The oestrous cycle stage affects mammary tumour sensitivity to chemotherapy

Abstract

The response of breast cancer to neoadjuvant chemotherapy (NAC) varies substantially, even when tumours belong to the same molecular or histological subtype¹. Here we identify the oestrous cycle as an important contributor to this heterogeneity. In three mouse models of breast cancer, we show reduced responses to NAC when treatment is initiated during the dioestrus stage, when compared with initiation during the oestrus stage. Similar findings were observed in retrospective premenopausal cohorts of human patients. Mechanistically, the dioestrus stage exhibits systemic and localized changes, including (1) an increased number of cells undergoing epithelial-to-mesenchymal transition linked to chemoresistance^2-4 and (2) decreased tumour vessel diameter, suggesting potential constraints to drug sensitivity and delivery. In addition, an elevated presence of macrophages, previously associated with chemoresistance induction⁵, characterizes the dioestrus phase. Whereas NAC disrupts the oestrous cycle, this elevated macrophage prevalence persists and depletion of macrophages mitigates the reduced therapy response observed when initiating treatment during dioestrus. Our data collectively demonstrate the oestrous cycle as a crucial infradian rhythm determining chemosensitivity, warranting future clinical studies to exploit optimal treatment initiation timing for enhanced chemotherapy outcomes.

Fig. 1. Oestrous cycle-dependent proliferation of tumour cells during oestrus stage in MMTV-PyMT tumours.

a, Schematic representation of multiday IVM set-up (left), with representative overview images (single z-plane) (right) of five mice in the first imaging session (top) and the following imaging session (bottom). Rectangles indicate example clones represented over multiple imaging days shown in b (left). b, Total clone size (number of cells in a clone) (right, top) and clone growth rate (right, bottom) of example clones quantified over time. c, Representative examples of clone growth rate in genetic MMTV-PyMT tumours imaged over multiple days with IVM. Graphs depict all growth rates (graph I) or a selection of clones (graph II) measured in physiologically cycling mice (left) and ovariectomized mice (right). Each line represents a clone, and coloured lines highlight the behaviour of individual clones. d, Violin plots depicting percentage of clones showing synchronized alternating growth rates, as determined by multiday IVM for physiological cycling mice (Ctrl; n = 5 mice) and ovariectomized mice (OVX; n = 4 mice). e,f, Representative immune fluorescent images and quantification of EdU incorporation (e) and expression of PHH3 (f) in genetic MMTV-PyMT mammary tumours during different stages of the oestrous cycle (n = 3 mice per oestrous stage, 14 and 11 individual tumours for EdU and PHH3 staining, respectively). d–f, Thicker lines represent median, thinner lines indicate the 25th and 75th percentiles and dots represent individual tumours. Statistical analysis by two-sided Wilcoxon–Mann–Whitney test (d) or one-sided hypothesis testing using linear mixed-effects model (e,f). *P < 0.05, **P < 0.01. Further details on statistical analysis are provided in Supplementary File 1. Scale bars, 500 µm (a), 100 µm (b,e,f). Source Data

Fig. 2. Decreased sensitivity to NAC in MMTV-PyMT tumours during dioestrus.

