Fasting-mimicking diet and hormone therapy induce breast cancer regression

Abstract

Approximately 75% of all breast cancers express the oestrogen and/or progesterone receptors. Endocrine therapy is usually effective in these hormone-receptor-positive tumours, but primary and acquired resistance limits its long-term benefit1,2. Here we show that in mouse models of hormone-receptor-positive breast cancer, periodic fasting or a fasting-mimicking diet3-5 enhances the activity of the endocrine therapeutics tamoxifen and fulvestrant by lowering circulating IGF1, insulin and leptin and by inhibiting AKT-mTOR signalling via upregulation of EGR1 and PTEN. When fulvestrant is combined with palbociclib (a cyclin-dependent kinase 4/6 inhibitor), adding periodic cycles of a fasting-mimicking diet promotes long-lasting tumour regression and reverts acquired resistance to drug treatment. Moreover, both fasting and a fasting-mimicking diet prevent tamoxifen-induced endometrial hyperplasia. In patients with hormone-receptor-positive breast cancer receiving oestrogen therapy, cycles of a fasting-mimicking diet cause metabolic changes analogous to those observed in mice, including reduced levels of insulin, leptin and IGF1, with the last two remaining low for extended periods. In mice, these long-lasting effects are associated with long-term anti-cancer activity. These results support further clinical studies of a fasting-mimicking diet as an adjuvant to oestrogen therapy in hormone-receptor-positive breast cancer.

Conflict of interest statement

Competing interests A.N. and I.C. hold intellectual property rights on clinical uses of fasting-mimicking diets. V.D.L. holds intellectual property rights on clinical uses of fasting-mimicking diets and equity interest in L-Nutra, a company that develops and markets medical food. The remaining authors declare no competing interests.

Figures

Extended Data Fig. 1 |. Fasting and FMD enhance ET anti-tumour activity in HR+ BC cells.

a, b, MCF7 cells were plated in 96-well plates and treated with STS conditions, tamoxifen or fulvestrant at the indicated concentrations, or their combinations. After 96 h, cells were imaged by light microscopy (b) and their viability was detected (a). c, MCF7 cells were seeded in 6-well plates and cultured with or without STS, tamoxifen (TMX) or fulvestrant (FULV), or their combinations for 24 h. Thereafter, cells were cultured in regular medium for an additional 14 d. Finally, cells were fixed and stained with sulforhodamine B and cell colonies were counted. d, T47D and ZR-75–1 cells were plated in 96-well plates and exposed to STS, TMX at the indicated concentrations or their combination. Cell viability was detected after 96 h. e, ZR-75–1 cells were seeded in 6-well plates and cultured with or without STS, FULV, or their combinations for 24 h before being cultured in regular medium for an additional 14 d. Finally, cells were fixed and stained with sulforhodamine B and cell colonies were counted. f, g, MCF7 xenografts were established in 6–8-week-old female BALB/c nude mice. Once tumours became palpable, mice were randomized to be treated with ad libitum diet (control; n = 5), TMX (n = 5), FULV (n = 5), weekly 48-h water-only fasting (n = 5), or combined TMX + fasting (n = 5) or FULV + fasting (n = 5). Tumour volume and mouse weight were monitored over time (f). At the end of the experiment, mice were killed and tumour masses were imaged and weighed (g). In a–e, one representative experiment out of three is presented. In a, c, e, data are from four, three, and four biological replicates (four wells), respectively. In d - left graph, data are from six (left) or four (right) biological replicates (wells).Data are presented as mean ± s.d. (a, c–e, g) or s.e.m. (f). Data were analysed by two-tailed Student’s t-test (a, c–f; tumour volume at day 32, g) or two-way ANOVA (f).

Extended Data Fig. 2 |. Enhancement of ET activity via fasting/FMD in mouse xenografts of HR+ BC cell lines and in human HR+ BC organoids.

