Pivotal Role for the Visceral Fat Compartment in the Release of Persistent Organic Pollutants During Weight Loss

Context:

Polychlorinated biphenyls (PCBs), are implicated as potential endocrine disruptors and obesogens. These lipophilic substances are preferentially stored in the fat compartment and released into the circulation during weight loss.

Objective:

The aim of this study was to examine the contribution of abdominal adiposity, and visceral adiposity in particular, to the increase of serum PCB levels during weight loss.

Materials and Methods:

Fourty-five obese women were prospectively recruited. Twenty individuals received dietary counseling and 25 underwent bariatric surgery. Anthropometric data were collected and intra-abdominal adiposity was assessed by measurement computed tomography scanning of the abdominal fat compartment, delineating the visceral and subcutaneous compartment. Serum levels of 27 PCBs were determined and the sum of all PCBs (ΣPCBs) calculated. Follow-up measurements of anthropometric data, computed tomography scanning, and PCB levels were performed after 6 months in all patients.

Results:

In patients who lost weight, serum ΣPCB levels displayed an increase after 6 months of approximately 50%. Both correlation and regression analysis, focusing on the relative contribution of the visceral vs the subcutaneous fat compartment, suggested that the increase in ΣPCB serum levels after 6 months of weight loss was more pronounced in patients losing relatively more visceral adipose tissue. This trend could be established in the diet-treated, but not the surgery-treated subgroup.

Conclusion:

Our study suggests that the contribution of PCBs released from the visceral fat compartment might be more pronounced compared with the subcutaneous fat compartment during weight loss. These findings are present in the entire study group whereas subanalysis of the diet vs surgery groups suggested the same effect in the diet group but failed to reach statistical significance in the surgery group. This suggests a possible weight-loss method-specific effect.

Traditionally, the onset of obesity is attributed to a decline in energy expenditure and the consumption of energy-dense foods (1). Recently, data have emerged that suggest a possible causal role for certain chemicals, the so-called endocrine disruptors or “obesogens,” in the occurrence and development of obesity (2–4). This obesogen hypothesis proposes that perturbations in metabolic signaling, resulting from exposure to novel environmental influences, are superimposed on the trends of energy intake and expenditure (5). Most candidate environmental obesogens represent mimickers of endogenous lipophilic hormones or inhibitors of endogenous hormone metabolism (5).

One of the most important groups of chemicals implicated as obesogens are the persistent organic pollutants (POPs), which include the polychlorinated biphenyls (PCBs). PCBs were used worldwide in various industrial electrical appliances but the production was banned in the 1970s in most countries. Initial data on the relationship between POPs and body weight were based on accidental, high-level exposure causing severe intoxication symptoms (6). Intoxication victims often display weight loss (7). In contrast, low levels of POPs are suspected to be promoters of obesity, suggesting a nonmonotonic dose-response relationship (8). Due to legal restrictions and bans, exposure to PCBs in the environment has decreased (9). However, given that PCBs are extremely resistant to biological degradation, they remain detectable in humans (10). Humans are predominantly exposed through the consumption of contaminated food, mainly meat, fish, and dairy products (5). Multiple cross-sectional studies have suggested a relationship between serum PCB levels and obesity (11). Most of these studies investigated the link between serum levels and body weight, body mass index (BMI), or waist circumference.

Given that PCBs are lipophilic compounds, their predominant storage site in the human body is the fat compartment. The human body has different fat compartments, the main sites being the subcutaneous and visceral fat compartments. These two depots are both anatomically and functionally distinct from one another. In particular, the visceral fat compartment is considered metabolically more active by releasing proinflammatory and prothrombotic cytokines and is associated with the development of metabolic disease states, such as diabetes mellitus, atherosclerosis, fatty liver disease, and cardiovascular disease (12, 13). For example, leptin has proatherogenic properties and is linked to enhanced platelet aggregation and arterial thrombosis (12). Both the functional aspect as the amount of visceral fat have their effect on obesity-related comorbidity and mortality. There is an independent curvilinear association between visceral adiposity and mortality, suggesting that a large amount of visceral fat is required for an increased risk of mortality (14).

This study aimed to investigate the contribution of the visceral vs the subcutaneous fat compartment to the dynamic of PCB release during a weight-loss phase. We hypothesized that the metabolically more active visceral fat compartment displays a different dynamic of PCB release during a weight-loss phase than the subcutaneous abdominal fat compartment.

