Abstract
The invention is based on the modulation of cancer-associated fibroblasts (CAFs) in the treatment of proliferative disorders by radiotherapy. By reducing CAF sensitivity to cellular senescence adverse immune reactions upon radiotherapy of solid tumours and subsequent treatment resistance could be avoided. The invention pertains to senolytics or Interleukin 1 (IL1) signalling inhibitors for use in a method of treating solid tumours and in the treatment or prevention of inflammatory adverse effects upon radiation therapy in context of a cancer treatment. The invention optionally also pertains to senolytics or Interleukin 1 (IL1) signalling inhibitors for use in a method of treating solid tumours and in the treatment or prevention of inflammatory adverse effects upon chemotherapy and/or radiation therapy in context of a cancer treatment.
Claims
1. A compound for use in the treatment of a solid tumour in a subject, wherein the subject is treated by reducing sensitivity of cancer-associated fibroblasts (CAFs) to cellular senescence and concomitant or subsequent irradiation therapy targeting the solid tumour in the subject.
2. The compound for use of claim 1, wherein the compound is an IL-1 signalling inhibitor or a senolytic.
3. The compound for use of claim 1 or 2, wherein the CAFs are located adjacent to, or within, the solid tumour in the subject.
4. The compound for use of any one of claims 1 to 3, wherein the compound increases tumour sensitivity of, or reduces tumour resistance to, irradiation therapy, preferably by inhibiting irradiation induced cell senescence in CAFs.
5. The compound for use of any one of claims 1 to 4, wherein the compound is an inhibitor/antagonist of inflammatory CAF polarization.
6. The compound for use of any one of claims 1 to 5, wherein the compound is an IL-1 inhibitor selected from rilonacept, anakinra, or biosimilars thereof, or an IL-1 or IL-1R antibody (preferably IL-1α/IL-1RA), such antibody may be selected from anti-IL-1R1 antibody AMG 108 (Amgen), chimeric, Canakinumab, humanized or human antibody directed to IL-1 alpha or beta (such as CDP-484, Celltech) or the IL-1 receptor (for example, AMG-108, Amgen; R-1599, Roche).
7. The compound for use of any one of claims 1 to 6, wherein the solid tumour is characterized by the presence of CAFs.
8. The compound for use of any one of claims 1 to 7, wherein the compound is a senolytic, selected from venetoclax, or Quercetin, Fisetin, Luteolin, Curcumin, A1331852, A1155463, Geldanamycin, Tanespimycin, Alvespimycin, Piperlongumine, FOXO4-related peptide, or Nutlin3a.
9. The compound for use of any one of claims 1 to 8, wherein the solid tumour is resistant to irradiation therapy.
10. The compound for use of any one of claims 1 to 9, wherein the subject has already received a first-line irradiation therapy, and has developed a resistance or reduced sensitivity to irradiation therapy.
11. A compound for use in treating inflammatory side effects of irradiation therapy in a subject suffering from a solid tumour disease, wherein the compound is an IL-1/IL-1R inhibitor or senolytic.
12. The compound for use of claim 11, wherein the compound is administered to the subject in an amount effective to inhibit irradiation induced cell senescence in CAFs of the tumour.
13. The compound for use of any one of the preceding claims, wherein the compound when administered to the subject in said therapeutically effective amount has no irradiation-independent anti-tumour effect.
14. A method for identifying or characterizing a compound suitable for increasing sensitivity of a solid tumour disease to irradiation therapy, the method comprising the steps of (a) Contacting a candidate compound and a cancer-associated fibroblast (CAF), (b) Inducing a p53 senescence program in the CAF contacted with the candidate compound, wherein a reduced pro-inflammatory phenotype in the CAF contacted with the candidate compound compared to a CAF not contacted with the candidate compound indicates the candidate compound suitable for increasing sensitivity of a solid tumour disease to irradiation therapy.
15. The method of claim 14, wherein step (b) involves induction of IL-1 signalling, for example using IL-1α, in the CAF or irradiation of the CAF.
16. A compound for use in the treatment of a solid tumour in a subject, wherein the subject is treated by reducing sensitivity of cancer-associated fibroblasts (CAFs) to cellular senescence and concomitant or subsequent cancer therapy targeting the solid tumour in the subject, wherein the cancer therapy is a chemotherapy and/or irradiation therapy.
17. The compound for use of claim 16, wherein the compound is the compound according to any one of claims 1 to 10.
18. The compound for use of claim 16 or 17, wherein the irradiation therapy and the chemotherapy are administered to the subject concomitantly or subsequently.
19. The compound for use of any one of claims 16 to 18, wherein the compound increases tumour sensitivity of, or reduces tumour resistance to, irradiation therapy and/or chemotherapy, preferably by inhibiting cell senescence in CAFs induced by irradiation therapy and or chemotherapy.
20. The compound for use of any one of claims 16 to 19, wherein the solid tumour is resistant to irradiation therapy and/or wherein the solid tumour is resistant to chemotherapy.
21. The compound for use of any one of claims 16 to 20, wherein the subject has already received a first-line irradiation therapy and/or chemotherapy, and has developed a resistance or reduced sensitivity to irradiation therapy and/or chemotherapy.
22. A compound for use in treating inflammatory side effects of cancer therapy in a subject suffering from a solid tumour disease, wherein the compound is an IL-1/IL-1R inhibitor or senolytic, wherein the cancer therapy is a irradiation therapy and/or chemotherapy.
23. The compound for use of claim 22, wherein the compound is administered to the subject in an amount effective to inhibit irradiation therapy and/or chemotherapy induced cell senescence in CAFs of the tumour.
24. A method for identifying or characterizing a compound suitable for increasing sensitivity of a solid tumour disease to cancer therapy, wherein the cancer therapy is irradiation therapy and/or chemotherapy, the method comprising the steps of (i) Contacting a candidate compound and a cancer-associated fibroblast (CAF), (ii) Inducing a p53 senescence program in the CAF contacted with the candidate compound, wherein a reduced pro-inflammatory phenotype in the CAF contacted with the candidate compound compared to a CAF not contacted with the candidate compound indicates the candidate compound suitable for increasing sensitivity of a solid tumour disease to irradiation therapy and/or chemotherapy.
25. The method of claim 24, wherein step (ii) involves induction of IL-1 signalling, for example using IL-1α, in the CAF or irradiation of the CAF.
