Single-cell pathology analysis of tumor samples

Abstract

The present invention relates to a method for indicating clinical outcomes of cancer patients by labeling cancer samples with labeled molecular probes, assaying the expression of multiple biomolecules at single-cell resolution, and assigning a cellular identity (CI) to each single cell in the sample based on its expression pattern; and then assigning single cell pathology (SCP) patient groups according to the proportion of each CI that the sample contains. In another aspect, the present invention relates to a method of treating a patient with an anti-cancer agent depending on the patient's assignment to a particular SCP. Alternatively, this aspect can be organized as providing a specific agent for the treatment of cancer in a patient characterized by a tumor assigned to a particular SCP.

Description

[Technical Field]

[0001] The present invention relates to the analysis of tumor tissue samples by multiplex staining of the samples with ligands, particularly antibodies, specific for cancer-specific biomolecules at the cellular level, and subsequent assignment of a likely clinical outcome. A particularly advantageous analytical method for carrying out the present invention is imaging mass cytometry.

[0002] The present invention further relates to the use of antineoplastic agents for the treatment of cancer in patients assigned to specific groups by the methods of the present invention. [Background technology]

[0003] Histological and phenotypic differences between tumors guide cancer diagnosis, prognosis, and treatment options. Currently, breast cancer patients are graded based on tumor architecture and cytomorphology, and subcategorized if more than 1% of tumor cells contain hormone receptors, or if more than 10% express high levels of human epidermal growth factor receptor type 2 (HER2), or if the HER2 gene is amplified. While additional molecular subclasses and morphological features have been identified as prognostic, the majority of cells remain uncharacterized (Curtis et al., Nature 486, 346-352 (2012); Sorlie et al., Proc Natl Acad Sci USA 98, 10869-10874 (2001); Nature 490, 61-70 (2012); Beck et al., Sci. Transl. Med. 3, 108ra113 (2011)).

[0004] Highly multiplexed imaging has enabled the identification of complex multicellular phenotypes in the context of the tumor microenvironment and enabled sophisticated histopathological classification of clinical tissue samples (Schapiro et al.; Nat. Methods 14, 873-876 (2017); Keren et al.; Cell 174, 1373-1387.e19 (2018); Carvajal-Hausdorf et al., Clin. Cancer Res. 25, 3054-3062 (2019); Damond et al., Cell Metab. 29, 755-768.e5 (2019)).

[0005] Although single-cell analysis has revealed extensive intra- and inter-patient heterogeneity among cancer tissue samples, the complex single-cell phenotype and its spatial context are not yet reflected in histological stratification that provides the basis for clinical decision-making. [Prior art documents] [Non-patent literature]

[0006] [Non-Patent Document 1] Curtis et al., Nature 486, 346-352 (2012) [Non-patent document 2] Sorlie et al., Proc Natl Acad Sci USA 98, 10869-10874 (2001) [Non-patent document 3] Nature 490,61-70(2012) [Non-patent document 4] Beck et al., Sci. Transl. Med. 3, 108ra113 (2011) [Non-Patent Document 5] Schapiro et al.; Nat. Methods 14, 873-876 (2017) [Non-patent document 6] Keren et al;Cell 174,1373-1387.e19(2018) [Non-Patent Document 7] Carvajal-Hausdorf et al., Clin. Cancer Res. 25, 3054-3062 (2019) [Non-patent document 8] Damond et al., Cell Metab. 29, 755-768.e5 (2019) Summary of the Invention [Problem to be solved by the invention]

[0007] Based on the above prior art, it is an object of the present invention to provide means and methods for identifying multicellular features of the tumor microenvironment that can be used to stratify cancer patients, particularly breast cancer patients, into novel subgroups to provide more accurate information on clinical outcome than this histological subcategory. [Means for solving the problem]

[0008] This object is achieved by the subject matter of the independent claims herein. [Brief explanation of the drawings]

[0009] [Figure 1] Figure 1 shows the results of the comparison and reproducibility analysis between IHC and IMC images. The scatter plot shows the correlation between the total quantified IMC signal (ion counts / μm²) in sections of the same tumor core compared to either the quantification of the IHC signal (optical density / μm²) or the number of positively stained cells in sections from the same tumor core (n = 319 cores). [Figure 2]Figure 2 shows a Bland-Altman plot demonstrating the reproducibility of IMC signal in positively stained cells between images from different regions of the same tumor. The plot is adapted to visualize the average across four samples on the X-axis and the difference between individual samples relative to the tumor mean on the Y-axis. For this analysis, only images containing positively stained cells and a total of more than 200 cells were considered (ER: n = 280 cores from 72 patients; PR: n = 213 cores from 66 patients; HER2: n = 291 cores from 72 patients; Ki67: n = 281 cores from 72 patients; E / P-cadherin: n = 200 cores from 65 patients). The red line represents the overall mean difference relative to the tumor mean, and the blue line represents the 95% confidence interval (1.96 × standard deviation). The percentage of observations falling within the confidence interval is displayed at the top of each plot. [Figure 3] Figure 3 shows a schematic illustrating the workflow for multiplexed imaging of 281 breast cancer patients acquired with IMC and analyzing single-cell phenotypes, cell communities, tumor and patient subclassification, and overall patient survival. [Figure 4] Figure 4 shows single-cell phenotypes in high-dimensional histopathology of breast cancer, illustrated by tSNE maps of 171,288 subsampled single cells from a high-dimensional image of a breast tumor colored according to cell metacluster identifier, patient identity, tumor or stroma classification, distance of each cell to the tumor-stroma interface, and number of neighbors of each cell. [Figure 5] Figure 5 shows heat maps of each cellular metacluster, with z-scores representing the mean marker expression level or distance to the tumor-stroma interface. The absolute cell counts for each cellular metacluster are displayed as bars. In the bubble plots, the size of the circles indicates the relative proportion of all cells of the clinical subtype derived from each cluster, and the opacity of the circles indicates the proportion of each cluster present in different clinical subtypes. The cellular metaclusters are colored as shown in Figures 9 and 10. [Figure 6]Figure 6 shows how we defined tumor cell metacluster cutoffs by hierarchically clustering the diverse tumor groups defined by PhenoGraph to reduce the patient population to 14 common tumor cell subtypes (epithelial cell metaclusters 14–27). The heatmap displays the z-scores of mean marker expression for single-cell phenotypic clusters identified by PhenoGraph. The color bar and hierarchical clustering indicate the corresponding metaclusters. Red stars in the hierarchical clustering tree indicate the 14 subgroups that robustly reappear as distinct groups using multiscale bootstrap resampling (R function pvclust, p<0.05). [Figure 7-1] Figure 7 shows the density of cell metaclusters in different clinical subtypes and subgroups of SCP patients. (a) Box plots of cell metacluster density for patients in each clinical subtype (HR+HER2-: n = 173, HR+HER2+: n = 29, HR-HER2+: n = 23, Triple Negative (TripleNeg): n = 48) and (b) each SCP subgroup (center line, median; box limits, first and third quartiles; whiskers, 1.5 times the interquartile range; points beyond the whiskers, outliers; SCP1:n =17, SCP2:n=21, SCP3:n=20, SCP4:n=12, SCP5:n=32, SCP6:n=10, SCP7:n=13, SCP8:n=11, SCP9:n=20, SCP10: n=24, SCP11:n=31, SCP12:n=14, SCP13:n=15, SCP14:n=11, SCP15:n=8, SCP16:n=10, SCP17:n=9, SCP18:n=3). [Figure 7-2]Figure 7 shows the density of cell metaclusters in different clinical subtypes and subgroups of SCP patients. (a) Box plots of cell metacluster density for patients in each clinical subtype (HR+HER2-: n = 173, HR+HER2+: n = 29, HR-HER2+: n = 23, Triple Negative (TripleNeg): n = 48) and (b) each SCP subgroup (center line, median; box limits, first and third quartiles; whiskers, 1.5 times the interquartile range; points beyond the whiskers, outliers; SCP1:n =17, SCP2:n=21, SCP3:n=20, SCP4:n=12, SCP5:n=32, SCP6:n=10, SCP7:n=13, SCP8:n=11, SCP9:n=20, SCP10: n=24, SCP11:n=31, SCP12:n=14, SCP13:n=15, SCP14:n=11, SCP15:n=8, SCP16:n=10, SCP17:n=9, SCP18:n=3). [Figure 7-3] Figure 7 shows the density of cell metaclusters in different clinical subtypes and subgroups of SCP patients. (a) Box plots of cell metacluster density for patients in each clinical subtype (HR+HER2-: n = 173, HR+HER2+: n = 29, HR-HER2+: n = 23, Triple Negative (TripleNeg): n = 48) and (b) each SCP subgroup (center line, median; box limits, first and third quartiles; whiskers, 1.5 times the interquartile range; points beyond the whiskers, outliers; SCP1:n =17, SCP2:n=21, SCP3:n=20, SCP4:n=12, SCP5:n=32, SCP6:n=10, SCP7:n=13, SCP8:n=11, SCP9:n=20, SCP10: n=24, SCP11:n=31, SCP12:n=14, SCP13:n=15, SCP14:n=11, SCP15:n=8, SCP16:n=10, SCP17:n=9, SCP18:n=3). [Figure 8]Figure 8 shows distinct patterns of multicellular structure in breast tumor tissue based on the defined cellular metaclusters. Cell communities were identified by constructing a topological neighbor-cell interaction network and applying a graph-based community detection approach using the Louvain algorithm. Applied only to tumor cells, community detection identified dense epithelial patches of various sizes, referred to as tumor communities (TCs). When applied to all cells, microenvironment communities (MCs) containing tumor and stromal cellular components were identified. Representative images illustrate various stages in the spatial analysis. From left to right, the images show a pseudocolored IMC, a single-cell mask of the same field labeled with a cellular metacluster identifier, neighboring cellular interactions detected by the topological cellular interaction network, modular regions of the tumor network identifying color-labeled epithelial communities, and modular regions of the tumor-stroma network identified as tumor microenvironment communities. Scale bar = 100 μm. [Figure 9] Figure 9 shows the classification of multicellular communities according to community size and tumor cell phenotype (tumor community, TC). Distinctively colored tumor communities (n=8495) were clustered in PhenoGraph based on the minimum and maximum normalized absolute number of cells from each cellular metacluster, visualized both as a tSNE map and as a stacked bar graph showing the average number of cells from each cellular metacluster. [Figure 10] Figure 10 shows a second method of grouping multicellular communities as in Figure 9, but referring to all cells and being agnostic to tumor cell type (Microenvironment Community, MC). Distinctively colored microenvironment communities (n=12,854) were clustered with PhenoGraph based on the minimum-maximum normalized absolute counts from each cellular metacluster and visualized both as tSNE maps and as stacked bar graphs showing the average cell counts from each cellular metacluster. [Figure 11]Figure 11 shows patient tumors grouped based on tumor cell metacluster organization using unsupervised clustering to identify 18 single-cell pathology (SCP) subgroups that separate classical clinical subtypes. (a) Hierarchically clustered stacked bar graphs show the cell metacluster density for each tumor sample. Colored columns indicate classical clinical subtypes and novel SCP subgroups. (b) Heatmaps show the proportion of distinct epithelial tumor communities present in each image. [Figure 12] Figure 12 shows the comparison and enrichment between the SCP breast cancer group and the histology-based clinical classification. Bubble plots visualize the overlap between subgroups of SCP breast cancer patients (SCP1: n = 17, SCP2: n = 21, SCP3: n = 20, SCP4: n = 12, SCP5: n = 32, SCP6: n = 10, SCP7: n = 13, SCP8: n = 11, SCP9: n = 20, SCP10: n = 24, SCP11: n = 31, SCP12: n = 14, SCP13: n = 15, SCP14: n = 11, SCP15: n = 8, SCP16: n = 10, SCP17: n = 9, SCP18: n = 3 (excluded due to small n values)) and clinical subtypes (HR+HER2-: n = 173, HR+HER2+: n = 29, HR-HER2+: n = 23, Triple Negative (TripleNeg): n = 48). One-sided Fisher's exact test for enrichment. ·p<0.1, *p<0.05, **p<0.01, ***p<0.001. [Figure 13ab] Figure 13 demonstrates the different clinical outcomes of the SCP groups. Kaplan-Meier curves show overall survival for each patient group (n = 278 patients total) based on (a) clinical subgroup, (b) clinical grade, and (c-f) SCP subgroup. *p<0.05 compared to all other samples, *p<0.05 compared to similar subgroups, and *p<0.05 compared to other HR+ / HER2- patients by two-sided log-rank test. [Figure 13cd]Figure 13 demonstrates the different clinical outcomes of the SCP groups. Kaplan-Meier curves show overall survival for each patient group (n = 278 patients total) based on (a) clinical subgroup, (b) clinical grade, and (c-f) SCP subgroup. *p<0.05 compared to all other samples, *p<0.05 compared to similar subgroups, and *p<0.05 compared to other HR+ / HER2- patients by two-sided log-rank test. [Figure 13ef] Figure 13 demonstrates the different clinical outcomes of the SCP groups. Kaplan-Meier curves show overall survival for each patient group (n = 278 patients total) based on (a) clinical subgroup, (b) clinical grade, and (c-f) SCP subgroup. *p<0.05 compared to all other samples, *p<0.05 compared to similar subgroups, and *p<0.05 compared to other HR+ / HER2- patients by two-sided log-rank test. [Figure 14ab] Figure 14 shows Kaplan-Meier survival curves for overall survival and disease-free survival. (a-b) Overall survival for specific stromal environments, and (c-l) disease-free survival for each patient group based on (c) clinical subtype, (d) grade, (e-h) SCP subgroup, and (i-l) stromal environment. Two-tailed log-rank test. *p<0.05 compared with all other samples. [Figure 14cd] Figure 14 shows Kaplan-Meier survival curves for overall survival and disease-free survival. (a-b) Overall survival for specific stromal environments, and (c) Kaplan-Meier survival curves for (c-l) disease-free survival for each patient group based on clinical subtype: (d) grade, (e-h) SCP subgroup, and (i-l) stromal environment. Two-tailed log-rank test. *p<0.05 compared with all other samples. [Figure 14ef] Figure 14 shows Kaplan-Meier survival curves for overall survival and disease-free survival. (a-b) Overall survival for specific stromal environments, and (c-l) disease-free survival for each patient group based on (c) clinical subtype, (d) grade, (e-h) SCP subgroup, and (i-l) stromal environment. Two-tailed log-rank test. *p<0.05 compared with all other samples. [Figure 14gh] Figure 14 shows Kaplan-Meier survival curves for overall survival and disease-free survival. (a-b) Overall survival for specific stromal environments, and (c-l) disease-free survival for each patient group based on (c) clinical subtype, (d) grade, (e-h) SCP subgroup, and (i-l) stromal environment. Two-tailed log-rank test. *p<0.05 compared with all other samples. [Figure 14ij] Figure 14 shows Kaplan-Meier survival curves for overall survival and disease-free survival. (a-b) Overall survival for specific stromal environments, and (c-l) disease-free survival for each patient group based on (c) clinical subtype, (d) grade, (e-h) SCP subgroup, and (i-l) stromal environment. Two-tailed log-rank test. *p<0.05 compared with all other samples. [Figure 14kl] Figure 14 shows Kaplan-Meier survival curves for overall survival and disease-free survival. (a-b) Overall survival for specific stromal environments, and (c-l) disease-free survival for each patient group based on (c) clinical subtype, (d) grade, (e-h) SCP subgroup, and (i-l) stromal environment. Two-tailed log-rank test. *p<0.05 compared with all other samples. [Figure 15] Figure 15 shows that spatially defined cellular communities are associated with patient outcomes. The graph shows the relative hazard ratios and 95% confidence intervals for disease-specific overall survival for tumor (T) and microenvironment (ME) cellular community densities and clinical categories (molecular subtype and grade) estimated by Cox proportional hazards model (n = 266 patients, excluding n = 15 patients containing only <10 cellular communities). [Figure 16-1] Figure 16 shows the antibody conjugates used in the staining panel. [Figure 16-2] Figure 16 shows the antibody conjugates used in the staining panel. [Figure 17a]Figure 17 shows (a-c) Coxph and (d-f) logrank tests for the difference in overall survival between each single-cell pathology subgroup and other patients, (a, d) within the cohort, (b, e) within the similar SCP subgroup, and (c, f) within patients clinically classified as HR+ / HER2-. [Figure 17bc] Figure 17 shows (a-c) Coxph and (d-f) logrank tests for the difference in overall survival between each single-cell pathology subgroup and other patients, (a, d) within the cohort, (b, e) within the similar SCP subgroup, and (c, f) within patients clinically classified as HR+ / HER2-. [Figure 17de] Figure 17 shows (a-c) Coxph and (d-f) logrank tests for the difference in overall survival between each single-cell pathology subgroup and other patients, (a, d) within the cohort, (b, e) within the similar SCP subgroup, and (c, f) within patients clinically classified as HR+ / HER2-. [Figure 17f] Figure 17 shows (a-c) Coxph and (d-f) logrank tests for the difference in overall survival between each single-cell pathology subgroup and other patients, (a, d) within the cohort, (b, e) within the similar SCP subgroup, and (c, f) within patients clinically classified as HR+ / HER2-. [Figure 18] Figure 18 shows that adding SCP grouping or tumor-stroma community information, when compared with clinically defined subtypes, improved the ability to predict patient overall survival using Cox proportional hazards modeling likelihood ratio tests between nested Cox models. The null hypothesis is that larger models (more variables) are not superior to smaller models. A P value <0.05 rejects the null hypothesis. [Figure 19-1] Figure 19 shows the relationship between the single cell pathology (SCP) group and the standard of care clinical histopathology group. [Figure 19-2] Figure 19 shows the relationship between the single cell pathology (SCP) group and the standard of care clinical histopathology group. DETAILED DESCRIPTION OF THE INVENTION