a, ZR-75–1 xenograft-bearing, 6–8-week-old female BALB/c nude mice were treated with ad libitum diet (control; n = 5), TMX (n = 5), weekly 48-h FMD (n = 6) or combined TMX and FMD (n = 6). Tumour volume (left) and mouse weight (right) were monitored over time. Mice from all treatment groups were killed at day 65, and tumours were isolated and weighed (middle). b, Growth of T47D xenografts in 6–8-week-old female BALB/c nude mice treated with ad libitum diet (n = 10), TMX (n = 7), fulvestrant (FULV; n = 9), or combined TMX + FMD (n = 5) or FULV + FMD (n = 5). c, Tumour organoids from patients with HR+ BC were cultured with or without TMX, STS conditions or their combination for 120 h, and then were imaged (left; one representative experiment out of three is presented) and their viability quantified (right). Viability of 209M, 213M and 33T BC organoids (graphs at right) was calculated from four biological replicates (wells) in each type of organoid for each treatment condition. d, MCF7 cells were grafted into 6–8-week-old female BALB/c nude mice, and once tumours became palpable, mice were treated with ad libitum diet (n = 14), TMX (n = 14), weekly FMD (n = 18) or their combination (n = 20). Tumour volume (left) and progression-free survival (right) were monitored for each group. n indicates the number of mice (with one tumour per animal; a) or the number of tumours per treatment group (b, d). Data are mean ± s.e.m. (a–d) or s.d. (c). Data were analysed by two-way ANOVA (a, left; b; d, left), by two-tailed Student’s t-test (a, right; b, tumour volume at day 105; c) or by log-rank test (d, right).

Extended Data Fig. 3 |. FMD-mediated increase in ET anti-cancer activity is mediated by the reduction in circulating insulin, IGF1 and leptin.

a, MCF7 cells were seeded in 96-well plates and cultured for 96 h with or without STS conditions (1% FBS, 0.5 g/l glucose), low-serum conditions (1% FBS, 1 g/l glucose), TMX or FULV at the indicated concentrations, or combinations of these treatments. Thereafter, cell viability was determined. One representative experiment out of three is presented. Cell viability in each treatment condition was calculated from three (MCF7 cells) or four (T47D, ZR-75–1 cells) biological replicates (wells). b, c, Serum β-hydroxybutyrate, IGFBP1, IGFBP3, adiponectin, TNF and IL-1β concentrations in female 6–8-week-old BALB/c nude mice treated with fasting/FMD (or ad libitum diet) with or without TMX or FULV. In all mouse groups, serum was collected at the end of the fasting/FMD. Data are from biological replicates. d, MCF7 cells were injected into 6–8-week-old female BALB/c nude mice; once tumours became palpable, mice were randomized to be treated with ad libitum diet (n = 6), FULV (n = 6), FULV + weekly FMD (n = 6), FULV + FMD + i.p. insulin (n = 5), FULV + FMD + IGF1 (n = 5), FULV + FMD + leptin (n = 5), or FULV + FMD + combined insulin, IGF1 and leptin (FRFs; n = 7). FRF administration was withdrawn at day 35 (crossover), while it was started in mice that had only been treated with FULV + FMD. Right, tumour volume in each treatment group at day 57; n, number of tumours per treatment group. e, MCF7 cells were seeded in 96-well plates and cultured for 96 h with or without STS, 10 μM FULV, insulin, IGF1, leptin (at the indicated concentrations), combined insulin, IGF1 and leptin, or combinations of these treatments. Cells were then imaged by light microscopy (left) and cell viability was determined (right). One representative experiment out of three is presented. f, At the end of the experiment shown in d, tumour masses were isolated for protein lysate generation. Total and phosphorylated AKT (Ser473) and p70S6K (Thr389) and vinculin (on the same gel) were assessed by immunoblotting. One representative experiment out of three (n = 5 or 6 tumour masses/treatment group were evaluated) is presented. For gel source data, see Supplementary Fig. 1. Data are from biological replicates and represent mean ± s.d. (a–c, e, right) or s.e.m. (d). Data were analysed by two-tailed Student’s t-test.

Extended Data Fig. 4 |. Fasting or FMD and oestrogen therapy cooperate to inhibit PI3K–AKT–mTOR and oestrogen receptor signalling in HR+ BC cells.

