Obesity Contributes to Ovarian Cancer Metastatic Success through Increased Lipogenesis, Enhanced Vascularity, and Decreased Infiltration of M1 Macrophages

Ovarian cancer is the leading cause of death from gynecologic malignancy and the fifth leading cause of cancer death in women in the United States, with a lifetime prevalence of 2.5% (1). The highest rates of disease have been reported in industrialized and northern European countries and incidence is most common in women older than 40 years (2–4). When diagnosed prior to metastatic dissemination, the overall 5-year survival rate is >90%; however, the vast majority of women with ovarian cancer are diagnosed with advanced-stage disease with intraperitoneal metastasis already present, resulting in a survival rate of <30% (5, 6).

Obesity is a recognized epidemic worldwide. Recent data show that 8 of 10 Americans are overweight (body mass index or BMI > 25), >35% of women are obese (BMI > 30), and 8% of women are extremely obese (BMI > 40; ref. 7). A large number of studies have examined the relationship between obesity and ovarian cancer risk (8–15). While the study methods and results vary, recent meta-analyses support the finding that increased BMI is associated with enhanced risk of ovarian cancer incidence (12, 13). Similar trends are observed in a smaller study evaluating only African-American women (16). Furthermore, several meta-analyses also describe an association between obesity and survival in women with ovarian cancer, reporting an adverse effect of obesity on survival (15, 17, 18).

The high mortality rate of ovarian cancer is attributable to widespread intraperitoneal metastasis as a result of direct extension from the primary ovarian or fallopian tube tumor or from exfoliation and subsequent intraperitoneal dissemination (19). Metastasizing ovarian cancer cells adhere avidly to the mesothelial cells that cover peritoneal organs, induce mesothelial cell retraction, anchor, and proliferate to produce secondary lesions (19–21). The omental fat pad that covers the bowel and the abdominal cavity is the most common site of metastatic colonization, with >60% of newly diagnosed patients displaying omental metastases (6, 19, 22, 23). In the omentum, immune aggregates known as milky spots, together with omental adipocytes, promote homing of tumor cells to this tissue whereupon adipocytes contribute to progressive tumor growth (24–27). Other common sites of metastasis are the ovary and peritoneum, particularly the diaphragm and small bowel mesentery (19). The accumulation of ascites fluid, frequently in high volumes, provides an additional vehicle for intraperitoneal tumor cell dissemination as well as a depot of soluble factors that influence tumor cell survival.

While the cumulative evidence supports a correlation between obesity and both ovarian cancer incidence and ovarian cancer survival, the influence of obesity on metastasis has not been evaluated. The objective of the current study was to test the hypothesis that obesity contributes to ovarian cancer metastatic success. We have used an integrative approach with three distinct murine models combining in vitro, ex vivo, and in vivo studies to address the impact of diet-induced obesity (DIO) on adhesion of ovarian cancer cells to peritoneal mesothelium and on metastatic colonization. Our results demonstrate that increased body fat enhances tumor cell–mesothelial cell adhesion, alters mesothelial ultrastructure, and promotes intraperitoneal metastatic dissemination. These data suggest that the negative impact of obesity on ovarian cancer survival may be due, in part, to enhanced metastatic success.

Type I collagen (rat-tail) was obtained from Corning. Oleic acid and linoleic acid for lipid loading was obtained from Sigma-Aldrich. Oil Red O and hematoxylin and eosin (H&E) stains were products of American MasterTech. Antibodies were purchased from various sources as indicated: proliferating cell nuclear antigen (stains proliferating cells, PCNA; catalog no. NBP1-40761; Novus), F4/80 (stains macrophages, catalog no. ab6640, Abcam), and CD31 (stains vessels, catalog #ab28364, Abcam). Rat monoclonal anti-Lipocalin 2 (Lcn-2, clone 29) antibody, rabbit polyclonal anti-fatty acid–binding protein-4 (FABP4), rabbit polyclonal anti-sterol regulatory element–binding protein 1 (SREBP1), rabbit polyclonal anti-inducible nitric oxide synthase (iNOS), rabbit polyclonal anti-mannose receptor (CD206), and rat monoclonal anti-CD45R (also known as protein tyrosine phosphatase receptor type C or lymphocyte common antigen; clone RA3-6B2) were purchased from Abcam. All peroxidase-conjugated secondary antibodies (anti-rabbit IgG, anti-mouse IgG, anti-rat IgG) were from Vector Lab. Peroxidase detection system reagents were obtained from BioGenex.

