Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Association of type 2 diabetes mellitus with plasma organochlorine compound concentrations

Abstract

The increased prevalence of type 2 diabetes mellitus (T2DM) is associated with obesity, age, and sedentary lifestyle, but exposure to some organochlorine (OC) compounds has also been recently implicated. The hypothesis tested is that higher concentrations of bioaccumulative OC compounds are associated with T2DM. Plasma samples were obtained from a cross-section of adult male and female Caucasians and African Americans, either with or without T2DM from two US Air Force medical facilities. A method of extracting OC compounds from human plasma using solid phase extraction was developed, and three OC compounds [p,p’-DDE (DDE), trans-nonachlor, and oxychlordane] were quantified by gas chromatography/mass spectrometry. Multivariable logistic regression modeling indicated that increasing body mass index (BMI) was associated with T2DM in Caucasians but not in African Americans, and African Americans were more likely to have T2DM than Caucasians with decreasing odds ratios as BMI increased. An association between T2DM and increasing plasma DDE (adjusted for age, base, race, and BMI) was observed. Increasing DDE concentrations were associated with T2DM in older individuals and those with lower BMIs. Thus, in this study sample there was a higher risk of T2DM with increasing DDE concentrations in older people of normal weight and relatively lower risk associated with increasing DDE concentrations in those who are overweight or obese.

INTRODUCTION

The prevalence of type 2 diabetes mellitus (T2DM) in the United States has increased significantly over the past several decades.1 In 2011, 8.3% of the US population (19.7 million people) had T2DM and 38.2% were considered prediabetic and at enhanced risk for developing T2DM.2 Known contributing factors to T2DM development include obesity, age, dietary glycemic intake, and sedentary lifestyle. Identification of the specific cause of the disease in any individual and the disparities in T2DM prevalence seen in some minority populations, such as African Americans,3 has been elusive because of the multi-factorial etiology of T2DM as well as the contributing factors implicated in disease development ranging from environmental toxicants to genetic susceptibility to geographical residence.4

Legacy organochlorine (OC) insecticides include both DDT and the cyclodienes, such as chlordane. These compounds were used predominately during the 1950s and 1960s in the United States. These OC insecticides are highly lipophilic and bioaccumulative,5 and they as well as selective metabolites/derivatives still maintain a background presence in the soil of many agricultural areas both within and outside the United States,6,7 and some are still used outside the United States. In fact, our laboratories have recently found that 67% of the soil samples collected in 2011 in the agricultural Delta region of Mississippi had measurable residues of 1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene (DDE; the bioaccumulative metabolite of DDT) with an average of 340 ng DDE/g soil (unpublished data). These OC compounds have been implicated in increases in adipocyte dysfunction and have been associated with metabolic disorders such as T2DM.8,9 Turyk et al.10 reported that over the course of 20 years, those with a higher blood concentration of DDE had a significantly higher prevalence of T2DM. In addition, Lee et al.11,12 analyzed the National Health and Nutrition Examination Survey (NHANES) data sets and identified the association of both T2DM and increased insulin resistance in non-diabetics with higher plasma concentrations of OC compounds, specifically DDE, and two compounds within the chlorinated cyclodiene class, trans-nonachlor (a component of the insecticide mixture chlordane) and oxychlordane (a metabolite of chlordane). A summary evaluation of epidemiological studies has indicated an overall association of some OC compounds in the blood with T2DM.13 Recently, an association of higher levels of OC compounds possessing less than five chlorines has been reported in adipose tissue from type 2 diabetics.14

An increasing percentage of the US adult population is suffering from T2DM. The present study was designed to investigate possible associations between the presence of T2DM and plasma concentrations of the OC compounds DDE, trans-nonachlor, and oxychlordane. Study subjects were adults of both sexes and either Caucasian or African American, and came from military populations or their family members at two US Air Force bases, Keesler Air Force Base (KAFB; Biloxi, MS, USA) and Wright-Patterson Air Force Base (WPAFB; Dayton, OH, USA). As members of the military and their families frequently change duty stations, military populations were considered representative of the US population and not reflective of specific geographical locations. Our approach was to identify associations between the presence of T2DM and the presence or concentration of OC compounds, as well as various clinical and demographic factors through multivariable logistic regression modeling, similar to modeling approaches we have used before.15 This approach has allowed us to identify novel interactions of OC compounds with obesity and with age. We also describe a solid phase extraction method for plasma that is faster and more efficient for extracting these OC compounds from human plasma than the traditional method.3