[0010] Summary of the Invention A first aspect of the present invention relates to a method for classifying tumors according to morphological characteristics, in particular for indicating clinical outcome or drug sensitivity in cancer patients, said method comprising the steps of: a. providing a cancer tissue sample obtained from a patient; b. labeling a cancer tissue sample with a plurality of molecular probes, each probe specific to a biomolecule, each of said molecular probes characterized by a detectable marker; c. obtaining information regarding the expression of each of said plurality of biomolecules at single-cell resolution; d. in a cell assignment step, assigning a cellular identity (CI) to each single cell in the labeled tissue sample based on the expression of the plurality of biomolecules, wherein the cellular identity is assigned according to the cellular expression of a particular biomolecule; e. In a pathology group assignment step, the cancer tissue samples are assigned to single cell pathology (SCP) patient groups according to the proportion of each cell identity contained in the samples assigned in the cell assignment step.

[0011] A method according to the present invention may include constructing an image of a cancer tissue sample.

[0012] Furthermore, in certain embodiments, the method comprises the steps of: f. partitioning the image of the cancer tissue sample into multicellular regions, wherein each single cell within the multicellular region is highly interconnected with neighboring cells, resulting in a cellular community; g. Assigning a cellular community identity (CCI) to each cellular community according to the number of cells in the cellular community and the proportion of each CI contained therein.

[0013] This aspect of the invention may further comprise assigning patients to predicted outcome groups according to the type and number of cellular communities.

[0014] In another aspect, the present invention relates to a method of treating a patient with an anti-cancer agent depending on the patient's assignment to a particular SCP (single cell pathology). Alternatively, this aspect can be organized as providing a specific agent for the treatment of cancer in a patient characterized by a tumor assigned to a particular SCP.

[0015] While the inventors have demonstrated the extraordinary utility of the method in a large cohort of breast cancer patients and the wealth of information that can be obtained to inform treatment decisions in cancer patients, those skilled in the art will recognize that the general methodology can be applied to many other cancer types, given appropriate clinical information.

[0016] Detailed Description of the Invention Terms and Definitions For the purposes of interpretation of this specification, the following definitions shall apply, and where appropriate, terms used in the singular shall include the plural and vice versa. In the event that any definition set forth below conflicts with any document incorporated herein by reference, the set forth definition shall control.

[0017] As used herein, the terms "comprising," "having," "containing," "including," and other similar forms, and their grammatical equivalents, as used herein, are intended to be equivalent and open-ended in that the one or more items following any one of these words are not intended to be an exhaustive list of such one or more items, or to be limited only to the listed one or more items. For example, an item "comprising" components A, B, and C can consist of components A, B, and C (i.e., contain only components A, B, and C), or it can include not only components A, B, and C, but also one or more other ingredients. Thus, "comprise" and its similar forms, and its grammatical equivalents, are intended and understood to include disclosure of "consisting essentially of" or "consisting of" embodiments.

[0018] Where a range of values ​​is given, unless the context clearly indicates otherwise, each intervening value to the tenth of the unit of the lower limit between the upper and lower limits of that range, and any other stated or intervening value within that stated range, is understood to be encompassed within the disclosure, subject to any specifically excluded limitation of the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the disclosure.

[0019] As used herein, reference to "about" a value or parameter includes (and describes) variations on the value or parameter itself. For example, a statement "about X" includes the statement "X."

[0020] As used in this specification, including the appended claims, the singular forms "a," "or," and "the" include plural references unless the context clearly dictates otherwise.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art (e.g., cell culture, molecular genetics, nucleic acid chemistry, hybridization techniques, and biochemistry). Molecular, genetic, and biochemical techniques employ standard techniques (see generally, Sambrook et al., Molecular Cloning: A Laboratory Manual, 4th ed. (2012) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, and Ausubel et al., Short Protocols in Molecular Biology (2002) 5th ed., John Wiley & Sons, Inc.) and chemical methods.

[0022] The term "membrane-associated" in the context of this specification relates to a molecule that is part of or interacts with a biological membrane. Common examples include Cluster of Differentiation (CD) proteins, which are cell surface molecules used to determine the phenotype of a cell.

[0023] The terms "gene expression" or "expression," or alternatively, "gene product," can refer to either or both of the process of producing a nucleic acid (RNA), or a peptide or polypeptide, and its product, and can refer to transcription and translation, respectively, or to any of the intermediate processes that regulate the processing of genetic information to produce a polypeptide product. The term "gene expression" can also apply to the transcription and processing of RNA gene products, such as regulatory RNA or structural (e.g., ribosomal) RNA. When the expressed polynucleotide is derived from genomic DNA, expression can include splicing of mRNA in eukaryotic cells. Expression can be assessed at both the level of transcription and translation, i.e., mRNA and / or protein product.

[0024] As used herein, the term "positive," when used in the context of marker expression, refers to expression of an antigen as assayed by a labeled (e.g., fluorescent or isotype-labeled) antibody, where the signal (fluorescent or otherwise) of the label in the structure (e.g., cell) designated as "positive" is at least 30% higher (≧30%), and particularly ≧50% or ≧80% higher, in median signal (fluorescent or otherwise) intensity, compared to staining with an isotype-matched labeled antibody that does not specifically bind to the same target. Expression of such markers is indicated by the superscript "plus" ( + ) or (e.g. CD4 + ), or by adding a plus sign after the name of the marker ("CD4+"). As used herein, when the term "expression" is used in the context of "gene expression" or "marker or biomolecule expression," and no further qualification is mentioned for "expression," it means "positive expression" as defined above.

[0025] As used herein, the term "negative," when used in the context of marker expression, refers to expression of an antigen, as assayed by a labeled (especially fluorescently or isotype-labeled) antibody, where the median label (fluorescence) intensity is less than 30% higher, and particularly less than 15% higher, than the median label (fluorescence) intensity of an isotype-matched antibody that does not specifically bind to the same target. Expression of such markers is indicated by a superscript minus ( - ) or (e.g. CD4 - ), or indicated by a minus sign after the marker name (“CD4−”).

[0026] "High expression" of a marker, e.g., high expression of a hormone receptor (e.g., progesterone or estrogen receptor, HR), refers to the expression level of that marker in a distinct cell population, as detected, e.g., by FACS, that exhibits the highest signal intensity per cell compared to other populations characterized by lower signal intensity per cell. High expression is indicated by a superscript "high" or "hi" after the marker name (e.g., HR). high The term "highly expressed" refers to this same characteristic.

[0027] "Low expression" of a marker, e.g., low expression of HR, refers to the expression level of that marker in a distinct cell population, as detected by FACS, that exhibits the lowest signal intensity per cell compared to other populations characterized by higher signal intensity per cell. Low expression is indicated by the superscript "low" or "lo" after the marker name (e.g., HR). low The term "low expressed" refers to this same characteristic.

[0028] Marker expression can be assayed using techniques such as fluorescence microscopy, flow cytometry, ELISPOT, ELISA, and multiplex analysis. Mass spectrometry is considered a particularly useful method for marker detection. Another particularly useful analytical method is imaging mass cytometry. An exemplary workflow for imaging mass cytometry (IMC) (see Giesen et al., Nature Methods, 2014 Apr;11(4):417-22) involves preparing tissue sections for metal-chelating antibody labeling using an IHC protocol. The tissue sample is then placed in a laser ablation chamber. The tissue is ablated and transported by a gas stream to a time-of-flight mass spectrometer, such as a CyTOF (Fluidigm) for mass cytometry analysis. The measured isotope signal is plotted using the coordinates of each single laser shot to generate a multidimensional tissue image. By measuring the characteristics and marker expression of single cells, cell subpopulation characteristics within the analyzed tissue can be investigated.