ID8 murine ovarian cancer cells were obtained from Dr. Katherine Roby, University of Kansas Medical Center (Kansas City, KS; ref. 28) and were transduced using lentiviral transduction (GenTarget Inc.) to express red fluorescent protein (designated ID8-RFP). Cells were maintained in DMEM supplemented with 4% FBS, 1% penicillin/streptomycin, 5 μg/mL insulin, 5 μg/mL transferrin, and 5 ng/mL sodium selenite. SKOVi.p. human ovarian cancer cells were obtained from Dr. Katherine Hale, University of Texas MD Anderson Cancer Center (Houston, TX) and were subjected to lentiviral transduction to express RFP to generate SKOVi.p.-RFP. Cells were maintained in 5% FBS, 1% GlutaMax, 1 mmol/L sodium pyruvate, 100 U/100 μg penicillin/streptomycin, 1% NEAA, and 10 mmol/L HEPES in RPMI-1640 medium. Human OVCA433 ovarian cancer cells were obtained from Dr. Robert Bast, Jr., MD Anderson Cancer Center, and were maintained in MEM (Gibco Invitrogen), 10% FBS (Gibco Invitrogen), penicillin/streptomycin (Gibco Invitrogen), amphotericin B, (Cellgro by Mediatech), nonessential amino acids (Cellgro by Mediatech), and sodium pyruvate (Cellgro by Mediatech)_. Cells were labeled with CellTracker Green (5-chloromethylfluorescein diacetate or CMFDA; Life Technologies Inc.) by incubation for 30 minutes at 37°C. SKOVi.p., and OVCA433 cell lines were tested and authenticated by Genetica DNA Laboratories using short tandem repeat (STR) DNA profiling and were found to be 95% and 100% concordant, respectively. Primary human peritoneal mesothelial cells were derived from patients undergoing bariatric surgery or lean controls and were obtained from Zen-Bio and were maintained in Medium 199 (M199, Earle's) with supplemented with 5% FBS and penicillin/streptomycin. LP9 human peritoneal mesothelial cell line was obtained from Coriell Aging Cell Repository and maintained in a 1:1 ratio of M199 (Earle's) with Ham F12 (Gibco) supplemented with 15% FBS, 10 ng/mL EGF (Sigma), 400 ng/mL hydrocortisone, 1% penicillin/streptomycin, 1% GlutaMax (Gibco), and 10 mmol/L HEPES.

Human peritoneal mesothelial cells were grown as above or lipid-loaded according to previously described protocols. Briefly, stock solutions of oleate and/or linoleate were conjugated to albumin, and cells were cultured in media containing 0.3 mmol/L oleate and 0.3 mmol/L linoleate or BSA alone for up to 10 days with daily replacement of medium (29). Lipid uptake was demonstrated by fixing cells in 4% paraformaldehyde followed by staining with Oil Red O (0.5%, 20 minutes, room temperature). Control and lipid-loaded mesothelial cells were used to generate mesomimetic cultures comprised of 3-dimensional type I collagen gels containing embedded human fibroblasts overlaid with a monolayer of mesothelial cells as described (30, 31). Mesomimetic cultures were incubated with fluorescently tagged ovarian cancer cells (OVCA433 or SKOVi.p., 1 × 10⁵ cells, 2 hours, 37°C), washed with PBS, and adherent cells quantified using fluorescence microscopy.

To evaluate adhesion to intact peritoneal mesothelium, an ex vivo peritoneal explant assay was used (31, 32). An optically clear solid support was prepared using Sylgard 184 Silicone Elastomer Kit (Fisher) according to manufacturer's specifications. Mice were first euthanized by CO₂ inhalation followed by cervical dislocation and then rapidly dissected using a ventral midline incision. After skin removal, the parietal peritoneum lining the ventral abdominal wall was dissected to remove a ∼1.2 × 1.2 cm² square of peritoneal tissue immediately adjacent to the midline in the lower two abdominal quadrants. The tissue explant was then pinned mesothelial-side up to the silastic resin using fine gauge dissecting pins and the explant immersed in PBS. Tissue integrity is maintained for up to 48 hours. Fluorescently tagged ovarian cancer cells (2.5 × 10⁵ cells/mL) were added to the explant in culture medium (2.5 mL) and allowed to adhere for 0.5 to 2 hours before gently washing and processing the explant for fluorescence microscopy using an EVOS FL microscope or scanning electron microscopy (below).