Materials and Methods

Population

A cohort of 45 overweight and obese women was prospectively selected from patients visiting the weight management clinic of the Department of Endocrinology, Diabetology and Metabolism of the Antwerp University Hospital between 2010 and 2012. All participants were older than 18 years of age. The subjects' characteristics are described in Tables 1 and 2. Participants were treated with a weight-loss protocol including dietary counseling and physical activity (n = 20) or with bariatric surgery (n = 25). According to Belgian law, patients are eligible for bariatric surgery if their BMI is at least 40 kg/m² or if their BMI at least 35 kg/m² with the presence of diabetes mellitus, therapy-resistant arterial hypertension, or obstructive sleep apnea. In the dietary intervention group, almost all (n = 20) patients started on a hypocaloric protein–enriched 700 kcal diet, two patients started on a hypocaloric 1500 kcal diet. For those patients on a 700 kcal diet, the caloric intake was gradually increased after 6 weeks, by 200 kcal per month. After bariatric surgery (n = 25), the patients received nutritional advice with low-fat, healthy dietary choices. Patients were followed up after 6 months. This study was approved by the ethical committee of the Antwerp University Hospital (Belgian Registry No. B30020097009 and registered at clinicaltrials.gov as trial number NCT01778868). All participants provided their written informed consent.

Table 1.

Descriptive Statistics of Study Population and Subpopulations: Descriptive Statistics at Baseline

Characteristic	Entire Group	Diet Group	Surgery Group	MWL Group	EWL Group
Diet/surgery	20/25	20/0	0/25	12/4	8/21
Age, y	40a	43a	38a	42a	39a
	(18–60)	(26–60)	(18–57)	(18–60)	(18–57)
Weight, kg	112	109a	114a	114a	111a
	(74–150)	(74–147)	(93–150)	(93–147)	(74–150)
BMI, kg/m2	40.3a	38.5a	42.5a,c	41.0a	40.6a
	(27.4–51.4)	(27.4–50.5)	(36.1–51.4)	(33.1–50.5)	(27.4–51.4)
Waist, cm	117a	114a	121a	119a	117a
	(80–150)	(80–135)	(106–150)	(101–133)	(80–150)
Fat mass, kg	60a	58a	62a	62a	59a
	(28–90)	(28–87)	(41–90)	(41–87)	(28–90)
CT-TAT, cm2	877a	774a	905a,c	869a	834a
	(341–1334)	(341–1073)	(702–1334)	(633–1334)	(341–1083)
CT-VAT, cm2	187a	171a	201a	183a	190a
	(49–481)	(49–326)	(52–481)	(52–326)	(49–481)
CT-SAT, cm2	659a	603a	704a,c	686a	644a
	(292–1055)	(292–899)	(458–1055)	(414–1055)	(292–945)
CT-VAT/TAT	0.22a	0.22a	0.22a	0.22a	0.23a
	(0.07–0.47)	(0.14–0.36)	(0.07–0.47)	(0.07–0.36)	(0.09–0.47)
CT-SAT/TAT	0.78a	0.78a	0.78a	0.78a	0.77a
	(0.53–0.93)	(0.64–0.86)	(0.53–0.93)	(0.64–0.93)	(0.53–0.91)
PCB153, ng/g lw	36b	43b	35b	36b	35b
	(4–403)	(7–402)	(4–403)	(7–402)	(4–403)
PCB138, ng/g lw	23b	24b	22b	24b	22b
	(5–317)	(5–263)	(5–317)	(5–263)	(5–317)
PCB180, ng/g lw	22b	24b	20b	24b	20b
	(2–222)	(4–222)	(2–191)	(3–222)	(2–191)
ΣPCBs, ng/g lw	129b	166b	215b	158b	122b
	(19–1690)	(27–1538)	(19–1690)	(27–1538)	(19–1690)

Abbreviation: lw, lipid weight.

^a

Data are represented as mean (minimum − maximum).

^b

Data are represented as median (minimum − maximum).

^c

Variable is significantly different between diet and surgery group with P < .05.

Table 2.

Descriptive Statistics of Study Population and Subpopulations: Descriptive Statistics at 6-Month followup