MATERIALS AND METHODS

Study Sample

The study sample consisted of 263 subjects, with 114 diabetics and 149 non-diabetics. Subjects included military personnel (active duty or retired) or their family members who were supplying fasting blood samples for measurement of glucose and a lipid panel as part of their standard care in the medical centers. Mississippi State University, KAFB and WPAFB Institutional Review Boards (IRBs) approved the study protocol. All information and samples supplied to Mississippi State University for analysis were de-identified. A convenience sample was comprised of plasma specimens from attendees from the previous 3 weeks at either KAFB Medical Center or WPAFB Medical Center. Samples and information from 251 subjects were obtained from pre-existing heparinized refrigerated blood samples at the hospital laboratories that were available on a single week as approved by the KAFB IRB or the WPAFB IRB, the majority of which were less than 1 week old, with only a few between 1 and 3 weeks. Plasma samples were obtained from KAFB and WPAFB on 8 April 2010 and 22 June 2010, respectively, and were frozen at <−20 °C for further analyses.The remaining 12 subjects were enrolled between 17 August 2010 and 9 November 2010 (following their written informed consent). All study subjects were either Caucasian or African American, and between the ages of 18 and 65 years. Subjects were designated as having T2DM either by electronic medical record ICD-9 codes for T2DM or by self-identification. Eighty-seven of the 114 T2DM subjects were on glycemic control medications. Use of lipid-controlling or other medications was not captured during data collection. Subjects from other racial groups or whose race could not be identified were excluded from this study.

Blood Samples and Clinical Chemistry

Plasma was derived from whole blood samples by centrifugation and tested for glucose and lipids (HDL, LDL, triglycerides, and total cholesterol) on the day of collection. The respective hospital laboratories performed the clinical chemistry analyses using either a Roche Hitachi Cobas 6000 (Indianapolis, IN, USA) or Siemens Dimension RxL Max (Iselin, NJ, USA) chemistry analyzer.

OC Compound Analysis

The extraction methodology described here for plasma was developed initially by DPX labs (Columbia, SC, USA) for analysis of pesticides in fruits and vegetables, and was modified in our laboratories for OC extraction from human plasma. An internal standard (100 μl) containing C13p,p’-DDT at 0.01 μg/ml and C13trans-nonachlor at 0.01 μg/ml in hexane was added to 1 ml of plasma. The sample was vortexed for 30 s and then was allowed to stand for 5 min to allow any binding of the internal standard to matrix to reflect accurate extraction efficiency. Acetonitrile (2 ml) was added to the sample to precipitate proteins. The mixture was vortexed for 1 min and centrifuged at 2,500 g for 10 min with the resultant supernatant decanted. Deionized water (2 ml) was added to the supernatant and the sample was vortexed for 30 s. The resultant mixture was aspirated into a DPX disposable pipette solid phase extraction column (DPX) for 30 s. The DPX column uses reverse-phase mechanisms to extract and isolate the OC compounds onto a styrene divinyl benzene sorbent stored in the tip of the column. Using a 10-ml syringe the aqueous portion was then forced out of the column and discarded. The sorbent was aspirated with a wash solution (0.5 ml, 33% acetonitrile/67% H2O) to remove any additional protein and artifacts from the column. OC compounds were eluted from the sorbent matrix by aspirating 10 s with 1 ml of 1:1 (v/v) ethyl acetate/hexane (EtAc/hex) and the eluent was then collected in a glass conical tube. This step was repeated twice. The combined eluents were then dried under a gentle stream of N2. The sample was resuspended with 100 μl of EtAc/hex (1:1, v/v) to optimize recovery and was dried. The sample was resuspended in a final volume of 100 μl in EtAc/hex (1:1, v/v) and was transferred to an autosampler vial for analysis. If the initial plasma sample was <1 ml, the final resuspension volume was adjusted accordingly. This solid phase extraction method improved the extraction efficiency from 81% to 90% for the OCs compared with our use of the traditional method with an accelerated solvent extractor as described in CDC method 6015.01 (CDC3).

Concentrations of the target OC compounds were determined by isotope dilution gas chromatography/mass spectrometry (GC/MS) using a modification of CDC method 6015.01 (CDC3). Extracts were analyzed using an Agilent Technologies 6,890 N GC connected to a 5975C triple-axis mass spectrometer. Two microliters of the concentrated extract were injected into the GC by splitless injection. Chromatographic separation of the individual compounds was achieved using a 30 m × 0.25 mm i.d. DB-5MS ([5%-phenyl]-methylpolysiloxane, 0.25 m film thickness) capillary column (J&W Scientific, Folsom, CA, USA) with helium as a carrier gas at a constant flow of 1 ml/min. The injector and transfer line temperatures were set at 275 °C and 270 °C, respectively. The initial column temperature, 100 °C, was held for 1 min, increased to 220 °C at 18 °C/min, held for 1 min, increased to 300 °C at 25 °C/min, held for 5 min and finally increased to 320 °C at 25 °C/min and held for 5 min.

