Eur J Nutr (2016) 55:1011–1020 DOI 10.1007/s00394-015-0915-0
ORIGINAL CONTRIBUTION
The modifying effect of vitamin C on the association between perfluorinated compounds and insulin resistance in the Korean elderly: a double‑blind, randomized, placebo‑controlled crossover trial Jin Hee Kim1 · Hye Yin Park2 · Jung Dae Jeon3 · Younglim Kho3 · Seung‑Kyu Kim4 · Min‑Seon Park5 · Yun‑Chul Hong2,6
Received: 5 December 2014 / Accepted: 25 April 2015 / Published online: 5 May 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract Purpose There is limited evidence whether environmental exposure to perfluorinated compounds (PFCs) affects insulin resistance (IR) and whether vitamin C intake protects against the adverse effect of PFCs. This study was carried out to investigate the effect of PFCs on IR through oxidative stress, and the effects of a 4-week consumption of vitamin C supplement compared placebo on development of IR by PFCs. Methods For a double-blind, community-based, randomized, placebo-controlled crossover intervention of vitamin C, we assigned 141 elderly subjects to both vitamin C and placebo treatments for 4 weeks. We measured
Electronic supplementary material The online version of this article (doi:10.1007/s00394-015-0915-0) contains supplementary material, which is available to authorized users. * Yun‑Chul Hong
[email protected] 1
Department of Environmental Health, Graduate School of Public Health, Seoul National University, Seoul, Republic of Korea
2
Department of Preventive Medicine, Seoul National University College of Medicine, Seoul 110‑799, Republic of Korea
3
Department of Health, Environment and Safety, Eulji University, Sungnam, Republic of Korea
4
Department of Marine Science, College of Natural Sciences, University of Incheon, Incheon, Republic of Korea
5
Department of Family Medicine, Seoul National University Hospital, Seoul, Republic of Korea
6
Institute of Environmental Medicine, Seoul National University Medical Research Center, Seoul, Republic of Korea
serum levels of PFCs to estimate PFC exposures and urinary levels of malondialdehyde (MDA) and 8-hydroxy-2′deoxyguanosine (8-OHdG) for oxidative stress. We also measured levels of fasting glucose and insulin and derived the homeostatic model assessment (HOMA) index to assess IR. Results Perfluorooctane sulfonate (PFOS) and perfluorododecanoic acid (PFDoDA) levels were found to be positively associated with HOMA index at the baseline and after placebo treatment. Risks of IR for the top decile of PFOS and PFDoDA exposures were significantly elevated compared with those with lower PFOS and PFDoDA exposures (both, P 0.30). Furthermore, PFOS and PFDoDA levels were also significantly associated with MDA and 8-OHdG levels, and MDA levels were positively associated with HOMA index. Conclusion PFOS and PFDoDA exposures were positively associated with IR and oxidative stress, and vitamin C supplementation protected against the adverse effects of PFOS and PFDoDA on IR. Keywords Perfluorinated compounds · Insulin resistance · Oxidative stress · Vitamin C supplementation
Introduction Perfluorinated compounds (PFCs) are persistent organic pollutants that do not occur naturally [1, 2]. PFCs are found in a variety of consumer and industrial products such as cookware and food packaging and are accumulated in the body due to their chemical stabilities and resistance to biodegradation [1, 3, 4]. Because of their ubiquitous
13
1012
exposure in modern living environment, PFCs are commonly detected in human samples [5, 6]. Based on a study for the general population in Busan, South Korea [7], perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) were the dominant PFCs detected in Korean serum samples with mean concentrations of 13 and 8.4 ng/mL, respectively, similar or a little higher compared with concentrations in the other Korean study [8]. PFCs have been reported to exhibit hormone disturbances in addition to hepatotoxicity, immunotoxicity, developmental toxicity, and tumorigenic potential in crosssectional epidemiologic studies as well as rat models [9]. Insulin resistance (IR) has been regarded as an important health problem because it affects the development of type 2 diabetes and metabolic syndrome [10, 11]. The number of patients with diabetes or metabolic syndrome has been growing very