Characteristic	Entire Group	Diet Group	Surgery Group	MWL Group	EWL Group
Diet/surgery	20/25	20/0	0/25	12/4	8/21
Age, y	40a	43a	38a	42a	39 a
	(18–60)	(26–60)	(18–57)	(18–60)	(18–57)
Weight, kg	90b	98a	88b	107a	86a,d
	(64–139)	(64–139)	(67–127)	(86–139)	(64–115)
ΔWeight, %	−16a	−10a	−21a,c	−6a	−22a,d
	(−41–7)	(−25 to −3)	(−41–7)	(−10–7)	(−41 to −10)
BMI, kg/m2	34.0a	34.7a	33.5a	38.6a	31.4a,d
	(23.6–48.9)	(23.6–46.3)	(25.8–48.9)	(32.1–48.9)	(23.6–41.0)
ΔBMI, %	−16a	−10a	−21a,c	−6a	−22a,d
	(−41–7)	(−25 to −3)	(−41–7)	(−10–7)	(−41 to −11)
Waist, cm	104a	106a	103a	113a	99a,d
	(73–138)	(73–127)	(83–138)	(98–127)	(73–138)
ΔWaist, %	−11a	−7a	−15a,c	−5a	−15a,d
	(−32–0)	(−15 to −0.5)	(−32–0)	(−10–0)	(−32 to −8)
Fat mass, kg	44a	48a	41b	56a	38a,d
	(19–83)	(19–80)	(20–83)	(35–83)	(19–59)
ΔFat mass, %	−13a	−10a	−16a,c	−4a	−18a,d
	(−49–9)	(−23–5)	(−49–9)	(−14–9)	(−49 to −4)
CT-TAT, cm2	636a	665a	613a	770a	563a,d
	(238–976)	(238–976)	(271–882)	(579–976)	(238–882)
ΔCT-TAT%, %	−25a	−11b	−32a	−10b	−32a,d
	(−72–5)	(−39–1)	(−72–5)	(−40–5)	(−72 to −6)
CT-VAT, cm2	120b	135a	92b	158a	82b,d
	(33–311)	(33–286)	(51–311)	(72–286)	(33–311)
ΔCT-VAT%, %	−29a	−21a	−36a,c	−12b	−40a,d
	(−76–38)	(−54–25)	(−76–38)	(−51–38)	(−76 to −9)
CT-SAT, cm2	507a	525a	494a	606a	453a,d
	(199–836)	(205–836)	(199–757)	(354–836)	(199–724)
ΔCT-SAT%, %	−22a	−14a	−29a,c	−11a	−29a,d
	(−70–10)	(−36–8)	(−70–10)	(−38–5)	(−70–10)
CT-VAT/TAT	0.20a	0.20a	0.16b	0.21a	0.19a
	(0.07–0.42)	(0.14–0.33)	(0.07–0.42)	(0.09–0.42)	(0.07–0.42)
ΔCT-VAT/TAT, %	−7a	−7a	−8a	0.33a	−11a,d
	(−43–35)	(−34–35)	(−43–32)	(−17.5–35)	(−43–13)
CT-SAT/TAT	0.80a	0.79a	0.84b	0.78a	0.80a
	(0.58–0.93)	(0.67–0.86)	(0.58–0.93)	(0.58–0.91)	(0.58–0.93)
ΔCT-SAT/TAT, %	3a	2a	3a	−0.5a	5a,d
	(−14–28)	(−15–16)	(−10–29)	(−15–8)	(−4–29)
PCB153, ng/g lw	58b	63a	60b	48a	71b
	(8–277)	(14–178)	(8–277)	(10–109)	(8–276)
ΔPCB153%, %	42b	36a	63b	8a	79b
	(−85–385)	(−85–165)	(−83–385)	(−85–89)	(−83–385)
PCB138, ng/g lw	33b	36a	32b	28a	40a
	(5–152)	(8–103)	(5–152)	(6–67)	(5–152)
ΔPCB138%	40b	35a	50b	5a	72a
	(−87–304)	(−87–174)	(−87–304)	(−87–105)	(−86–304)
PCB180, ng/g lw	36b	39a	36b	29a	38b
	(4–205)	(7–108)	(4–205)	(5–71)	(4–205)
ΔPCB180%, %	55b	52a	76b	16a	104a,d
	(−86–532)	(−86–225)	(−82–532)	(−86–117)	(−82–532)
ΣPCBs, ng/g lw	239b	232a	220b	176a	274b
	(32–1249)	(50–626)	(32–1250)	(41–394)	(32–1249)
ΔΣPCBs%, %	47b	42a	70b	11a	82b,d
	(−87–422)	(−87–209)	(−85–422)	(−87–138)	(−85–422)

Abbreviation: lw, lipid weight.

^a

Data are represented as mean (minimum − maximum).

^b

Data are represented as median (minimum − maximum).

^c

Variable is significantly different between diet and surgery group with P < .05.

^d

Variable is significantly different between MWL and EWL group with P < .05.

