obesogens (Dirinck et al., 2011; Pereira-Fernandes et al., 2014), compounds that increase the risk of obesity. As obesity itself is known as a risk factor of many chronic diseases including cardiovascular diseases and cancer, POPs as obesogens can be harmful as well. Second, POPs provoke an inflammatory state in adipose tissue, a condition associated with the metabolic side effects of obesity (La Merrill et al., 2013). From these two viewpoints, POPs and fat mass might be expected to show synergic interaction on the risk of obesity-related chronic diseases and so do not explain the non-synergistic interactions between POPs and fat mass which were observed in the current study. Taking the data presented here literally, low fat mass was associated with higher CVD risk at high dose PCBs or OC pesticides; whereas high fat mass was suggestively associated with higher cancer risk at low dose PCBs or OC pesticides. Given limited understanding of the biologic interactions between POPs and fat mass, it is difficult to provide any reasonable mechanistic interpretation of these epidemiological findings at this point. However, whatever the biologic roles of fat mass in the associations between POPs and mortality, at least we can say that it is difficult to explain our findings under the traditional toxicology paradigm of linear dose–response relations considering the strong inverse association between POPs and cancer mortality among elders with high fat mass. Recently, POPs have been suggested to show non-monotonic dose response relations with many biological outcomes (Vandenberg et al., 2012). Also, the current findings at least present the importance of fat mass in the relations between POPs and clinical outcomes. At present, almost all studies on obesity have not considered the presence of POPs in adipose tissue while almost all studies on POPs have considered obesity only as one of confounders. This study clearly shows that the interaction between POPs and fat mass should be considered in the future studies. This study has several limitations. First, it has a limited power for analysis of cause-specific mortality due to the small numbers of deaths. Nevertheless, it is important to present the different patterns of associations between CVD and cancer mortality for purposes of hypothesis generation. In addition, despite small sample sizes and numbers of fatal outcomes, most P values for interactions on cause-specific mortality reached at least marginal statistical significance. Second, we evaluated the associations with mortality, not disease incidence. POPs may affect the process of disease development differently from how they affect disease survival. Third, as many in the U.S. population are obese, there may not have been a sufficient number of elders who represented those with absolutely small fat mass. In this study, low and high fat mass were not absolute terms, but were simply classified based on the relative distribution of fat mass in the current sample.

5. Conclusions In conclusion, different associations of PCBs or OC pesticides with total mortality by amount of fat mass were observed in the elderly U.S. population. PCBs associated inversely with total mortality for high fat mass, but not for lower fat mass. On the contrary, OC pesticides associated positively with total mortality for low fat mass and this association weakened at higher fat mass. The interaction was also observed with CVD mortality. PCBs or OC pesticides were associated with higher risk of CVD mortality only among the elderly with low fat mass. The possibility that the amount of fat mass can modify the associations between POPs and


mortality has important implications in modern societies that suffer from an obesity epidemic. The possibility of interaction between POPs and the amount of fat mass on morbidity and mortality risk of chronic diseases should be tested in other populations with a larger sample size and a wider range of fat mass. Also, experimental studies on molecular mechanisms are needed.

Conflict of interest/financial disclosure Nothing to declare.

Acknowledgments This work was partly supported by grants from the Korean Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (HI13C0715), and the National Research Foundation of Korea (No. 2013R1A2A2A01068254).

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2015.01. 021.

References Baker, N.A., et al., 2013. Coplanar polychlorinated biphenyls impair glucose homeostasis in lean C57BL/6 mice and mitigate beneficial effects of weight loss on glucose homeostasis in obese mice. Environ. Health Perspect. 121, 105–110. Brown, D.P., 1992. Mortality of workers employed at organochlorine pesticide manufacturing plants – an update. Scand. J. Work Environ. Health 18, 155–161. CDC, 1999–2000. Laboratory Procedure Manual: PCBs and Persistent Pesticides. Vol.