rapidly and is expected to further increase worldwide in the future [10, 12, 13]. Recently environmental chemicals including PFOS and PFOA were reported to be associated with development of insulin resistance and metabolic syndrome or diabetes mellitus mortality [14, 15]. However, there is limited evidence whether environmental exposure to PFCs contributes to the development of IR [14, 16, 17], although there have been reports on the relations between PFC exposure and hormone disturbances or mortality from diabetes mellitus [5, 6, 15]. Oxidative stress has been reported to play an important role in a variety of pathological conditions including diabetes mellitus. It is defined as an impaired balance between free radical production and antioxidant capacity leading to the presence of excess oxidative products. Recently, a few studies have demonstrated that exposure to environmental pollutants, such as air pollutants and phthalates, affects IR through the oxidative stress pathway [18–23]. Even though oxidative stress is suggested as a mechanism contributing to the development of IR, little information is available regarding the role of oxidative stress induced by community levels of exposure to PFCs. Vitamin C, a well-known antioxidant, was known to play an important role in antioxidation by capture of ·H and ·OH radicals [24]. Supplementation of vitamin C as an antioxidant has been reported to decrease oxidative stress produced in the body [25, 26]. Because oxidative stress plays an important role in pathological conditions such as diabetes mellitus and insulin resistance, intake of vitamin C as an antioxidant may reduce oxidative stress and the adverse effect of PFCs on IR, if PFCs affect IR through the oxidative stress mechanism. In this community clinical trial, we confirmed the association of environmental exposure to PFCs with oxidative stress and IR and evaluated the effect of vitamin C as an antioxidant on oxidative stress and IR produced by environmental exposure to PFCs in the Korean elderly.
13
Eur J Nutr (2016) 55:1011–1020
Subjects and methods Study population and treatment strategies The vitamin C intervention study in the elderly was launched in June 2011 to evaluate the effect of vitamin C intervention on relations between environmental exposure and health in the elderly population. One hundred and twenty of sample size was calculated to have power more than 80 % at 5 % significance level for detection of mean difference at 5 pmol/L of insulin levels in a 2-by-2 crossover design using PASS program version 12 (NCSS Institute Inc., Kaysville, Utah, USA) [27]. Target subject number for recruitment (140 subjects) was finally calculated considering 16.3 % of diabetes prevalence surveyed in Korean Elderly Environmental Panel (KEEP) study [22]. From its start to November 2011, this study recruited a total of 141 people aged 60 or over without a history of serious cardiovascular complications such as ischemic heart diseases or stroke for community-based randomized crossover clinical trial. For a follow-up survey, all participants were examined to January 2012. Because all subjects recruited in our study were followed up, we have no dropout, and any adverse event was not shown during whole study period. Participants visited a community welfare center in the Seongbuk-gu area in Seoul, Korea, for three times of medical examinations (first visit for baseline measurement and second and third crossover visits after placebo or vitamin C treatment) (Fig. 1). Total participants were randomly divided into two groups using PROC SURVEYSELECT statement with random number function (RANUNI) in SAS 9.4 (SAS Institute Inc., NC, USA). For one group (vitamin C–placebo group, n = 71), vitamin C and placebo were supplemented sequentially, each for 4 weeks, and there was a 2-week non-treatment period (determined based on 6–8 h of half-life for vitamin C) between vitamin C and placebo supplementation to flush out the effect of the first treatment (total of 10 weeks—4 + 2 + 4). The sequence of supplementation was reversed for the other group (placebo–vitamin C group, n = 70). Dosage and length of vitamin C treatment were derived from a previously reported research on the effect of high-dose vitamin C supplementation on insulin sensitivity and glucose homeostasis [27]. All