Anthropometric data

Anthropometric measures were taken in the morning with patients in a fasting state and undressed. Height was measured to the nearest 0.5 cm and body weight was measured with a digital scale to the nearest 0.2 kg. Waist circumference was measured at the midlevel between the lower rib margin and the iliac crest. Body composition was determined by bio-impedance analysis as described by Lukaski (15), and fat mass was calculated, using the formula of Deurenberg (16). A computed tomography (CT) scan at the L4–L5 level was performed to measure the amount of total abdominal adipose tissue (CT-TAT), visceral abdominal adipose tissue (CT-VAT), and subcutaneous abdominal adipose tissue (CT-SAT) (17). First, the total area of abdominal adipose tissue (CT-TAT) was measured at −190 to −30 Hounsfield Units. Subsequently, the area of CT-VAT was distinguished from CT-SAT by manually tracking the abdominal muscular wall separating the two adipose tissue compartments. The area of CT-VAT was measured and the area of CT-SAT was calculated by subtracting the area of CT-VAT form the area of CT-TAT. Anthropometric measures and CT scanning were performed at baseline and 6 months.

Blood sampling and determination of PCBs

Venous blood samples were obtained from fasting subjects from an anticubital vein between 0800 and 1000 hours into sterile BD Vacutainer tubes. Patients started fasting at 2200 hours, resulting in a fasting time of at least 10 hours. Blood samples for toxicological analysis were immediately centrifuged at 2500–3000 rpm during 15 minutes. Serum was then stored in glass vials in a −20°C freezer. Analyses of PCBs were performed at the Toxicology Centre (University of Antwerp). The samples were analyzed for 27 PCB congeners (IUPAC No. 28, 74, 95, 99, 101, 105, 118, 149, 146, 153, 138, 187, 183, 128, 167, 174, 177, 171, 172, 156, 180, 170, 199, 196/203, 194, 206, and 209). PCB levels were expressed on a lipid-adjusted basis. All levels were added to create the sum of all PCBs (ΣPCBs). Total lipids were calculated using the formula: Total lipids (g/L) = total cholesterol (g/L) * 2.27 + triglycerides (g/L) + 0.62 (18). Total cholesterol and triglycerides were measured at the Antwerp University Hospital, using a validated method (Dimension Vista 1500 Systems, Siemens). Levels below the method level of quantification (LOQ) were assigned a value of DF × LOQ, with DF being the proportion (%) of measurements with levels above the LOQ or the detection frequency. The analytical methods and quality assurance and quality control have been published previously (19).

Statistical analysis

Statistical calculations were performed using SPSS, version 21.0 (SPSS). The change in anthropometric data and POP serum levels at 6 months was calculated as a percent change, each time compared with baseline values ([followup − baseline]/baseline) × 100. Data reflecting such a percent change are presented as Δdata%. A decrease in weight or adipose tissue is thus reflected by a negative number whereas an increase in PCB serum levels is reflected in a positive number. ΔCT-VAT% and ΔCT-SAT% represent the percent decline in visceral and subcutaneous adipose tissue. ΔCT-VAT/TAT% and ΔCTSAT/TAT% provide information on the regression of the adipose tissue compartment in question, compared with the overall decline in abdominal adipose tissue. For additional clarity, an overview of used formulas is provided in Supplemental Table 1. Normality of distribution was verified using the Kolmogorov-Smirnov test. Normally distributed data are presented with their mean value and not-normally distributed values are presented with their median values. ΔPCB levels were all not normally distributed, and were not transformable to normality. Anthropometric data were normally distributed or transformable to normality using Log(x²). Individuals with a weight loss exceeding 10% were labeled as having excessive weight loss (EWL), those with a weight loss less then 10% were labeled as having moderate weight loss (MWL). The correlation between ΔPCB% and all anthropometric data was assessed using Spearman's correlation. The correlation analysis was performed in the total study population, the diet group, the surgery group, the EWL group, and the MWL group. To correct for multiple testing, a false discovery rate correction procedure was applied (with q = 0.05 and m = 9). A standard linear regression was performed to assess the individual effect of decline in CT-VAT and CT-SAT on a PCB serum level increase in the total study population, the diet group, the surgery group, the EWL group and the MWL group.

Results

The study population included 45 women with a mean age of 40 ± 11 years and a mean BMI of 41 ± 6 kg/m². Twenty-five patients (56%) underwent bariatric surgery (Table 1). After 6 months, all 45 patients were re-evaluated. All patients but one had lost weight (+8.6 kg for this single patient). Sixteen individuals displayed a moderate weight loss of less than 10% weight reduction (MWL group), of whom 12 of these individuals were part of the diet group and four were part of the surgery group. Twenty-nine individuals had an excessive weight loss of greater than 10% of their initial weight. Eight of them were part of the diet group and 21 were part of the surgery group (Table 1).

We measured 27 PCB congeners (Supplemental Table 2). Among the PCB compounds detected in our population, PCB153 was the most prevalent congener, followed by PCB180 and PCB138 (Tables 1, 2, and Supplemental Table 2.) These three congeners were detected in all participants. Representation of individual PCB congeners is therefore restricted to these three congeners. All serum levels of these PCBs, with the exceptions of PCB101, PCB149, and PCB174, increased during the study period in those patients with weight loss. The median increase in ΣPCB serum levels after 6 months was 47% (Table 2).