A targeted mass analysis was performed in electron ionization (EI+) mode using single ion monitoring (SIM) for the analytes described above. The MS parameters included an ion source temperature of 230 °C and electron energy of 40 eV. Quantification and confirmation ions16 were monitored for each analyte and its respective isotopically labeled internal standard. Analyte peaks acquired by SIM were quantified using Agilent ChemStation software. Limits of detection (LOD) were 100 pg/ml plasma (14.8 ng/g lipid) for DDE, trans-nonachlor, and oxychlordane. Pooled samples of human plasma (purchased from Sigma, St. Louis, MO, USA) that did not contain detectable concentrations of the target compounds were spiked with various concentrations of OC standards and carried through the extraction procedures to determine standard curves for each OC compound of interest. Isotopically labeled internal standards were used to account for variations in extraction efficiency. Recoveries for oxychlordane, trans-nonachlor, and DDE were 87%, 89%, and 90%, respectively, and are comparable to methods used by CDC for the NHANES.16,17 Extraction ion masses along with the retention times used for analysis are provided in Table 1. Areas under the curve were converted to ng/ml plasma using a standard curve of each of the target OC compounds.

Table 1 GC retention times and mass spectral ions for electron-impact ionization GC/MS detection of organochlorine compounds.

Lipid Adjustment

Plasma OC compound levels were used as a proxy for adipose tissue concentration/body burden18 and were adjusted for plasma lipid levels. Plasma lipid concentrations for each subject were calculated from their lipid panel (using the total cholesterol and triglyceride values) and expressed as g lipid/l of plasma.19,20 Concentrations of DDE, trans-nonachlor, and oxychlordane were lipid adjusted and were expressed as ng analyte/g lipid.

Statistical Analysis

All data analyses were performed using SAS for Windows version 9.2 (SAS Institute, Cary, NC, USA). A t-test, using Proc TTEST, was conducted to compare continuous variables between T2DM and non-diabetics. A χ2-analysis, using PROC FREQ, was conducted to compare categorical variables between T2DM and non-diabetics. A univariable logistic regression analysis, using PROC LOGISTIC, was conducted to measure the strength of associations between the occurrence of T2DM and various demographic and clinical laboratory test values. Univariable logistic regression was also used to investigate potential associations between T2DM presence and the plasma concentration for each OC compound. Total cholesterol and triglycerides were not included in the analysis, as they were used to calculate the plasma lipid concentration for standardizing plasma OC concentrations per gram lipid. LDL cholesterol also was not included, as its value was calculated from total cholesterol, triglycerides, and HDL cholesterol. Because of the small proportion of subjects who possessed detectable concentrations of trans-nonachlor and oxychlordane, these two OC compounds were analyzed as dichotomous variables with the respective pesticide classified as absent (i.e., nondetectable, ND) or present (i.e., detectable). Samples with no detectable DDE were assigned a value of 0 ng/gm lipid. DDE was then log transformed by first adding 1 to all DDE ng/g lipid values and taking the base 10 logarithm.

Following the univariable analysis, multivariable logistic regression modeling of T2DM occurrence and the demographic and clinical variables, as well as each one of the OC compounds, was done to determine the association, if any, of the OC compounds with T2DM. First, a basic multivariable logistic regression model without any of the OC compound variables was constructed. To assess co-linearity among explanatory variables, Spearman rank correlation among the variables was determined by PROC CORR. If the Spearman correlation coefficient was >0.8, the two variables would be evaluated on the significance of the association with T2DM, number of missing observations, and relationship with other explanatory variables to determine which one would be included in the analysis. The basic multivariable model was developed by starting with the demographic and clinical candidate variables, and considered main effects only, that is, no interaction terms were included at this step. A manual backward selection process was performed in which the variable with the highest P-value was removed and the model was refit. This process was continued until only demographic and clinical variables with a P-value ≤0.05 remained. All of the possible two way interactions between these remaining variables were then added to the model and the manual backward selection process was repeated until only variables or two-way interaction terms with P-value ≤0.05 remained. Individual variables that were part of an interaction term that was significant were retained in the model regardless of their individual P-value. This model was termed the basic multivariable logistic regression model or basic multivariable model. To assess the possible association of each of the OC compounds with T2DM, each of the three OC compounds was added individually to the basic multivariable model. Two-way interaction terms between the OC compound and the other variables in the basic multivariable model were also included. A manual backward selection process was performed in which the variable or two-way interaction term with the highest P-value was removed and the model was refit until only variables or two-way interaction terms with a P-value ≤0.05 remained.

Laboratory Safety

All appropriate chemical safety procedures were employed and universal precautions were used for procedures involving human samples.