participants took a vitamin C (or placebo) once a day on prescription. For vitamin C intake, one tablet with 1000 mg of vitamin C (C-ALL TAB.; Ildong Pharmaceutical Co., Ltd.) was supplemented once a day during the 4-week period. For the placebo intake, a tablet with 10.7 mg of sodium citrate, 7.4 mg of magnesium stearate, and 1047.0 mg of lactose, packed in an identical manner to the vitamin C, was supplemented. Medical examinations were conducted on the day of first visit and at the day after completing placebo and vitamin C treatment for 4 weeks. All steps including random assignments of sequence and packaging of vitamin C and placebo
Eur J Nutr (2016) 55:1011–1020
1013
Fig. 1 Design for vitamin C intervention study
were blinded until the study was finished. A third person, not informed of researchers’ objectives or in direct contact with subjects, packed vitamin C and placebo tablets according to the divided groups. Another third person, not knowing whether the tablets were vitamin C or placebo, handed the identically packed tablets to the participants. Fasting blood and morning urine samples were obtained between 9 a.m. and 10 a.m. of each visit day from each individual. Blood and urine samples were collected in serum separation tube (SST) and 50-mL falcon tube, respectively. All bloods were centrifuged at 3000 rpm for 10 min, and then, serums were transferred to 1 mL of cryotube with screw cap. Serum and urine were stored at −80 and −20 °C, respectively, until they were analyzed. We also obtained detailed information of the participants, using a structured questionnaire which included demographics, lifestyle habit, and medical history, and using a food frequency questionnaire related to dietary habit. Each study participant provided written informed consent. The study protocol was conducted in accordance with guidelines laid down in the Declaration of Helsinki, and all procedures were approved by the Institutional Review Board at Seoul National University Hospital (IRB no., C-1105-043361) and registered in Clinical Research Information Service
(http://www.who.int/ictrp/network/primary/en/index.html), a primary registry that participates in the WHO International Clinical Trial Registry Platform (Identifier; KCT0000749). Measurement of serum PFCs We measured serum levels of four perfluoroalkyl sulfonates and eleven perfluorocarboxylic acids. Target PFCs were perfluorobutane sulfonate (PFBS), perfluorohexane sulfonate (PFHxS), PFOS, perfluorodecane sulfonate (PFDS), perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), PFOA, perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluoroundecanoic acid (PFUnDA), perfluorododecanoic acid (PFDoDA), perfluorotridecanoic acid (PFTrDA), and perfluorotetradecanoic acid (PFTeDA). Fifteen native PFC compounds and nine mass-labeled compounds (13C4–PFHxS, 13C4–PFOS, 13C4–PFBA, 13C4–PFHxA, 13 C4–PFOA, 13C4–PFNA, 13C4–PFDA, 13C4–PFUnDA, and 13 C4–PFDoDA) were purchased from Wellington Laboratories (Guelph, ON, Canada). Mass-labeled compounds were used as internal standards for recovery determination and correction of individual native compounds.
13
1014
Sample preparation for PFCs were performed according to the methods outlined in Kim et al. [8] with minor modifications. Briefly, 20 µL of labeled internal standards were spiked into 200 µL serum and were added 500 µL of 0.1 % formic acid in water. Then, serum samples were loaded onto a solid-phase extraction (SPE) cartridge (WAX; 1 cc, 30 mg; Waters, Milford, MA). Target PFC compounds on the SPE cartridge were eluted using 4 mL 0.1 % ammonium hydroxide in methanol, evaporated to dryness, and reconstituted to 0.2 mL methanol. Bovine serum samples, spiked with native and internal standards (or not spiked by native compounds), were prepared as quality control samples, allowing for the determination of accuracy and precision, and blank contamination, respectively. Bovine serum was also used as a matrix for calibration. Target analytes were identified and quantified using high-performance liquid chromatography (Series 1100, Agilent Technologies, Palo Alto, CA) coupled with an API 4000 triple–quadruple mass spectrometer (Applied Biosystems, Foster city, CA). PFCs in injected sample were separated on a 2.0 × 150 mm, 3-m YMC C18 column (Waters) on HPLC. The