We used the percentage increase in serum levels of PCB153 (ΔPCB153%), PCB180 (ΔPCB180%), PCB138 (ΔPCB138%), and ΣPCB (ΔΣPCB%) to investigate its relationship with the percent change in BMI (ΔBMI%), weight (Δweight%), waist (Δwaist%), fat mass (Δfat mass%), total adipose tissue (ΔCT-TAT%), visceral adipose tissue (ΔCT-VAT%), subcutaneous adipose tissue (ΔCT-SAT%), and the ratio VAT/TAT (ΔCT-VAT/TAT%) and SAT/TAT (ΔCT-SAT/TAT%) (Table 3). In the total study group, the percent changes in BMI, weight, waist, fat mass, CT-TAT, CT-VAT, and CT-SAT all correlated significantly negative with ΔPCB153%, ΔPCB180%, ΔPCB138%, and ΔΣPCB%. ΔCT-SAT/TAT correlated significantly positive with ΔPCB153%, ΔPCB180%, ΔPCB138%, and ΔΣPCB% whereas ΔCT-VAT/TAT was significantly negatively correlated with ΔΣPCB% and ΔPCB180% but failed to reach statistical significance with ΔPCB153% and ΔPCB138%. Given that we calculated Δ as ([followup − baseline]/baseline) × 100, a negative ΔSAT/TAT% suggests a relative stronger decline in SAT compared with VAT (see Supplemental Table 1 for an illustration). So both the significantly positive correlations with ΔSAT/TAT% as the negative correlations with ΔVAT/TAT% thus suggest that the increase in serum PCB levels is more pronounced in those patients who lose relatively more visceral adipose tissue during weight loss.

Table 3.

Correlation Analyses of Changes in Anthropometric Measures and Changes in Serum PCB Levels, as Expressed by Percent Change

	Statistic	ΔBMI%	ΔWeight%	ΔWaist%	ΔFat Mass%	ΔCT-TAT%	ΔCT-VAT%	ΔCT-SAT%	ΔCT-VAT/TAT%	ΔCT-SAT/TAT%
Total study group (n = 45)
ΔPCB	r	−0.571	−0.570	−0.544	−0.648	−0.532	−0.562	−0.429	−0.287	0.350
153%	P	< .001	< .001	< .001	< .001	< .001	< .001	.003	.056	.018
ΔPCB	r	−0.545	−0.546	−0.514	−0.603	−0.499	−0.500	−0.386	−0.272	0.350
138%	P	< .001	< .001	< .001	< .001	< .001	< .001	.009	.071	.019
ΔPCB	r	−0.477	−0.477	−0.454	−0.569	−0.419	−0.489	−0.301	−0.295	0.367
180%	P	.001	.001	.002	< .001	.004	.001	.045	.049	.013
ΔΣPCB	r	−0.555	−0.555	−0.525	−0.631	−0.501	−0.541	−0.392	−0.297	0.361
%	P	< .001	< .001	< .001	< .001	< .001	< .001	.008	.047	.015
Diet group (n = 20)
ΔPCB	r	−0.692	−0.708	−0.564	−0.674	−0.468	−0.698	−0.214	−0.465	0.486
153%	P	.001	< .001	.010	.001	.038	.001	.366	.039	.030
ΔPCB	r	−0.645	−0.663	−0.469	−0.513	−0.432	−0.583	−0.156	−0.412	0.453
138%	P	.002	.001	.037	.021	.057	.007	.510	.071	.045
ΔPCB	r	−0.598	−0.617	0.501	0.597	−0.383	−0.678	−0.117	−0.535	0.577
180%	P	.008	.004	.025	.005	.095	.001	.622	.015	.008
ΔΣPCB	r	−0.648	−0.669	−0.520	−0.632	−0.427	−0.662	−0.170	−0.466	0.484
%	P	.002	.001	.019	.003	.060	.001	.474	.038	.031
Surgery group (n = 25)
ΔPCB	r	−0.507	−0.495	−0.460	−0.665	−0.556	−0.457	−0.532	−0.181	0.253
153%	P	.010	.012	.021	< .001	.004	.022	.006	.387	.222
ΔPCB	r	−0.566	−0.556	0.538	−0.683	−0.562	−0.458	−0.533	−0.175	0.269
138%	P	.003	.004	.005	< .001	.003	.021	.006	.402	.193
ΔPCB	r	−0.376	−0.364	−0.318	−0.559	−0.403	−0.357	−0.380	−0.138	0.223
180%	P	.064	.074	.121	.004	.046	.080	.061	.512	.284
ΔΣPCB	r	−0.484	−0.472	−0.428	−0.660	−0.517	−0.445	−0.484	−0.192	0.278
%	P	.014	.017	.033	< .001	.008	.026	.014	.359	.179
Moderate weight loss group (n = 16)
ΔPCB	r	−0.062	−0.097	0.115	−0.385	−0.188	−0.271	0.076	−0.274	0.359
153%	P	.820	.721	.672	.141	.485	.311	.778	.305	.172
ΔPCB	r	−0.300	−0.335	−0.012	−0.509	−0.485	−0.468	−0.115	−0.379	0.453
138%	P	.259	.204	.966	.044	.057	.068	.672	.147	.078
ΔPCB	r	0.106	0.071	0.191	−0.226	−0.044	−0.315	0.253	−0.350	0.429
180%	P	.696	.795	.478	.399	.871	.235	.345	.184	.097
ΔΣPCB	r	−0.103	−0.138	0.079	−0.456	−0.300	−0.397	0.032	−0.391	0.450
%	P	.704	.610	.770	.076	.259	.128	.905	.134	.080
Excess weigh loss group (n = 29)
ΔPCB	r	−0.264	−0.257	−0.226	−0.426	−0.299	−0.234	−0.315	−0.136	0.198
153%	P	.166	.179	.239	.021	.115	.222	.096	.096	.304
ΔPCB	r	−0.317	−0.311	−0.301	−0.425	−0.338	−0.269	−0.337	−0.121	0.202
138%	P	.094	.100	.112	.022	.073	.158	.073	.533	.292
ΔPCB	r	−0.210	−0.202	−0.185	−0.356	−0.195	−0.209	−0.198	−0.168	0.243
180%	P	.275	.293	.337	.058	.311	.277	.303	.382	.203
ΔΣPCB	r	−0.277	−0.270	−0.215	−0.400	−0.277	−0.243	−0.283	−0.144	0.214
%	P	.145	.156	.263	.031	.145	.204	.137	.455	.264