Description
BRIEF DESCRIPTION OF THE FIGURES
[0103] The figures show:
[0104] FIG. 1: shows that fibroblasts predict survival of rectal cancer patients. (A) Relative distribution and Kaplan-Meier-Survival curves of 212 rectal cancer patients according to their CMS classification. Note that only patients that were identified unanimously by at least two of the different algorithms were included into analysis. (B and C) Global proteomics analysis of laser capture microdissected tumour cells from pre-therapeutic biopsies of 29 pCR and 32 non-pCR patients using mass spectrometry: (B) heatmap; (C) PCA plot. (D and E) Representative multiplex immunofluorescent analysis of tumour biopsies from pCR and non-pCR patients before CRT; blue: DAPI; red: αSMA; white: CD45; green: Vimentin; magenta: panCK; scale bars=200 μm; (E) quantification of: CD163, CD3, CD4, CD8, FOXP3, αSMA Vim and Ki67 panCK % of positive cells. Data are mean±SD. N=16 pCR and 19 non pCR patients, ** p<0.01 by t-test. (F) Gene set enrichment analysis (GSEA) for CAF, EMT, inflammatory, CMS4, endothelial cells and leukocytes signatures using pre-CRT RNA sequencing data of 105 patients' biopsies: 26 pCR & 79 non-pCR. Significance is for NES≤−1; FDR q-value<0.05. (G) Kaplan-Meier Disease Free Survival (DFS) curve of patients with high (n=23) and low (n=23) CAFs (αSMA+Vim+double positive cells) score based on the multiplexed immunohistochemical analysis of pre-CRT sections. Median split for high and low scores. * p<0.05 by Log-rank (Mantel-Cox) test.
[0105] FIG. 2: shows orthotopic mouse model of rectal cancer. (A and B) Schematic representation of the orthotopic transplantation of APTKA & APTK mouse organoids into C57BL/6 mice followed by local fractionated radiotherapy (5×2 Gy) of established tumours and end point colonoscopy 21 days after the last irradiation. (C) Heat map analysis of RNA sequencing data of APTKA (n=4) and APTK (n=5) orthotopic tumours. (D) GSEA of CAF signature using RNA sequencing data of APTKA and APTK orthotopic tumours. Significant NES≤−1; FDR q-value<0.05. (E) H&E stained sections of APTK and APTKA untreated and irradiated orthotopic tumours 21 days after last irradiation. Scale bars=500 μm. (F and G) Overall tumour (mm3) and invasive area (%) of untreated (n=5) and irradiated (n=6) APTK orthotopic tumours. Data are mean±SD. *** p<0.001 and **** p<0.0001 by t-test. (H and I) Overall tumour (mm3) and invasive area (%) of untreated (n=5) and irradiated (n=5) APTKA orthotopic tumours. ** p<0.01 by t-test. Data are mean±SD. (J) Frequency of mice with liver metastasis in untreated or irradiated APTKA orthotopic tumours (n=5 mice per group). (K) H&E staining of untreated and irradiated APTKA orthotopic tumours. Scale bars=300 μm. Representative images of αSMA (Green), Vimentin (Blue) and Collagen (Yellow) multiplex immunofluorescence of untreated and irradiated APTKA tumours. Scale bars=400 μm. Sirius red staining of APTKA orthotopic tumours upon radiotherapy. Scale bars=300 μm. (L) Percent of αSMA, Vim and Col1 triple positive cells in untreated (n=9) and irradiated (n=10) APTKA tumours. Data are mean±SD. ** p<0.01 by t-test. (M) Sirius red pixel positivity in untreated (n=7) and irradiated (n=6) APTKA tumours. Data are mean±SD. * p<0.05 by t-test. (N-P) Ex vivo irradiation (5×2Gy) of APTKA and APTK organoids: (N) relative fluorescent units of untreated and irradiated organoids, 6 h post last dose, using resazurin cell viability kit (n=4 independent wells per condition); (O) quantification of the number of fully grown untreated and irradiated organoids 4 days' post first reseeding (n=7 to 9 independent wells per condition); (P) subcutaneous tumour growth in C57BL/6 mice transplanted with untreated or irradiated organoids. Data are mean±SD. N=6 mice per group; p<0.05 by one-way ANOVA followed by Tukey's multiple comparisons test. Representative results from one experiment which has been independently repeated least 3 times.
[0106] FIG. 3: shows that resistant tumours induce inflammatory CAF polarization in an IL-1α dependent manner. (A) Schematic illustration of APTKA or APTK organoids supernatant collection and treatment of primary mouse colon fibroblasts of unchallenged wildtype C57BL/6 mice. (B-E) RNA sequencing analysis of unchallenged (n=3), APTKA (n=4) or APTK supernatant (n=4) treated fibroblasts for 24 h: (B) Heat map of differential transcriptional profiles; (C and D) GSEA of Gene Ontology (GO) and CAF: inflammatory (iCAF) and myofibroblasts (myCAF) signatures, significantly enriched in APTKA compared to APTK media treated fibroblasts (15). NES≤−1 and FDR q-Value<0.05; (E) Ingenuity-pathway interactive analysis of APTKA polarized fibroblasts transcriptomic profile suggesting NF-κB (activation z-score=−4.64, p-value<0.05) and p38 (activation z-score=−4.236, p<0.05) as upstream regulators of differentially expressed genes compared to APTK polarized fibroblasts. (F) Relative expression analysis by RT-qPCR of inflammatory genes in untreated wildtype and APTKA conditioned medium treated fibroblasts in the presence of IKKß inhibitor ML120B (30 μM) and p38 MAPK inhibitor SB203580 (10 μM) for 24 h (n=4 independent wells per group). Data are mean±SD. p<0.05 by one-way ANOVA followed by Tukey's multiple comparisons test. Representative results from one experiment which has been independently repeated 3 times. (G and H) RNA sequencing analysis of APTKA (n=3) and APTK (n=3) organoids: (G) heat map analysis; (H) differentially expressed cytokines in APTKA (Log 2 fold change≤−1) compared to APTK organoids (log 2 fold Change≥1). p-value<0.05. (I) RT-qPCR analysis of inflammatory genes in APTKA conditioned fibroblasts treated with control IgG, α-IL-1α, α-IL-33, α-TNF-α, α-BAFF or α-CXCL16 neutralizing antibodies (2 μg/mL each) for 24 h (n=4 independent wells per group). Data are mean±SD. **** p<0.0001 by one-way ANOVA followed by Tukey's multiple comparisons test. Experiment performed twice. (J) RNA expression profiles of inflammatory genes in APTKA conditioned fibroblasts stimulated for 24 h then subjected to anakinra (10 μg/mL), APTKA fresh or regular medium treatments for another 24 h (n=4 independent wells per condition). Data are mean±SD; p<0.05 by one-way ANOVA followed by Tukey's multiple comparisons test. Representative results from one experiment which has been independently repeated 2 times.