a, b, MCF7, T47D and ZR-75–1 cells were seeded in 6-well plates and cultured for 48 h with or without STS conditions in the presence or absence of tamoxifen (5 μM) or fulvestrant (10 μM). Thereafter, cells were subjected to protein lysate generation, and PI3K–AKT–mTOR signalling and EGR1, PTEN and β-actin (on the same gel) levels were detected by immunoblotting. For gel source data, see Supplementary Fig. 1. In b, protein bands were quantified and normalized to vinculin levels; data are from biological replicates and were obtained by three different experiments. c, MCF7 cells were injected into 6–8-week-old female BALB/c nude mice. Once tumours became palpable, mice were randomized to be treated with ad libitum diet (n = 3), FULV (n = 3), TMX (n = 3), weekly 48-h FMD (n = 4), FULV + FMD (n = 4) or TMX + FMD (n = 4). Mice were killed at the end of the fourth FMD cycle. Tumour masses were isolated and EGR1 and PTEN expression were detected by qPCR. Data are from biological replicates. n, number of tumours isolated and used for this experiment. d, Metastasis-derived HR+ BC organoids were cultured with or without TMX, STS or their combination for 48 h. Thereafter, organoids were isolated and subjected to RNA isolation, and EGR1 and PTEN expression was quantified by qPCR. Data are from biological replicates. One representative experiment out of three is presented. e, MCF7, T47D and ZR-75–1 cells were seeded in 96-well plates and cultured for 48 h with or without 1 μg/ml cycloheximide (CYCLO), STS conditions, TMX (5 μM), FULV (10 μM) or their combinations. Thereafter, protein synthesis was detected by OPP assay. Left, cells were imaged by fluorescence microscopy (one representative image out of three biological replicates (wells) is shown). Right, single-cell analysis was performed by acquiring the fluorescence signal from >1,000 cells per treatment condition (range 11548290). In b–e, data represent mean ± s.d. P values were calculated by two-tailed Student’s t-test. b, ns: non-significant; *P < 0.05; **P < 0.01; ***P < 0.001 (for P values, see Source Data file).

Extended Data Fig. 5 |. EGR1, PTEN and reduced AKT activation mediate the cooperation between ET and fasting/FMD in BC cells.

a–d, MCF7 cells were transduced with either one of two independent EGR1-targeting shRNAs (#1 or #2). In a, d, cells were seeded in 96-well plates and cultured with or without STS conditions, TMX or FULV at the indicated concentrations (5 μM TMX or 10 μM FULV in a) or their combinations for 96 h and then cell viability was detected (a, d). In a (upper panel), b, c, cells were plated in 6-well plates and cultured with or without STS, 5 μM TMX, 10 μM FULV or their combinations for 48 h. Afterwards, cells were subjected to protein lysate generation, and EGR1, PTEN and phosphorylated (Ser473) and total AKT, as well as GAPDH, were detected by immunoblotting. e, MCF7 cells transduced with a control vector (n = 12), an EGR1- (n = 10) or a PTEN-targeting (n = 10) shRNA, or myr-AKT (n = 11) were injected into 6–8-week-old female BALB/c nude mice. Once tumours became palpable, mice were treated with FULV and weekly cycles of FMD, and tumour volume was monitored. n, number of tumours per treatment group. f, MCF7 cells engineered to express either a control scrambled shRNA or an EGR1-targeting shRNA (shRNA#1) were seeded in 96-well plates. Twenty-four hours later, cells were cultured with or without TMX (5 μM), STS, GDC0068 (1 μM), AZD5363 (500 nM), LY294002 (2 μM) or their combinations. Cell viability was detected after 96 h. g, h, MCF7 cells were transduced with either a PTEN-targeting shRNA or myr-AKT. Cells were plated in 6-well plates and subjected to protein lysate generation. PTEN (g), AKT (h) and vinculin (g, h) were detected by immunoblotting. Thereafter, cells were seeded in 96-well plates and cultured with or without STS, TMX (5 μM), FULV (10 μM) or their combinations for 96 h and then cell viability was detected. i, MCF7 cells were seeded in 6-well plates and 24 h later were cultured with or without STS + TMX (5 μM), STS + TMX + combined insulin (400 pM), IGF1 (5 ng/ml) and leptin (50 mg/ml) or STS + TMX + 17β-oestradiol (100 nM). Forty-eight hours later, cells were subjected to protein lysate generation and EGR1, PTEN, 4E-BP1 and vinculin were detected by immunoblotting. j, k, MCF7 cells that were engineered with either a control vector or myr-AKT were cultured for 48 h with or without TMX (5 μM), FULV (10 μM), STS or their combinations. Thereafter, cells were subjected to protein lysate generation and monitoring of EGR1 and GAPDH levels by immunoblotting (j) or for RNA extraction and EGR1 and PTEN mRNA quantification (k). In a (lower panel), d, f, g (lower panel), h (lower panel), k, data are from biological replicates (in d, data are from four biological replicates (wells) per treatment condition). In a (upper inset), b, c, g (upper inset), h (upper inset), i, j, one representative experiment out of three is presented. Loading controls (GAPDH, vinculin) were always run on the same gels that were used to detect other proteins. For gel source data, see Supplementary Fig. 1. Data are presented as mean ± s.d. (a, d, f–h, k) or s.e.m. (e). In a, d, f–h, k, P values were determined by two-tailed Student’s t-test. In e, data were analysed by two-way ANOVA with Bonferroni post-hoc test and by two-tailed Student’s t-test (tumour volumes at day 40).