Spearman rank correlation analysis was performed. Correction for multiple testing with the false discovery rate correction procedure with q = 0.05 and m = 9.

P values that remained significant after this correction are indicated in bold.

The correlation analyses were repeated in the diet and surgery subgroups (Table 3). In the diet group, the analysis was very similar to that in the total group, albeit that the significant correlation with ΔCT-SAT% was lost for all congeners and the correlation with ΔCT-TAT% was lost for ΔPCB180%, ΔPCB138%, and ΔΣPCB%. After correction for multiple analyses, the significant correlation with ΔCT-VAT/TAT% and ΔCT-SAT/TAT% was lost for ΔPCB138%. In the surgery group, significant correlations with all anthropometric data but ΔCT-VAT/TAT% and ΔCT-SAT/TAT% were seen for ΔPCB153%, ΔPCB138%, and ΔΣPCB% (Table 3). The correlation analyses were repeated in the EWL and MWL groups, but no significant correlation with any of the anthropometric data could be established (Table 3).

To further explore the relative contribution of each adipose tissue compartment to the increase in PCB serum levels, we performed a multiple regression analysis in which CT-VAT% and CT-SAT% were introduced at the same level (Table 4). In the total and diet group, ΔCT-VAT% was the unique significant contributor for ΔPCB153%, ΔPCB180%, ΔPCB138%, and ΔΣPCB%. In the surgery group, the regression analysis identified ΔCT-SAT% as the significant contributor for ΔPCB138% and ΔΣPCB% (Table 4). Regression models were not significant in the EWL and MWL group (data not shown).

Table 4.