a, Experimental set-up; stages: oestrus (O) and dioestrus (D). b,c, Percentage of PI⁺ tumour cells (b) and change in tumour burden (c). n: O^Doxo = 5 mice/30 tumours; D^Doxo = 3 mice/22 tumours; O^Cyclo = 6 mice/47 tumours; D^Cyclo = 5 mice/40 tumours. d, Experimental set-up. e, Proportion of mice exhibiting physiological or disrupted oestrous cycle following doxorubicin treatment initiated at the indicated stage; n: O = 13, D = 10 mice. f,g, Number of preantral follicles per square millimetre in ovaries (f) and progesterone levels in serum (g) at pretreatment, 7 days following initial treatment and long term (LT) at the endpoint of experiment. Follicle count, n: O⁰ = 10; O⁷ = 7; O^LT = 4; D⁰ = 10; D⁷ = 7; D^LT = 8 mice. Hormone measurements, n: O⁰ = 9; O⁷ = 10; O^LT = 8; D⁰ = 10; D⁷ = 7; D^LT = 9 mice. h, Left, effect of oxorubicin treatment. Symbols represent mean relative tumour size per mouse; n = 8–9 mice per group, O = 69 and D = 84 tumours at first treatment. Lines represent mean of all tumours, shading represents s.e.m. Treatment effect is shown until the point at which over 40% of mice reached humane endpoint. Middle, fitted growth curves of mixed-effects model representing tumour volume. Right, Kaplan–Meier survival analysis; n: O = 8, D = 9 mice. i. Representative immunofluorescent lung section images, arrows indicate metastatic nodules. j–l, Quantification of macrometastatic nodules (j), tumour volume (k) and time elapsed between first treatment and analysis (l); n = 5 mice. b,c,f,g,j,k, Thicker solid lines represent median, and thinner solid lines the 25th and 75th percentiles. Individual dots represent either a tumour (b) or a mouse (c,f,g,h,j–l). Statistical analysis by either one-sided hypothesis testing using linear mixed-effects model (b), one-sided Wilcoxon–Mann–Whitney (c,j), Kruskal–Wallis (f,g), nonlinear modelling of tumour growth (h, left), a joint model (h, middle), a regression model (h, right) or two-sided Wilcoxon–Mann–Whitney test (k,l). ^#P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. Further details on statistical analysis are provided in Supplementary Files 1 and 2. D0, day 0; NS, not significant; PI, propidium iodide. Scale bars, 2 mm. Source Data

Fig. 3. Reduced chemosensitivity during dioestrus in hormone receptor-positive and hormone receptor-negative tumour models.

a,c,e,f, Immunohistochemical staining for ER (left) and PR (right), representative of more than five treatment-naive tumours in more than five mice of the genetic MMTV-Wnt1 (a), transplanted Brca1^−/−;Trp53^−/− (c) and genetic (e) and transplanted MMTV-PyMT (f) models at oestrus (top) and dioestrus (bottom). b,d, Left, treatment effect in transplanted MMTV-Wnt1 (b) and Brca1^−/−;Trp53^−/− (d) models, represented as tumour size normalized to initial treatment size. Arrows indicate rounds of treatment, lines the tumour means and shading s.e.m. MMTV-Wnt1 model: n = 5–6 mice per group with O = 6 and D = 6 tumours at first treatment. Brca1^−/−;Trp53^−/− model: n = 6–8 mice per group, with O = 12 and D = 14 tumours at first treatment. Middle, fitted growth curves of mixed-effects model representing tumour volume for transplanted MMTV-Wnt1 (b) and Brca1^−/−;Trp53^−/− (d) models. Right, Kaplan–Meier curves for MMTV-Wnt1 (b) and Brca1^−/−;Trp53^−/− (d) mice treated with doxorubicin initiated during oestrus or dioestrus. MMTV-Wnt1: oestrus, n = 6 mice; dioestrus, n = 5 mice; Brca1^−/−;Trp53^−/−: oestrus, n = 6 mice; dioestrus n = 8 mice. g, Left, treatment effect in transplanted MMTV-PyMT model, represented as tumour size normalized to initial treatment size. Arrows indicate rounds of treatment, lines the tumour means and shading s.e.m.; n = 8–13 mice per group, with O = 25, D = 15 and OVX = 20 tumours at first treatment. Right, fitted growth curves of mixed-effects model representing tumour volume. b,d,g, Left, significance was determined using nonlinear modelling of tumour growth. Treatment effect is shown until over 40% (b,g) or over 20% (d) of mice reached humane endpoint. Response between groups was statistically evaluated using a joint model (b,d,g, middle); survival analysis was performed using a regression model (b,d,g, right). ^#P < 0.1, *P < 0.05; **P < 0.01. Scale bars, 150 µm. Source Data

Fig. 4. Mechanisms contributing to reduced chemosensitivity in dioestrus.