[0107] FIG. 4: shows that IL-1 is responsible for therapy resistance in orthotopic tumours. (A) Schematic representation of the orthotopic transplantation of APTKA mouse organoids into C57BL/6 mice followed by 5×2 Gy radiotherapy of established tumours untreated or treated daily with anakinra (500 μg/day) for 21 days. (B and C) Overall tumour size (mm3) and % invasive area of 5×2Gy radiated APTKA orthotopic tumours 21 days after last irradiation without (n=5) or with (n=4) anakinra treatment (500 μg/day). Data are mean±SD; p<0.05 by t-test. (D) Frequency of mice with liver metastasis upon irradiation of APTKA orthotopic tumours without (n=5) or with (n=4) anakinra treatment (500 μg/day). (E and F) Sirius red staining of untreated (n=3) or anakinra treated (n=4) 5×2 Gy irradiated APTKA tumours. Scale bars=200 μm. Data are mean±SD. *** p<0.001 by t-test. (G, H) αSMA immunofluorescent staining and its quantification of irradiated APTKA orthotopic tumours without (n=5) or with (n=4) anakinra treatment (500 μg/day). Scale bars=200 μm. Data are mean±SD; p<0.05 by t-test. (I) Schematic representation of IL-1α overexpression in APTK organoids by CRISPRa/Cas9. (J) ELISA of IL-1α protein levels in conditioned media from APTK sgctrl and APTK sgIL-1α organoids (n=4 independent wells per group). Data are mean±SD. **** p<0.0001 by t-test. (K) Relative RNA expression of inflammatory genes in fibroblasts treated with conditioned media from APTK sgctrl and APTK sgIL-1α organoids for 24 h (n=5 independent wells per condition). Data are mean±SD. **** p<0.0001 by t-test. Representative results from one experiment which has been independently repeated 2 times. (L) Colonoscopy of APTK sgIL-1α orthotopic tumour before (untreated) and 21 days after irradiation (5×2 Gy). (M) % of invasive area in untreated and 5×2 Gy irradiated APTK sgIL-1α orthotopic tumours (n=3 mice per group). Data are mean±SD. ** p<0.01 by t-test. (N) Quantification of sirius red staining of untreated and irradiated APTK sgIL-1α orthotopic tumours (n=3 mice per group). Data are mean±SD. ** p-value=0.01 by t-test.
[0108] FIG. 5: shows that irradiation of inflammatory fibroblasts induces senescence. (A) Schematic illustration of APTKA organoids supernatant collection and treatment of colon fibroblasts from unchallenged wildtype C57BL/6 mice for 12 h followed by 3×2 Gy irradiation. Microscopic, proteomic & transcriptomic analysis were performed 72 h post first irradiation. (B) Microscopic captures of non-irradiated wildtype (WT) and APTKA conditioned fibroblasts. Remaining captures belong to irradiated fibroblasts: WT, IL-1α (1 μg/mL) stimulated WT, and APTKA conditioned media minus or plus anakinra (20 μg/mL) treated fibroblasts. Scale bars=200 μm. (C) Mass spectrometric analysis of matrisome profiles of 3×2 Gy irradiated (n=8) and non-irradiated (n=8) APTKA conditioned fibroblasts. Heat map represents significantly enriched (student's t-test q-value<0.05) ECM related proteins. (D-F) RNA sequencing analysis of untreated (n=4) and irradiated (n=4) APTKA polarized fibroblasts: (D) heat map analysis; (E) GSEA in irradiated (NES≥1, FDR q-Value<0.05) compared to untreated (NES≤−1, FDR q-Value<0.05) APTKA polarized fibroblasts; (F) ingenuity-pathway interactive analysis of radiated APTKA fibroblasts transcriptomic profile suggesting TP53 (activation z-score=5.882, p-value<0.05) as upstream regulator. (G) Immunoblot analysis of untreated and irradiated APTKA conditioned fibroblasts. Representative results from one experiment which has been independently repeated 2 times. (H) SA-ß gal staining of untreated and irradiated APTKA fibroblasts. Scale bars=100 μm. (I and J) p21 and pH2AX immunofluorescent staining of untreated and irradiated APTKA fibroblasts. Scale bars=200 m. Data are mean±SD. ** p<0.001 and *** p<0.001 by t-test. (K) Nitrite concentration determined by Griess assay in conditioned media from WT, APTK media, APTKA media, IL-1α (1 μg/mL) stimulated and APTKA plus anakinra (20 μg/mL) treated fibroblasts for 72 h (n=10 independent wells per group). Data are mean±SD. p-value<0.05 by one-way ANOVA followed by Tukey's multiple comparisons test. Representative results from three independent experiments. (L) 8-OHDG immunofluorescent staining of WT, IL-1α (1 μg/mL) stimulated, APTK media, APTKA media minus or plus NOS2 inhibitor W1400 (1 mM) treated fibroblasts for 72 h (n=5 independent wells per group). Scale bars=200 μm. Data are mean±SD. p<0.05 by one-way ANOVA followed by Tukey's multiple comparisons test. (M) p21 immunohistochemistry of untreated (n=7) and irradiated (n=9) APTKA orthotopic tumours 21 days' post last dose RT. Scale bars represent 200 μm. Data are mean±SD. * p<0.05 by t-test. (N) Colonoscopy of APTKA orthotopic tumour before therapy and 21 days' post 5×2 Gy irradiation and daily venetoclax treatment (100 mg/kg/day). (O and P) Overall tumour size and % of invasive area of venetoclax treated non irradiated (n=5) and irradiated (n=6) APTKA orthotopic tumours 21 days' post RT. Data are mean f SD. P-value<0.05 by t-test. (Q) Frequency of liver metastasis in venetoclax treated APTKA tumours without (n=5) or with (n=6) radiotherapy. (R) Sirius red staining of venetoclax treated APTKA tumours without (n=5) or with (n=5) 5×2 Gy radiotherapy. Data are mean±SD. p<0.05 by t-test. (S and T) αSMA (red) Vimentin (green) immunofluorescent staining of venetoclax treated APTKA tumours without (n=5) or with (n=5) irradiation. Scale bars=200 μm. Data are mean±SD. **p<0.001 by t-test.