Multiple Linear Regression Analysis

Outcome	R2	P the Model	Variable	B (95% CI)	P
Total group
ΔΣPCB%	0.291	<0.001	ΔCT-VAT%	−0.346 (−2.720–0.070)	.040
			ΔCT-SAT%	−0.255 (−3.182–0.399)	.124
ΔPCB153%	0.322	<0.001	ΔCT-VAT%	−0.380 (−2.404–0.201)	.022
			ΔCT-SAT%	−0.251 (−2.652–0.324)	.122
ΔPCB138%	0.299	<0.001	ΔCT-VAT%	−0.397 (−2.532–0.246)	.018
			ΔCT-SAT%	−0.207 (−2.523–0.566)	.208
ΔPCB180%	0.269	0.001	ΔCT-VAT%	−0.365 (−3.221–0.144)	.033
			ΔCT-SAT%	−0.209 (−3.383–0.775)	.213
Diet group
ΔΣPCB%	0.365	0.021	ΔCT-VAT%	−0.649 (−4.596–0.802)	.008
			ΔCT-SAT%	0.122 (−1.854–3.212)	.579
ΔPCB153%	0.467	0.005	ΔCT-VAT%	−0.717 (−4.004–1.056)	.002
			ΔCT-SAT%	0.084 (−1.571–2.364)	.676
ΔPCB138%	0.395	0.014	ΔCT-VAT%	−0.675 (−4.303–0.883)	.005
			ΔCT-SAT%	0.125 (−1.642–2.924)	.561
ΔPCB180%	0.441	0.007	ΔCT-VAT%	−0.715 (−5.404–1.356)	.003
			ΔCT-SAT%	0.139 (−1.824–3.581)	.502
Surgery group
ΔΣPCB%	0.315	0.016	ΔCT-VAT%	−0.142 (−2.481–1.295)	.522
			ΔCT-SAT%	−0.466 (−5.694–0.091)	.044
ΔPCB153%	0.307	0.018	ΔCT-VAT%	−0.165 (−2.195–1.024)	.459
			ΔCT-SAT%	−0.441 (−4.712–0.067)	.056
ΔPCB138%	0.336	0.011	ΔCT-VAT%	−0.193 (−2.280–0.898)	.377
			ΔCT-SAT%	−2.075 (−4.717–0.001)	.050
ΔPCB180%	0.262	0.036	ΔCT-VAT%	− 0.143 (−2.951−1.569)	.533
			ΔCT-SAT%	−0.414 (−6.318–0.390)	.080

Abbreviation: CI, confidence interval.

Multiple regression analysis was performed with ΔCT-VAT% and ΔCT-SAT%, to assess the individual effect of decline in CT-VAT and CT-SAT on a PCB serum level increase.

Discussion

The aim of this study was to examine the role of abdominal adiposity, and visceral adiposity in particular, in the dynamics of serum POP levels during weight loss.

Our results confirm the increase in serum levels of PCBs after weight loss. The source of this increase is likely elevated internal exposure. People who have lost weight, irrespective of the method by which this is established, take in a significantly smaller amount of food then they formerly did. Given that the main route for PCB exposure in humans is oral intake of contaminated food (5), their ongoing external exposure declines and cannot account for the increase in serum levels. In concordance with previous reports, the increase in serum PCB levels correlates well with changes in “crude” measures of body composition such as BMI. Three studies have reported a significant increase in plasma levels of PCBs after diet-induced weight loss, expressed in BMI or fat mass (20–22). The follow-up measurements were performed after 15 weeks, which is notably shorter than the 6-month followup in our study. Mullerova and coworkers (23) reported no significant increase in plasma levels of PCB153 after a 3-month low-calorie diet. Others focused on surgically induced weight loss, and all reported a positive relation between weight loss (reduction in BMI) and serum levels of PCBs (24–26).

From a clinical point of view, however, the reduction in abdominal adiposity is most interesting, given that abdominal adiposity in particular is strongly linked with metabolic complications of obesity such as cardiovascular disease and type 2 diabetes mellitus. We are the first to focus on sequentially performed detailed analysis by CT, strengthening the VAT vs SAT association with PCB dynamics in serum (2, 27). In our total study group, the correlations with ΔCT-SAT/TAT% suggest the dominant contribution of visceral adipose tissue in increasing PCB serum levels during weight loss. These findings suggest that the dynamics of PCBs during weight loss in humans might differ between several types of fat compartments, with a relatively larger release from the visceral fat compartment. Interestingly, this phenomenon has already been identified in seals, with a significantly different trend in PCB concentrations between inner and outer blubber during fasting (28). In humans, repetitive sampling of the visceral adipose tissue compartment is not feasible. The waist circumference, as a substitute for abdominal adiposity, is not able to correctly distinguish visceral from subcutaneous fat. In the study by Imbeault et al (21), visceral and subcutaneous adipose tissue is measured by CT scan, but the authors do not investigate the specific influence of these compartments on the detected increase in serum PCB levels after a 15-week hypocaloric diet. De Roos et al (29) studied 109 women with CT-data on VAT and SAT. No weight-loss intervention was performed, although the researchers established a positive relation between self-reported weight loss in the last 20 years and ΣPCB plasma levels. In this study, PCBs with six or more chlorine atoms (IUPAC No. ≥ 146) were inversely associated with body weight, SAT, and VAT. PCBs with five or fewer chlorine atoms showed no association or an inverse association with body composition characteristics (29). It should be noted that the women included in the study were weight stable for the previous 3 months. In another study by this group, the less-chlorinated PCBs 105 and 118 were positively related to both VAT and SAT, whereas the more highly chlorinated PCBs (153, 156, 157, 169, 170, 180, 194, 206, and 209) were inversely related to both VAT and SAT in a weight-stable, elderly population (30). Our analyses of the different PCB congeners did not reveal such a divergence between less and highly chlorinated PCBs.