Extended Data Fig. 6 |. STS conditions cooperate with ET to activate AMPK and to dampen oestrogen receptor transcriptional activity in BC cells.

a, MCF7 cells transduced with a control vector, a PTEN-targeting shRNA or myr-AKT were plated into 6-well plates and cultured with or without STS conditions, TMX (5 μM), FULV (10 μM) or their combinations for 48 h. Afterwards, cells were subjected to protein lysate generation, and total and phosphorylated AMPK (Thr172) were detected by immunoblotting. The loading control for the immunoblots with PTEN-silenced MCF7 cells (vinculin) is shown in Supplementary Fig. 1. The loading control for the immunoblots with myr-AKT-expressing MCF7 (GAPDH) is shown in Extended Data Fig. 5j. Both loading controls were run on the same gels that were used to detect total and phosphorylated AMPK. For gel source data, see Supplementary Fig. 1. b, MCF7 cells were plated into 24-well plates and transfected with a pGL3 promoter plasmid or with a pS2/TFF1 reporter vector containing 1.3 kb of the proximal promoter of the oestrogen-responsive gene TFF1 cloned in the pGL3-basic backbone. Afterwards, cells were cultured with or without STS, TMX (5 μM), FULV (10 μM) or their combinations for 48 h. Finally, luciferase reporter activity was measured. c, MCF7, T47D and ZR-75–1 cells were plated in 6-well plates and cultured with or without STS, TMX (5 μM) or their combinations for 48 h. Thereafter, cells were subjected to RNA extraction, and PGR, TFF1 and GREB1 expression was detected by qPCR. In a–c, one representative experiment out of three is presented. In b, c, data are from biological replicates and are presented as mean ± s.d. P values were determined by two-tailed Student’s t-test.

Extended Data Fig. 7 |. FMD cooperates with ETs to induce cell cycle arrest in HR+ BC.

a–c, MCF7 cells were plated in 6-well plates and cultured with or without STS conditions, 5 μM tamoxifen (TMX) or their combination. After 24 h, RNA was isolated and used for gene expression microarray experiments (three biological replicates for each treatment condition were used). GSEA, represented by a heat map (a), was done by performing 10,000 permutations and using the REACTOME Pathways data set. Box plots (b) were generated using the ggplot2 package; to evaluate the differences between the groups, a non-parametric two-sided Wilcoxon test was used. The box-plot centre lines indicate the median value. c, Enrichment scores for G1-specific transcription genes of STS + TMX versus CTR, STS, or TMX groups. NES: normalized enrichment score. d, MCF7, T47D and ZR-75–1 cells were treated with TMX (5 μM), STS or their combinations. After 48 h, cells were subjected to RNA extraction or protein lysate generation. E2F1, E2F2, CCND1 and CCNE1 mRNA expression was detected by qPCR. e, CCND1, phosphorylated (Ser807/811) and total RB protein and vinculin (on the same gel) were detected by immunoblotting in the MCF7 cells. f, Tumour masses from mice that were treated for 4 weeks with ad libitum diet (n = 5), FULV (n = 4), weekly 48 h FMD (n = 4) or their combination (n = 4) were isolated at the end of the last FMD cycle. CCND1 and vinculin (run on the same gel) were detected by immunoblotting (left). CCND1 levels were quantified and normalized to vinculin (right). n signifies the number of tumours isolated per treatment group. g, T47D and ZR-75–1 cells were cultured in vitro for 48 h with or without TMX (5 μM), STS conditions or their combination, and then used to generate protein lysates. Phosphorylated (Ser807/811) RB protein bands were quantified and normalized to total RB. In d, f (right), g, data are from biological replicates. In g (right), data were obtained through three independent experiments. In d–f (left), one representative experiment out of three (d, e) or out of two (f, left) is presented. In d, f, (right), g, data are presented as mean ± s.d. P values were determined by two-tailed Student’s t-test. For gel source data, see Supplementary Fig. 1.