7

2014. CDC, 2001–2002. Laboratory Procedure Manual: PCBs and Persistent Pesticides. Vol. 2014. CDC, 2003–2004. Laboratory Procedure Manual: PCBs and Persistent Pesticides. Vol. 2014. CDC, 2014. NCHS Data Linked to Mortality Files. Vol. 2014. Dirinck, E., et al., 2011. Obesity and persistent organic pollutants: possible obesogenic effect of organochlorine pesticides and polychlorinated biphenyls. Obesity 19, 709–714. Graubard, B.I., Korn, E.L., 1999. Analyzing health surveys for cancer-related objectives. J. Natl. Cancer Inst. 91, 1005–1016. Hong, N.S., et al., 2012. The association between obesity and mortality in the elderly differs by serum concentrations of persistent organic pollutants: a possible explanation for the obesity paradox. Int. J. Obes. 36, 1170–1175. Kimbrough, R.D., Krouskas, C.A., 2003. Human exposure to polychlorinated biphenyls and health effects: a critical synopsis. Toxicol. Rev. 22, 217–233. Kimbrough, R.D., et al., 2014. Mortality among capacitor workers exposed to polychlorinated biphenyls (PCBs), a long-term update. Int. Arch. Occup. Environ. Health. Korn, E.L., Graubard, B.I., 1991. Epidemiologic studies utilizing surveys: accounting for the sampling design. Am. J. Public Health 81, 1166–1173. La Merrill, M., et al., 2013. Toxicological function of adipose tissue: focus on persistent organic pollutants. Environ. Health Perspect. 121, 162–169. Lee, D.H., et al., 2014. Chlorinated persistent organic pollutants, obesity, and type 2 diabetes. Endocr. Rev. 35 (4), 557–601. Lin, Y.S., et al., 2012. Environmental exposure to dioxin-like compounds and the mortality risk in the U.S. population. Int. J. Hyg. Environ. Health 215, 541–546. Pereira-Fernandes, A., et al., 2014. Expression of obesity markers and Persistent Organic Pollutants levels in adipose tissue of obese patients: reinforcing the obesogen hypothesis? PLoS One 9, e84816. Prince, M.M., et al., 2006. Update: cohort mortality study of workers highly exposed to polychlorinated biphenyls (PCBs) during the manufacture of electrical capacitors, 1940–1998. Environ. Health 5, 13. Ruder, A.M., et al., 2006. Mortality among workers exposed to polychlorinated biphenyls (PCBs) in an electrical capacitor manufacturing plant in Indiana: an update. Environ. Health Perspect. 114, 18–23. Swaen, G.M., et al., 2002. Cancer mortality in workers exposed to dieldrin and aldrin: an update. Toxicol. Ind. Health 18, 63–70. Tuomisto, J.T., et al., 1999. TCDD-induced anorexia and wasting syndrome in rats: effects of diet-induced obesity and nutrition. Pharmacol. Biochem. Behav. 62, 735–742. Vandenberg, L.N., et al., 2012. Hormones and endocrine-disrupting chemicals: lowdose effects and nonmonotonic dose responses. Endocr. Rev. 33, 378–455.

Organochlorine pesticides and polychlorinated biphenyls in human adipose tissue.

Selected organochlorine pesticides and polychlorinated biphenyls in urban atmosphere of Pakistan: concentration, spatial variation and sources.

Time trends in serum organochlorine pesticides and polychlorinated biphenyls in the general population of Biscay, Spain.

Transplacental Transfer of Polychlorinated Biphenyls, Polybrominated Diphenylethers, and Organochlorine Pesticides in Ringed Seals (Pusa hispida).

Risk of female breast cancer and serum concentrations of organochlorine pesticides and polychlorinated biphenyls: a case-control study in Tunisia.

Persistent organochlorine pesticides and polychlorinated biphenyls in air of the North Sea region and air-sea exchange.

Concentrations of Polychlorinated Biphenyls and Organochlorine Pesticides in Umbilical Cord Blood Serum of Newborns in Kingston, Jamaica.

Polychlorinated biphenyls and selected organochlorine pesticides in serum of Slovak population from industrial and non-industrial areas.

Levels of polycyclic aromatic hydrocarbons, polychlorinated biphenyls, and organochlorine pesticides in various tissues of white-backed vulture in India.

Vertical distributions of organochlorine pesticides and polychlorinated biphenyls in an agricultural soil core from the Guanzhong Basin, China.

Residues of selected polychlorinated biphenyls (PCB) and organochlorine pesticides (OCP) in postmortem lungs from Epirus, northwestern Greece.

Organochlorine pesticides and polychlorinated biphenyls in surface soils from Ruoergai high altitude prairie, east edge of Qinghai-Tibet Plateau.

Organochlorine pesticides and polychlorinated biphenyls in fish from Lake Awassa in the Ethiopian Rift Valley: human health risks.

Estimation of measurement uncertainty of polychlorinated biphenyls, polycyclic aromatic hydrocarbons and organochlorine pesticides in the atmosphere using gas chromatography-mass spectrometry and gas chromatography-electron capture detector.

Polychlorinated biphenyls and links to cardiovascular disease.

Organochlorine pesticides and polychlorinated biphenyls along an east-to-west gradient in subtropical North Atlantic surface water.

Polychlorinated biphenyls, organochlorine pesticides and chlorobenzenes content of livers from Atlantic cod (Gadus morhua) caught off Halifax Nova Scotia.

Greater cognitive decline with aging among elders with high serum concentrations of organochlorine pesticides.

Supercritical fluid extraction of polychlorinated biphenyls and pesticides from soil. Comparison with other extraction methods.

Mortality among workers exposed to polychlorinated biphenyls.

Associations of Infant Subcutaneous Fat Mass with Total and Abdominal Fat Mass at School-Age: The Generation R Study.

Can Inconsistent Association between Hypertension and Cognition in Elders be Explained by Levels of Organochlorine Pesticides?

Paradoxical Associations of Insulin Resistance With Total and Cardiovascular Mortality in Humans.

Polychlorinated biphenyls and organochlorine pesticides in surface sediments from the sand flats of Shuangtaizi Estuary, China: levels, distribution, and possible sources.