Three distinct animal models were used in this study: two DIO models in female athymic nude mice (nu/nu; Charles River) or C56BL6 mice (Jackson Laboratories) and one genetic obesity model in female leptin-mutant ob/ob mice (B6.Cg-Lep^ob/J; Jackson Laboratories). Cohort one was a DIO cohort composed of two groups of nu/nu mice (n = 13 per group) fed either control diet (control; Research Diets #98121701) or Western diet (Western; Research Diets #D12079Bi, 40% fat) for 16 weeks. Nude mice were maintained in temperature-controlled housing at 84°F, 27% humidity. Cohort two was a DIO cohort composed of two groups of C56BL/6 mice (n = 10 per group) fed either control or Western diet for 26 weeks. Cohort three was a genetic obesity cohort composed of two groups of mice (n = 10 per group): B6.Cg-Lep^ob (designated ob/ob) or wild-type BL6 colony controls were both fed a diet of normal mouse chow for 13 weeks. Animals in all cohorts were weighed biweekly or weekly until a statistically significant difference in group weight within the cohort was achieved (Supplementary Fig. S1). Each cohort study was performed in duplicate. All animal procedures were conducted in accordance with University of Notre Dame or University of Missouri Institutional Animal Care and Use Committee regulations.

To evaluate intraperitoneal metastatic colonization, mice in each cohort were injected intraperitoneally with ovarian cancer cells. Nude mice in cohort one were injected with SKOV3i.p-RPF (5 × 10⁶ cells in 2-mL PBS) human ovarian cancer cells and allowed to grow for 3 weeks. Animals were euthanized using CO₂ anesthesia followed by cervical dislocation and dissected anteriorly along the ventral midline followed by lateral dissection to create two skin flaps. Organs were then dissected individually and visually evaluated for tumor burden. Visible tumors were measured with calipers in the longest direction and designated as small (<2 mm), medium (2–5 mm), or large (>5 mm). C56BL/6 mice in cohort two were injected with ID8-RFP murine ovarian cancer cells (1 × 10⁷) and allowed to grow for 10 weeks. Ob/ob and control mice in cohort three were injected with ID8-RPF (8 × 10⁶ in 2-mL PBS) and allowed to grow for 7 weeks. Animals were monitored visually for signs of lethargy and abdominal distension or via live fluorescence imaging using a Bruker Xtreme in vivo imaging system (Bruker Corp.). When necessary, a depilatory cream was used to remove the hair on the torso and abdomen before imaging. Endpoint imaging was performed as described in detail (33). Briefly, animals were euthanized using CO₂ anesthesia followed by cervical dislocation and dissected anteriorly along the ventral midline followed by lateral dissection to create two skin flaps. Prior to dissection, mice were imaged by placing the mouse ventral side downward with skin flaps open to scan organs in situ. Organs were then dissected individually and visually evaluated for tumor burden as described above. Organs were then placed on a template and scanned ex vivo. Tumor burden was quantified from fluorescence images following spectral unmixing using Bruker Multispectral software and ImageJ and reported as relative tumor area/organ surface area (33). After scanning, organs were flash-frozen or fixed in 10% formaldehyde, paraffin-embedded, and processed for histology and immunohistochemistry as described below.

Peritoneal explants were dissected from the murine ventral surface and either processed directly for scanning electron microscopy (SEM) or used for the ex vivo peritoneal explant assay described above. Tissues were then fixed for SEM (2% glutaraldehyde, 2% paraformaldehyde in 0.1 mol/L cacodylate, pH 7.35, 1 hour), washed using 2-mercaptoethanol (2-ME) buffer (0.1 mol/L sodium cacodylate, pH 7.35, containing 0.13 mol/L sucrose, 0.01 mol/L 2-ME; 3 × 20 minutes), and fixed in osmium tetroxide (2% in cacodylate buffer) with microwave processing. After rinsing with cacodylate buffer and washing with ultrapure water (3 × 5 minutes), tissues were dehydrated in a graded series of ethanol, subjected to critical point drying, placed on carbon stubs, and sputter-coated with platinum. Samples were examined using a Hitachi S-4700 field emission scanning microscope or a FEI-Magellan 400 Field Emission Scanning Electron Microscope.