[0109] FIG. 6: shows that low IL-1RA levels enhance IL-1 signaling in rectal cancer patients and sensitize CAFs to therapy-induced senescence. (A and B) Quantification of IL-1 and IL-1RA baseline serum levels in non-pCR (n=25) and pCR (n=7) patients before CRT. Data are mean±SD. ** p<0.01 by t-test. (C) IL-1RA serum level association with two SNPs: rs4251961 (T/C) and rs579543 (A/G) using DNA isolated from patients (n=31) pre-CRT peripheral blood mononuclear cells. Data are mean±SD. ** p=0.0052 by t-test. (D) Kaplan-Meier DFS curve of patients with high (n=37) and low (n=37) IL-1RN expression score (median split) based on the RNA sequencing data of pre-CRT biopsies. * p<0.05 by Log-rank (Mantel-Cox) test. (E and G) Paired analysis of non-pCR patients (n=12) post-CRT versus pre-CRT RNA sequencing data: (E) Heat map; (F) PCA Plot and (G) GSEA of EMT, ECM, collagen formation, senescence & CAF matrisome signatures, significantly enriched post-therapy (NES≥1 & FDR q-value<0.05). (H) Paired analysis of sirius red staining of non-pCR patients (n=11) pre versus post-CRT. Scale bars=300 μm. (I) Paired quantification of p21 immunohistochemistry staining of non-pCR patients (n=10) pre versus post CRT. (J) Correlation analysis of pre-CRT serum IL-1RA levels with ratio of p21 induction (post/pre-CRT p21 RNA expression level) (n=10 non pCR patients). * p<0.05 by Spearman r correlation test. (K) IL-1RA and IL-1α protein levels in conditioned media from T3 and T13 human rectal cancer organoids (n=4 independent wells per group). Data are mean±SD. ** p-value<0.05 by t-test. (L) RT-qPCR analysis of inflammatory genes in untreated, IL-1RA high & IL-1RA low conditioned media treated human fibroblasts (n=4 independent wells per group). Data are mean±SD. Significant p<0.05 by one-way ANOVA followed by Tukey's multiple comparisons test. Representative results from one experiment which has been independently repeated at 2 times. (M) Nitrite concentration determined by Griess assay in supernatants from WT, IL-1RA high & IL-1RA low conditioned media treated human fibroblasts for 72 h (n=13 independent wells per group). Data are mean±SD. Significant p-value<0.05 by one-way ANOVA followed by Tukey's multiple comparisons test. (N) Nitrite concentration determined by Griess assay in supernatants from untreated and IL-1α (1 μg/mL) stimulated human fibroblasts for 72 h (n=32 independent wells per group representing 3 independent experiments). Data are mean±SD. Significant p-value<0.05 by t-test. (O) SA-ß gal staining of 3×2 Gy irradiated: WT, IL-1Ra high, IL1-Ra low with or without anakinra (20 μg/mL) conditioned media treated human fibroblasts for 72 h. Scale bars=100 μm. (P) Summarizing scheme: In patients with low circulating IL-1RA levels, tumour cell derived IL-1α induces iCAF polarization and upregulation of iNOS, which elevates nitrite production in CAFs resulting in oxidative DNA damage. Upon irradiation DNA damage is further enhanced and triggers CAF senescence. Senescent CAFs change their secretory profile and produce more ECM, which counteracts irradiation-induced cell death and tumours, thus promoting tumour growth and inducing therapy resistance.
[0110] FIG. 7: (A) APTKA organoids orthotopic transplantation into C57BL/6 mice followed by 5-fluorouracil (5-FU) chemotherapy (20 mg/Kg/day every 4 days for 21 days—experiment end point). Tumour area (mm3), % invasive area and liver metastasis frequency in untreated and 5-FU treated APTKA tumours (n=2 mice per group). Significant p-value<0.05 by t-test. (B and C) p21, pH2Ax IF and SA-ß gal staining quantifications and representative SA-ß gal staining images among untreated and 5-FU treated IL-1α (1 μg/mL) stimulated human fibroblasts and APTKA conditioned medium treated mouse fibroblasts. Significant p-value<0.05. Results representative of 4 independent experiments. Scale bars=100 μm.
EXAMPLES
[0111] Certain aspects and embodiments of the invention will now be illustrated by way of example and with reference to the description, figures and tables set out herein. Such examples of the methods, uses and other aspects of the present invention are representative only, and should not be taken to limit the scope of the present invention to only such representative examples.
[0112] The examples show:
Example 1: Investigation of the Role of Fibroblasts for CRT Response in Distinct Patient Cohorts in Comparison to Other Cells and Biomarkers
[0113] The recently established consensus molecular subtype (CMS) classification for colon cancer based on transcriptomic profiles of whole tumours demonstrated that tumour subtyping allows prognosis prediction to some extent (8, 9).
[0114] Yet, when survival of 212 rectal cancer patients was compared, treated with neoadjuvant CRT and surgery and grouped according to three CMS classifier algorithms (CMSclassifier-random forest prediction, CMSclassifier-single sample prediction and CMScaller), a comparable outcome for the different subgroups (FIG. 1A) was noticed indicating that subtyping of rectal cancer patients according to the CMS criteria does not allow DFS prediction (8, 9).
[0115] To identify possible predictive markers and signaling pathways that may be responsible for therapy response, a comprehensive global proteomic analysis of laser capture microdissected tumour cells derived from pre-therapeutic biopsies of 29 pCR (i.e. excellent prognosis) was performed after CRT and 32 patients with non-pCR (i.e. residual lymph node metastases with poor prognosis) after CRT using mass spectrometry. A total of 2747 proteins could be identified, however, principal component and heat map analysis of protein expression profiles did not reveal a distinct separation of the two patient cohorts (FIG. 1B, C) suggesting that possibly cells in the tumour microenvironment rather than the actual tumour cells may determine CRT response.