injection volume was 3 µL, and the flow rate was 200 µL/min in gradient mode, starting with 70 % of mobile phases A (5 mM ammonium acetate in water, pH 3) and 30 % of B (methanol), increasing to 100 % B for 10 min, and being kept constant for 7 min. The mass spectrometer was operated in the electrospray ionization (ESI) negative mode with multiple reaction monitoring (MRM). The ESI conditions for the analysis of the targeted PFCs were optimized to the following conditions: ion source voltage −4.5 kV, ESI temperature 400 °C, curtain gas 15 psi, nebulizer gas (GS1 40 psi and GS2 60 psi), entrance potential −10.0 V, dwell time 45 ms. The settings for declustering potentials and collision energies were optimized individually for each chemical. The mass analyzer was operated in the multiple reaction monitoring (MRM) mode: PFBS (m/z 299 → 80), PFHxS (m/z 399 → 80), PFOS (m/z 499 → 80), PFDS (m/z 599 → 80), PFBA (m/z 213 → 169), PFPeA (m/z 263 → 219), PFHxA (m/z 313 → 269), PFHpA (m/z 363 → 319), PFOA (m/z 413 → 369), PFNA (m/z 463 → 419), PFDA (m/z 513 → 469), PFUnDA (m/z 563 → 519), PFDoDA (m/z 613 → 569), PFTrDA (m/z 663 → 619), and PFTeDA (m/z 713 → 669). Measurement of urinary oxidative stress biomarkers, malondialdehyde (MDA) and 8‑hydroxy‑2′‑deoxyguanosine (8‑OHdG)
Eur J Nutr (2016) 55:1011–1020
substances as follows: 50 μL of urine was mixed with 300 μL of 0.5 M phosphoric acid solution and 150 μL of 23 mM thiobarbituric acid solution (Sigma-Aldrich T-5500, Steinheim, Germany) and was heated at 95 °C for 1 h. After cooling on ice, the mixture was vortexed with 500 μL of methanol and was centrifuged at 5000×g. Absorbance of the supernatant was measured at 532 nm using HPLC–UV. The mobile phase was potassium phosphate (0.05 mol/L; pH 6.8) and methanol (58:42, v/v). As a DNA damage marker, urinary 8-OHdG levels were measured using HPLC/MS/MS according to following instructions. Five hundred milliliter of urine was first diluted with 500 mL of distilled water, followed by an addition of 20 μL of 15N5–8-OHdG solution (42.6 ng/mL) as an internal standard. After the addition of 150 μL of 1 M ammonium acetate buffer (pH 5.25), the solution was vigorously mixed and loaded into an Oasis HLB (10 mg/ mL; Waters) which had been preconditioned with 1 mL methanol and 1 mL of distilled water. The column was then washed with 1 mL of distilled water. The fraction containing 8-OHdG was eluted with 1 mL of methanol, dried under vacuum for 30 min, and dissolved in 100 μL of 10 mM ammonium acetate buffer in methanol. Twenty microliter of the dissolved solution was injected into the HPLC/MS/ MS instrument. The HPLC system consisted of an Agilent 6460 (Agilent, CT, USA) equipped with the column Hypercarb Thermo 100 × 2.1 mm × 5 μm and Hypercarb precolumn (Thermo Electron Corporation, USA) was used. The mobile phase was 10 mM ammonium acetate buffer in methanol (pH 10.5) and methanol (90:10, v/v). The eluent of the HPLC system was connected to a triple–quadruple mass spectrometer (ESI Agilent, CA, USA) equipped with a Turbo ion Spray TM source. Electrospray ionization was performed in the positive mode. Measurement of IR markers We measured fasting glucose and insulin levels in serum collected at each visit to evaluate IR. The measurement methods for glucose and insulin levels are available in a previously reported paper [22]. We also calculated the homeostatic model assessment (HOMA) as an index of IR, according to the following equation: fasting insulin (in microunits per milliliter) × fasting glucose (milligrams per deciliter)/405. IR was defined as a HOMA ≥2.5 according to the criteria determined by the Japan Diabetes Society [28]. Air pollution concentrations and meteorological factors
We measured urinary levels of MDA and 8-OHdG to estimate oxidative stress level. As a lipid peroxidation marker, urinary MDA levels were determined by measuring thiobarbituric-acid-reactive
13
Because ambient air pollutants (PM10, O3, and NO2) were found to affect IR in a previous paper [22], concentration of ambient air pollutants measured at the monitoring
Eur J Nutr (2016) 55:1011–1020