a, Percentage of mesenchymal tumour cells determined in treatment-naive tumours of genetic MMTV-PyMT model at different oestrous cycle stages; n = 3 mice/24 tumours per group. b, Left, imunohistochemical staining for CD31; right, quantified diameter of intratumoural vasculature in treatment-naive genetic MMTV-PyMT tumours during stages of oestrous cycle. n = 3 mice/12 tumours per oestrous stage. c, Left, immunohistochemical staining for F4/80 (left). Quantification of percentage of F4/80⁺ cells in treatment-naive tumours (middle) or 7 days post treatment (right) in the genetic MMTV-PyMT model at oestrus and dioestrus. n: O^pre = 5 mice/21 tumours; D^pre = 7 mice/27 tumours; O^post = 3 mice/12 tumours; D^post = 3 mice/11 tumours. d, Schematic representation of experimental set-up. e, Immunohistochemical staining of chemotherapy treatment-naive tumours treated with either control antibody (anti-IgG2a) (top, representative of 15 tumours/3 mice) or macrophage depletion antibody (Ab) (anti-CSF1R) (bottom, representative of 21 tumours/5 mice). f,g, Percentage of PI⁺ tumour cells (f) and tumour volume change compared with baseline (g) for mice receiving control antibody (anti-IgG2a) or macrophage depletion antibody (anti-CSF1R) and treatment with doxorubicin. n: O^a-IgG2a = 6 mice/24 tumours; D^a-IgG2a = 6 mice/28 tumours; O^a-CSF1R = 4 mice/17 tumours; D^a-CSF1R = 5 mice/21 tumours. a,b,c,f,g, Thicker solid lines represent median, and thinner solid lines the 25th and 75th percentiles. Dots represent either individual tumours (a–c,f) or mice (g). Statistical analysis was performed using a linear mixed-effects model (a,c) or a two-sided (b) or one-sided (f,g) Wilcoxon–Mann–Whitney test. *P < 0.05, **P < 0.01, ***P < 0.001. Further details on statistical analysis are provided in Supplementary File 1. Scale bars, 500 µm (b (left)), 50 µm (b (right), c (right)), 200 µm (c (left), e). Source Data

Fig. 5. Retrospective analysis of menstrual cycle-dependent sensitivity to NAC in patients with BC.

a, Flowchart of case-selection processes for the HR⁺HER2⁻ BC cohort (cohort 1) and TNBC cohort (cohort2 ), showing every selection step. b,e, Evaluation of BC treatment response to NAC initiated during the progesterone-low (top) and progesterone-high phase (bottom) in patients with HR⁺HER2⁻ (b) and TNBC (e), based on Response Evaluation Criteria in Solid Tumours 1.1111 criteria. c,f, Change in size of target lesion post treatment compared with pretreatment, as a percentage, in patients with HR⁺HER2⁻ BC (c) and TNBC (f). Blue and red represent progesterone-low and -high conditions, respectively. Solid bars indicate patients from the Antoni van Leeuwenhoek Hospital-Netherlands Cancer Institute (AvL-NKI), striped bars indicate patients from the Dutch Breast Cancer Research Group DIRECT study, with data collected at the Clinical Research centre of Leiden University Medical Center (LUMC). Tumour (T), lymph node (N), metastases (M) staging of individual patients is indicated below bars. Dashed lines indicate −30% and +20%, which represent the thresholds for response and progression in HR⁺HER2⁻ tumours (c); −30% indicates minimal reduction in tumour size for a partial response classification in TNBC (f) based on Response Evaluation Criteria in Solid Tumours 1.1111 guidelines. d,g, Change in primary tumour size following NAC relative to tumour size before treatment, as a percentage in patients with HR⁺HER2⁻ BC (d) and TNBC (g). HR⁺HER2⁻: n = 17 (progesterone-low) and n = 13 (progesterone-high) at first NAC treatment; TNBC: n = 17 (progesterone-low) and n = 8 (progesterone-high) at first NAC treatment. Data shown as violin plots; thicker solid lines represent median, thinner solid lines indicate the 25th and 75th percentiles and dots represent individual patients. Statistical analysis was performed using two-sided Wilcoxon–Mann–Whitney testing. *P < 0.05, **P < 0.01. Further details on statistical analysis are provided in Supplementary File 1. Source Data