To visualize lipids in cultured mesothelial cells, cells were fixed in formalin (10%) for 5 minutes. Fixed cells or frozen sections of tissues were stained with Oil Red O for 6 minutes at 60°C, washed, and counterstained with hematoxylin for 1 minute. For H&E staining, tissues were deparaffinized according to standard protocols and stained with H&E according to manufacturer's specifications. Immunohistochemical analysis was performed on 5-μm tissue sections that were deparaffinized. Antigen retrieval was accomplished by heat induction at 99°C in an antigen retrieval solution (10 mmol/L Tris, 1 mmol/L EDTA, pH 9.0) for 30 minutes followed by cooling for 20 minutes. Immunohistochemical staining with anti-PCNA (1:200 dilution, proliferation marker), anti-CD31 (1:600, vasculature marker), anti-CD45 (1:800, B lymphocyte marker), anti-F4/80 (1:50 dilution, general macrophage marker), anti-iNOS (1:2,000, stains M1 polarized macrophages), anti-CD206 (1:64,000, stains M2 polarized macrophages), anti-LCN2 (1:1,000 dilution), anti-SREBP1 (1:600 dilution), and anti-FABP4 (1:6,400 dilution) was performed according to standard procedures. Peroxidase-conjugated secondary antibody with 2-component DAB chromogen detection was used to detect bound antigen. Staining in metastatic tumors was quantified with an Aperio ePathology system and macros built using the Image Analysis Tools package (Aperio) or by manual enumeration of a minimum of 5 high-power fields/slide. Statistical comparisons were made using the Student t test or Mann–Whitney U test (Sigmaplot).

All organs within the peritoneal cavity, including the omentum, are covered with a monolayer of peritoneal mesothelium. Metastasizing ovarian cancer cells adhere avidly to peritoneal mesothelial cells, induce mesothelial cell retraction, and anchor and proliferate within the submesothelial matrix to generate widely disseminated secondary lesions (19–21). To evaluate the effect of enhanced lipid content on tumor cell–mesothelial cell interaction, primary human peritoneal mesothelial cells were lipid-loaded by culturing in the presence of oleate and linoleate. Staining with Oil Red O demonstrates uptake of lipid by mesothelial cells (Fig. 1A and B). Lipid-loaded or control mesothelial cells were then assembled into 3-dimensional organotypic mesomimetic cultures to be used as an adhesive substratum for fluorescently labeled ovarian cancer cells (30, 31). A 1.5- to 4-fold increase in adhesion of fluorescently labeled ovarian cancer cells to lipid-loaded mesomimetics was observed relative to control mesothelium (Fig. 1C–F).

To examine the effect of dietary lipid on mesothelial adhesion in an intact tissue context, a DIO protocol was used wherein mice (athymic nude or C57BL6) were fed control chow diet or a “Western” diet (40% fat), weighed weekly, and sacrificed when weight differences were statistically significantly different for at least 2 weeks (Supplementary Fig. S1). Peritoneal tissues were then used for a recently described ex vivo adhesion assay. In this protocol, peritoneum from DIO or control mice was pinned mesothelium-side up to an optically clear silastic resin prior to incubation with fluorescently tagged ovarian cancer cells. The tissue explant was then washed to remove unbound cells and examined under a fluorescent microscope (31, 32). Significantly enhanced adhesion to peritoneal explants from DIO mice relative to controls was evident when tissue explants were visualized by either fluorescence microscopy or SEM (Fig. 2A–F). Furthermore, high-resolution examination of peritoneal tissues by SEM demonstrated a striking change in peritoneal ultrastructure, with a substantial increase in mesothelial microvilli content in tissue from DIO mice, indicative of a marked increase in mesothelial surface area (Fig. 2G–J).