Separate analysis of the diet and surgery group suggested that this dominant contribution of the visceral fat compartment could also be established in the diet group but not in the surgery group (Table 3). Moreover, the correlation with ΔCT-SAT% was not significant for any of the PCB congeners in the diet group. Naturally, the diet group is distinctly different from the surgery group by the weight-loss method, but also by the average weight loss achieved (Tables 1 and 2). To differentiate whether the differences in the correlation analyses were weight-loss method specific or related to the amount of weight loss, two groups were created based on a cutoff of 10% weight loss (moderate vs excessive weight-loss group) (31, 32). Each group contained both diet and surgery group individuals. Interestingly, all significant relations with anthropometric data were lost in both these groups. This seems to suggest that the dominant contribution of the visceral fat compartment is restricted to individuals with diet-induced weight loss. It is theoretically possible that the speed at which weight loss occurs is an important determinant, with the VAT mainly contributing during the initial kilograms. However, the dominant contribution of the visceral fat compartment is also established in the total study group, albeit that the surgery group is numerically greater. It should be noted that our study population is rather small, so repetition of the analyses in a larger group is warranted. In that case, CT-scanning at 3 months might help to elucidate the importance of contribution of the speed of weight loss.

In the surgically treated subgroup of our population, sampling of the visceral and subcutaneous fat compartment was performed during surgery and adipose tissue levels of PCBs were measured (33). As previously reported, our group detected no difference between the absolute adipose tissue levels of PCBs in the visceral vs the subcutaneous compartment (33). This is in concordance with other studies that suggest no difference in absolute PCB levels between the subcutaneous and visceral fat compartment in weight-stable individuals (26). Therefore, sampling of the subcutaneous fat compartment has often been used as a valid surrogate to estimate the burden of PCBs in different fat compartments (26). Moreover, multiple studies have demonstrated that weight loss, regardless of the method by which it is achieved, does not target visceral fat preferentially. In a recent review it seemed that modest weight loss is associated with preferential loss of visceral fat compared subcutaneous fat but this effect is attenuated with greater weight loss (34). Despite the significant reduction in both visceral and subcutaneous adipose tissue, our data suggest that the dynamic of PCB release by the visceral fat compartment during weight loss, in particular if achieved by dietary measures, might be distinguishably different from that of the subcutaneous fat compartment. Therefore, absolute PCB levels in subcutaneous fat vs visceral fat might start to diverge due to weight loss and subcutaneous fat levels may not reflect actual PCB levels in visceral fat during weight loss.

Our findings might help explain the so-called “obesity paradox.” Despite the established causative role of obesity in numerous diseases, several studies have suggested a better survival in obese patients suffering from hypertension, heart failure, chronic obstructive disease, etc. (35). Recently, it has been suggested that this obesity paradox is only present in an obese population with low levels of POPs, but absent in an obese population with high levels of POPs (36). It is conceivable that weight cycling with repetitive release of large amounts of POPs from the fat compartment and subsequent reabsorption in a re-expanding fat mass, might cause endocrine-disrupting effects on metabolic and cardiovascular function, thus leading to poorer health outcomes in later life.

Conclusion

We performed a follow-up study of overweight and obese subjects before and after 6 months of diet or surgery-induced weight loss. In all participants with weight loss, irrespective of the method, serum PCB levels increased significantly at followup. Analysis of sequential CT measurements of the abdominal subcutaneous and visceral fat compartment suggests that the visceral fat compartment in particular might play a dominant role in the release of PCBs during weight loss. These findings are mainly present after weight loss induced by dietary measures, but could not be established in the subgroup with surgically induced weight loss. Additional studies are warranted to investigate the long-term release patterns of PCBs during and after weight loss, and to study the ensuing health effects.

Acknowledgments

This study was registered in ClinicalTrials.gov as trial number NCT01778868.

The project was funded by the University of Antwerp through a GOA project (Endocrine disrupting environmental chemicals: from accumulation to their role in the global ‘neuro-endocrine’ epidemic of obesity and its metabolic consequences; FA020000/2/3565). A.C.D. acknowledges a postdoctoral fellowship from the Research Scientific Foundation-Flanders (FWO).

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• BMI

  body mass index

 
• CT

  computed tomography

 
• EWL

  excessive weight loss

 
• LOQ

  level of quantification

 
• MWL

  moderate weight loss

 
• PCB

  polychlorinated biphenyl

 
• POP

  persistent organic pollutant

 
• SAT

  subcutaneous abdominal adipose tissue

 
• TAT

  total abdominal adipose tissue

 
• VAT

  visceral abdominal adipose tissue.