RESULTS

Characteristics of the Study Sample

The study sample of 263 subjects had the following numbers of subjects with detectable concentrations of OC compounds: 193 for DDE (73%), 49 for trans-nonachlor (19%) and 80 for oxychlordane (30%). Comparing the arithmetic means to obtain an overview of the study sample, there was 2.3 times more DDE and 1.4 times more trans-nonachlor in diabetics than in non-diabetics, but there was no difference between the groups for oxychlordane (Table 2). An increasingly greater proportion of diabetics comprises the group of subjects possessing more of the three OC compounds with 41% (23/56), 47% (66/148), 59% (44/75), and 62% (13/21) possessing 0, 1, 2, and 3 detectable OC compound residues, respectively.

Table 2 Lipid-adjusted detectable concentrations of DDE, trans-nonachlor, and oxychlordane.

Characteristics of the Study Sample used in Statistical Modeling

The characteristics of the study sample are presented in Table 3. Height and weight were used to calculate body mass index (BMI). The mean fasting glucose value (139.3 mg/dl) for the diabetic subjects is only slightly greater than the fasting glucose value used to define diabetes (i.e., >125 mg/dl), and reflects the likelihood that the diabetics were being treated with diet and/or medication at the time of sample collection. Of the 114 diabetic subjects, 87 were known to be taking some type of medication for their diabetes. The non-diabetics had a fasting glucose level appropriate for non-diabetics (94.8 mg/dl). Diabetics were noted to have lower total cholesterol (156.2 versus 189.4 mg/dl P<0.001) and LDL cholesterol (89.7 versus 117.9 mg/dl P<0.001) than non-diabetics, suggesting that a significant proportion of diabetic subjects were receiving treatment for dyslipidemia A higher proportion of subjects from the KAFB Medical Center were diabetic than those from the WPAFB Medical Center.

Table 3 Demographic and clinical laboratory information for the study sample used to construct the multivariable model (n=263).

Univariable Association of T2DM with Demographic and Clinical Variables and OC Compounds

Associations between the presence of T2DM and each of the demographic variables, clinical variables and plasma OC compound concentrations were determined by logistic regression (Table 4). Fasting blood glucose concentrations were not included in the analysis, as these values were used to diagnose T2DM. Statistically significant associations (P≤0.05) between the presence of T2DM and increasing age, base location (KAFB had more T2DM than WPAFB), increasing BMI, race (African Americans displayed more T2DM than Caucasians), decreasing HDL, increasing logDDE, and detectable concentrations of trans-nonachlor were observed (Table 4). The presence of oxychlordane was not significantly associated with T2DM.

Table 4 Univariable logistic regression analysis.

Multivariable Logistic Regression Models

Spearman rank correlation coefficients were determined for all the candidate continuous variables to check for collinearity. None of the continuous variables were noted to be correlated and hence all were retained for consideration in the basic multivariable model. The basic multivariable model constructed as described in the Materials and Methods section is shown in Table 5. This model showed an association of T2DM with the interaction term for BMI and race (BMI*race in Table 5). Because of BMI’s interaction with race in this model, the interaction was examined in greater detail (Table 6). BMI was not statistically significantly associated with T2DM in African Americans but was in Caucasians (Table 6). The odds ratios for African Americans versus Caucasians are shown for three separate BMI values (arbitrarily selected to be a normal value 23.5 kg/m2, an overweight value 27.5 kg/m2, and an obese value 31.5 kg/m2, Table 6). All three odds ratios are >1, indicating African Americans are more likely to be diabetic than Caucasians at the three BMIs assessed. The odds ratios decreased as the BMIs progressed from normal to obese.

Table 5 Basic multivariable logistic regression model of the association of type 2 diabetes with demographic variables and clinical variables.
Table 6 Odds ratios and 95% CIs

A multivariable model was then constructed for each of the three OC compounds by adding an individual OC compound and all possible two-way interactions containing the OC compound to the basic multivariable model (shown in Table 5) as described in the Materials and Methods. The multivariable models constructed with either oxychlordane or trans-nonachlor showed no association of these two OC compounds with T2DM and were identical to the basic multivariable model. The model constructed by adding logDDE and its two-way interaction terms to the basic multivariable model is shown in Table 7. This model contained logDDE as well as interaction terms for logDDE and age, and for logDDE and BMI (Table 7). Odds ratios, 95% confidence intervals (CIs), and P-values were determined for age and BMI using the default setting of logDDE=mean logDDE (i.e., 1.6108). The interaction terms age and logDDE (age*logDDE in Table 7) and logDDE and BMI (BMI*logDDE in Table 7) were both significantly associated with T2DM.

