Detection of Persistent Organic Pollutants in Omental Adipose Tissue from Patients with Diffuse-Gastric Cancer: A Pilot Study

Simple Summary This pilot study reported the observation that great omentum could be analyzed to detect persistent organic pollutants (POPs). Diffuse gastric cancer is an increasing disease that could be associated with pollutants’ exposition. Here, we report a specific POP profile regarding a patient not affected by cancer, nor by diffuse gastric cancer or other abdominal cancers. The widespread presence of a substantial list of POPs (PCDDs/Fs, PCBs, and brominated flame retardants) was found in the omentum from patients with diffuse gastric cancer with minor presence of some organochlorine pesticides. Abstract The greater omentum represents a specific adipose tissue resected with gastric surgery for cancer. Diffuse gastric adenocarcinoma (diffuse-GC) is of major relevance among gastric cancers due to its unknown origin, aggressiveness, and metastasis in the peritoneal cavity. We postulated that persistent organic pollutants (POPs) could be detected in the greater omentum. Great omentum from patients with (i) diffuse-GC, or (ii) with other peritoneal metastatic cancer, and (iii) control group without cancer disease were analyzed for the distribution of a large panel of 96 POPs. POPs include polychlorinated dioxins/furans (PCDD/Fs), polychlorobiphenyls (PCBs), polybrominated diphenyl ethers (PBDE), polybrominated biphenyls (PBB), hexabromocyclododecanes, organochlorine pesticides, and polycyclic aromatic hydrocarbons (PAHs). The widespread presence of a substantial list of POPs (PCDDs/Fs, PCBs, and brominated flame retardants) was found in the omentum from patients with aggressive diffuse-GC, with minor presence of some organochlorine pesticides and PAHs at the low analyzed levels. Some chemicals appeared in larger concentrations in diffuse-GC or other cancer groups, including some PCDDs, PCB105, 123, 138, PBDE209, and PBB153. Overall, the present pilot study provides novel information regarding POPs levels in the omental fat, which is an understudied fat depot in terms of POPs load, and diffuse-GC association.


Introduction
Gastric cancer (GC) is a major health problem, the fourth among the most common cancers worldwide and the third in mortality rates [1,2]. The different sub-types include intestinal-, diffuse-, and mixed types according to the Lauren classification [3]. Unlike the decreasing incidence of the intestinal-type GC, the prevalence of the diffuse-type is increasing worldwide, especially in the USA and Europe [4]. Diffuse-type GC is a poorly differentiated, infiltrating, and scattered type cancer, and it is generally diagnosed at an

Patients and Study Design
Thirty-two patients were enrolled from 2013 to 2015 in the Department of Digestive and Oncology Surgery at Lariboisière Hospital, Paris (France), including 14 patients with an independent cell adenocarcinoma (diffuse-GC), 10 patients with other cancers (ovary or colon) that had metastasized in the peritoneal cavity, and 8 patients operated for non-cancer disease, as wound abdominal surgery. If patients met the eligibility criteria, they were completely informed about the research protocol and gave informed consent to participate in the study, which was approved by the 'Comité de protection des personnes' in 2013 in France (French equivalent of an Institutional Review Board, IRB). The protocol of this study was approved by the Bioethics Committee of the GNEDS ('Groupe Nantais d'Ethique dans le Domaine de la Santé'). The greater omentum is typically resected during surgery for gastric, colon, and ovary cancer. In case of gastrectomy, a systematic sample (one centimeter) of the greater omentum without a macroscopic tumor implant or lymph node was resected. In case of other peritoneal extended cancers (ovary, colon, peritoneal pseudomyxoma), the cytoreductive surgery for carcinomatosis also includes the resection of the greater omentum. All diagnoses were confirmed by histopathology, and gastric cancers were classified as late stages III or IV according to the Lauren classification [3]. One patient was the source of 4 specimens for analysis of intra-person variability of samples. Control individuals were operated for hernia or surgical hernia. After informed consent, about 10 g of omental fat was removed from the abdominal cavity during surgery or under local anesthesia. The specimen was frozen (−80 • C) for later chemical analysis. All patient data were anonymized with a unique id-code that remained blind for its status as a case or control during the chemical analysis. Detailed information, such as anthropometric variables (age, sex, body mass index), and other factors potentially associated with exposure to POPs was recorded into a database. These variables had been used prior in the toxicological literature to adjust the results of POPs concentration.