[0116] Indeed, a detailed multiplex fluorescent immunohistochemical analysis of pre-therapeutic biopsy samples of patients that either showed pCR or not revealed a significant enrichment of alpha-smooth muscle actin α-SMA+/Vimentin+fibroblasts in pre-therapeutic biopsies from non-pCR patients (FIG. 1D, E), while there was no difference in the number of T cells (CD3+, CD4+, CD8+ or Foxp3), macrophages (CD163) or Ki-67+epithelial cells (panCK) (FIG. 1D, E). Moreover, an enrichment of a cancer-associated fibroblast (CAF) signature as well as an EMT and an inflammatory signature in transcriptomes of pretreatment biopsies from non-pCR patients (FIG. 1F) could be identified.
[0117] In contrast, there was no enrichment for an endothelial signature or a leukocyte signature in non-pCR patients at baseline further highlighting the relevance of fibroblasts for an impaired therapy response of these patients (10). Importantly, patients with high tumour expression of alpha-smooth muscle actin α-SMA+/Vimentin+fibroblasts in the pre-therapeutic biopsies had a significantly worse DFS compared to patients with low expression (FIG. 1G). Surprisingly, yet in line with the lack of difference in DFS (FIG. 1A), no enrichment in a mesenchymal CMS4 signature was also found in the two patient cohorts (11). However, the presence of an enriched inflammatory signature suggested that a distinct CAF polarization profile may confer resistance to CRT (12).
Example 2: Confirmation of Importance of Inflammatory CAFs by a Novel Preclinical Model of Rectal Cancer
[0118] To functionally confirm the importance of inflammatory CAFs for rectal cancer therapy and to dissect the underlying mechanism, a novel preclinical model of rectal cancer that allows monitoring of the effects of local irradiation in vivo was established. In this model tumour organoids are orthotopically transplanted into the distal lumen of C57BL/6 mice. Once growth of single tumours is confirmed by mini-endoscopy, mice undergo an image-guided radiation therapy using a small animal radiation research platform (SARRP) (FIG. 2A).
[0119] The inventors could recently demonstrate that orthotopic transplantation of organoids mutant for Apc, Trp53, Tgfbr2 and K-rasG12D (APTK) induce single invasive tumours in the distal colorectum that spontaneously metastasize to the liver in about 10% (13). Yet, transplantation of APTK organoids that in addition express myristoylated AKT (APTKA) develop tumours that are characterized by a more pronounced stromal response associated with a more aggressive phenotype and a higher frequency of spontaneous liver metastases (˜60%) (13). Remarkably, APTKA-tumours display an enrichment of CAF-associated genes (FIG. 2B-D) resembling the enhanced stroma reaction seen in the pre-therapeutic biopsies of patients with non-pCR after CRT.
[0120] The response of either APTK or APTKA-derived rectal tumours to a fractionated irradiation schedule (5×2 Gy) was compared in vivo, APTK-derived tumours responded well to radiation therapy (FIG. 2E) and showed a massive reduction in tumour size 21 days after therapy initiation (FIG. 2F, G). In stark contrast, APTKA-induced tumours were completely resistant to local irradiation (FIG. 2E) and displayed even more invasion and a higher frequency of liver metastases after irradiation when compared to untreated APTKA tumours (FIG. 2H-J). Interestingly, this was paralleled by a further increase of α-SMA+/vimentin+/collagen-1+(Col-1) fibroblasts and a change in fibroblast morphology that appeared in a more aligned pattern (FIG. 2K, L), which is usually associated with a change in extracellular matrix constituents and enhanced tumour progression (14). Increased collagen content was confirmed by sirius red staining (FIG. 2K, M). Importantly, however, APTK and APTKA organoids were equally sensitive to 5×2 Gy irradiation when treated ex vivo (FIG. 2N, O) and tumour growth was markedly retarded in both cases when organoids were injected subcutaneously following ex vivo irradiation (FIG. 2P) strongly supporting the notion that APTKA organoids were not intrinsically resistant to irradiation, but that this was rather mediated by stromal cells when tumour organoids were grown and treated in vivo.
[0121] To examine whether APTK and APTKA tumour organoids differentially affect surrounding fibroblasts in tumours directly in a paracrine manner, supernatants were collected from both tumour organoid lines and treated primary intestinal fibroblasts of unchallenged wildtype mice for 24 hours with the two different tumour-conditioned media ex vivo (FIG. 3A). RNAseq analysis confirmed distinct gene expression patterns in fibroblasts exposed to the two organoid supernatants (FIG. 3B). Gene set enrichment analysis (GSEA) revealed a strong inflammatory polarization in fibroblasts exposed to APTKA conditioned supernatants (FIG. 3C, D) and a distinct enrichment of inflammatory CAF (iCAF) but not myofibroblasts (myCAF)-associated genes (FIG. 3D), thus resembling the phenotype which was found in whole tumours in vivo (15).
[0122] Ingenuity-pathway analysis (IPA) suggested NF-κB and p38 activation as the main signaling pathways responsible for the inflammatory gene expression profile (FIG. 3E). Indeed, treating wildtype fibroblasts with APTKA conditioned medium in the presence of the specific IKKß inhibitors ML120B or the p38 inhibitor SB203580 significantly inhibited the induction of pro-inflammatory genes such as Cxcl1, Cxcl2, Cxcl5, Mmp9, Mmp13, Il1a, 116, Tnfa and Nos2 (FIG. 3F).
[0123] The inventors then aimed to identify the upstream factors secreted by APTKA organoids that were responsible for activating NF-κB and p38 in fibroblasts. To this end RNAseq analysis of APTK and APTKA tumour organoids were performed and identified several differentially expressed genes encoding cytokines (FIG. 3G, H). Using neutralizing antibodies, NF-κB and p38 inducing cytokines TNFα, IL-1α, IL-33, BAFF or CXCL16 were blocked in APTKA supernatants and expression of Cxcl1, Cxcl2, Cxcl5 and Ilia as inflammatory marker genes in fibroblasts was examined. In line with the previously described role of IL-1 in inflammatory CAF polarization in pancreatic cancer (16), only inhibition of IL-1α prevented upregulation of pro-inflammatory genes in fibroblasts challenged with conditioned medium from APTKA organoids (FIG. 3I), which could be confirmed using recombinant IL-ra (anakinra) or Il-1r−/− fibroblasts (data not shown). Conversely, recombinant IL-1α alone sufficed to induce a similar transcriptomic profile as APTKA supernatant in fibroblasts (data not shown).