As ex vivo adhesion to tissue explants supported in vitro mesomimetic data and suggested an effect of DIO on mesothelial adhesion, in vivo models of ovarian cancer metastasis were examined. In initial experiments, nude mice were placed on a DIO protocol wherein they were fed a control or Western diet, as above, prior to intraperitoneal injection with SKOV3i.p. human ovarian cancer cells. Tumor burden was quantified after 3 weeks by necropsy, counting and measurement of tumor size with calipers. At the time of injection, a 2.2-g difference in average cohort weight was observed (Supplementary Fig. S1; P = 0.005). Intraperitoneal implants on all organs (liver, stomach, omentum, diaphragm, intestine, colon, mesentery, spleen) were enumerated, measured with calipers in the longest direction, and characterized as small (<2 mm), medium (2–5 mm), or large (>5 mm). In the DIO cohort, the fraction of implants categorized as medium or large was enhanced on the omentum, ovaries, and liver relative to cohort controls (70.0%, 100.0%, and 59.4% vs. 54.0%, 25.0%, and 36.4%, respectively).

To confirm these results in a model of syngeneic ovarian cancer, a murine C57BL6 model was used for allograft studies. Compared with athymic nude mice, C57BL6 mice have an intact immune system and exhibit larger relative weight differences when fed on the DIO protocol (Supplementary Fig. S1). To examine the effect of DIO in this model, mice were fed control versus Western diet prior to intraperitoneal injection with ID8-RFP murine ovarian cancer cells. At the time of injection, a 5-g difference in average cohort weight was observed (P = 0.001). After 10 weeks, mice were sacrificed and fluorescent tumor implants on intraperitoneal organs imaged in situ and ex vivo. Tumor burden was quantified by spectral unmixing of fluorescent images obtained from individual organs ex vivo and reported as relative tumor area/organ surface area (33). Fluorescent images from representative control and Western diet subjects are shown in Fig. 3A–D, and relative tumor burden at each organ site is quantified in Fig. 3E. Results show that DIO leads to enhanced metastatic tumor burden in the omentum and ovaries relative to mice on a control diet. Representative H&E stains from all organ sites show serous histology (Supplementary Fig. S2).

The ob/ob mouse arose from a spontaneous mutation in leptin and gives rise to an obese phenotype even on a normal chow diet (34, 35). To examine the effect of marked obesity on intraperitoneal metastasis, ob/ob or control mice were fed a normal chow diet for 6 weeks prior to injection with ID8-RFP murine ovarian cancer cells. At the time of injection, a 38.2-g difference in average cohort weight was observed (P < 0.001). At sacrifice (7 weeks), organs were imaged in situ and ex vivo and tumor burden was quantified as described above. Fluorescent images from representative ob/ob and control subjects are shown in Fig. 4A–D, and relative tumor burden at each organ site is quantified in Fig. 4E. Results show that obese ob/ob mice exhibit enhanced metastatic growth at all sites examined, with substantial tumor burden in the omentum, ovaries, and mesentery. Representative H&E stains from all organ sites show serous histology (Supplementary Fig. S3). No significant differences in proliferation are observed between groups, as shown by immunohistochemical staining with PCNA (Supplementary Fig. S4), suggesting an alternative mechanism(s) for enhanced metastatic success in DIO and ob/ob mice.

To examine potential mechanisms by which obesity may contribute to metastatic success, intraperitoneal metastatic lesions from cohorts 2 and 3 [C56Bl/6 control diet vs. Western diet and wild-type BL6 (WT) vs. B6.Cg-Lep^ob (ob/ob)] were examined by quantitative immunohistochemical analysis. Immunostaining against CD31 demonstrated significantly increased vascularity in lesions from Western diet and ob/ob mice relative to their respective controls (Fig. 5A–E). As immune cell infiltration has been shown to both promote and inhibit tumor progression depending on the composition of the infiltrate, inflammatory cell influx was assessed via staining for the pan-B-cell marker CD45R. Both Western diet and ob/ob mice showed an ∼3-fold increase in CD45⁺ lymphocytes relative to control diet and WT controls, respectively (Fig. 6A–E). Interestingly, however, metastatic lesions show decreased overall macrophage infiltration (F4/80 immunoreactivity) in Western diet and ob/ob mice relative to their respective control mice (Fig. 6F–J). To evaluate whether this decrease was due to alteration levels of in M1- or M2-polarized macrophages, additional analyses were performed. Similar to results obtained with F4/80, decreased immunoreactivity against iNOS (M1-polarized macrophage marker) was observed in Western diet and ob/ob mice relative to their respective controls (Fig. 6K–O), whereas no significant differences in CD206 levels (M2-polarized macrophage marker) were observed among groups (Fig. 6P–T).