Chemical Analysis
The methodologies applied to isolate, detect, and quantify the targeted POPs, including 17 dioxins (PCDD/F), 18 PCBs (dioxin-like PCBs + non-dioxin-like PCBs), 8 polybromodiphenylethers (PBDE), 6 polybromobiphenyls (PBBS), hexabromocyclododecanes (HBCDs), and 30 organochlorine pesticides (OCs), have been described previously [32,36,37]. Briefly, 13C-labeled congeners were added to each sample for quantification according to the isotopic dilution method. Lipids were extracted from adipose tissue samples under high pressure and temperature (ASE Dionex, Sunnyvale, CA, USA). The resulting extracts were dried and weighed to measure fat content using the gravimetric method. Gel permeation chromatography was used for isolating OCs, while three purification steps using successively acid silica, florisil, and celite/carbon columns were applied for other targeted substances. PCDD/F, PCB, PBDE, PBB, and OC measurements were performed by gas chromatography (Agilent 7890A, Nantes, France) coupled with high-resolution mass spectrometry (GC-HRMS) on electromagnetic sector instruments (JEOL MS 700D or 800D, Nantes, France), operating at a resolution of 10,000 and in the single ion monitoring (SIM) acquisition mode [32]. HBCD isomers were quantified using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) on a triple quadrupole instrument (Agilent 6410). All data were generated in a manner blind to the control or case status of individuals. QA/QC procedures included systematic analysis of negative (blank, n = 3) and positive (standard reference material, n = 1) control samples in each batch of analyzed samples, and several inter-laboratory assays that were realized both nationally and internationally confirmed the robustness and accuracy of the protocols. All analyses were developed with performant methods and according to current European criteria in the field of routine control of foodstuffs of animal origin (EU, 2014). The analyses have been conducted in an ISO 17025:2005 accredited laboratory. Recoveries were in the 80-120% range, and the method's extended uncertainty was lower than 20%. The omental content of POPs was expressed on lipid basis content as previously reported, measuring the fat content in adipose tissue gravimetrically.

Statistical Analysis
Distributions of participants' characteristics and concentration data of POPs in adipose tissue were summarized as medians and interquartile ranges. The distribution of continuous variables and POP concentrations of cases were compared with controls using the Mann-Whitney U test. Chemicals with detection rates below 50% were excluded, and LOD/2 was attributed to non-detected samples. Correlation between POPs was evaluated using Spearman's rank correlation coefficient and displayed with heatmaps. A preliminary exploratory analysis to identify the most relevant POPs contributing to the discrimination between groups was conducted with sparse partial least square analysis (sPLS-A) [38]. All statistical analyses were conducted with R software v.3.5.0 (Free Software Foundation)

Demographic Characteristics
All patients (n = 32, 16 men and 16 women) were Caucasian and recruited in Europe (France). The clinical characteristics of case and control individuals in terms of age and body mass index (BMI), major factors determining the internal content of lipophilic POPs were similar among groups (Table 1). Patients with cancer had diffuse-GC or other tumors that metastasize in the peritoneal cavity.