[0029] The term "molecular probe" in the context of this specification refers to a specific ligand, in particular an antibody, an antibody fragment, an antibody-like molecule or an aptamer, more in particular an antibody or an antibody fragment, which binds to a target molecule, such as a specific surface protein of a T cell or a specific transcription factor, with a binding affinity of ≦10 -7 mol / l, especially ≦10 -8 It refers to what can be bound with a dissociation constant in mol / l. Molecular probes include detectable markers such as particles, beads, dyes, or enzymes.

[0030] The term "set of molecular probes" relates to a panel of molecular probes for positive and / or negative selection of marker expression.

[0031] The term "fluorescent dye" in the present context refers to a small molecule capable of fluorescing in the visible or near-infrared spectrum. Examples of fluorescent labels or labels that exhibit visible color include, but are not limited to, fluorescein isothiocyanate (FITC), rhodamine, allophycocyanin (APC), peridinin chlorophyll (PerCP), phycoerythrin (PE), Alexa Fluor (Life Technologies, Carlsbad, CA, USA), Dylight Fluor (Thermo Fisher Scientific, Waltham, MA, USA), ATTO Dye (ATTO-TEC GmbH, Siegen, Germany), BODIPY Dye (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene-based dye), etc.

[0032] The term "aptamer" refers to an oligonucleotide or peptide molecule that binds to a specific target molecule. Aptamers can be created by selecting them from a large pool of random sequences. Nucleic acid aptamers can be generated by repeated in vitro selection or, equivalently, by SELEX (Systematic Evolution of Ligands by Exponential Enrichment), which bind to molecular targets such as small molecules, proteins, or nucleic acids through non-covalent interactions. Aptamers have molecular recognition properties comparable to antibodies.

[0033] The term "specific binding" in the context of the present invention refers to the property of a ligand to bind to its target with particular affinity and target specificity. The affinity of such a ligand is indicated by the dissociation constant of the ligand. A specifically reactive ligand has a dissociation constant of ≦10 upon binding to the target. -7 mol / L, but has a dissociation constant at least three orders of magnitude higher when interacting with a molecule that has nearly the same chemical composition as the target but a different conformation.

[0034] As used herein, the term "treating" or "treatment" of any disease or disorder (e.g., cancer) refers, in one embodiment, to ameliorating the disease or disorder (e.g., delaying or preventing or reducing the onset of the disease or at least one of its clinical symptoms). In another embodiment, "treating" or "treatment" refers to alleviating or improving at least one physical parameter, including physical parameters that may not be discernible to the patient. In yet another embodiment, "treating" or "treatment" refers to modulating the disease or disorder either physically (e.g., stabilization of discernible symptoms), physiologically (e.g., stabilization of physical parameters), or both. Methods for assessing the treatment and / or prevention of disease are generally known in the art, unless otherwise specified below.

[0035] A first aspect of the present invention relates to a method for characterizing tumor samples from cancer patients according to single cell characteristics, particularly as part of a method for indicating a likely prognosis or clinical outcome for the cancer patient. The method of the present invention is particularly advantageous for patients whose cancer expresses hormone receptors, and even more particularly advantageous for patients with breast cancer or ovarian cancer.

[0036] Certain aspects of the methods of the present invention allow for the identification of breast cancer patients with different survival outcomes or assist clinicians in selecting the most effective treatment for a particular tumor subtype.

[0037] In the most general terms, the method of the present invention comprises the steps of: a. Providing a cancer tissue sample obtained from a patient, said sample comprising cells. In many embodiments of the methods of the invention, the sample will be in the form of a slide or histological preparation, however, the invention encompasses embodiments in which, at an optional step, a single cell preparation is obtained from the cell sample, information about cell types is obtained on the single cell preparation, and subsequent analysis is performed on the single cell preparation. b. In a labeling step, a cancer tissue sample or single-cell preparation is labeled with multiple molecular probes, each specific to a biomolecule of interest that may or may not be present in the sample. The term "molecular probe" is used synonymously with "ligand bearing a unique detectable label," where the label may be, for example, an atom of distinct mass or a dye with distinct spectroscopic properties. The probe may also be a nucleic acid sequence specific to mRNA present in the cell. Each of the molecular probes is characterized by a detectable marker (primer, dye, isotope) that can label the cells of the sample. This labeling step results in a "labeled tissue sample."

[0038] The biomolecules targeted by the probe in the labeling step are selected from the list comprising or consisting of: i. epithelial cadherin (E-cadherin), ii. cytokeratin (CK) 18 and / or 19, iii. CK7, iv. estrogen receptor (ER) and / or progesterone receptor (PR), v. cell proliferation markers, particularly Ki-67 and / or PCNA; vi. CK5 and / or p63 and / or CK14, vii. p53, viii. steroid hormone receptors (HR), in particular receptors for estrogen and / or progesterone (ER and PR, respectively); ix. Markers of apoptosis, particularly cleaved form of poly ADP-ribose polymerase (cPARP) and / or cleaved form of caspase 3 (cC3), other markers of apoptosis, including but not limited to Annexin V, or other permeable dyes, are used in FACS; x. epidermal growth factor receptor (EGFR), and xi. Markers of hypoxia, especially carbonic anhydrase (CAIX). Other markers could be HIF1a or oxygen-sensing chemical probes.

[0039] Optionally, additional markers that can be used in this step include: i. trimethylated H3K27, ii. phosphorylated ribosomal protein S6 (p-S6), iii. phosphorylated mechanistic target of rapamycin mTOR (p-mTOR), iv. human epidermal growth factor receptor 2 (HER2), v. c-myc, vi. histone 3 (H3), and / or vii. Markers of DNA content, especially DNA intercalating dyes including but not limited to ethidium bromide, propidium iodide or cyanine dimers CAS numbers 169454-15-3, 169454-13-1, 143413-84-7.

[0040] Other markers useful in subsequent steps include: TWIST (Uniprot ID Q15672), SLUG (Uniprot ID O43623), placental cadherin (P-cadherin), GATA3, and SMA.

[0041] In certain embodiments, it is advantageous to also obtain spatial parameters at this stage, such as area, size, range eccentricity, and number of adjacent or contacting cells, defined as within 4 pixels, which corresponds to approximately 4 uM.

[0042] A subsequent read step obtains information about the per-cell expression of each of the plurality of biomolecules in the labeled tissue sample at single-cell resolution. Expression can be assessed relative to the (average) expression of a positive control tissue sample in which the biomolecule is expressed, or relative to a negative control tissue sample in which the biomarker is not expressed (healthy tissue or an organ such as the liver that is hormone receptor negative, a probe-negative sample).

[0043] In certain embodiments, the classification of the biomolecules as negative, low expression, positive and high expression in the cell reading step is performed relative to or based on absolute measurements from a reference sample containing all biomolecules used in the classification step, in particular a sample pre-assigned to SCP2, and SCP1 and / or SCP3, and SCP4 and / or SCP5, and SCP6, SCP9 and / or SCP10.

[0044] A subsequent cell assignment step assigns a mutually exclusive cell identity (CI) to essentially each single cell in the labeled tissue sample based on high-dimensional clustering analysis of the expression of the plurality of biomolecules (present in the case of DNA, high or positive expression in the case of p53), where cellular identity is assigned as a function of the cell's expression of biomolecules identified by markers according to the following list: CI1:CIAX hi , EGFR-, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)+; CI2:p53 hi , (cC3 and / or cPARP)+, (ER and / or PR)-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-; CI3: (Ki-67 and / or PCNA)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI4:p53 hi ,EGFR+, CIAX hi , (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI5: (CK5 and / or p63 and / or CK14)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-; CI6:E-cadherin hi , (CK18 and / or CK19) hi , CK7+, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)-; CI7: CK7+, (CK18 and / or CK19)+, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI8: E-cadherin-, CK7-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-, (ER and / or PR)-; CI9:(E-cadherin lo or E-cadherin-), ((CK18 and / or CK19)- or (CK18 and / or CK19) lo ), (ER and / or PR) lo , (CK5 and / or p63 and / or CK14)-; CI10: (CK18 and / or CK19) hi or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR) hi , CK5 and / or p63 and / or CK14)-; CI11:(CK18 and / or CK19) hi or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR)+, CK5 and / or p63 and / or CK14)-; CI12:(E-cadherin lo or E-cadherin+), ((ER and / or PR) lo or (ER and / or PR)-), (CK5 and / or p63 and / or CK14)-; CI13:p53 hi , EGFR+, (ER and / or PR) hi , ((CK5 and / or p63 and / or CK14) lo and / or (CK5 and / or p63 and / or CK14)-); CI14: CK7+, (CK18 and / or CK19)+, (CK5 and / or p63 and / or CK14)+, (ER and / or PR)-.

[0045] Clusters represent groups of cells that are similar to each other in a high-dimensional marker space.

[0046] In certain embodiments, annotation of clusters was based on the average marker expression of the clusters relative to each other. Some populations exhibit relatively low marker averages, thus representing populations where most markers are negative (noise only), and other populations exhibit relatively high averages that are considered positive. In these embodiments, to identify a group of cells as positive, there must be another group of cells that are negative / low for this marker. The difference between + and hi is more subtle and can only be distinguished when there is, for example, a negative cell population and two distinct populations where one positive cell population has a higher average value than the other positive cell population.

[0047] Next, in a pathology group assignment step, the cancer tissue samples are assigned to single cell pathology (SCP) patient groups according to the proportion or frequency of each cell identity that the samples contain as assigned in the cell assignment step, wherein the list of SCP patient groups includes or consists of: SCP1.>70% of single cells are CI10; SCP2.>70% of single cells are CI11; SCP3. ≤70% of single cells are CI10; >70% single cells for SCP4.CI12; SCP5. ≤70% of single cells are CI12; SCP6.>80% of single cells are CI9; SCP7.>80% of single cells are CI8; SCP8. ≤70% of single cells are CI9; or CI10; or CI12; SCP9, >60% of single cells are CI9; SCP10. >70% of single cells are CI9; or CI10; or CI12; SCP11.>60% of single cells are CI7; SCP12.>70% of single cells are CI6; SCP13.>50% of single cells are CI5; SCP14.>60% of single cells are CI3; SCP15.>70% of single cells are CI4; SCP16.>50% of single cells are CI2; SCP17.>50% of single cells are CI1; SCP 18. >90% of single cells are CI 14.

[0048] In certain embodiments, the cancer tissue sample comprises or essentially consists of neoplastic cells derived from tissue characterized by expression of steroid hormone receptors, particularly neoplastic cells derived from breast, ovarian or endometrial tissue, more particularly neoplastic cells derived from human breast or human mammary epithelium.

[0049] In certain embodiments, the method for obtaining information regarding the average expression of a plurality of biomolecules comprises constructing an image of the cancer tissue sample, which can be achieved, for example, by immunohistochemistry using dye-labeled ligands, or by mass spectrometry-based methods such as, but not limited to, imaging mass cytometry.

[0050] In certain embodiments, the cancer tissue sample is a section, particularly a histological slide.

[0051] In certain embodiments, the cancer tissue sample is a monolayer of adherent cells or cells otherwise immobilized on a solid surface.

[0052] In certain embodiments, the plurality of biomolecules is labeled by contacting the sample with a plurality of metal- or fluorochrome-conjugated antibodies and / or nucleic acid probes.

[0053] In a particular embodiment, the method for obtaining information about the average expression of a plurality of biomolecules is directed to imaging mass cytometry at subcellular resolution, particularly at a resolution of ≦5 μm, or even ≦1 μm.

[0054] WO2015128490A1, the entirety of which is incorporated herein by reference, discloses subjecting a plurality of cells of a labeled tissue sample to laser ablation at a plurality of known locations using a laser spot size of 4 μm or less to form a plurality of plumes; and subjecting the plumes to inductively coupled plasma mass spectrometry, thereby enabling construction of an image of the tissue sample by detecting the labeled atoms in the plumes.

[0055] In certain embodiments, the labeling step includes an additional marker selected from: - CD3 or CD90, and - CD20 or CD19, and - CD68, and - CD44 and / or CD45, and - fibronectin, and - Vimentin, and - CD31 and / or von Willebrand factor (vWF) and / or CD34.