[0124] It was then examined whether polarization towards an inflammatory CAF represented a stable state or whether blocking IL-1α does not only prevent but may also allow reversion of the established inflammatory phenotype as this would have implications for the therapy of tumours with a distinct CAF profile. To do so, fibroblasts were stimulated for 24 hours with APTKA conditioned medium before adding anakinra, adding fresh APTKA medium or changing medium to regular medium. Interestingly, simply replacing APTKA medium with regular medium reduced gene expression of Cxcl1, Cxcl5 and Mmp13 in fibroblasts (FIG. 3J). Moreover, also addition of anakinra to APTKA medium completely reverted expression of these genes indicating that organoid-derived IL-1α was required to maintain CAF polarization.
Example 3: Investigation of Improving Response to Radiation Therapy by IL-1α Blockade
[0125] To examine whether IL-1α blockade would improve the response to radiation therapy of APTKA tumours in vivo, mice were treated with anakinra (500 μg/day) once tumours had established (FIG. 4A). While anakinra administration alone did not affect APTKA tumour growth significantly (data not shown), it led to a marked reduction in tumour size and invasion (FIG. 4B, C) and blocked metastasis formation when APTKA tumours were irradiated (FIG. 4D). This was accompanied by a significant decrease in sirius red staining (FIG. 4E, F) as well as a decreased number of αSMA+cells (FIG. 4G, H) indicating that anakinra had a pronounced impact on CAF activation and suggested that tumour cell derived IL-1α represented a key factor in the development of therapy resistance of murine rectal tumours. To further confirm this, Ilia transcription was activated in irradiation-sensitive APTK organoids by CRISPRa/Cas9 (FIG. 4I, J). When fibroblasts were exposed to conditioned medium of APTK-sgIL-1α organoids they started expressing an inflammatory gene expression profile that was comparable to the APTKA-induced one (FIG. 4K). Moreover, increased IL-1α release by APTK organoids rendered the corresponding tumours resistant to radiation therapy in the orthotopic in vivo model (FIG. 4L). APTK-sgIL-1α tumours did not decrease in size but became more invasive (FIG. 4M). Similar to APTKA-tumours, APTK-sgIL-1α tumours displayed an enhanced stroma reaction upon irradiation, characterized by an increased sirius red staining (FIG. 4N). Collectively, these data supported the conclusion that tumour cell-derived IL-1α polarizes CAFs towards an inflammatory phenotype, which then causes resistance to irradiation.
[0126] To explore how CAFs would confer therapy resistance in our model, it was examined whether exposure of fibroblasts to conditioned medium of irradiated tumour organoids would change the CAF phenotype. However, RNAseq analysis did not reveal any marked changes in gene expression when fibroblasts were compared that were subjected to medium derived from irradiated or non-irradiated organoids (Ext FIG. S2B). Therefore, it was tested whether APTK and APTKA-polarized fibroblasts themselves may be affected differently by irradiation. Primary fibroblasts were treated with APTK or APTKA conditioned medium or left them untreated and then subjected them to a fractionated irradiation (3×2Gy; FIG. 5A). This led to marked morphological differences between the differently treated fibroblasts and when compared to unchallenged or APTK-polarized fibroblasts, APTKA-educated fibroblasts developed a more elongated, spindle like shape (FIG. 5B) that clearly resembled the phenotype observed in vivo (FIG. 2K). Administration of anakinra to wildtype fibroblasts or use of Il1-r−/− fibroblasts prevented the APTKA-induced phenotype, while single application of recombinant IL-1α was sufficient to cause a comparable morphology after irradiation (FIG. 5B and data not shown), thus, clearly demonstrating the IL-1α-dependent morphological change. Elongation of fibroblasts is associated with changes in ECM constituents (14). Therefore, supernatants of irradiated and non-irradiated APTKA-polarized fibroblasts were collected and the ECM components (matrisome) were characterized by mass spectrometry (17, 18). In line with the morphological changes, irradiation led to a significant increase in various ECM glycoproteins, collagens and proteoglycans as well as ECM-regulators and growth factors (FIG. 5C) many of which are known to confer substantial support for tumour progression (19, 20).
[0127] Furthermore, transcriptomes of irradiated and non-irradiated APTKA-educated fibroblasts were compared by RNAseq analysis, which revealed marked changes and a downregulation of cell cycle and cell cycle checkpoint associated genes as well as induction of genes controlling senescence (FIG. 5D, E). Moreover, IPA indicated that many of the induced genes including those encoding various secreted factors (extracellular proteins: collagens, proteoglycans, growth factors, ECM regulators, ECM affiliated proteins) that also had been identified in the matrisome analysis as well as the cell cycle inhibitor Cdkn1a were regulated in a p53-dependent manner (FIG. 5F) suggesting induction of a p53-dependent senescence in APTKA-polarized fibroblasts upon irradiation.
[0128] In agreement with this notion and the presence of a growth arrest of fibroblasts, immunoblot analysis confirmed upregulation of p21, cdk4, cyclin D2 and downregulation of cdk2 and cyclin A (FIG. 5G), while SA-βgal staining validated senescence in irradiated fibroblasts that had been exposed to APTKA conditioned medium before (FIG. 5H). Senescence induction was dependent on IL-1 and was accompanied by markedly enhanced DNA damage (FIG. 5I). To understand how IL-1α would sensitize fibroblasts to irradiation-induced senescence, the inventors hypothesized that IL-1α induced formation of reactive oxygen and nitrogen species (ROS and RNS), which in turn triggered oxidative DNA damage. Subsequent irradiation would enhance DNA damage further and initiate a p53-dependent senescence program.
[0129] However, it was not able to detect a higher level of ROS in fibroblasts exposed to APTKA-conditioned medium (data not shown). In line with meeting the altered energetic demand, due to decreased proliferation and increased quiescence, mitochondrial function in APTKA-conditioned fibroblasts was declined. Cells displayed lower oxygen consumption and extracellular acidification rate as well as lowered maximal respiration and reserve respiratory capacity. Interestingly, also basal glycolysis as well as compensatory glycolysis, when mitochondrial respiration was blocked, remained lower in these cells. In contrast, 72 hours after exposure to APTKA-conditioned medium or to recombinant IL-1α and in line with enhanced Nos2 expression (FIG. 3F) nitrite levels in fibroblast supernatants were significantly elevated (FIG. 5K) which was paralleled by an increase in oxidative DNA damage determined by 8-OHdG immunofluorescence (FIG. 5L). Administration of the Nos2 inhibitor W1400 blocked nitrite production and prevented oxidative DNA damage as well as senescence induction in fibroblasts that had been treated with APTKA-conditioned medium (FIG. 5L). In agreement with these findings, a significant increase in nuclear p21 in stroma cells of APTKA orthotopic tumours upon irradiation was detected (FIG. 5M) indicating that a senescence program was initiated also in vivo.