Extended Data Fig. 8 |. Combined FMD and ET downregulate CCND1 via EGR1 upregulation and AKT inhibition and revert acquired resistance to fulvestrant plus palbociclib.

a, b, MCF7, T47D and ZR-75–1 cells were plated into 96-well plates and cultured with or without TMX (5 μM) or FULV (10 μM), STS conditions or their combinations for 48 h. They were then subjected to cell cycle analysis by propidium iodide staining of isolated cell nuclei and flow cytometry. c, Upper panel, MCF7 cells were transduced with either a control scrambled shRNA or an EGR1-targeting shRNA (EGR1-shRNA#1); lower panel, MCF7 cells were transduced with a control vector or myr-AKT. Cells were cultured with or without TMX (5 μM) or FULV (10 μM), STS or their combinations for 48 h. Afterwards, cells were subjected to protein lysate generation, and CCND1 and vinculin (on the same gel) levels were detected by immunoblotting. For gel source data, see Supplementary Fig. 1. d, MCF7 cells transduced with a control vector or myr-AKT were cultured with or without TMX (5 μM) or FULV (10 μM), STS or their combinations for 48 h and then subjected to cell cycle analysis. e, MCF7 cells were injected orthotopically into 6–8-week-old female NOD/SCIDγ mice. Once tumours became palpable, mice were treated with ad libitum diet (n = 15), FULV (n = 16), cyclic FMD (n = 15), palbociclib (n = 15), FULV + PALB (n = 18), FULV + FMD (n = 16), PALB + FMD (n = 10) or FULV + PALB + FMD (n = 18), and their progression-free survival was monitored over time. f, g, MCF7 cells with acquired resistance to combined FULV + PALB (from Fig. 2b) were isolated and expanded ex vivo. Resistant MCF7 as well as parental MCF7 cells (control) were injected into 6–8-week-old female NOD/SCIDγ mice. Once tumours became palpable, mice were randomly assigned to be treated with or without FULV + PALB (mice with control MCF7 treated with vehicle: n = 5, with FULV + PALB: n = 4; mice with resistant MCF7 treated with vehicle: n = 4, with FULV + PALB: n = 5). Tumour volumes were monitored, and at the end of the experiment (day 38) tumour masses were isolated and subjected to histology (haematoxylin and eosin, HE) and immunohistochemical detection of Ki67+ cells (g). In g, right, 3–5 slices were prepared from each tumour (technical replicates) and subjected to Ki67+ cell enumeration. In e, f, n indicates the number of mice per treatment group (one tumour per mouse). In a–d, one representative experiment out of three is presented. Data are mean ± s.e.m. (f) or mean ± s.d. (g, right). In e, progression-free survival was analysed by log-rank test. In f, data were analysed by two-way ANOVA with Bonferroni post-hoc test and by two-tailed Student’s t-test (day 38). In g, P values were determined by two-tailed Student’s t-test.

Extended Data Fig. 9 |. Fasting or FMD prevents tamoxifen-induced endometrial hyperplasia and reduces intra-abdominal fat.

a–d, Six–eight-week-old female BALB/c mice were treated for 5 weeks with ad libitum diet (n = 11), tamoxifen (TMX; n = 11), weekly 48-h water-only fasting (n = 11) or FMD (n = 8), TMX + fasting (n = 11) or TMX + FMD (n = 8). Mice from all treatment groups were killed at the end of the last fasting/FMD cycle. Uteri were collected, imaged (a), fixed for histology (b) and subjected to protein lysate generation and RNA extraction. In addition, intra-abdominal fat depots were also isolated. Tff1, Pten and Egr1 mRNA levels in mice uteri were determined by qPCR (c), and total and phosphorylated AKT (Ser473), EGR1, PTEN and vinculin in mouse uteri were detected by immunoblotting (d; vinculin was detected on the same gel as EGR1 and PTEN). For gel source data, see Supplementary Fig. 1. e, Intra-abdominal fat depots isolated from the mice were imaged. In c, d, data are from biological replicates. P values were determined by two-tailed Student’s t-test.