[0056] In certain embodiments, in the cell assignment step, additional cell identities are assigned depending on the cellular expression of biomolecules identified by markers according to the following list: CI15. CD44+, CD45+, (CD3 or CD90)+, fibronectin-, E-cadherin-, (CK5 and / or p63 and / or CK14) lo or (CK5 and / or p63 and / or CK14)-); CI16. CD20+, (fibronectin lo or fibronectin-), ((E-cadherin lo or E-cadherin-), ((CK5 and / or p63 and / or CK14) lo or (CK5 and / or p63 and / or CK14)-); CI17. (CD3 or CD90)+, (CD20 or CD19)+; CI18.CD68+; CI19. Vimentin+, (CD34 and / or VWF and / or CD31)+; CI20. Vimentin-, (Fibronectin+ or Fibronectin hi ), (CD3 or CD90)-, (CD20 or CD19)-, CD45-, CD44-.

[0057] In certain embodiments, the single cell image used to assign cell identity is a fragment of an image of a cancer tissue sample, hi certain embodiments, the single cell image is a fragment, which consists of pixels within a region where membrane-bound molecules surround a single nucleus.

[0058] Single-cell segmentation can be performed using the programs Ilastik and CellProfiler. Ilastik is used to train a pixel classifier that distinguishes between nuclear, membrane, and background regions based on all markers. The resulting probability map (an image in which every pixel is assigned to one of three categories) is then segmented into single-cell masks (in which every pixel is assigned to an individual cell) using CellProfiler.

[0059] In certain embodiments, the method comprises two steps that take into account the cellular community or network in which an individual cell is located, hereinafter referred to as the "cellular community detection step" and the "cellular community assignment step."

[0060] The cell community detection process involves partitioning an image of a cancer tissue sample into multicellular regions, where each single cell within the multicellular region is highly interconnected as determined to provide a cell community within physical proximity of its neighboring cells, within 4 pixels / 4 uM of the outermost pixel assigned to the cell by a suitable graph-based community detection algorithm, e.g., the Louvain algorithm. In particular, in certain embodiments, a single cell within the cell community is only weakly connected to single-cell regions from neighboring cell communities compared to single-cell regions within its own cell community.

[0061] The cellular community assignment step assigns a cellular community identity (CCI) to each cellular community associated with a clinical outcome according to the number of cells in the cellular community and the proportion of each CI contained therein, where the list of CCIs includes or consists of: CCI1. Among cells with identities CI1-CI14, >10% of single cells are CI6 and the average cell community size is >25 cells (favorable hazard ratio); or CCI2. Among cells with identities CI1–CI14, >10% of single cells are CI6 and the average cell community size is ≤25 cells (poor hazard ratio); or CCI3. Among cells with identities CI1-CI14, >10% of single cells are CI2 and the average cell community size is >25 cells (good hazard ratio); or CCI4. Among cells with identities CI1-CI14, >10% of single cells are CI2 and the average cell community size is ≤25 cells (poor hazard ratio); or CCI5. Of cells with identities CI1-CI14, >10% of the cells are CI3 and the average cell community size is ≤25 cells (good hazard ratio); or CCI6. Of cells with identities CI1-CI14, >5% of cells are CI19, >10% of cells are CI20, and >3% of cells are CI18, and the average cell community size is <50 cells (poor hazard ratio); or CCI7. >80% of all cells are any of CI1-CI15 identities and <10% are CI20, and the average cell community size is <75 cells (poor hazard ratio); or CCI8. >20% of all cells are CI15, and / or CI16, and / or CI17, and <40% of cells are any of CI1-CI14 identities, and the average cell community size is >75 cells (good hazard ratio; includes small fibroblasts (or pericytes) with compacted, elongated nuclei); or CCI9. >5% of all cells are CI18 and the average cell community size is >25% (favorable hazard ratio); or CCI10: >80% of all cells are CI1-CI14 identities, <2% of cells are CI20, and the average cell community size is >115 cells but <125 cells (good hazard ratio).

[0062] Many combinations of community definitions from CCI1 to 10 may exist in a single patient sample (cell preparation, section, or image of the sample). While not all such combinations are represented in a patient population, patient outcomes can be linked to the presence or absence of these constructs in specific situations, in the sense that increasing numbers of each community are associated with increased or decreased hazard ratios.

[0063] In another aspect of the invention, SCP (single cell pathology) information is used to select the most appropriate course of treatment for patients with said survival outcome.

[0064] In certain embodiments, patients are assigned to drug resistance or predicted outcomes specific to the SCP group according to the classification of the sample in the pathology group assignment step: - High likelihood of insensitivity to anti-cancer drugs targeting SCP4 or SCP18:ER; - SCP7: likely to lack sensitivity to anti-angiogenic antineoplastic agents: - SCP1, SCP11: High probability of favorable outcome, significantly higher probability of favorable outcome compared to clinical HR+HER- group, log-rank score p<0.05. - SCP8, SCP14, SCP17: High probability of poor outcome, log-rank score P<0.05 compared with all other SCP groups.

[0065] The SCP associated with the favorable clinical outcome may exhibit a log-rank score of P<0.05 compared to any clinical group, such as HR+HER- or HR+HER+, or all other SCP groups. The SCP associated with the unfavorable clinical outcome may exhibit a log-rank score of P<0.05 compared to all other SCP groups, or the HR+HER- or HR+HER+ groups.

[0066] In certain embodiments, in the patient assignment step, patients are assigned to a drug sensitivity group, or predicted outcome, according to the SCP classification of the sample in the pathology group assignment step: - SCP1, high probability of favorable outcome and high probability of sensitivity to: Selective estrogen receptor modulator (SERM) antineoplastic agents, in particular SERM drugs selected from raloxifene, toremifene and tamoxifen; Selective estrogen receptor degrader (SERD) antineoplastic agents, in particular SERD agents selected from fulvestrant, brilandestrant and elacestrant; aromatase inhibitor antineoplastic agents, in particular aromatase inhibitor antineoplastic agents selected from exemestane, letrozole, vorozole, formestane, fadrozole and anastrozole; and / or PI3K pathway inhibitors, particularly PI3K pathway inhibitors selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235SCP2; - SCP2, likely susceptible to: antiangiogenic antineoplastic agents, in particular antiangiogenic antineoplastic agents selected from bevacizumab, thalidomide and lenalidomide; and / or HER2-targeted antineoplastic agents, in particular HER2-targeted antineoplastic agents selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, H2NVAC, HER2.taNK cells, HER2-pulsed dendritic cells, and HER2-targeted T cells; - SCP3, high likelihood of poor outcome and likely susceptibility to: anthracycline antineoplastic agents, in particular anthracycline antineoplastic agents selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; anti-mitotic drugs, in particular anti-mitotic drugs selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes, in particular those selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; alkylating antineoplastic agents, in particular those selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbaine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, thiptepa and trabectedin; antimetabolite antineoplastic agents, in particular antimetabolite antineoplastic agents selected from: azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate and phototrexate; Selective SERM antineoplastic agents, in particular SERM agents selected from raloxifene, toremifene and tamoxifen; SERD antineoplastic agents, in particular SERD antineoplastic agents selected from fulvestrant, brilanestrant and elacestrant; aromatase inhibitors, in particular aromatase inhibitor antineoplastic agents selected from exemestane, letrozole, vorozole, formestane, fadrozole and anastrozole; and / or PI3K pathway inhibitors, particularly PI3K pathway inhibitors selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235; - High risk of insensitivity to anti-cancer drugs targeting SCP4 and ER, especially to: SERM antineoplastic drugs, more particularly SERM drugs selected from raloxifene, toremifene and tamoxifen; a SERD antineoplastic agent, more particularly a SERD antineoplastic agent selected from fulvestrant, brilanestrant and elacestrant; and / or aromatase inhibitor antineoplastic agents, in particular aromatase inhibitor antineoplastic agents selected from exemestane, letrozole, vorozole, formestane, fadrozole and anastrozole; - SCP5, likely susceptible to: EZH2 methyltransferase inhibitor antineoplastic agents, in particular EZH2 methyltransferase inhibitor antineoplastic agents selected from 3-deazaneplanocin A (DZNep), tazemetostat, EPZ005687, EI1, GSK126, or UNC199; - SCP6, high likelihood of poor outcome and likely susceptibility to: HER2-targeted anti-cancer drugs, in particular HER2-targeted anti-cancer drugs selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102 and H2NVAC; - SCP7, likely insensitive to: antiangiogenic antineoplastic agents, in particular antiangiogenic antineoplastic agents selected from bevacizumab, thalidomide and lenalidomide; and likely susceptible to: HER2-targeted anti-cancer drugs, in particular HER2-targeted anti-cancer drugs selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102 and H2NVAC; - SCP8, high probability of poor outcome; - SCP9, likely susceptible to: HER2-targeted anti-cancer drugs, in particular HER2-targeted anti-cancer drugs selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102 and H2NVAC; - SCP 10, likely susceptible to: HER2-targeted anti-cancer drugs, in particular HER2-targeted anti-cancer drugs selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102 and H2NVAC; - SCP11, likely to have a good outcome and likely to be sensitive to: HER2-targeted anti-cancer drugs, in particular HER2-targeted anti-cancer drugs selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102 and H2NVAC, and / or PI3K pathway inhibitors, particularly PI3K pathway inhibitors selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235; - SCP12, high likelihood of poor outcome and likely susceptibility to: EZH2 methyltransferase inhibitor antineoplastic agents, in particular EZH2 methyltransferase inhibitor antineoplastic agents selected from 3-deazaneplanocin A (DZNep), tazemetostat, EPZ005687, EI1, GSK126, and UNC199; - SCP-13, likely susceptible to: HER2-targeted anti-cancer drugs, in particular HER2-targeted anti-cancer drugs selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102 and H2NVAC; SCP14: High likelihood of poor outcome AND high likelihood of susceptibility to: anthracycline antineoplastic agents, in particular anthracycline antineoplastic agents selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; anti-mitotic drugs, in particular anti-mitotic drugs selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes, in particular those selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; alkylating antineoplastic agents, in particular selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbaine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, thiptepa, and trabectedin; and / or antimetabolite antineoplastic agents, in particular antimetabolite antineoplastic agents selected from: azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate and phototrexate; - SCP-15, likely susceptible to: anthracycline antineoplastic agents, in particular anthracycline antineoplastic agents selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; anti-mitotic drugs, in particular anti-mitotic drugs selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes, in particular those selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; alkylating antineoplastic agents, in particular those selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbaine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, thiptepa and trabectedin; antimetabolite antineoplastic agents, in particular antimetabolite antineoplastic agents selected from: azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate and phototrexate; and / or Antineoplastic agents that are EGFR bioactive inhibitors, in particular EGFR bioactive inhibitors selected from gefetinib, erlotinib, lapatinib, cetuximib, neratinib, osimeratib, panitumamib, vandetanib, necitumumab, and dacomitinib; - SCP16: High likelihood of favorable outcome, likely susceptible to: anthracycline antineoplastic agents, in particular anthracycline antineoplastic agents selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; anti-mitotic drugs, in particular anti-mitotic drugs selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes, in particular those selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; alkylating antineoplastic agents, in particular selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbaine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, thiptepa, or trabectedin; or antimetabolite antineoplastic agents, in particular antimetabolite antineoplastic agents selected from: azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate or phototrexate; - SCP17, high potential for poor outcome and high potential for susceptibility to: Quinone alkylating antineoplastic agents, especially the quinone alkylating antineoplastic agent mitomycin C; - High risk of insensitivity to anti-cancer drugs that target SCP18 and ER, especially to: SERM antineoplastic drugs, more particularly SERM drugs selected from raloxifene, toremifene or tamoxifen; a SERD antineoplastic agent, more particularly a SERD antineoplastic agent selected from fulvestrant, brilandestrant and elacestrant; and / or aromatase inhibitor antineoplastic agents, in particular aromatase inhibitor antineoplastic agents selected from exemestane, letrozole, vorozole, formestane, fadrozole and anastrozole; and likely susceptible to: PI3K pathway inhibitors, particularly PI3K pathway inhibitors selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235.

[0067] The SCP classification provides clinicians and patients with two types of information: a prediction of prognosis in terms of overall survival, and / or, in some cases, a prediction of the likelihood that a tumor of a given SCP class will be sensitive or resistant to one or more classes of antineoplastic drug treatment.

[0068] Drug treatment recommendations for SCP can be informed by the specific cellular characteristics observed in this class of tumor. In some embodiments, only one class of drug is recommended for patients in one SCP group. Using SCP17 as an example, in addition to predicting a poor prognosis for these patients, the high prevalence of EGFR cells expressing markers of hypoxia (e.g., CIAX) that define this group suggests that these tumors may be sensitive to drugs designed to be activated in a hypoxic environment, such as the quinone alkylating antineoplastic drug class. In some embodiments, SCP classification can more effectively select candidates for trials of specific classes of drugs; for example, patients with SCP17 may be good candidates for novel hypoxia-targeting drugs.