Table 7 Multivariable logistic regression model of the association of type 2 diabetes with demographic variables, clinical variables, and logDDE.

Because of logDDE’s interaction with both age and BMI in this model, the interactions were examined in greater detail by calculating odds ratios and 95% CIs for increasing logDDE at selected values of age and BMI (Table 8). Values across rows going from a lower BMI to a higher BMI for an individual age show decreasing odds ratios for T2DM with each one log increase in logDDE for all of the ages examined. Likewise, values going down columns from a younger age to an older age for an individual BMI show increasing odds ratios for T2DM with each one log increase in logDDE for each BMI examined. A similar pattern of results was found when DDE values not standardized by plasma lipid concentration were analyzed.

Table 8 Odds ratios and 95% CIs for association of increasing logDDE and type 2 diabetes at selected values of age and BMI.

DISCUSSION

The study sample was obtained from two geographically different US Air Force medical facilities, one in the southern United States (KAFB) and the other in the Midwestern United States (WPAFB). Military members and their families typically move every 3–4 years on average; hence, it is assumed that the majority of the subjects had probably lived in multiple locations and the results were not expected to reflect particular geographic regions in the United States. Nevertheless, base displayed a statistically significant difference in prevalence of T2DM in the univariable analysis (Table 4), with KAFB subjects having a greater proportion of diabetics than WPAFB subjects, but the reasons for this difference are unknown. Further, base had a significant effect when adjusted for all the demographic factors placed into the multivariable model (Table 5), but the study was not designed to distinguish the cause for the difference in T2DM prevalence in the populations served by the two bases.

The other factors that were associated with T2DM, that is, age, BMI, African American race, and decreasing HDL cholesterol, were as expected. We did not have information on the family history of diabetes and thus were not able to analyze the effect of a genetic component on the prevalence of T2DM. Military personnel and their dependents will most likely have greater access to medical care than the population of the United States as a whole, such as the population in the NHANES study.11 In addition, African Americans made up 22% of the study population, a larger amount than reported in some previous studies.11,21 It is understood that a convenience sample with a cross-sectional study design can be subject to inadvertent selection bias.

Arithmetic means for DDE in diabetic and non-diabetic subjects in the present study were 317.5 and 136.9 ng/g lipid, respectively, and for the entire study sample was 227.2 ng/g lipid; thus, the values observed in this study sample were in a similar range to the NHANES level from a national sample, as expected. For both DDE and trans-nonachlor, the lower end of the range of values was similar between diabetics and non-diabetics, while the upper end of the range was considerably higher in diabetics than in non-diabetics, indicating that some of the diabetics had very high plasma concentrations. Ranges were the same in both groups for oxychlordane. The percentage of the overall NHANES samples that were non-detectable for oxychlordane was not stated; however, many demographic groups were indicated lower than the LOD for over 50% of the study sample tested, so it is assumed that the percentage nationwide of people carrying this compound is relatively low. The latest NHANES report did not include these three OC compounds among the reported analytes.22 The high percentage of our study sample that had detectable concentrations of at least one of the 3 OC compounds agrees with a previous study in Spain on healthy populations.21

Counterintuitively, it appeared that elevated total cholesterol level was protective of T2DM development. However, because a common first step in T2DM management is alteration of the patient’s diet and exercise regimen, a presumed change in diet, along with treatment with cholesterol-lowering statins, are the more likely causes of the lower cholesterol in the T2DM subjects. Recommendations at the time the samples were collected for treatment of hyperlipidemia contain lower target values for LDL cholesterol in diabetics than in non-diabetics, because diabetes is such a strong risk factor for atherosclerosis. Thus, diabetics with lower LDL cholesterol values (compared with non-diabetics) will still meet recommendations for dyslipidemia treatment.

Logistic regression modeling for both main effects and interactions was performed similar to the approach we have previously used to identify associations of risk factors with atherosclerosis.15 The basic model developed here included age, race, BMI (all known factors associated with T2DM), and Air Force base. Despite the association seen for trans-nonachlor and T2DM observed with the univariable modeling, no association was observed when the relevant demographic/clinical factors were adjusted for, indicating that trans-nonachlor was not associated with T2DM independently of the known risk factors.

An unexpected observation arising from the modeling was the significant association of BMI and T2DM in Caucasians but not in African Americans. Furthermore, African Americans had a greater association with T2DM than Caucasians with the greatest odds ratio at the low (normal) BMI value and decreasing as the BMI increased through moderate (overweight) and high (obese) values. These odds ratios suggest that obesity is a risk factor for the Caucasian population, as is widely known,23 but that the higher prevalence of T2DM in African Americans may be the result of factors other than increased adiposity, such as genetic profile or pattern of disease progression. The results suggest that more study of risk factors in minority populations would be useful.