centers nearest to the residence of each participant was obtained from the Korea National Institute of Environmental Research and was adjusted in the analysis. Outdoor temperature and dew point obtained from the Korea Meteorological Administration were adjusted as well. Cotinine Urinary cotinine levels were measured for monitoring tobacco exposure. Cotinine level was analyzed by an enzyme-linked immunosorbent assay method [23]. Creatinine Urinary creatinine levels were measured to capture variation of urinary flow. Creatinine level was analyzed by a kinetic colorimetric assay with CREA (Roche, Indianapolis, Indiana, USA) reagent. A Hitachi 7600 (Hitachi, Tokyo, Japan) was used in the measurement. Statistical analysis PFC concentrations under the limit of detection (LOD) were assigned as a default value of LOD divided by 2. Creatinine-adjusted levels of PFCs, MDA, and 8-OHdG were used to capture variation of urinary flow. We estimated the effects of each PFC on MDA, 8-OHdG, glucose, insulin, and HOMA, using a linear mixed-effect model. In the analysis, we adjusted for age, sex, body mass index [BMI, weight (kg)/height2 (m2)], cotinine level, treatment arm, and vitamin C treatment and included air pollutants and meteorological factors in the statistical model as well. Sex, treatment arm, and vitamin C treatment were treated as categorical variables, and the other variables (age, BMI, cotinine level, PM10, O3, NO2, temperature, and dew point) were treated as continuous variables in these models. The relationships among exposures to each PFC were also estimated using the Pearson correlation. To estimate risk conferred by high level of PFCs, we used a generalized estimating equation model for the risk of IR by deciles of each PFC. SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) and R version 2.15.1 (The Comprehensive R Archive Network: http://cran.r-project.org) were used for statistical analyses.
Results At the baseline, characteristics of subjects without (n = 126) and with (n = 15) a history of diabetes mellitus were compared. All characteristics, except BMI, were not different between subjects with and without a history of diabetes mellitus (see Supplemental Material, Table S1).
1015
Because we aimed to evaluate relations among PFC exposures, oxidative stress, and IR in this study, and treatment of diabetes mellitus might modify the relations among them, we excluded 15 subjects with a history of diabetes mellitus among 141 elderly peoples recruited, and finally, 126 elderly were used in the statistical analysis. Included participants (n = 126) were aged ≥60, and the mean age was 73.3 years. Nineteen (15.1 %) participants were males and one hundred and seven (84.9 %) were females (Table 1). There were 26 elderly with HOMA ≥2.5 at the baseline, although they were not diagnosed as diabetes patients (Table 1). Total participants were randomly divided into two treatment arms. Vitamin C–placebo group and placebo–vitamin C group were 65 and 61 of 126 participants, respectively (Fig. 1). Because percent of male participants were bigger (20 %) in vitamin C–placebo group than that (9.8 %) in placebo–vitamin C group, height was different between both groups (P = 0.04). However, there was no difference of basic characteristics except height between the two groups (all baseline characteristics, P > 0.1). We measured each participant’s exposure to PFCs at the baseline (see Supplemental Material, Table S2). Of 15 PFCs, PFPeA and PFHxA were not detected in serum of any participants. Because only eight PFCs (PFHxS, PFOS, PFOA, PFNA, PFDA, PFUnDA, PFDoDA, and PFTrDA) among 15 PFCs showed a detection rate above 90 %, we analyzed relations among the eight PFCs. In the analyses, the eight PFCs were found to be strongly correlated with each other (all, P 0.1), we pooled results by treatment arm and compared the effect of PFCs on IR between vitamin C treatment and vitamin C non-treatment at the baseline and after placebo. We estimated the effect of the eight PFCs on HOMA index by treatments (Table 2 and Supplemental Material, Table S4). Among the eight PFCs, PFOS and PFDoDA levels were positively associated with HOMA index at the baseline and
13
1016
Eur J Nutr (2016) 55:1011–1020
Table 1 Baseline characteristics of participants without a history of diabetes Characteristic
Total subjects (n = 126) Vitamin C–placebo (n = 65) Placebo–vitamin C (n = 61) P value
No. of males (%) Mean age (min–max) Height [mean ± SD (cm)] Weight [mean ± SD (kg)] BMI (kg/m2), no. (%) ≥30 25 ~