[0069] In another embodiment, the SCP assignment involves assigning the likelihood of drug resistance rather than sensitivity. The present tumor classifications, such as hormone receptor-positive or triple-negative, used to assign or exclude ER-targeted drugs have been shown to include several novel SCP groupings with discriminatory markers that can indicate significantly different clinical outcomes or unique drug sensitivity. The novel thresholds provided herein distinguish SCP4 or SCP18 subsets of patients containing many ER-negative cells likely to have ER drug resistance, representing small subsets of this clinical classification. The inventors propose that classification based on the likelihood of drug resistance, such as SCP4 or SCP18, rather than sensitivity, provides an important clinical benefit by excluding patients from drug trials or treatment plans involving the indicated agents. Detection of KRAS mutations to predict tyrosine kinase therapy resistance in EGFR+ lung cancer is another example of a biomarker of drug resistance used by clinicians.

[0070] In some embodiments, multiple recommended drugs are assigned to the SCP, and these may be prescribed singly, in combination, or sequentially. Taking SCP3 as an example, first, when comparing the survival rates of patients belonging to this SCP within the clinical classification of parental HR+HER2- / +, both SCP1 and SCP3 showed a high proportion of CK+HR high These findings suggest that SCP1 patients may have a favorable outcome, whereas SCP3 patients may have a poor outcome. SCP offers a nuanced prognosis compared with traditional groupings that do not distinguish between SCP1 and SCP3, and may influence, for example, the clinician's urgency in managing cases, the selection of more aggressive drug treatments, or the consideration of combination therapy or a series of targeted agents.

[0071] Second, SCP3 tumors are characterized by a heterogeneous, mixed phenotype in which multiple tumor cells express markers indicative of sensitivity to different drugs. SCP3 tumors are defined by high levels of hormone receptor (HR) expression in more than 70% of cells of cell identity 10. High HR expression indicates that these tumor cells are likely to be sensitive to hormone-targeted drugs, such as aromatase inhibitors, SERMs, or SERD drugs (assigned to SCP3 in claim 9). However, SCP3 tumors still have a poor prognosis, suggesting that combinations of drugs may be effective. The additional presence of HR+ HER2- cells in SCP patients suggests that a proportion of SCP3 tumor cells will respond to PIK3 pathway inhibitors (assigned to SCP3 in claim 15). Finally, SCP3 tumors are also characterized by a smaller, but significant, proportion of HR- cells. Patients with aggressive HR- tumors are often prescribed drugs that target cell proliferation, such as antimitotic drugs or antimetabolites (assigned to SCP3 in claim 8). The SCP3 classification is associated with three different types of treatment, so clinicians should consider monotherapy, combination therapy, or sequential treatment with these agents.

[0072] In some circumstances, because SCP groups are relative and mutually exclusive, classification can provide implicit information about clinical outcomes that inform drug selection. In the example of SCP2, a high-risk or low-risk clinical outcome is not explicitly assigned, but patients are excluded from the similar SCP1, which is associated with a very favorable outcome, indicating an intermediate chance of survival. This may lead clinicians to select an aggressive second-line agent appropriate for SCP2 morphology, such as an antiangiogenic agent (assigned to SCP2 in claim 10), instead of a first-line hormone-targeted therapy, despite the presence of HR+ cells.

[0073] Information such as drug availability or cost, mutations, tumor severity, age, or comorbidities may also be considered in combination with the characterization of the tumor according to the present invention as broadly drug sensitive when selecting a treatment strategy for such patients.

[0074] Even a high probability of a poor prognosis, such as that associated with SCP8, is useful in clinical practice. Increased certainty in the setting of a poor prognosis can still be beneficial to patients, allowing them to choose alternative treatments best suited to prolonging survival, even if they are unlikely to cure the patient (e.g., SCP6), or, in certain cases where the risk of a poor prognosis is very high, such as highly heterogeneous SCP8 tumors, to choose palliative care given the unpleasant side effects of many breast cancer treatment regimens or comorbidities.

[0075] In certain embodiments, patients are assigned to putative outcome groups according to the presence of the cellular communities assigned in said cellular community assignment step: - CCI1, CCI3, CCI5, CCI8, CCI9, or CCI10: Likely a good outcome. The presence of these CCIs can indicate a significantly better outcome than the mean value for the entire sample. - CCI2, CCI4, CCI6, or CCI7: Likely poor outcome. The presence of more of these CCIs can indicate a significantly poorer outcome than the mean value for the entire sample.

[0076] To determine the most appropriate indicators of patient outcome, log-transformed community or single-cell densities, along with clinical subgrouping and grading, were introduced into Coxph survival models to find significant associations between specific community or single-cell types and patient risk, and hazard ratios were examined.

[0077] A further aspect of the present invention relates to a pharmaceutical formulation for use in the treatment of cancer, particularly cancer derived from tissue characterized by expression of steroid hormone receptors, particularly neoplastic cells derived from breast, ovarian or endometrial tissue, more particularly neoplastic cells derived from human breast or human mammary epithelium, wherein patients with said cancer have been assigned to a subgroup of patients defined by SCP or tumor identity according to the methods described herein.

[0078] A further set of embodiments relates to the use of pharmaceutical preparations for the treatment of cancer, particularly breast cancer, informed by morphological features characterizing the SCP grouping of patient samples.

[0079] A series of embodiments of this aspect of the invention relates to pharmaceutical formulations for use in the treatment of cancer, particularly breast cancer, wherein a patient with said cancer has been assigned an SCP selected from SCP3, SCP14, SCP15, and SCP16 according to the methods described herein.

[0080] In one embodiment, the pharmaceutical formulation for use in treating a cancer assigned to SCP3, SCP14, SCP15, or SCP16 comprises an anthracycline antineoplastic agent, particularly an anthracycline antineoplastic agent selected from daunorubicin, doxorubicin, epirubicin, idarubicin, mitoxantrone, and pixantrone.

[0081] In one embodiment, the pharmaceutical formulation for use in treating a cancer assigned to SCP3, SCP14, SCP15, or SCP16 comprises an anti-mitotic taxane antineoplastic agent, particularly an anti-mitotic taxane antineoplastic agent selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine.

[0082] In one embodiment, the pharmaceutical formulation for use in treating a cancer assigned to SCP3, SCP14, SCP15, or SCP16 comprises an antineoplastic agent selected from vincristine, vinblastine, vinorelbine, vinflunine, cryptophycin 52, halichondrin, dolastatin, and hemiasterlin, or an antineoplastic agent selected from the group including colchicine, podophyllotoxin, rigosertib, steganacin, ABT-751, combretastatin, and 2-methoxyestradiol.

[0083] In another embodiment, the pharmaceutical preparation for use in treating a cancer assigned to SCP3, SCP14, SCP15, or SCP16 comprises an anti-cancer platinum complex, in particular carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin tetranitrate.

[0084] In yet another embodiment, the pharmaceutical formulation for use in treating a cancer assigned to SCP3, SCP14, SCP15, or SCP16 comprises an alkylating antineoplastic agent, particularly an alkylating antineoplastic agent selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbayine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, thiptepa, or trabectedin.

[0085] In another embodiment, the pharmaceutical preparation for use in the treatment of a cancer assigned to SCP3, SCP14, SCP15, or SCP16 comprises an antimetabolite antineoplastic agent, particularly an antimetabolite antineoplastic agent selected from: azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate, or phototrexate.

[0086] The SCP information obtained by the method according to the above aspect of the invention is used to identify patients who may benefit from chemotherapeutic interventions that target proliferation and cell division.

[0087] Another aspect of this aspect of the invention relates to a pharmaceutical formulation for use in the treatment of cancer, wherein said cancer patient has been assigned an SCP characterized by high HR expression selected from SCP1, or SCP3, according to the methods described herein.

[0088] In one embodiment, the pharmaceutical formulation for use in treating a cancer assigned to SCP1 or SCP3 comprises a selective estrogen receptor modulator (SERM) antineoplastic agent, in particular a SERM agent selected from raloxifene, toremifene, or tamoxifen.

[0089] In one embodiment, the pharmaceutical formulation for use in treating a cancer assigned to SCP1 or SCP3 comprises a selective estrogen receptor degrader (SERD) antineoplastic agent, particularly fulvestrant, brilanestrant, and elacestrant.

[0090] In one embodiment, the pharmaceutical formulation for use in the treatment of a cancer assigned to SCP1 or SCP3 comprises an aromatase inhibitor antineoplastic agent, in particular an aromatase inhibitor antineoplastic agent selected from exemestane, letrozole, vorozole, formestane, fadrozole, and anastrozole.

[0091] Another alternative embodiment of this aspect of the invention relates to a pharmaceutical formulation for use in the treatment of cancer, wherein said cancer patient has been assigned to SCP2 according to the methods described herein. The pharmaceutical formulation according to this alternative embodiment comprises an anti-angiogenic antineoplastic agent, particularly an anti-angiogenic antineoplastic agent selected from bevacizumab, thalidomide, or lenalidomide.

[0092] Similarly, the present invention relates to a pharmaceutical formulation comprising an EZH2 methyltransferase inhibitor antineoplastic agent, in particular an EZH2 methyltransferase inhibitor antineoplastic agent selected from 3-deazaneplanocin A (DZNep), tazemetostat, EPZ005687, EI1, GSK126, or UNC1999, for use in the treatment of cancers assigned to SCP, in particular an SCP selected from SCP5 or SCP12, H3K27me3+ cells, or SCP characterized by intermediate HR expression.

[0093] Another alternative embodiment of this aspect of the present invention relates to a pharmaceutical formulation comprising an EGFR bioactive inhibitor antineoplastic agent, particularly an EGFR bioactive inhibitor antineoplastic agent selected from gefetinib, erlotinib, lapatinib, cetuximib, neratinib, osimeratib, panitumamib, vandetanib, necitumab, or dacomitinib, for use in the treatment of cancer, wherein the cancer patient is assigned to SCP15.

[0094] Yet another alternative embodiment of this aspect of the invention relates to a pharmaceutical formulation comprising a hypoxia-activated drug, such as a quinone alkylating antineoplastic drug, particularly mitomycin C, for use in the treatment of cancer, wherein the cancer patient is assigned to SCP17.

[0095] Yet another alternative embodiment of this aspect of the present invention relates to a pharmaceutical formulation comprising a HER2-targeted antineoplastic drug, therapy or vaccine, in particular a HER2-targeted antineoplastic drug selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, H2NVAC, and / or HER2.taNK cells, HER2-pulsed dendritic cells, or HER2-targeted T cells for HER2-targeted cell therapy, for use in the treatment of cancer, wherein cancer patients have been assigned to an SCP comprising HER2+ cells selected from SCP2, SCP6, SCP7, SCP9, SCP10, SCP11, or SCP13. In comparison to the clinical classifications presented in the examples that illustrate the state of the art, the present invention reclassifies HER2+ tumors into SCPs 11, 6, 7, 9, 10, 13, and 2, each of which contains additional markers that can be used to predict the efficacy of different clinical approaches.

[0096] Yet another alternative embodiment of this aspect of the invention relates to a pharmaceutical formulation comprising a PI3K pathway inhibitor, in particular rapamycin, dactolisib, BGT226, SF1126, PKI-587, NVPBE235, for use in treating cancer, wherein the cancer patient has been assigned to an SCP comprising HR high / +HER2- cells selected from SCP1, SCP3, SCP11, or SCP18.

[0097] Similarly, it is within the scope of the present invention to administer an effective amount of an anti-cancer drug to a patient according to the patient's assignment to a particular subgroup according to the above description.

[0098] Another aspect relates to a method of diagnosis whereby certain patients are assigned to a group for which chemotherapy discontinuation is recommended. The inventors speculate that SCP1 may not require treatment, even though current assessment would place these patients in a group for which conventional chemotherapy is recommended.

[0099] Another aspect of the present invention relates to a system for use in characterizing tumors in breast cancer patients in order to classify or stratify breast cancer patients into groups sharing similar cellular characteristics, the system being configured to implement the methods for tumor stratification disclosed herein.

[0100] The system components may include: a device for analyzing a cancer tissue sample obtained from a patient, in particular, the cancer tissue originating from human breast or human mammary epithelium; a plurality of molecular probes for labeling the cancer tissue sample, each probe specific for a biomolecule, each of the molecular probes characterized by a detectable marker, and the (labeled) biomolecules selected from the list comprising or consisting of biomolecules i to xii as defined according to the first aspect of the present invention; the device is capable of obtaining information on the expression of each of the plurality of biomolecules at single-cell resolution, and is configured to assign a cell identity (CI) to each single cell in the labeled tissue sample based on the expression of the plurality of biomolecules, wherein the cell identity is assigned according to the cellular expression of biomolecules identified by markers according to CI1 to CI14 as defined according to the first aspect of the present invention.