Extended Data Fig. 10 |. Effects of periodic FMD on circulating FRFs and on disease in patients with HR+ BC and in HR+ BC mouse xenografts.

a, Body weight, hand grip, phase angle, fat-free and fat mass in patients (n = 23) with HR+ BC treated with ET and cyclic FMD in the NCT03595540 clinical trial. To evaluate changes in these parameters, we fitted a linear mixed-effects model taking into account absolute values as a function of time with a random covariate represented by subject ID. b, Quantification of total muscle area (highlighted in red) at L3 level (patient 1) and at the Louis angle level (patient 3) in patients with HR+ BC treated with ET and cyclic FMD (NCT03595540). c, Serum IGFBP1 and IFGBP3 (n = 6), blood glucose (n = 7) and ketone bodies (n = 11) before and after an FMD cycle in patients with BC treated with ET and FMD in the NCT03595540 trial. d, Serum leptin, adiponectin, C-peptide, IGF1, IGFBP1 and IFGBP3 (n = 7) and blood glucose (n = 12) before and after an FMD cycle in patients with BC treated with ET and FMD in the NCT03340935 clinical trial. e, Serum C-peptide, IGFBP1 and IGFBP3 in patients with HR+ BC treated with ET and cyclic FMD before and 3 weeks after an FMD cycle (n = 23; NCT03595540 trial). f, 6–8-week-old BALB/c nude mice were treated with ad libitum diet (n = 7), fulvestrant (n = 7), tamoxifen (n = 6), weekly FMD (n = 7), FULV + FMD (n = 6) or TMX + FMD (n = 7) for 2 weeks; 1 week after the end of the last FMD cycle, mice were killed, serum was isolated and serum C-peptide, IGF1 and leptin were measured by ELISA. g, h, MCF7 cells were injected into 6–8-week-old female BALB/c nude mice. Once tumours became palpable, mice were randomly assigned to be treated for 1 month with ad libitum diet (g, n = 9; h, n = 8), TMX (g, n = 9; h, n = 7), FULV (g, h, n = 9), FMD (g, n = 7), TMX + FMD (g, n = 9), FULV + FMD (g, n = 9), 48 h water-only fasting (h, n = 8), TMX + fasting (h, n = 14) or FULV + fasting (h, n = 10), followed by observation. Tumour volume (h) and mouse survival (g, h) were monitored. n, number of mice per treatment group (f, g) or of tumours per treatment group (h). In c–f, data are from biological replicates. In f, data are mean ± s.d. In h (left), data are mean ± s.e.m. In c–e, data were analysed by two-tailed Student’s paired t-test. In f, P values were determined by two-tailed Student’s t-test. In g, h (right), data were analysed by log-rank test. In h (left), results were analysed by two-way ANOVA with Bonferroni post-hoc test.

Fig. 1 |. Fasting or FMD potentiates the activity of ET in HR+ BC by reducing circulating growth-promoting factors.

a, Growth of MCF7 xenografts in 6–8-week-old female BALB/c nude mice treated with ad libitum diet (control, CTR; n = 6), weekly 48-h FMD (n = 6), tamoxifen (TMX; n = 6), fulvestrant (FULV; n = 8), or combined TMX + FMD (n = 8) or FULV + FMD (n = 10). b, Changes in serum IGF1, C-peptide and leptin concentration in female 6–8-week-old BALB/c nude mice treated with fasting or FMD (or ad libitum diet) with or without TMX or FULV. Serum was collected at the end of the fast or FMD. c, Six-to-eight-week-old female BALB/c nude mice were inoculated with MCF7 cells; when tumours became palpable, mice were randomized to be treated with ad libitum diet (n = 6), FULV (n = 6), FULV plus weekly FMD (n = 6), FULV plus weekly FMD plus intraperitoneal (i.p.) insulin (ins.; n = 5), IGF1 (n = 5), leptin (lept.; n = 5) or combined insulin + IGF1 + leptin (FRFs comb.; n = 7). At day 35 (crossover), FRF administration was withdrawn, whereas it was started in mice that had received only FULV + FMD. Left, MCF7 xenograft growth in response to ad libitum diet, fulvestrant, fulvestrant plus FMD, or fulvestrant, FMD and re-addition of the three FRFs (for the first 35 days or starting from day 35). Right, volume of MCF7 xenografts in response to the different treatments at day 35. d, Six-to-eight-week-old female BALB/c nude mice were inoculated with MCF7 cells. When tumours became palpable, mice were randomized to be treated with ad libitum diet, FULV, 48-h FMD or FULV + FMD. Mice were killed at the end of the fourth FMD cycle. Left, phosphorylated (p-) (Ser473 for AKT; Thr389 for p70S6K) and total AKT and p70S6K, EGR1, PTEN and vinculin (on the same gel) in the tumours were detected by immunoblotting. Right, phosphorylated AKT bands were quantified and normalized to total AKT (data points are biological replicates). For gel source data, see Supplementary Fig. 1. n, number of tumours per treatment group. Data are mean ± s.e.m. (a, c) or s.d. (b, d, right). a, P values determined by two-way ANOVA with Bonferroni post-hoc test and two-tailed Student’s t-test (day 45); b–d, by two-tailed Student’s t-test.