A third observation in this study related to the main question being asked, that is, the association of OC compounds with T2DM. Neither of the chlordane-associated compounds, trans-nonachlor or oxychlordane, showed an association with T2DM in the study sample when adjusted for the other demographic and clinical factors, even though both did in the NHANES analyses.11 This negative finding in the current study may have been the result of our study sample being smaller than that of the NHANES study.3

As noted in Table 8, which presents odds ratios for T2DM with increasing logDDE for each age/BMI group, there is a consistent trend of decreasing odds ratios with increasing adiposity for each of the age classes, indicating greater association of T2DM with increasing DDE concentrations in leaner individuals when adjusted for race. There is also a consistent trend of increasing odds ratios with increasing age within each of the BMI classes, indicating an increasing risk of T2DM with increasing DDE with aging regardless of adiposity. Considering both risk factors together, DDE seems to be having more of an influence on the odds ratio in older people of normal weight and relatively less of an influence on the odds ratio in those who are overweight or obese, especially when young. In contrast to our results, Gasull et al.24 found no association of diabetes with serum DDE or DDT (adjusted for sex, age, and BMI) in a Catalonian population. An additional study of a Spanish population indicated an association of increased levels of DDE in adipose tissue of non-obese individuals with T2DM, but not in obese individuals; however, these results were based on only 34 diabetics out of a study sample of 386 subjects.25 Both of these studies were done on Spanish populations presumably of entirely European ancestry dramatically different than our study population, which contained 22% African Americans. Lee et al.11 reported the association of diabetes with a sum of persistent organic pollutants, which included DDE, and that association tended to be stronger in younger individuals and obese subjects. The reason for the difference in results regarding obesity between their study and ours is unknown, although we examined DDE individually and Lee et al.11 examined DDE along with five other persistent organic pollutants. Moreover, the population in this study had ready and at times required access to health care with the likelihood of earlier detection and treatment of T2DM and dyslipidemia than the NHANES study population. Airaksinen et al.26 also reported a weak association of diabetes with DDE (P=0.087) but only in obese individuals. This study investigated individuals born in Helsinki from 1934 until 1944 and was a much different group of subjects with respect to age range and other demographics than we studied.

This study has a small sample size for multivariable modeling (n=263) when compared with several well-known studies that developed models for the prediction of incident T2DM.27, 28, 29 Systolic blood pressure and family history of diabetes, two variables included in many models predicting incident T2DM, were not available in the clinical information we were able to obtain. Fasting glucose concentrations were not included in the model, as these values were used in diagnosis of diabetes. Lipid panels were used to calculate serum DDE/ng of lipid and hence were not included as a separate variable in the model. Nevertheless, there is still an association of DDE with T2DM in this study sample. Its role, if any, in T2DM remains to be determined. Further, studies prospective in nature with larger sample sizes and more complete clinical information would be of interest, as would studies on mechanisms by which DDE might alter lipid metabolism and/or influence insulin resistance. In addition, the possibility of reverse causation has recently been suggested for dioxin and by extension all persistent organic pollutants.30 This hypothesis developed with dioxin data proposes that some alteration in lipid metabolism occurs as a result of diabetes, which elevates the serum levels of dioxin. The reverse causation hypothesis could be the reason for some of our observed results, but our study was designed to assess association, not causation or reverse causation.

In conclusion, this study found that increased plasma concentrations of DDE are associated with increased prevalence of T2DM, with the greatest association in older people of normal BMI. An increase in persistent organic pollutants has been detected in cross-sectional analyses (the majority of studies published thus far) and is responsible for the association of elevated levels of persistent organic pollutants with T2DM prevalence reported in the majority of these studies. Although the association of elevated levels of persistent organic pollutants with the increased prevalence of T2DM has been reported multiple times, including here, the mechanism responsible for the association has not been determined.

References

  1. 1

    Cowie CC, Rust KF, Ford ES, Eberhardt MS, Byrd-Holt DD, Li C et al. Full accounting of diabetes and pre-diabetes in the U.S. population in 1988-1994 and 2005-2006. Diabetes Care 2009: 32: 287–294.

    Article  Google Scholar 

  2. 2

    Go AS, Mozaffarian D, Roger VL, Benjamin EJ, Berry JD, Borden WB et al. Heart disease and stroke statistics—2013 update: a report from the American Heart Association. Circulation 2013: 127: e6–e245.

    PubMed  Google Scholar 

  3. 3

    CDC. Fourth national report on human exposure to environmental chemicals. Centers for Disease Control and Prevention. http://www.cdc.gov/exposurereport/pdf/fourthreport.pdf. 2009. Accessed November 20, 2013.