[0130] The inventors then aimed to confirm the functional relevance of CAF senescence for therapy resistance of APTKA-tumours and treated tumour bearing mice with the senolytic compound venetoclax (100 mg/kg/day) in addition to RT. Indeed, targeting senescent cells sensitized APTKA-tumours to irradiation and led to a significant decrease in tumour load (FIG. 5N, O) and suppressed invasion and formation of liver metastases (FIG. 5P, Q) which was accompanied by a decrease in sirius red staining and α-SMA+/vimentin+cell number (FIG. 5R-T). Collectively, these results confirmed that IL-1α-dependent polarization of inflammatory CAFs elevated nitrite production and oxidative DNA damage in these cells. Subsequent irradiation enhanced DNA damage further and triggered a p53-dependent senescence program including the secretion of cytokines and ECM constituents that support tumour invasion and metastasis of cancer cells to counteract the irradiation-induced tumour cell death.
[0131] Finally, the inventors aimed to examine whether the observed IL-1α-induced changes in the tumour microenvironment of our murine model would be associated with poor therapy response in rectal cancer patients. To this end, serum samples were analyzed using Luminex Bio-Plex assay and determined serum levels of IL-1α, IL-1β, IL-1RA as well as IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12, IL-13, IL-15, IL-17, CCL2, CCL3, CCL4, CCL5, TNFα, FGF, PDGF, GM-CSF and Eotaxin in non-pCR patients (n=25) and compared them to patients that showed a pCR (n=7) after CRT. While IL-1α levels were below the detection limit in all patients examined, it could not be observed a significant difference for IL-1P or for any of 25 other cytokines (FIG. 6A). However, serum IL-1RA levels were markedly reduced in non-pCR patients with residual lymph node metastases and poor prognosis after CRT (FIG. 6B), suggesting that these patients indeed were characterized by an enhanced IL-1 signaling due to lower expression of the antagonist. IL-RA levels in vivo have been reported to be frequently associated with two single nucleotide polymorphisms (SNPs) in IL-1RA (rs4251961 T/C and rs579543 C/T) with higher IL-RA levels being observed in carriers of the homozygous rs4251961 T/T and rs579543 G/G genotypes (21).
[0132] An elevated IL-1RA serum levels in homozygous rs4251961 T/T patients could be confirmed, while no correlation between IL-1RA serum levels and the different rs579543 genotypes (FIG. 6C) was observed. Moreover, all pCR patients analyzed were carriers of the protective rs4251961 T/T or T/C alleles and none of them was a homozygous C/C carrier. Importantly, lower baseline IL1RN gene expression in pre-therapeutic tumour biopsies was associated with a significantly worse DFS (FIG. 6D) supporting the relevance of enhanced IL-1 signaling in therapy response. Then RNA expression was compared before and after CRT in 12 non-pCR patients (FIG. 6E). Principle component analysis revealed that CRT led to a clear clustering of RNA expression profiles (FIG. 6F). Furthermore, CRT induced an EMT, ECM and collagen formation signature in these patients (FIG. 6G). When a cross-species comparison was performed non-pCR patients showed an enrichment of a senescence signature that was established in irradiated APTKA-conditioned fibroblasts as well as a clear expression of the genes identified in the murine CAF matrisome (FIG. 6G). CRT induced a pronounced stromal response determined by sirius red staining (FIG. 6H) and in line with induction of stromal cell senescence, p21 expression increased in all patients in response to CRT (FIG. 6I), while the ratio of transcriptional p21 induction (post CRT/pre CRT) correlated inversely with the respective serum IL-1RA levels (FIG. 6J), thus strongly supporting a link between IL-1 signaling and therapy-induced stromal senescence in rectal cancer patients.
[0133] To further corroborate this, patient derived organoids (PDO) from biopsies taken before initiation of CRT from non-pCR patients were established and it was focused on organoids that were characterized by different IL-1RA but comparable IL-1α levels (FIG. 6K). Incubation of human intestinal fibroblasts with conditioned supernatants from “IL-1RA low” organoids induced a stronger pro-inflammatory gene expression profile than supernatants from “IL-1RA high” organoids (FIG. 6L) and led to higher nitrite production by fibroblasts (FIG. 6M), which could also be confirmed when fibroblasts were stimulated with IL-1α only (FIG. 6N). Importantly, SA-β-gal staining of fibroblasts demonstrated that this was associated with a stronger senescence induction in “IL-1RA low” pre-conditioned fibroblasts upon irradiation, which could be blocked by anakinra confirming the IL-1 dependence of therapy-induced senescence in CAFs (FIG. 6O).
Example 4: Investigation of IL-1 Mediated Senescence Mechanism Associated with Chemotherapy
[0134] The inventors further investigated, whether therapy resistance of tumours during treatment with cytotoxic chemotherapy is also connected to the senescence behavior of CAFs. For this, the orthotopic APTKA organoid C57BL/6 mice of the inventors' murine model were divided in three groups and subjected to different treatments. The groups consisted of a control group of untreated mice, a second group that was administered 20 mg/kg/day of the cytotoxic chemotherapeutic agent 4-fluorouracil (5-FU) every 4 days and a third group that was treated with a therapy based on a concomitant administration of 5-FU and radiation therapy based on a fractionated irradiation schedule (5×2 Gy). After 21 days of treatment, parameters of tumour progression were determined. The mice treated with only 5-FU showed a significant increase in tumour area and tumour invasiveness compared to the untreated control group. Furthermore, the mice treated with 5-FU have developed liver metastases while the control group did not. Similarly, the group of mice treated with concomitant 5-FU administration and radiotherapy showed significant increase in tumour area and invasive area compared to the control group and every mouse of this group suffered from liver metastases (FIG. 7A). Overall, in vivo administration of the chemotherapy aggravated invasive growth and stimulated formation of liver metastases similarly to irradiation.