Extended Data Fig. 1. Oestrous cycle dependent proliferation of tumour cells during oestrus stage in genetic MMTV-PyMT model.

a, Left panel: Schematic of healthy mammary gland turnover during the oestrous cycle (centre) and representative immunofluorescent images (3D rendered) of tumour-free mammary ducts during dioestrus (left) and oestrus (right) stage. Right panel: representative immunofluorescent image of tumour-free mammary ducts in an ovariectomized mouse. Mammary ducts are labelled with smooth muscle actin (white). Scale bars, 100 µm. b, Schematic representation of R26-CreERT and R26R-Confetti constructs, which were crossed with different tumour models to enable lineage tracing in a subset of tumour cells. Upon injection of low dose of tamoxifen, some tumour cells will recombine the confetti construct leading to stochastic expression of one of the four confetti colours. These single cells can be followed over multiple days using IVM. c, Clone growth rate of all quantified clones from multiday IVM imaging of the genetic MMTV-PyMT tumour model in physiologically cycling mice (top) or ovariectomized mice (bottom). Each line represents a clone, and coloured lines highlight growth of individual clones. Top panels show coordinated clonal growth, which is abolished in the ovariectomized conditions (bottom panels). d, Images of crystal violet stained vaginal smears in oestrus (left) and dioestrus (right) representative for n > 30 mice. The proportion of epithelial cells and leukocytes was used to determine the cycle stage. Scale bars, 100 µm. e, Serum levels of progesterone in different oestrous cycle stages, determined by cytology of the vaginal smear. n = 9 (oestrus) and n = 10 (dioestrus) mice. f, Representative immunofluorescent images of EdU incorporation in tumours derived from the genetic MMTV-PyMT mouse model during oestrus and dioestrus phase. Scale bars, 1 mm. g, Expression of the gene PyMT in different oestrous stages depicted as violin plot. n = 10 mice per oestrous stage. Data is depicted as violin plots thick solid line represents median and solid lines indicate the 25th and 75th percentiles (e,g). Dots represent different mice (e) and tumours from different mice (g). Statistical analysis was performed using one-sided (e) and two-sided (g) Wilcoxon-Mann-Whitney test. *P < 0.05. More details on statistical analysis can be found in Supplementary File 1. Source Data

Extended Data Fig. 2. Oestrous cycle dependent proliferation of tumour cells during oestrus stage can be reproduced in the MMTV-Wnt1 tumour model.

a, Schematic representation of the multiday intravital microscopy setup (left) with overview images (single Z-plane) (centre) of the first imaging session (top) and the following imaging session (bottom) that are presentative for 6 experiments. Rectangles indicate example clones represented over multiple imaging days (right). Scale bar, 100 µm (overview), 50 µm (zoom). b, Clone growth rate of all quantified clones from multiday IVM imaging of transplanted MMTV-Wnt1 model. Each line represents a clone, and coloured lines highlight coordinated growth of individual clones. c, Immunofluorescent images (left) and violin plots depicting quantification (right) of EdU incorporation in the genetic MMTV-Wnt1 model representative for n = 5 (oestrus) and n = 4 (dioestrus) mice. Scale bar, 100 µm. Data is depicted as violin plots, thick solid line represents median and solid lines indicate the 25th and 75th percentiles, each dot represents a tumour (c). Statistical analysis was performed using one-sided Wilcoxon-Mann-Whitney test (c). *P < 0.05. More details on statistical analysis can be found in Supplementary File 1. Source Data

Extended Data Fig. 3. Disruption of physiological oestrous cycle after chemotherapy.