[0101] The system may optionally be further configured to, in the pathology group assignment step, assign a single cell pathology (SCP) patient group to the sample according to the proportion of each cell identity assigned in the cell assignment step that the sample contains, in accordance with the constituent cell types characterized by SCP1 to 18 defined in the first aspect of the present invention.

[0102] Similarly, the present invention encompasses a method of treating a patient diagnosed with a cancer disease characterized by a high likelihood of drug sensitivity associated with one of the SCP patient groups according to the above-defined aspects of the invention, which method involves administering to the patient an effective amount of an anti-neoplastic agent, or a pharmaceutically acceptable salt thereof, as defined in detail herein, which is likely to be effective in the patient group as identified according to any of the above-defined aspects of the invention.

[0103] The present invention further encompasses the use of the following features in a kit or in the manufacture of a kit for detecting subsets of cancer patients with different likelihoods of drug sensitivity and / or clinical outcome: First, a list comprising or consisting of: E-cadherin, CK18 and / or 19, CK7, ER and / or PR, markers of cell proliferation, in particular Ki-67 and / or PCNA, CK5 and / or p63 and / or CK14, p53, HR, in particular ER and / or PR, markers of apoptosis, in particular cPARP and / or cC3, EGFR, markers of hypoxia, in particular CAIX; and markers of DNA content, in particular DNA intercalating dyes; and optionally CD3 or CD90; CD20 or CD19; CD68; CD44 and / or CD45; fibronectin; vimentin, and CD31 and / or vWF and / or CD34; the use of molecular probes specific to the biomolecules of wherein said molecular probe is conjugated to an isotope or a fluorescent dye, in particular said molecular probe is selected from the list comprising or consisting of: Indium-113, Lanthanum-139, Praseodymium-141, Neodymium-142, Neodymium-143, Neodymium-144, Neodymium-145, Neodymium-146, Neodymium-148, Neodymium-150, Samarium-147, Samarium-149, Samarium-152, Europium-151, Europium-153, Gadolinium-155, Gadolinium-156, Gadolinium-158, Gadolinium-160, Terbium-159, Dysprosium-162, Dysprosium-163, Dysprosium-164, Erbium-166, Erbium-167, Erbium-168, Thulium-169, Ytterbium-170, Ytterbium-172, Ytterbium-173, Ytterbium-174, Ytterbium-176, Lutetium-175, is bonded to an isotope of

[0104] Some embodiments of this aspect of the invention relate to the use of antibodies as molecular probes, while other embodiments relate to the use of nucleic acids as molecular probes.

[0105] Whenever alternatives to a single separable feature are described herein as "embodiments," it is to be understood that such alternatives can be freely combined to form separate embodiments of the invention disclosed herein.

[0106] The present invention further encompasses the following:

[0107] A. A method for indicating a clinical outcome in a cancer patient, said method comprising the steps of: a. providing a cancer tissue sample obtained from a patient; b. Labeling the cancer tissue sample with a plurality of molecular probes, each probe specific for a biomolecule, wherein each of said molecular probes is characterized by a detectable marker, and said biomolecule is selected from the list comprising or consisting of: i. epithelial cadherin (E-cadherin), ii. cytokeratin (CK) 18 and / or 19, iii. CK7, iv. estrogen receptor (ER) and / or progesterone receptor (PR), v. Markers of cell proliferation, particularly Ki-67 and / or PCNA; vi. CK5 and / or p63 and / or CK14, vii. p53, viii. Hormone receptors (HR), in particular receptors for estrogen and / or progesterone (ER and PR, respectively); ix. Markers of apoptosis, in particular cleaved form of poly ADP-ribose polymerase (cPARP) and / or cleaved form of caspase 3 (cC3); x. epidermal growth factor receptor (EGFR), and xi. Markers of hypoxia, especially carbonic anhydrase (CAIX); and xii. Markers of DNA content, especially DNA intercalating dyes; c. obtaining information regarding the expression of each of the plurality of biomolecules at single-cell resolution in the reading step; d. In a cell assignment step, assigning a cellular identity (CI) to each single cell in the labeled tissue sample based on the expression of the plurality of biomolecules, wherein the cellular identity is assigned according to the cellular expression of biomolecules identified by markers according to the following list: CI1. CIAX hi , EGFR-, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)+; CI2.p53 hi , (cC3 and / or cPARP)+, (ER and / or PR)-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-; CI3. (Ki-67 and / or PCNA)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI4. p53 hi , EGFR+, CIAX hi, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI5. (CK5 and / or p63 and / or CK14)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-; CI6. E-cadherin hi , (CK18 and / or CK19) hi , CK7+, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)-; CI7. CK7+, (CK18 and / or CK19)+, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI8. E-cadherin-, CK7-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-, (ER and / or PR)-; CI9. (E-cadherin lo or E-cadherin-), ((CK18 and / or CK19)- or (CK18 and / or CK19) lo ), (ER and / or PR) lo , (CK5 and / or p63 and / or CK14)-; CI10. ((CK18 and / or CK19) hi or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR) hi , CK5 and / or p63 and / or CK14)-; CI11. ((CK18 and / or CK19) hi or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR)+, CK5 and / or p63 and / or CK14)-; CI12. (E-cadherin lo or E-cadherin+), ((ER and / or PR) lo or (ER and / or PR)-), (CK5 and / or p63 and / or CK14)-; CI13. p53 hi , EGFR+, (ER and / or PR) hi , ((CK5 and / or p63 and / or CK14) loand / or (CK5 and / or p63 and / or CK14)-); CI14. CK7+, (CK18 and / or CK19)+, (CK5 and / or p63 and / or CK14)+, (ER and / or PR)-, In the pathology group assignment step, assigning the cancer tissue samples to single cell pathology (SCP) patient groups according to the proportion of each cell identity contained in the samples assigned in the cell assignment step, wherein the list of SCP patient groups includes or consists of: SCP1. >70% of single cells are CI10; SCP2. >70% of single cells are CI11; SCP3. ≤70% of single cells are CI10; SCP4. >70% of single cells against CI12; SCP5. ≤70% of single cells are CI12; SCP6. >80% of single cells are CI9; SCP7. >80% of single cells are CI8; SCP8. ≤70% of single cells are CI9; or CI10; or CI12; SCP9, >60% of single cells are CI9; SCP10. >70% of single cells are CI9; or CI10; or CI12; SCP11. >60% of single cells are CI7; SCP12. >70% of single cells are CI6; SCP13. >50% of single cells are CI5; SCP14. >60% of single cells are CI3; SCP15. >70% of single cells are CI4; SCP16. >50% of single cells are CI2; SCP17. >50% of single cells are CI1;

[0108] B. The method described in item A, wherein the method for obtaining information regarding the average expression of the plurality of biomolecules comprises constructing an image of the cancer tissue sample.

[0109] C. The method according to item A or B, wherein the cancer tissue sample is a section.

[0110] D. The method of any one of paragraphs B or C, wherein the cancer tissue sample is a monolayer of adherent cells or cells otherwise immobilized on a solid surface.

[0111] E. The method of any one of paragraphs A to D, wherein the plurality of biomolecules is labeled by contacting the sample with a plurality of metal-conjugated or fluorescent dye-conjugated antibodies and / or nucleic acid probes.

[0112] F. The method of any one of items C to E, wherein the method for obtaining information about the average expression of the plurality of biomolecules is imaging mass cytometry at subcellular resolution, particularly at a resolution of ≦5 μm, and even more particularly ≦1 μm.

[0113] G. said method comprising: a. In the labeling step, xiii. CD3 or CD90; xiv. CD20 or CD19; xv. CD68; xvi. CD44 and / or CD45; xvii. Fibronectin; xviii Vimentin, and xix. CD31 and / or von Willebrand factor (vWF) and / or CD34; and further comprising a further marker selected from: b. In the cell allocation step, i.CI15. CD44+, CD45+, (CD3 or CD90)+, fibronectin-, E-cadherin-, (CK5 and / or p63 and / or CK14)lo or (CK5 and / or p63 and / or CK14)-); ii.CI16. CD20+, (fibronectin lo or fibronectin-), (E-cadherin lo or E-cadherin-), ((CK5 and / or p63 and / or CK14) lo or (CK5 and / or p63 and / or CK14)-); iii.CI17. (CD3 or CD90)+, (CD20 or CD19)+; iv.CI18.CD68+; v.CI19. Vimentin+, (CD34 and / or VWF and / or CD31)+; vi.CI20. Vimentin-, (fibronectin+ or fibronectin hi ), (CD3 or CD90)-, (CD20 or CD19)-, CD45-, CD44-, and a further cell identity selected from The method according to any one of items C to F, comprising:

[0114] H. The method of any one of items C to G, wherein the single cell is a fragment of an image of the cancer tissue sample, in particular a fragment consisting of pixels within an area where membrane-bound molecules surround a single nucleus.

[0115] I. The method comprises the steps of: a. a cellular community detection step, partitioning an image of a cancer tissue sample into multicellular regions, wherein each single cell within the multicellular region is highly interconnected with neighboring cells to provide a cellular community; b. In the cell community assignment step, assigning a cell community identity (CCI) to each cell community according to the number of cells in the cell community and the proportion of each cell community it contains, wherein the list of CCIs includes or consists of: CCI1. Among cells with identities CI1-CI14, >10% of single cells are CI6 and the average size of the cell community is >25 cells; or CCI2. Among cells with identities CI1-CI14, >10% of single cells are CI6 and the average size of the cell community is ≤25 cells; or CCI3. Of cells with identities CI1-CI14, >10% of single cells are CI2 and the average size of the cell community is >25 cells; or CCI4. Among cells with identities CI1-CI14, >10% of single cells are CI2 and the average size of the cell community is ≤25 cells; or CCI5. Of cells with identities CI1-CI14, >10% of the cells are CI3 and the average size of the cell community is ≤25 cells; or CCI6. >5% of the total cells are CI19, and >10% of the cells are CI20, and >3% of the cells are CI18, and the average size of the cell community is <50 cells; or CCI7. >80% of all cells are of any of the CI1-CI15 identities, and <10% of cells are of CI20 identity, and the average size of the cell community is <75 cells; or CCI8. >20% of all cells are CI15, and / or CI16, and / or CI17, and <40% of cells are any of CI1-CI14 identities, and the average size of the cell community is >75 cells; or CCI9. >5% of all cells are CI18 and the average size of the cell community is >25%; or CCI10: >80% of all cells are CI1-CI14 identities, and <2% of cells are CI20 identities, and the average size of the cell community is >115 cells and <125 cells. The method according to any one of items C to H, comprising:

[0116] J. In the pathology group assignment step, the patient is assigned to a predicted outcome group according to the SCP classification of the sample: - High likelihood of insensitivity to anti-cancer drugs targeting SCP4 or SCP18:ER; - SCP7: likely to lack sensitivity to antiangiogenic antineoplastic agents; - SCP1, SCP11: high probability of a good outcome; - SCP8, SCP14, or SCP17: High probability of adverse outcome; The method according to any one of items A to H, wherein

[0117] K. According to the number of cell communities assigned in the cell community assignment step, the patient is assigned to an estimated outcome group: - CCI1, CCI3, CCI5, CCI8, CCI9, or CCI10: High probability of a good outcome; - CCI2, CCI4, CCI6, or CCI7: High probability of poor outcome; The method according to item I, assigned to

[0118] L. The method of any one of paragraphs A to K, wherein the cancer tissue sample comprises or consists essentially of neoplastic cells derived from tissue characterized by expression of a steroid hormone receptor, particularly neoplastic cells derived from breast, ovarian or endometrial tissue, more particularly neoplastic cells derived from human breast or human mammary epithelium.

[0119] M. Pharmaceutical formulations comprising: a. an anthracycline antitumor drug, particularly an anthracycline antitumor drug selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; or b. Anti-mitotic antineoplastic agents, particularly those selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine (see UZ341WO); or c. Antitumor platinum complexes, especially carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin tetranitrate; or d. alkylating antineoplastic agents, particularly alkylating antineoplastic agents selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbaine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, thiptepa, or trabectedin; or e. Antimetabolite antineoplastic agents, in particular antimetabolite antineoplastic agents selected from the following: azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate or phototrexate; 1. A pharmaceutical formulation for use in the treatment of cancer, comprising: f. The pharmaceutical formulation wherein the cancer patient is assigned to an SCP selected from SCP3, SCP14, SCP15, and SCP16.