Distribution of Persistent Organic Pollutants in Omental Fat
Summary distributions of POPs levels determined in omental fat tissue are displayed in Tables 2-5. Detection frequencies were very high for PCDD/Fs, PCBs, PBDEs, or PBB153. Lower rates of detection were achieved for organochlorine pesticides, especially for endrine, aldrine, heptachlor, trans-heptachlor epoxyde, α-chlordane, γ-chlordane, αendosulfan, β-endosulfan, endosulfan sulfate, methoxychlor, o,p-DDE, o,p-DDD, with detection frequencies ranging from 3 to 41%, thus being excluded from the summary table. None of the 20 screened PAHs were detected in omental adipose tissue.
The total concentrations of PCDDs, as normalized by lipid content, in omentum adipose tissue was significantly higher in patients with diffuse-GC (median = 166 pg/g lipid, 2.45×, p = 0.005) compared to the control group (median = 67.5 pg/g lipid) ( Table 2). The toxic equivalence (WHO-TEQ-2005) of PCDD/PCDF in omental tissue in the two groups was 14.1 (9-20.4) for diffuse-GC versus 9.5 (7.2-11.7) for the control (p = 0.15) ( Table 2). Statistically significantly higher levels were mainly observed for OCDD (112.8 pg/g, 2.5×, p = 0.003), followed by 1.2.3.6.7.8 HxCDD (19.3 pg/g, 2×, p = 0.035) and 1.2.3.4.6.7.8 HpCDD (18.2 pg/g, 3×, p = 0.029) ( Table 2). In contrast, concentrations of total PCBs (including dioxin-like (dl) and non-dl PCB), PBDEs, or organochlorine pesticides, were similar between patients with diffuse-GC and controls (Tables 3 and 4). The main fraction of dioxin-like PCBs was not significantly different between diffuse-GC (median 54,709 pg/g) and controls, and represent a WHO-TEQ (TEF2005) equivalent of 9.5 versus 9.4 (p = 0.53) (Table 3). Moreover, a polybromodiphenyl-ether (PBDE209) and one polybromobiphenyl (PBB153) were significantly higher in patients with diffuse-GC in comparison to the control group (p = 0.005 and p = 0.042, respectively) ( Table 4). A few organochlorine pesticides, such as p,p-DDE and b-hexachlorobenzene, were detected in the human omentum, although not statistically different in patients with diffuse-GC vs controls at p < 0.05 (Table 5). In addition to analysis performed in patients with diffuse-GC, the concentrations of POPs were analyzed in another group of patients with peritoneal metastatic cancers (including ovarian and colon cancers, as defined by 'other tumors'). A few PCBs (PCB105, PCB123, and PCB138) were significantly higher in the group 'other tumors' as compared to controls (Table 3). Table 2. Summary of polychlorinated dioxins/furans (PCDDs/PCDFs) distributions and detection frequencies (DF) measured in omental fat from cases of diffuse-gastric cancer, other cancers with peritoneal metastasis, and non-cancer as control patients. Results are represented by median (interquartile range) (pg/g lipid). The case groups were compared to the control group with the Mann-Whitney U test (* p < 0.05; ** p < 0.01).

An Exploratory Mixture Analysis
The correlation between POPs, mostly detected in omental fat (PCDD/Fs, PCBs, PBDEs, PBB153), is displayed in Figure 1. The heatmap illustrates the Spearman's rank correlation coefficient between two congeners. The results showed a very strong correlation between most PCDD/Fs, PCBs and few PBDEs (e.g., PBDE 153 and 209) and PBB153. Some congeners depicted weak or inverse correlation pattern with the rest of POPs, including PBDE 28, 47, 99 or PCB52, 101 (Figure 1).
Cancers 2021, 13 8 of 14 Figure 1. Heatmap displaying the Spearman's rank correlation coefficients between main congeners of PCDD/Fs, PCBs, PBDEs, and PBB153. The color intensity relates to the correlation coefficient magnitude, and the color to the direction (blue, positive associations; red, negative associations). Of note, and unsurprisingly, PCDD correlates with each other, as does PCDF, in our study; certain 'volatile' POPs, such as PBDE28 (and PBDE47), PCB28, and PCB52, known to be metabolized, are negatively correlated to the lipophilic POPs.
The principal component analysis further indicated that the first and second components accounted for most of model variance (54% and 12%, respectively, for Dim1 and Dim2) (Figure 2A). The individuals displayed in the first two principal components were not homogeneously distributed (Figure 2A). The groups of "diffuse-GC" and "control" appeared to be the most disassociated. Moreover, the sPLS-DA ( Figure 2B) showed thatthe OCDD and Sum of PCDDs (more abundant in the "diffuse-GC" group), -1.2.3.4.7.8-HxCDD (more abundant in the "other metastatic cancer" group), and -PCB52 (more abundant in the "control" group), were the most discriminant variables. Heatmap displaying the Spearman's rank correlation coefficients between main congeners of PCDD/Fs, PCBs, PBDEs, and PBB153. The color intensity relates to the correlation coefficient magnitude, and the color to the direction (blue, positive associations; red, negative associations). Of note, and unsurprisingly, PCDD correlates with each other, as does PCDF, in our study; certain 'volatile' POPs, such as PBDE28 (and PBDE47), PCB28, and PCB52, known to be metabolized, are negatively correlated to the lipophilic POPs.
The principal component analysis further indicated that the first and second components accounted for most of model variance (54% and 12%, respectively, for Dim1 and Dim2) (Figure 2A). The individuals displayed in the first two principal components were not homogeneously distributed (Figure 2A). The groups of "diffuse-GC" and "control" appeared to be the most disassociated. Moreover, the sPLS-DA ( Figure 2B) showed that -the OCDD and Sum of PCDDs (more abundant in the "diffuse-GC" group), -1.2.3.4.7.8-HxCDD (more abundant in the "other metastatic cancer" group), and -PCB52 (more abundant in the "control" group), were the most discriminant variables.