Previous studies have implicated proteins involved in lipid regulation in the intraperitoneal growth of metastatic ovarian cancer, including the adipokine lipocalin-2 (LCN2) and the lipid transport protein fatty acid–binding protein-4 (FABP4; refs. 27, 36). While Affymetrix-based gene expression profiling of tumors formed in the KpB murine model of ovarian cancer subjected to a high fat diet showed enhanced mRNA expression of LCN2 (36); we did not observe expression of this adipokine at the protein level in our models (Fig. 7A–D). FABP4 showed enhanced immunoreactivity in Western diet and ob/ob mice relative to controls (Fig. 7E–H), with intense staining present in adipocytes and in tumor cells immediately adjacent to adipocytes (see, e.g., Fig. 7F; ref. 27). Similar patterns were seen with Oil Red O staining of lipids in omental tissues in frozen sections of Western diet versus control diet mice (not shown). As cancer cells engage in extensive de novo lipogenesis, expression of SREBP1, a transcription factor that functions as a master regulator of lipid metabolism (37), was also examined. SREBP1 staining intensity was highly enhanced in tumors from mice fed a Western diet relative to a control diet (Fig. 7I), with tumors in Western diet mice exhibiting intense nuclear staining (Fig. 7J). Similar, albeit less intense, staining patterns were observed in tumors from ob/ob relative to WT mice (Fig. 7K and L).

Obesity is a recognized epidemic worldwide and recent meta-analyses have shown that obesity correlates with both ovarian cancer incidence and poor survival (8–18). Indeed, a growing body of evidence identifies obesity as a risk factor for cancer incidence and mortality in a large variety of malignancies (38). Using an integrative analysis of three distinct in vivo models together with ex vivo and organotypic data, results from the current study demonstrate that obesity influences ovarian cancer metastatic success. As the high mortality in women with ovarian cancer is directly attributable to widespread intraperitoneal metastasis, these data support a mechanistic link between obesity and poor survival.

In support of the epidemiologic data, it is increasingly recognized that adipose tissue is a key component of the ovarian cancer metastatic microenvironment. Ovarian cancer metastasizes preferentially to the omental fat pad (19, 21, 23), and omental adipocytes induce growth of ovarian tumor cells in vitro and in vivo via release and transport of free fatty acids, thereby functioning as a lipid reservoir to sustain the high-energy demands of cancer cells (27, 39, 40). The current study provides additional mechanistic insight into the enhanced tumor burden observed in intraperitoneal ovarian cancer xenografts grown in murine models of obesity. In both the DIO and mutational obesity models, our data demonstrate enhanced expression of SREBP-1, a master transcription factor that regulates genes involved in fatty acid synthesis, and thereby control de novo lipogenesis (37). SREBP-1 activity is tightly controlled at the posttranslational level via trafficking of the membrane-bound transcription factor from the endoplasmic reticulum to the Golgi, whereupon it is cleaved by the subtilisin-like membrane-bound serine proteinases sites-1 and -2 (37, 41). This cleavage releases the amino terminus of SREBP-1 for transport to the nucleus where it induces transcription of genes involved in lipogenesis. It is interesting to note that SREBP-1 immunoreactivity in tumors formed in Western diet and ob/ob mice is predominantly nuclear, supporting a mechanism through which elevated fatty acid synthesis, in addition to fatty acid transport (27), contributes to enhanced growth of peritoneal lesions in models of obesity. This further suggests alternative novel pharmacologic targets, including inhibitors of site-1 and site-2 proteinases, in control of metastatic tumor growth, particularly in obese women (37, 42).

In addition to regulation of lipid-based energy reservoirs, enhanced vascularity was observed in metastatic tumors in DIO and ob/ob mice relative to controls, suggesting that lipid-induced activation of cellular stress pathways contributes to development of a prometastatic tumor microenvironment. In this context, proangiogenic growth factors such as VEGF are known to be elevated in overweight and obese individuals (43), and patients with ovarian cancer with a high BMI treated with bevacizumab (anti-VEGF antibody) therapy had a shorter progression-free survival than women with low BMI (43). Recent data suggest that B cells also promote angiogenesis in ovarian cancer models (44) and correlate B-cell expression with poor survival in women (45). The current study shows an ∼3-fold increase in B cells in tumors grown in DIO and ob/ob mice. Together, these data suggest that adiposity warrants further consideration as a prognostic biomarker for patients receiving antiangiogenic therapy.