[0120] N. A pharmaceutical preparation comprising: a. selective estrogen receptor modulator (SERM) antineoplastic agents, in particular SERM drugs selected from raloxifene, toremifene, and tamoxifen; b. selective estrogen receptor degrader (SERD) antineoplastic agents, particularly fulvestrant, brilandestrant, and elacestrant; and / or c. Aromatase inhibitor antineoplastic agents, in particular aromatase inhibitor antineoplastic agents selected from exemestane, letrozole, vorozole, formestane, fadrozole and anastrozole; 1. A pharmaceutical formulation for use in the treatment of cancer, comprising:

[0121] O. A pharmaceutical formulation comprising an anti-angiogenic antineoplastic agent, particularly an anti-angiogenic antineoplastic agent selected from bevacizumab, thalidomide, or lenalidomide, for use in the treatment of cancer, wherein the cancer patient is assigned to SCP2.

[0122] P. A pharmaceutical formulation for use in the treatment of cancer, comprising an EZH2 methyltransferase inhibitor antineoplastic agent, particularly an EZH2 methyltransferase inhibitor antineoplastic agent selected from 3-deazaneplanocin A (DZNep), tazemetostat, EPZ005687, EI1, GSK126, or UNC1999, wherein the cancer patient is assigned to an SCP selected from SCP5 or SCP12.

[0123] Q. A pharmaceutical formulation for use in the treatment of cancer, comprising an EGFR bioactive inhibitor anti-neoplastic agent, particularly an EGFR bioactive inhibitor anti-neoplastic agent selected from gefetinib, erlotinib, lapatinib, cetuximib, neratinib, osimeratib, panitumamib, vandetanib, necitumab, or dacomitinib, wherein the cancer patient is assigned to SCP15.

[0124] R. A pharmaceutical formulation comprising a quinone alkylating antitumor drug, particularly mitomycin C, for use in the treatment of cancer, wherein the cancer patient is assigned to SCP17.

[0125] S. A pharmaceutical formulation for use in the treatment of cancer, comprising a HER2-targeted anti-cancer drug, particularly a HER2-targeted anti-cancer drug selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, and H2NVAC, and HER2-targeted NK cells (HER2.taNK cells), HER2-pulsed dendritic cells, or HER2-targeted T cells, wherein the cancer patient has been assigned to an SCP selected from SCP2, SCP6, SCP7, SCP9, SCP10, SCP11, or SCP13.

[0126] T. A pharmaceutical formulation comprising a PI3K pathway inhibitor, particularly rapamycin, dactolisib, BGT226, SF1126, PKI-587, NVPBE235, for use in the treatment of cancer, wherein the cancer patient is assigned to an SCP selected from SCP1, SCP3, SCP11, or SCP18.

[0127] Item 1. A method for indicating a clinical outcome in a cancer patient, comprising the steps of: a. providing a cancer tissue sample obtained from a patient; b. labeling a cancer tissue sample with a plurality of molecular probes, each probe specific to a biomolecule, each of said molecular probes characterized by a detectable marker, particularly wherein said plurality of molecular probes can distinguish at least 20 different biomolecules, more particularly wherein said plurality of molecular probes can distinguish at least 25 different biomolecules; c. obtaining information regarding the expression of each of the plurality of biomolecules at single-cell resolution in the reading step; d. in a cell assignment step, assigning a cellular identity (CI) to each single cell in the labeled tissue sample based on the expression of the plurality of biomolecules, wherein the cellular identity is assigned according to the cellular expression of biomolecules identified by markers; e. In a pathology group assignment step, the cancer tissue samples are assigned to single cell pathology (SCP) patient groups according to the proportion of each cell identity contained in the samples assigned in the cell assignment step. The method comprising:

[0128] Item 2. The method described in Item 1, wherein the method for obtaining information regarding the average expression of the plurality of biomolecules comprises constructing an image of the cancer tissue sample.

[0129] Item 3. The method according to item 1 or 2, wherein the cancer tissue sample is a section, or a monolayer of adherent cells, or cells immobilized on a solid surface.

[0130] Item 4. The method according to any one of Items 1 to 3, wherein the method for obtaining information about the average expression of a plurality of biomolecules is imaging mass cytometry at a subcellular resolution, particularly at a resolution of ≦5 μm, and even more particularly at a resolution of ≦1 μm.

[0131] Item 5. The method according to any one of Items 1 to 4, wherein the plurality of biomolecules are labeled by contacting the sample with a plurality of metal-conjugated or fluorescent dye-conjugated antibodies and / or nucleic acid probes.

[0132] Item 6. The method according to any one of Items 3 to 5, wherein the method for obtaining information about the average expression of a plurality of biomolecules is imaging mass cytometry at subcellular resolution, particularly at a resolution of ≦5 μm, and even more particularly at a resolution of ≦1 μm.

[0133] Item 7. The method according to any one of Items 2 to 6, wherein the single cell is a fragment of an image of the cancer tissue sample, in particular a fragment consisting of pixels within an area where membrane-bound molecules surround a single nucleus.

[0134] Item 8. The method comprises the following steps: a. a cellular community detection step, partitioning an image of a cancer tissue sample into multicellular regions, wherein each single cell within the multicellular region is highly interconnected with neighboring cells to provide a cellular community; b. In a cellular community assignment step, assigning a cellular community identity (CCI) to each cellular community according to the number of cells in the cellular community and the proportion of each CI it contains; The method according to any one of items 2 to 7, comprising:

[0135] Item 9. The method according to any one of Items 1 to 8, wherein in the pathology group assignment step, the patient is assigned to an estimated outcome group according to the SCP classification of the sample.

[0136] Item 10. The biomolecule for which the molecular probe is specific is one of the following: i. epithelial cadherin (E-cadherin), ii. cytokeratin (CK) 18 and / or 19, iii. CK7, iv. estrogen receptor (ER) and / or progesterone receptor (PR), v. Markers of cell proliferation, particularly Ki-67 and / or PCNA; vi. CK5 and / or p63 and / or CK14, vii. p53, viii. Hormone receptors (HR), in particular receptors for estrogen and / or progesterone (ER and PR, respectively); ix. Markers of apoptosis, in particular cleaved form of poly ADP-ribose polymerase (cPARP) and / or cleaved form of caspase 3 (cC3); x. epidermal growth factor receptor (EGFR), and xi. Markers of hypoxia, especially carbonic anhydrase (CAIX); and xii. Markers of DNA content, especially DNA intercalating dyes; xiii. CD3 or CD90; xiv. CD20 or CD19; xv. CD68; xvi. CD44 and / or CD45; xvii. Fibronectin; xviii. Vimentin, and xix. CD31 and / or von Willebrand factor (vWF) and / or CD34 10. The method according to any one of items 1 to 9, selected from the list comprising or consisting of:

[0137] Item 11. The cell identity assigned in the cell assignment step is one of the following: CI1. CIAX hi , EGFR-, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)+; CI2.p53 hi , (cC3 and / or cPARP)+, (ER and / or PR)-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-; CI3. (Ki-67 and / or PCNA)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI4. p53 hi , EGFR+, CIAX hi , (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI5. (CK5 and / or p63 and / or CK14)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-; CI6. E-cadherin hi , (CK18 and / or CK19) hi , CK7+, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)-; CI7. CK7+, (CK18 and / or CK19)+, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI8. E-cadherin-, CK7-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-, (ER and / or PR)-; CI9. (E-cadherin lo or E-cadherin-), ((CK18 and / or CK19)- or (CK18 and / or CK19) lo ), (ER and / or PR) lo , (CK5 and / or p63 and / or CK14)-; CI10. ((CK18 and / or CK19) hi or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR) hi , CK5 and / or p63 and / or CK14)-; CI11. ((CK18 and / or CK19) hi or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR)+, CK5 and / or p63 and / or CK14)-; CI12. (E-cadherin lo or E-cadherin+), ((ER and / or PR) lo or (ER and / or PR)-), (CK5 and / or p63 and / or CK14)-; CI13. p53 hi , EGFR+, (ER and / or PR) hi , ((CK5 and / or p63 and / or CK14) lo and / or (CK5 and / or p63 and / or CK14)-); CI14. CK7+, (CK18 and / or CK19)+, (CK5 and / or p63 and / or CK14)+, (ER and / or PR)- CI15. CD44+, CD45+, (CD3 or CD90)+, fibronectin-, E-cadherin-, (CK5 and / or p63 and / or CK14 )lo or (CK5 and / or p63 and / or CK14)-); CI16. CD20+, (fibronectin lo or fibronectin-), (E-cadherin lo or E-cadherin-), ((CK5 and / or p63 and / or CK14) lo or (CK5 and / or p63 and / or CK14)-); CI17. (CD3 or CD90)+, (CD20 or CD19)+; CI18.CD68+; CI19. Vimentin+, (CD34 and / or VWF and / or CD31)+; CI20. Vimentin-, (Fibronectin+ or Fibronectin hi ), (CD3 or CD90)-, (CD20 or CD19)-, CD45-, CD44- 11. The method according to any one of items 1 to 10, selected from the list comprising or consisting of:

[0138] Item 12. The SCP patient group assigned in the pathology group assignment step is one of the following: SCP19. >70% of single cells are CI10; SCP20. >70% of single cells are CI11; SCP21. ≤70% of single cells are CI10; SCP22. >70% single cells for CI12; SCP23. ≤70% of single cells are CI12; SCP24. >80% of single cells are CI9; SCP25. >80% of single cells are CI8; SCP26. ≤70% of single cells are CI9; or CI10; or CI12; SCP27. >60% of single cells are CI9; SCP28. >70% of single cells are CI9; or CI10; or CI12; SCP29. >60% of single cells are CI7; SCP30. >70% of single cells are CI6; SCP31. >50% of single cells are CI5; SCP32. >60% of single cells are CI3; SCP33. >70% of single cells are CI4; SCP34. >50% of single cells are CI2; SCP35. >50% of single cells are CI1; SCP36. >90% of single cells are CI14 12. The method according to any one of items 1 to 11, selected from the list comprising or consisting of:

[0139] Item 13. The cell community identity (CCI) assigned in the cell community assignment step is one of the following: CCI1. Of cells with identities CI1-CI14, >10% of single cells are CI6 and the average size of the cell community is >25 cells; or CCI2. Among cells with identities CI1-CI14, >10% of single cells are CI6 and the average size of the cell community is ≤25 cells; or CCI3. Of cells with identities CI1-CI14, >10% of single cells are CI2 and the average size of the cell community is >25 cells; or CCI4. Of cells with identities CI1-CI14, >10% of single cells are CI2 and the average size of the cell community is ≤25 cells; or CCI5. Of cells with identities CI1-CI14, >10% of the cells are CI3 and the average size of the cell community is ≤25 cells; or CCI6. >5% of the total cells are CI19, and >10% of the cells are CI20, and >3% of the cells are CI18, and the average size of the cell community is <50 cells; or CCI7. >80% of all cells are of any of the CI1-CI15 identities and <10% are of CI20 identity, and the average size of the cell community is <75 cells; or CCI8. >20% of all cells are CI15, and / or CI16, and / or CI17, and <40% of cells are any of CI1-CI14 identities, and the average cell community size is greater than 75 cells; or CCI9. >5% of all cells are CI18 and the average size of the cell community is >25%; or CCI10: >80% of all cells are CI1-CI14 identities, <2% of cells are CI20, and the average cell community size is >115 cells but <125 cells. 10. The method according to any one of items 3 to 9, selected from the list comprising or consisting of:

[0140] Item 14. The method of any one of Items 1 to 13, wherein the cancer tissue sample comprises or essentially consists of neoplastic cells derived from tissue characterized by expression of a steroid hormone receptor, particularly neoplastic cells derived from breast, ovarian or endometrial tissue, more particularly neoplastic cells derived from human breast or human mammary epithelium.

[0141] The present invention is further illustrated by the following examples and figures, from which further embodiments and advantages can be derived, which are intended to illustrate the invention without limiting its scope. [Example]

[0142] method Clinical Data Tumor samples and patient metadata were collected from cohorts obtained from the University Hospitals of Basel and Zurich. The Basel cohort included 281 patients who were not selected for any clinical or histological features. Pathologists assessed the suitability of tissue sections for tissue microarray (TMA) construction (Kononen, J. et al., Nat. Med. 4, 844-847 (1998)). TMAs contained one 0.8 mm tumor core per patient, possibly matched healthy breast tissue samples, and a small number of control samples (liver tissue). The Zurich cohort consisted of 72 patients, and samples included four 0.6 mm cores from four different regions of each tumor, as described (Kundig, P. et al., J. Translat. Med., 16(1), 118 (2018)). Tumor cores were punched from two central and two peripheral areas, with an average distance between areas of 1 cm. Samples were selected to represent equal representation of different tumor grades as well as patients with and without lymph node metastasis. A total of 720 images of tumors of different sizes and locations were acquired. This project was approved by the local ethics committee (reference numbers: 2014-397 and 2012-0553).