[0135] In staining experiments, the inventors quantified the presence of nuclear p21, nuclear phosphorylated H2AX (pH2AX) and SA-βgal in APTKA medium conditioned mouse fibroblasts that were either left untreated or treated with 5-FU. The inventors were able to detect a significant upregulation of SA-βgal and nuclear p21 in 5-FU treated CAFs when compared to the non-treated fibroblasts, thus indicating their senescent state. Furthermore, the significant increase of pH2AX in 5-FU treated CAFs when compared to the non-treated cells, indicates DNA damage in the treated 5-FU treated samples. (FIG. 7B). Overall, these experiments confirmed a chemotherapy induced senescence of the fibroblasts, when exposed to 5-FU.
[0136] Moreover, the inventors investigated the influence of chemotherapy on the IL-1α mediated changes to the tumour microenvironment in human cancer patients. Using staining experiments, the upregulation of pH2AX, p21, and SA-βgal in untreated and 5-FU treated IL-1α (1 μg/mL) stimulated human fibroblasts was quantified. Under these conditions, the IL-1α stimulated human fibroblasts exhibited a similar behavior as the APTKA medium conditioned mouse fibroblasts. The addition of the chemotherapeutic resulted in a significant increase of all three biomarkers and in view of the above examples confirming that the administration of 5-FU induces DNA damage and senescence in cancer-associated fibroblasts in an IL-1 dependent manner (FIG. 7C). Overall, the inventors were able to demonstrate that their previous findings were not limited to the application of local irradiation but could also be observed in cases when cytotoxic chemotherapy is applied. Thus, targeting IL-1 in combination with chemotherapy may provide a substantial benefit particularly in tumours that are known to contain inflammatory cancer-associated fibroblasts such as pancreatic cancer, breast cancer or head and neck cancers.
REFERENCES
[0137] The references are: [0138] 1. 1. F. R. Greten, S. I. Grivennikov, Inflammation and Cancer: Triggers, Mechanisms, and Consequences. Immunity 51, 27-41 (2019). [0139] 2. C. Rodel, R. Hofheinz, E. Fokas, Rectal cancer: Neoadjuvant chemoradiotherapy. Best Pract Res Clin Gastroenterol 30, 629-639 (2016). [0140] 3. R. Sauer et al., Preoperative versus postoperative chemoradiotherapy for locally advanced rectal cancer: results of the German CAO/ARO/AIO-94 randomized phase III trial after a median follow-up of 11 years. J Clin Oncol 30, 1926-1933 (2012). [0141] 4. W. van Gijn et al., Preoperative radiotherapy combined with total mesorectal excision for resectable rectal cancer: 12-year follow-up of the multicentre, randomised controlled TME trial. Lancet Oncol 12, 575-582 (2011). [0142] 5. E. Fokas et al., Outcome measures in multimodal rectal cancer trials. Lancet Oncol 21, e252-e264 (2020). [0143] 6. E. Fokas et al., Neoadjuvant rectal score as individual-level surrogate for disease-free survival in rectal cancer in the CAO/ARO/AIO-04 randomized phase III trial. Ann Oncol 29, 1521-1527 (2018). [0144] 7. E. Fokas et al., Tumour Regression Grading After Preoperative Chemoradiotherapy as a Prognostic Factor and Individual-Level Surrogate for Disease-Free Survival in Rectal Cancer. J Natl Cancer Inst 109, (2017). [0145] 8. P. W. Eide, J. Bruun, R. A. Lothe, A. Sveen, CMScaller: an R package for consensus molecular subtyping of colorectal cancer pre-clinical models. Sci Rep 7, 16618 (2017). [0146] 9. J. Guinney et al., The consensus molecular subtypes of colorectal cancer. Nat Med 21, 1350-1356 (2015). [0147] 10. C. Isella et al., Stromal contribution to the colorectal cancer transcriptome. Nat Genet 47, 312-319 (2015). [0148] 11. B. E. Michels et al., Human colon organoids reveal distinct physiologic and oncogenic Wnt responses. J Exp Med 216, 704-720 (2019). [0149] 12. E. Sahai et al., A framework for advancing our understanding of cancer-associated fibroblasts. Nat Rev Cancer 20, 174-186 (2020). [0150] 13. J. Varga et al., AKT-dependent NOTCH3 activation drives tumour progression in a model of mesenchymal colorectal cancer. J Exp Med 217, (2020). [0151] 14. C. Gaggioli et al., Fibroblast-led collective invasion of carcinoma cells with differing roles for RhoGTPases in leading and following cells. Nat Cell Biol 9, 1392-1400 (2007). [0152] 15. D. Ohlund et al., Distinct populations of inflammatory fibroblasts and myofibroblasts in pancreatic cancer. J Exp Med 214, 579-596 (2017). [0153] 16. G. Biffi et al., IL1-Induced JAK/STAT Signaling Is Antagonized by TGFbeta to Shape CAF Heterogeneity in Pancreatic Ductal Adenocarcinoma. Cancer Discov 9, 282-301 (2019). [0154] 17. A. Naba, K. R. Clauser, R. O. Hynes, Enrichment of Extracellular Matrix Proteins from Tissues and Digestion into Peptides for Mass Spectrometry Analysis. J Vis Exp, e53057 (2015). [0155] 18. A. Naba et al., Characterization of the Extracellular Matrix of Normal and Diseased Tissues Using Proteomics. J Proteome Res 16, 3083-3091 (2017). [0156] 19. A. Naba et al., Extracellular matrix signatures of human primary metastatic colon cancers and their metastases to liver. BMC Cancer 14, 518 (2014). [0157] 20. C. Tian et al., Proteomic analyses of ECM during pancreatic ductal adenocarcinoma progression reveal different contributions by tumour and stromal cells. Proc Natl Acad Sci USA 116, 19609-19618 (2019). [0158] 21. S. Rafiq et al., Common genetic variation in the gene encoding interleukin-1-receptor antagonist (IL-1RA) is associated with altered circulating IL-1RA levels. Genes Immun 8, 344-351 (2007). [0159] 22. K. Ganesh et al., A rectal cancer organoid platform to study individual responses to chemoradiation. Nat Med 25, 1607-1614 (2019). [0160] 23. Y. Yao et al., Patient-Derived Organoids Predict Chemoradiation Responses of Locally Advanced Rectal Cancer. Cell Stem Cell 26, 17-26 e16 (2020). [0161] 24. H. E. Barker, J. T. Paget, A. A. Khan, K. J. Harrington, The tumour microenvironment after radiotherapy: mechanisms of resistance and recurrence. Nat Rev Cancer 15, 409-425 (2015).