Fig. 2 |. FMD prevents resistance to combined fulvestrant and palbociclib and reduces tamoxifen-induced endometrial hyperplasia.

a, Growth of orthotopic MCF7 xenografts in 6–8-week-old female NOD/SCIDγ mice treated with ad libitum diet (n = 15), cyclic FMD (n = 15), fulvestrant (n = 16), palbociclib (n = 15), FULV + PALB (n = 18), FULV + FMD (n = 16), PALB + FMD (n = 10) or FULV + PALB + FMD (n = 18). b, Left, MCF7-xenograft-bearing 6–8-week-old female NOD/SCIDγ mice were treated with ad libitum diet (n = 3) or combined FULV + PALB (n = 6). At tumour progression, mice that were treated with FULV + PALB received weekly FMD cycles. Right, MCF7 cells with acquired resistance to FULV + PALB were expanded ex vivo and re-transplanted; xenograft-bearing mice were treated with FULV + PALB (n = 7) or FULV + PALB + FMD (n = 10). c–e, Six-to-eight-week-old female BALB/c mice were treated for 5 weeks with ad libitum diet (n = 5), TMX (n = 5), weekly 48-h fasting (n = 5), FMD (n = 5), fasting + TMX (n = 5) or TMX + FMD (n = 5). Mice from all groups were killed at the end of the last FMD or fasting cycle. Uteri were collected, weighed (c; uterus weight was normalized to the mean uterus weight of mice from the control arm) and fixed for histology (d), and intra-abdominal fat (gonadal, retroperitoneal and mesenteric depots) was collected and weighed (e). In d, black arrows indicate the tufts or blebs budding from the epithelium. Data are mean ± s.e.m. (a, b) or s.d. (c, e). a, b, P values determined by two-way ANOVA with Bonferroni post-hoc test and two-tailed Student’s t-test (last day); c, e, data points are biological replicates; P values, two-tailed Student’s t-test.

Fig. 3 |. Effects of periodic FMD on disease control and circulating FRFs in patients with HR+ BC and in mice.

a, Representative positron-emission tomography (PET) scans from a patient with HR+/HER2− metastatic BC (mBC; patient 26) exhibiting complete metabolic response upon treatment with fulvestrant, palbociclib and FMD. b, c, Serum IGF1, leptin and (in b) C-peptide in patients with HR+ BC treated with ET and cyclic FMD (NCT03595540) before and immediately (b; n = 6) or three weeks (c; n = 23) after an FMD cycle. d, MCF7 cells were grafted into 6–8-week-old female BALB/c nude mice. Tumour-bearing mice were treated for one month with ad libitum diet (n = 16), TMX (n = 16), FULV (n = 18), FMD (n = 16), TMX + FMD (n = 16) or FULV + FMD (n = 17) and observed thereafter. n, number of tumours per treatment group. e, f, MCF7 engraftment and growth in 6–8-week-old female BALB/c nude mice that were pre-treated for one month with ad libitum diet (n = 7), TMX (n = 8), FULV (n = 8), FMD (n = 8), TMX + FMD (n = 8) or FULV + FMD (n = 8). Mice were inoculated with MCF7 cells one week after the end of pre-treatment. g, Putative model for the cooperation between fasting or FMD and ET at the level of PI3K–AKT–mTOR and CCND1 signalling. ER, oestrogen receptor; IRS, insulin receptor substrate. b, c, P value determined by two-tailed Student’s paired t-test; d, f, mean ± s.e.m.; P values, two-way ANOVA and Student’s t-test (f, day 90); e, P values, log-rank test.