  4. 4

    Adeghate E, Schattner P, Dunn E . An update on the etiology and epidemiology of diabetes mellitus. Ann NY Acad Sciences 2006; 1084: 1–29.

    Article  Google Scholar 

  5. 5

    Ennaceur S, Gandoura N, Driss MR . Distribution of polychlorinated biphenyls and organochlorine pesticides in human breast milk from various locations in Tunisia: concentrations of contamination, influencing factors, and infant risk assessment. Environ Res 2008; 108: 86–93.

    CAS  Article  Google Scholar 

  6. 6

    Bidleman TF, Leone AD . Soil–air exchange of organochlorine pesticides in the Southern United States. Environ Pollut 2003; 128: 49–57.

    Article  Google Scholar 

  7. 7

    Jiang YF, Wang XT, Jia Y, Wang F, Wu MH, Sheng GY et al. Occurrence, distribution and possible sources of organochlorine pesticides in agricultural soil of Shanghai, China. J Hazard Mater 2009; 170: 989–997.

    CAS  Article  Google Scholar 

  8. 8

    Rignell-Hydbom A, Rylander L, Hagmar L . Exposure to persistent organochlorine pollutants and type 2 diabetes mellitus. Hum Exp Toxicol 2007; 26: 447–452.

    CAS  Article  Google Scholar 

  9. 9

    Ukropec J, Radikova Z, Huckova M, Koska J, Kocan A, Sebokova E et al. High prevalence of prediabetes and diabetes in a population exposed to high concentrations of an organochlorine cocktail. Diabetologia 2010; 53: 899–906.

    CAS  Article  Google Scholar 

  10. 10

    Turyk M, Anderson H, Knobleloch L, Imm P, Persky V . Organochlorine exposure and incidence of diabetes in a cohort of Great Lakes sport fish consumers. Environ Health Perspect 2009; 117: 1076–1082.

    CAS  Article  Google Scholar 

  11. 11

    Lee DH, Lee IK, Song K, Steffes M, Toscano W, Baker BA et al. 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 2006; 29: 1638–1644.

    CAS  Article  Google Scholar 

  12. 12

    Lee DH, Lee IK, Jin SH, Steffes M, Jacobs DR Jr . Association between serum concentrations of persistant organic pollutants and insulin resistance among nondiabetic adults: results from the National Health and Nutrition Examination Survey 1999-2002. Diabetes Care 2007; 30: 622–628.

    CAS  Article  Google Scholar 

  13. 13

    Taylor KW, Novak RF, Anderson HA, Birnbaum LS, Blystone C, DeVito M et al. Evaluation of the association between persistent organic pollutants (POPs) and diabetes in epidemiological studies: a national toxicology program workshop review. Environ Health Perspect 2013; 121: 774–783.

    Article  Google Scholar 

  14. 14

    Kim K-S, Lee Y-M, Kim SG, Lee I-K, Lee H-J, Kim J-H et al. Associations of organochlorine pesticides and polychlorinated biphenyls in visceral vs. subcutaneous adipose tissue with type 2 diabetes and insulin resistance. Chemosphere 2014; 94: 151–157.

    CAS  Article  Google Scholar 

  15. 15

    Coombes RH, Crow JA, Dail MB, Chambers HW, Wills RW, Bertolet BD et al. Relationship of human paraoxonase-1 serum activity and genotype with atherosclerosis in individuals from the Deep South. Pharmacogenet Genomics 2011; 21: 867–875.

    CAS  Article  Google Scholar 

  16. 16

    Barr JR, Maggio VL, Barr DB, Turner WE, Sjödin A, Sandau CD et al. New high-resolution mass spectrometric approach for the measurement of polychlorinated biphenyls and organochlorine pesticides in human serum. J Chromatogr B Analyt Technol Biomed Life Sci 2003; 794: 137–148.

    CAS  Article  Google Scholar 

  17. 17

    Sandau CD, Sjödin A, Davis MD, Barr JR, Maggio VL, Waterman AL et al. Comprehensive solid-phase extraction method for persistent organic pollutants. Validation and application to the analysis of persistent chlorinated pesticides. Anal Chem 2003; 75: 71–77.

    CAS  Article  Google Scholar 

  18. 18

    Schisterman EF, Whitcomb BW, Louis GM, Louis TA . Lipid adjustment in the analysis of environmental contaminants and human health risks. Environ Health Perspect 2005; 113: 853–857.

    CAS  Article  Google Scholar 

  19. 19

    Phillips DL, Pirkle JL, Burse VW, Bernert JT Jr, Henderson LO, Needham LL . Chlorinated hydrocarbon levels in human serum: effects of fasting and feeding. Arch Environ Contam Toxicol 1989; 18: 495–500.