Discussion
This pilot study explored, for the first time, the concentration of almost one hundred persistent organic pollutants (POPs) in the human omental tissue from French patients with diffuse-gastric adenocarcinoma, other cancers, and of a control group without the disease. The data reveal a widespread presence of low amounts of most POPs (PCDDs/Fs, PCBs, and brominated flame retardants) in the human omentum with minor presence of some organochlorine pesticides and PAHs at the low analyzed levels.
The amounts of POPs in the human omentum still remains scarcely explored compared to other adipose tissue locations, such as peripheral fat. Whereas some studies have reported high correlations between concentrations levels of POPs across different fat pads (i.e., sub-cutaneous, visceral), other studies have found substantial differences and low correlations [31,[39][40][41]. Methodological limitations on the biological collection omental fat may prevent the use of this matrix in research. The feasible application of omental fat in research on the associations between POPs and endometriosis has been conducted within the innovative framework of the ENDO study where a population-based cohort was matched to a clinical one, favoring a valuable biocollection [42]. In a previous study conducted within this framework, we found comparable concentrations and high bivariate correlations for most POPs between fat pads (sub-cutaneous vs omental) [32]. The greater omentum (i.e., Epiplon) represents a specific large flat adipose tissue that covers digestive organs and moves around the peritoneal cavity. The omentum is a common site for advanced gastro-intestinal cancer and is usually affected by peritoneal metastasis [43]. Thus, the specific concentrations of POPs in the greater omentum might directly reflect the target microenvironment, highlighting its valuable application in research in gastro-intestinal cancers.

Discussion
This pilot study explored, for the first time, the concentration of almost one hundred persistent organic pollutants (POPs) in the human omental tissue from French patients with diffuse-gastric adenocarcinoma, other cancers, and of a control group without the disease. The data reveal a widespread presence of low amounts of most POPs (PCDDs/Fs, PCBs, and brominated flame retardants) in the human omentum with minor presence of some organochlorine pesticides and PAHs at the low analyzed levels.
The amounts of POPs in the human omentum still remains scarcely explored compared to other adipose tissue locations, such as peripheral fat. Whereas some studies have reported high correlations between concentrations levels of POPs across different fat pads (i.e., sub-cutaneous, visceral), other studies have found substantial differences and low correlations [31,[39][40][41]. Methodological limitations on the biological collection omental fat may prevent the use of this matrix in research. The feasible application of omental fat in research on the associations between POPs and endometriosis has been conducted within the innovative framework of the ENDO study where a population-based cohort was matched to a clinical one, favoring a valuable biocollection [42]. In a previous study conducted within this framework, we found comparable concentrations and high bivariate correlations for most POPs between fat pads (sub-cutaneous vs omental) [32]. The greater omentum (i.e., Epiplon) represents a specific large flat adipose tissue that covers digestive organs and moves around the peritoneal cavity. The omentum is a common site for advanced gastro-intestinal cancer and is usually affected by peritoneal metastasis [43]. Thus, the specific concentrations of POPs in the greater omentum might directly reflect the target microenvironment, highlighting its valuable application in research in gastrointestinal cancers.
In the present study, the analytical method was selective and sensitive enough to have high detection frequency. The concentrations of POPs reported in our study are similar to those previously reported in France, indicating a background exposure profile comparable with that of the general population [32]. For instance, the median concentration of TCDD ranged between 1.2 and 1.3 pg/g in the present study, whereas median concentrations s ranging between 0.7 and 1.4 pg/g were reported in adipose tissue from French women recruited during a similar period [23,32]. Similar comparable concentrations and detection frequencies were also found for PCBs, PBDEs, or most organochlorine pesticides.
Noticeably higher concentrations (about 3-fold) were observed in the present study for hexachlorobenzene, b-HCH, trans-nanochlor, or p,p'-DDE compared with the endometriosis study [32]. Similar bivariate correlation signatures have been reported before for the list of analyzed POPs. Strong positive correlations between PCDD/Fs and PCBs and negative correlations were found for PBDE28 or 47 [38].
The goals and design of the present study do not allow for addressing more complex questions, such as the potential associations between POPs and cancer risk. Nonetheless, the preliminary analysis suggested some statistical differences among groups. For instance, significantly higher levels of PCDD/F were observed in diffuse-GC as compared to the control group, mainly due to OCDD (median >112.8 pg/g omental adipose tissue, p = 0.003), followed by 1.2.3.4.6.7.8 HpCDD and 1.2.3.6.7.8 HxCDD (Table 2). Other chemicals were also found in larger concentrations in the group of "other cancers", including 1.2.3.4.7.8-HxCDF, 1.2.3.6.7.8-HxCDF, OMS-TEQ dl-PCBs, Total-TEQ, or the sum of 6 non-dioxin-like PCBs, among others (Tables 2 and 3). These results are not conclusive themselves due to the limited sample size of patients and lack of adjustment for confounding variables. However, a larger study that include both diffuse-GC and other aggressive tumors that metastasize in the omentum (n = 32) showed significant adjusted OR for WHO-TEQ PCDD/PCDF and dlPCB (TEF2005) when compared to controls (patients without tumors) (p < 0.05). Crude ORs, adjusted OR, and corresponding 95% confidence intervals were estimated in an associated analysis (additional Tables S1-S3). Further research with large scale studies are needed to evaluate these associations, all influenced by geographic patients' locations.
Several epidemiological studies have reported an increased mortality rate from cancer, including stomach cancer after accidental or occupational exposure to POPs, such as dioxin, and in cohorts of chemical manufacturing workers producing PCDD/PCDF, PCBs, and PAHs [24,25,28,[44][45][46]. The associations between PCDD/PCDF and diffuse-GC are supported by several studies using experimental models. For instance, mice treated with TCDD or PCDD/HCDF all had hyperplasia in the fundus of glandular stomach [47]. Subchronic exposure to OCDD appears to cause effects similar to those observed following exposure to low levels of TCDD in a rat model [48,49], although with less potency in human cells [49]. TCDD, the most toxic compound among PCDD/F family, also triggers invasive processes in human gastric tumor cell lines [50]. The proposed molecular mechanism of PCDD/PCDFs envisages the binding of the dioxin to the aryl hydrocarbon receptor, which is a transcriptional regulator of cell growth, migration, and invasion of cancer cells, as well as inflammation and immune suppression in several cancers [51][52][53]. Besides this classical pathway, a number of papers are now dealing with the role of epigenetic mechanisms in response to environmental xenobiotics, such as TCDD and OCDD, or PCB compounds [54][55][56][57].