a, Representative plots depicting the gating strategy to determine PI⁺ tumour cells by flow cytometry analysis. b, Representative examples of crystal violet stained vaginal smears in mice with a disrupted cycle after treatment with chemotherapy. c, H&E image of physiological murine ovary, indicating the different follicle stages: PrF: Primordial follicle; PF: Primary follicle; SF: Secondary follicle; AF: Antral follicle and CL: Corpus luteum. Images are representative for n = 10 treatment naïve mice per stage. d, H&E images of ovaries from mice pretreatment, 7 days after initial treatment or long-term (LT) after treatment at endpoint, oestrus (top) or dioestrus (bottom). Images are representative for n = 10 mice (oestrus, Day 0), n = 7 mice (oestrus, Day 7), n = 4 mice (oestrus, LT), n = 10 mice (dioestrus, Day 0), n = 7 mice (dioestrus, Day 7), n = 8 mice (dioestrus, LT). e, Bar graph depicting the amount of PrF: Primordial follicles; PF: Primary follicles; SF: Secondary follicles per mm² in ovaries of mice, pretreatment, 7 days after initial treatment or at endpoint of long-term (LT) experiment. n = O⁰ = 10; O⁷ = 7; O^LT = 4; D⁰ = 10; D⁷ = 7; D^LT = 8. Scale bars, 100 µm (b), 1 mm (c, overview image, d), 100 µm (c, zoom).

Extended Data Fig. 4. Differential response in oestrous cycle stages independent of T cell presence.

a,b, Representative immunohistochemistry staining and quantification of CD4⁺ T cells (a) and CD8⁺ T cells (b) in treatment naïve MMTV-PyMT mammary tumours during different stages of the oestrous cycle. n = 3 mice per oestrous stage and 2 tumours per animal. c, Schematic representation of the experimental setup. Mice were treated with depletion antibodies prior to doxorubicin treatment, which was initiated at a cumulative tumour mass of 500–750 mm³ in either oestrus (O) or dioestrus (D) stage. d, Representative immunohistochemistry staining of CD4⁺ T cells (top) and CD8⁺ T cells (bottom) in 36 tumours of 9 mice receiving T cell depletion antibodies (α-CD4 + α-CD8). e,f, Violin plots depicting percentage of dying tumour cells identified by positive staining for propidium iodide (e) and tumour volume change compared to baseline (f) for mice receiving T cell depletion antibodies (α-CD4 + α-CD8) and treated with doxorubicin. n = O^a-CD4/8 = 5 mice/20 tumours, D^a-CD4/8 = 4 mice/16 tumours. Data in a,b,e,f is depicted as violin plots, thick solid line represents median and solid lines indicate the 25th and 75th percentiles. Dots represent individual tumours (a,b,e) or mice (f). Scale bars, 500 µm (a,b,d). Statistical analysis was performed using linear mixed effects models (e), a two way anova (a,b) and a one-sided Wilcoxon-Mann-Whitney test (f). *P < 0.05. More details on statistical analysis can be found in Supplementary File 1. Source Data

Extended Data Fig. 5. Gating strategy for flow cytometry analysis of depletion experiments.

Representative plots depicting the gating strategy to determine PI⁺ tumour cells by flow cytometry analysis of depletion experiments.

Extended Data Fig. 6. Retrospective analysis of menstrual cycle-dependent sensitivity to neoadjuvant chemotherapy in BC patients.

a, Patients with HR⁺HER2⁻ BC were divided into the progesterone-low (<2.2 nmol/L) or progesterone-high (>2.2 nmol/L) group based on the progesterone level in their sera within five days of initiating NAC. b-f, Table and comparison of standard clinicopathological parameters of 30 premenopausal patients with HR⁺HER2⁻ BC. g, Patients with TNBC were divided into the progesterone-low (<2.2 nmol/L) or progesterone-high (>2.2 nmol/L) group based on the progesterone level in their sera within five days of initiating NAC. h-l, Table and comparison of standard clinicopathological parameters of 25 premenopausal patients with TNBC Data in c,d,e,i,j is depicted as violin plots, thick solid line represents median and solid lines indicate the 25th and 75th percentiles. Statistical analysis was performed using two way anova (c,d,e), or two-sided Wilcoxon-Mann-Whitney test (i,j). ^#P < 0.1. More details on statistical analysis can be found in Supplementary File 1. Source Data