    CAS  Article  Google Scholar 

  20. 20

    Bergonzi R, De Palma G, Tomasi C, Ricossa MC, Apostoli P . Evaluation of different methods to determine total serum lipids for normalization of circulating organochlorine compounds. Int Arch Occup Environ Health 2009; 82: 1241–1247.

    CAS  Article  Google Scholar 

  21. 21

    Jakszyn P, Goñi F, Etxeandia A, Vives A, Millan E, López R et al. Serum concentrations of organochlorine pesticides in healthy adults from five regions of Spain. Chemosphere 2009; 76: 1518–1524.

    CAS  Article  Google Scholar 

  22. 22

    Centers for Disease Control and Prevention. NHANES 2011/12 report http://www.n.cdc.gov/nchs/nhanes/search/nhanes11_12.aspx. Accessed 20 November 2013.

  23. 23

    CDC/Diabetes Data & Trends. Centers for Disease Control and Prevention: National Diabetes Survelliance System. http://www.cdc.gov/diabetes/statistics. 2010. Accessed 20 November 2013.

  24. 24

    Gasull M, Pumarega J, Téllez-Plaza M, Castell C, Tresserras R, Lee DH et al. Blood concentrations of persistent organic pollutants and prediabetes and diabetes in the general population of Catalonia. Environ Sci Technol 2012; 46: 7799–7810.

    CAS  Article  Google Scholar 

  25. 25

    Arrebola JP, Pumarega J, Gasull M, Fernandez MF, Martin-Olmedo P, Molina-Molina JM et al. Adipose tissue concentrations of persistent organic pollutants and prevalence of type 2 diabetes in adults from Southern Spain. Environ Res 2013; 122: 31–37.

    CAS  Article  Google Scholar 

  26. 26

    Airaksinen R, Rantakokko P, Eriksson JG, Blomstedt P, Kajantie E, Kiviranta H . Association between type 2 diabetes and exposure to persistent organic pollutants. Diabetes Care 2011; 34: 1972–1979.

    CAS  Article  Google Scholar 

  27. 27

    Schmidt MI, Duncan BB, Bang H, Pankow JS, Ballantyne CM, Golden SH et al. Identifying individuals at high risk for diabetes: the Atherosclerosis Risk in Communities study. Diabetes Care 2005; 28: 2013–2018.

    Article  Google Scholar 

  28. 28

    Stern MP, Williams K, Haffner SM . Identification of persons at high risk for type 2 diabetes mellitus: do we need the oral glucose tolerance test? Ann Intern Med 2002; 136: 575–581.

    Article  Google Scholar 

  29. 29

    Wilson PW, Meigs JB, Sullivan L, Fox CS, Nathan DM, D’Agostino RB Sr . Prediction of incident diabetes mellitus in middle-aged adults: the Framingham Offspring Study. Arch Intern Med 2007; 167: 1068–1074.

    Article  Google Scholar 

  30. 30

    Kerger BD, Scott PK, Pavuk M, Gough M, Paustenbach DJ . Re-analysis of Ranch Hand study supports reverse causation hypothesis between dioxin and diabetes. Crit Rev Toxicol 2012; 42 (8): 669–687.

    CAS  Article  Google Scholar 

Download references

Acknowledgements

We express appreciation to Dr. Matthew Ross, Dr. George Howell III, Lauren Mangum and MaryBeth Dail of Mississippi State University for their support in aspects of sample and data analysis and interpretation. We also thank Ashley Iovieno at Keesler Air Force Base and Tina Thomason at Wright-Patterson Air Force Base for their support in sample collection and transport. This work was supported by the Center for Environmental Health Sciences, College of Veterinary Medicine, Mississippi State University and Keesler Air Force Base. Samples and clinical information were provided by Keesler Air Force Base and Wright-Patterson Air Force Base. The views expressed in this material are those of the authors and do not reflect the official policy or position of the US Government, the Department of Defense, or the United States Air Force. The work herein was performed under United States Air Force Surgeon General approved Clinical Investigation No. FKE-20100017H, FKE-20100018E, and FWP-20100035H. The voluntary, fully informed consent of the subjects used in this research was obtained as required by 32 CFR 219 and AFI 40–402, Protection of Human Subjects in Biomedical and Behavioral Research. This is Center for Environmental Health Sciences publication 131.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Janice E Chambers.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Eden, P., Meek, E., Wills, R. et al. Association of type 2 diabetes mellitus with plasma organochlorine compound concentrations. J Expo Sci Environ Epidemiol 26, 207–213 (2016). https://doi.org/10.1038/jes.2014.69

Download citation

Keywords

  • analytical methods
  • biomonitoring
  • empirical/statistical models
  • endocrine disruptors
  • epidemiology
  • pesticides

Further reading

Search

Quick links