Conclusions
To sum up, the present pilot study provides, for the first time, essential information regarding POPs levels in the omental fat, which is a relatively understudied fat depot in terms of POPs load. The study confirms the widespread presence of most relevant PCDD/Fs, PCBs, PBDEs, and organochlorine pesticides, despite the fact that they were banned in the 1970s. The information provided by the present study will support the prioritization of chemicals, biological matrix sampling strategy, and sample size calculations for future studies in patient's cohorts with gastric cancers. Capturing the complexity of chemicals mixtures should remain a priority over the analysis of single chemicals as confirmed in the present study. Environmental chemicals may be involved in the carcinogenic processes through multiple direct and indirect pathways, including the alteration of the poorly un-derstood adipocyte-tumoral cell cross-talk from cases with gastric cancers. Omental fat represents a target tissue of critical relevance on the pathogenesis of gastric cancer that should be seriously considered in the research of environmental risk factors including persistent pollutants.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cancers13194874/s1, Table S1: Summary of PCDD/PCDF distributions (median, interquartile range) in the omental fat from individuals with metastatic, Table S2: Summary of the distribution (median, interquartile range) of PCBs in the omental fat from individuals with metastatic adenocarcinoma (Diffuse gastric cancer, other tumors, and all tumors groups) (pg/g), Table S3: Summary of the distribution (median, interquartile range) of polybromodiphenylethers (PBDE) and polybromobiphenyls.(PBB) in the omental fat from individuals with metastatic diffuse adenocarcinoma (Diffuse gastric cancer, other tumors, and all tumors groups),(ng/g). adenocarcinoma (Diffuse gastric cancer, other tumors, and all tumors groups) (pg /g).

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the "Comité de protection des personnes" in 2013. The protocol of this study was approved by the Bioethics Committee of the GNEDS ('Groupe Nantais d'Ethique dans le Domaine de la Santé').
Informed Consent Statement: Inform consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.