In other tumor models, adipose tissue contributes actively to metastatic progression not only by providing an energy reservoir for anchored tumor cells but also by secretion of proinflammatory cytokines (reviewed in refs. 38, 39). An unexpected result from the current study is the decrease in tumor-associated macrophages (TAM) in metastatic implants in both DIO and ob/ob mice relative to controls. TAMs are the most abundant immune cell type in the ovarian tumor microenvironment and are generally categorized as classically activated or M1-polarized (cytotoxic to tumor cells) and alternatively activated or M2-polarized (growth promoting; reviewed in refs. 39, 46). Immunohistochemical analyses showed that while levels of M2 polarized macrophages remained relatively constant across cohorts, levels of tumor-associated M1 macrophages were decreased in Western diet and ob/ob tissues, resulting in an overall 1.5- to 4.2-fold decrease in the M1:M2 ratio in obesity models. To this end, it is interesting that a recent study of TAMs in human ovarian cancers showed a decrease in M1:M2 TAMs with increasing tumor stage that correlated with poor overall survival (47). On the basis of the combined evidence, future studies evaluating the potential link between obesity and M1:M2 TAM ratios in human ovarian tumors are warranted.

An additional interesting observation arising from the current study is the striking increase in microvilli present on peritoneal mesothelial cells in mice fed a Western diet, which correlates with enhanced tumor cell adhesion. It has been reported that the distribution of microvilli on B cells is modulated in an inducible fashion by T cells and regulates the 3-dimensional surface presentation of receptors important in cell–cell adhesion (48). Topographical remodeling of microvilli, characterized by altered length, number, and/or clustering, contributes to modulation of leukocyte adhesion under flow (48–50). As intraperitoneal implantation of metastasizing ovarian cancer cells predominantly occurs within the context of an ascites fluid–filled peritoneal cavity (19, 21, 31), these data suggest a mechanism whereby diet-induced changes in microvilli topography may contribute to early cell–cell adhesive events in metastatic colonization.

In summary, the current study used an integrative approach to demonstrate that obesity impacts ovarian cancer metastatic success via mechanisms involving lipogenesis and lipid transport, enhanced vascularity, and decreased infiltration of M1 macrophages. In light of the strong preclinical and clinical evidence supporting an association between obesity and ovarian cancer incidence (12, 13, 16), progression (15, 17, 18), and metastasis (current report), additional studies investigating obesity as an epigenetic modulator of ovarian cancer are warranted and may provide new targets for dietary or therapeutic intervention.

No potential conflicts of interest were disclosed.

Conception and design: Y. Liu, A. Slusarz, N.J. Dovichi, B. Cheatham, W.M. Leevy, M.S. Stack

Development of methodology: Y. Liu, Z. Shi, A. Slusarz, Z. Liang, W.M. Leevy, M.S. Stack

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Liu, M.N. Metzinger, K.A. Lewellen, S.N. Cripps, K.D. Carey, E.I. Harper, Z. Shi, A. Grisoli, E. Lee, A. Slusarz, J. Yang, E.A. Loughran, K. Conley, J.J. Johnson, Y. Klymenko, L. Bruney, Z. Liang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Liu, M.N. Metzinger, K.A. Lewellen, S.N. Cripps, K.D. Carey, E.I. Harper, Z. Shi, J. Yang, Z. Liang, M.S. Stack

Writing, review, and/or revision of the manuscript: Y. Liu, E.A. Loughran, M.S. Stack

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Liu, Z. Shi, J. Yang, L. Bruney

Study supervision: M.S. Stack

Other (histology): L. Tarwater

This research was supported in part by the Leo & Ann Albert Charitable Trust (M.S. Stack); NIH grants RO1CA109545 (M.S. Stack), RO1CA086984 (M.S. Stack), F31CA159804 (L. Bruney); and National Science Foundation grant DGE1313583 (E. Lee).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.