Calorie restriction outperforms bariatric surgery in a murine model of obesity and triple-negative breast cancer

Abstract

Obesity promotes triple-negative breast cancer (TNBC), and effective interventions are urgently needed to break the obesity-TNBC link. Epidemiologic studies indicate that bariatric surgery reduces TNBC risk, while evidence is limited or conflicted for weight loss via low-fat diet (LFD) or calorie restriction (CR). Using a murine model of obesity-driven TNBC, we compared the antitumor effects of vertical sleeve gastrectomy (VSG) with LFD, chronic CR, and intermittent CR. Each intervention generated weight and fat loss and suppressed tumor growth relative to obese mice (greatest suppression with CR). VSG and CR regimens exerted both similar and unique effects, as assessed using multiomics approaches, in reversing obesity-associated transcript, epigenetics, secretome, and microbiota changes and restoring antitumor immunity. Thus, in a murine model of TNBC, bariatric surgery and CR each reverse obesity-driven tumor growth via shared and distinct antitumor mechanisms, and CR is superior to VSG in reversing obesity's procancer effects.

Keywords: Breast cancer; Metabolism; Obesity; Oncology.

Figure 1. Dietary and surgical weight loss blunt obesity-driven tumor growth.

(A) Schematic of study design. CON, control; DIO, diet-induced obesity; HFD, high-fat diet; LFD, low-fat diet; VSG, vertical sleeve gastrectomy. (B) Body mass prior to weight loss interventions. (C) Change in body weight over time following weight-loss interventions. (D) Body mass prior to tumor cell injection. (E) Terminal body mass. (F) Ex vivo tumor mass. (G and H) Body composition following weight loss interventions. (I) Mediation analysis of weight change following weight loss intervention on tumor mass. (J) Hallmark gene sets determined significant by GSEA of tumor transcriptomics in pairwise comparisons with DIO-HFD. Gene sets grouped and colored as immune, other, and signaling related. (B–F and I) n = 21 CON-LFD, 21 DIO-HFD, 24 DIO-VSG, 19 DIO-LFD. (G and H) n = 8 CON-LFD, 8 DIO-HFD, 9 DIO-VSG, 8 DIO-LFD. (J) n = 6 CON-LFD, 6 DIO-HFD, 6 DIO-VSG, 5 DIO-LFD. (B–H) One-way ANOVA with Tukey’s post hoc test. NES, normalized enrichment score.

Figure 2. Transcriptomics analysis of mammary adipose tissue following dietary and surgical weight loss reveals discordant metabolic and immune signaling.

(A) Distribution of differentially expressed genes (DEGs) relative to DIO-HFD. (B–D) Volcano plots of DEGs generated in the comparison between CON-LFD, DIO-VSG, and DIO-LFD relative to DIO-HFD, respectively. (E) Hallmark gene sets determined significant by GSEA of adipose tissue transcriptomics in pairwise comparisons with DIO-HFD. Gene sets grouped and colored as signaling, metabolism, other, differentiation, and immune related. n = 4/group.

Figure 3. Epigenetic regulation through DNA methylation of mammary adipose tissue reveals transcriptional mediators of the gene expression profile conserved between human and mouse adipose tissue following surgical but not dietary weight loss.

(A–C) Volcano plots of DMGs generated in the comparison between CON-LFD, DIO-VSG, and DIO-LFD relative to DIO-HFD, respectively. (D) Distribution of DMGs relative to DIO-HFD. (E) MSigDB C3 gene sets determined significant by methylGSA of adipose tissue RRBS in comparison of DIO-VSG with DIO-HFD. (F) Regulator enrichment analysis of adipose tissue from patients who were never obese or obese before/after bariatric surgery (GSE59034). (A–E) n = 4/group, (F) n = 16/group. CPG, cytosine-phosphate-guanine; FET, Fisher’s exact test.

Figure 4. Dietary weight loss via caloric restriction outperforms surgical weight loss to blunt obesity-driven tumor growth.

(A) Schematic of study design. (B) Body mass prior to weight loss interventions. (C) Change in body weight over time following weight loss interventions. (D) Body mass prior to tumor cell injection. (E) Terminal body mass. (F) Ex vivo tumor mass. (G and H) Body composition following weight loss interventions. (I) Mediation analysis of weight change following weight loss intervention on tumor mass. (J) Mediation analysis of fat mass change following weight loss intervention on tumor mass. n = 20 CON-LFD, 18 DIO-HFD, 14 DIO-VSG, 19 DIO-ICR, 16 DIO-CCR. (B–H) One-way ANOVA with Tukey’s post hoc test.

Figure 5. Body weight loss and adiposity mediate blunting of obesity-driven tumor growth by dietary and surgical weight loss interventions.

(A) Significant GSEA Hallmark gene sets for tumor transcriptomics data in pairwise comparisons with DIO-HFD. (B–G) Circulating adipokines determined by multiplex ELISA. (H) Mammary adipose tissue oxylipin levels determined by UPLC-MS (z score). (A) n = 6/group. (B–G) n = 19 CON-LFD, 16 DIO-HFD, 14 DIO-VSG, 14 DIO-ICR, 17 DIO-CCR. (H) n = 11 CON-LFD, 10 DIO-HFD, 9 DIO-VSG, 10 DIO-ICR, 13 DIO-CCR.

Figure 6. Cecal Hungatella abundance associates with both body weight loss and tumor mass.

(A) Observed SVs. (B) Shannon index. (C) NMDS plot of Bray-Curtis distances. (D) Relative contribution of the 10 most frequent genera to each group. Spearman correlation between all genera and (E) percentage body weight change and (F) tumor mass. (G) Spearman correlation coefficients of the 20 genera showing the highest correlation coefficients with percentage body weight change and tumor mass. n = 10/group. *FDRq < 0.05, **FDRq < 0.01, ***FDRq < 0.001.