[0143] Antibody Panel A panel of antibodies was designed that targets breast cancer-specific epitopes as well as markers of cell cycle and phosphorylation signaling, and distinguishes between epithelial, endothelial, mesenchymal, and immune cell types (Figure 16).

[0144] Tissue preparation and staining Tissue samples were formalin-fixed and paraffin-embedded at the University Hospitals of Basel and Zurich. Tissue sections were stained using a panel of antibodies. Tissue sections were delipidated overnight in xylene and rehydrated in graded alcohols (ethanol:deionized water 100:0, 90:10, 80:20, 70:30, 50:50, 0:100; 5 min each). Heat-induced epitope retrieval was performed in a 95°C water bath using Tris-EDTA buffer, pH 9, for 20 min. Tissue microarrays were immediately cooled and blocked with 3% BSA, 5% goat serum in TBS for 1 h. Samples were incubated overnight at 4°C with 7.5 g / L primary antibody diluted in TBS / 0.1% Triton X-100 / 1% BSA. Tissue samples were washed twice with TBS / 0.1% Triton X-100 and twice with TBS and then dried before imaging mass cytometry measurements.

[0145] In combined immunofluorescence staining and imaging mass cytometry, tissues were stained with metal-conjugated mouse HER2 ( 151 Eu) and rabbit pan-cytokeratin ( 175 Lu) primary antibody at 4°C overnight, then washed and stained with fluorescent-conjugated and metal-conjugated anti-mouse (AF488, 165 Ho) and anti-rabbit (AF555, 159 Secondary staining with Tb was performed by adding the mixture for 1 hour at room temperature. A coverslip was added and the tissue was imaged for fluorescent signal. The coverslip was then removed, and the sample was washed, dried, and acquired by mass cytometry laser ablation.

[0146] Imaging Mass Cytometry Images were acquired using a Hyperion imaging system (Fluidigm). The largest square area from each core of the TMA was laser ablated in a 200 Hz raster pattern, and raw data preprocessing was completed using commercially available acquisition software (Fluidigm). IMC acquisition stability was monitored by interspersed acquisitions of isotope-containing polymers (Fluidigm). All successful image acquisitions were processed, and images containing variations in pan-marker staining specific to the TMA location were removed. If the acquisition was interrupted and then continued, two tumor images from the same patient were included. Therefore, for the 281-patient cohort, 289 tumor, 87 healthy breast, and 5 liver control images were acquired. Where applicable, signal spillover between channels was corrected using functions in the CATALYST R package (version 1.5.6, Chevrier, S. et al. Cell Syst. 6, 612-620 (2018)). In the 72-patient cohort, 263 tumor, 68 healthy breast, and 6 control images were obtained and used for analysis.

[0147] Data Processing Data were converted to .tiff format and segmented into single cells using a flexible analysis pipeline available at https: / / github.com / BodenmillerGroup / ImcSegmentationPipeline. Briefly, individual cells and tumor / stroma regions were segmented using Ilastik 1.1.9. ( Segmentation was performed using a combination of Sommer, C. et al., From Nano to Macro 230-233 (IEEE, 2011) and CellProfiler 2.1.1 (Carpenter et al., Genome Biol. 7:R100 (2006)). Ilastik was used to create probability maps by classifying pixels (single cells—nucleus, membrane, and background; tumor / stroma—tumor, stroma, and background) based on combinations of antibody staining that distinguish membranes and nuclei. CellProfiler was then used to segment the single cells or tumor and stroma object masks based on the probability maps.

[0148] Single-cell segmentation masks were overlaid on the 35-channel tiff images, and single-cell marker expression measures and spatial features were extracted using the Matlab toolbox regionprops, as implemented in histoCAT (Schapiro, D. et al. Nat. Biotechnol. 31, 545-552 (2013)).

[0149] For each cell, the single-cell identity (ID) of the direct neighbors within 4 pixels (4 µm) of the target cell was detected and recorded using histoCAT software. The number of pixels to extend for the neighbor search was chosen to fill small gaps in the segmentation and not record cells after the direct neighbors (cell minor axis length: 5th-95th percentile 4.84-14.59 pixels, mean 9.51 pixels).

[0150] Individual cells inside or outside the tumor mask were located, and the distance of each cell to the tumor boundary (from inside and outside the tumor area) was calculated using the Matlab toolbox regionprops. Distance was measured between the nearest pixels of the object of interest.

[0151] Data transformation and normalization The data shown were not transformed, and all analyses were based on raw IMC measurements. Single-cell marker expression was summarized by the mean pixel value for each channel. Single-cell data were censored at the 99th percentile to remove outliers, and Z-scored cluster means were visualized in heatmaps. For t-distributed stochastic neighbor embedding (t-SNE) and PhenoGraph, data were normalized to the 99th percentile. To visualize cell counts per image or patient and perform viability modeling, counts were normalized by image area (total number of pixels) and displayed as cell densities. For Coxph viability modeling, these densities were multiplied by 10 to obtain values ​​greater than 1. 31The values ​​were multiplied by a factor of and then logarithmic transformation was performed.

[0152] Clustering and Metaclustering Single cells from a large cohort from the University Hospital of Basel were subjected to initial unsupervised clustering and aggregation of these clusters into larger groups based on average marker correlations, using PhenoGraph (Levine, JH et al. Cell 162, 184-197 (2015)), (Bodenmiller, Cell Syst. 2, 225-238 (2016)) were used to cluster the data into phenotypically similar cell groups. In the first step, the data were overclustered to detect and separate rare cell subpopulations. PhenoGraph (version 2.0) was used with default parameters (as implemented in histoCAT / Cyt) and 20 nearest neighbors. For high-dimensional clustering, 29 markers and four cell shape features were used: iridium, histone, phosphohistone, CK14, CK5, CK8 / 18, CK19, CK7, panCK, E / P-cadherin, ER, PR, HER2, GATA3, SMA, vimentin, fibronectin, vWF / CD31, CD44, CD45, CD68, CD3, CD20, cleaved caspase 3 / cleaved PARP, and carbonic anhydrase. Anhydrase, phospho-S6, Ki67, p53, EGFR, area, eccentricity, extent, and number of neighbors. Of the 71 clusters obtained, 59 epithelial / tumor groups were collapsed into larger groups by hierarchical clustering of average marker correlations (Euclidean distance and Ward linkage). Uncertainty in each subtree was assessed using a multiscale bootstrap resampling method (R package pvclust, version 2.0), and hierarchical separation was assigned to maintain significant epithelial subtrees and separate known biological differences. This resulted in 14 tumor cell metaclusters, varying in size and subtree robustness. Clusters showing typical marker expression of stromal and immune cells, restricted by a panel focused on tumor markers, were retained, similar to the original PhenoGraph clustering, and were not collapsed into larger groups. This metaclustering yielded 27 cell subgroups representing various immune, stromal, and epithelial / tumor cell types.The level of phenotyping or clustering of the studied cell types depends on both the panel and the parameter selection. Finer classification of cell types may reveal subtle differences in cellular marker expression, but will limit comparability between tumors, as many tumor cell types are patient-specific.

[0153] Cluster matching between cohorts Single cells from the second cohort at University Hospital Zurich were unsupervised and independently clustered using PhenoGraph with the same settings as the first cohort and a nearest neighbor parameter of 30. These clusters were matched to the most similar metaclusters from the previous cohort using Pearson correlation of z-scored mean marker expression. In two special cases (clusters 8 and 15), the clusters of interest correlated somewhat poorly with all metaclusters but most highly with stromal cell types. Visual inspection of the images indicated that these clusters represented cells forming distinct tumor masses, and these clusters were manually reassigned.

[0154] bh-tSNE Algorithm for Visualizing High-Dimensional Data For visualization, the high-dimensional single-cell data were reduced to two dimensions using the nonlinear dimensionality reduction algorithm t-distributed stochastic neighbor embedding (tSNE) (Amir AD et al., Nat Biotechnol. 31, 545–552 (2013)). The Barnes-Hut implementation of tSNE (bh-tSNE) was applied to the 99th percentile normalized data with default parameters (initial dimension, 110; perplexity, 30; theta, 0.5). The algorithm was run on a randomly subsampled set of cells (20% from each image) to avoid obscuring visible patterns in the dense plot and to improve computational performance.

[0155] Neighborhood analysis To identify significantly enriched or depleted pairwise neighbor interactions between cell types, a permutation test-based analysis of spatial single-cell neighbors was performed using the histoCAT function (Shapiro, 2017). Neighboring cells were defined as cells within 4 pixels (4 μm). A p-value cutoff of <0.01 was used for significance.

[0156] Single-cell pathology patient group classification Patients were grouped based on the proportion of tumor cell metaclusters using the cytofkit R implementation of PhenoGraph (version 1.10.0, Levine, 2015) with 8 nearest neighbors and default parameters. The number of nearest neighbors was selected to allow for the separation of small groups of patients consisting of distinct predominant cell types. Selecting a higher value for this parameter would result in fewer groups, resulting in patients with completely unrelated predominant phenotypes being grouped together. Lower values ​​of the nearest neighbor parameter may capture subtle differences in cellular composition by tumor type, but would severely limit statistical power for group comparisons and survival analyses. Patient group 18 exhibited rare HR + / CK - Due to lack of statistical power due to including only three patients with well-defined tumors with strong cell type predominance, they were excluded from further downstream analysis.

[0157] Single-cell pathology patient matching Tumor cores from a second cohort at University Hospital Zurich were assigned to the most similar previously defined single-cell pathology group based on the matched tumor cell type composition, using the inverse Pearson correlation as the distance metric.

[0158] Spatial Community The image was converted into a topological adjacency graph, in which every cell was represented by a node (visualized by its centroid), and nodes were connected by edges if the cells were directly adjacent. Neighbors were defined as cells within 4 pixels (4 μm) of the cell's assigned outermost pixel. We then applied the Louvain community detection algorithm (Blondel, VD et al. J. Stat. Mech. P 10008 (2008)) (C implementation by Lefebvre and Guillaume, version 0.2, wrapped in Matlab as used by the PhenoGraph 2.0 implementation used by histoCAT / Cyt) to identify highly interconnected spatial subunits in the tissue graph. Although it is known that applying community detection algorithms to spatially constrained networks can mask the underlying non-space-driven solutions, the purpose of applying the algorithm here was solely to extract spatial information and identify communities based on physical proximity (Expert P. et al., PNAS 108(19), 7663-7668 (2011)). This analysis was performed on epithelial cells alone (without including stromal or immune cells in the graph) to identify tumor communities, and again on all cells in the tissue to identify tumor microenvironment communities. A tumor-specific cohesion score was calculated based on the average size of the identified tumor communities. To focus on cohesive cell patches rather than individual disconnected cells, communities containing fewer than 10 cells were excluded from further analysis. Fifteen patients were excluded from tumor community-based analysis because the imaged area did not contain any tumor communities consisting of at least 10 cells. To identify recurrent, similar spatial cell-type communities, we ran cytofkit PhenoGraph (Levine, 2015, version 1.10.0) on the min-max normalized absolute number of cells in each cell metacluster in each community.This analysis was performed separately for tumor communities based only on epithelial cell types (k = 80) and microenvironment communities based on all cells but considering only individual stromal cell types and aggregating all tumor cell types into a single label (cell type group 100: including all tumor cells, k = 30). This analysis was performed separately for each cohort but based on matched metacluster cell types.

[0159] interstitial environment Based on the composition of the microenvironment community, hierarchical clustering (Euclidean distance and Ward linkage) was used to group the images into 11 stromal environment types. This analysis was performed separately for each cohort, but based on matched metacluster cell types.

[0160] Overlapping taxonomy and enrichment To identify patient groups with single-cell pathology enriched in specific stromal environments, Fisher's exact test was used. This test was performed for all potential stromal regions in the patient group using the R function fisher.test (parameter enrichment = "greater"). p-values ​​were corrected for multiple testing using the Bonferroni method. This enrichment analysis was also performed for different combinations of single-cell pathology subgroups, stromal environments, and clinical classifications.

[0161] Survival curves and Coxph regression models Kaplan-Meier survival curves and Coxph survival regression models were generated using the survival package in R (version 2.42-4). Overall survival and disease-free survival were analyzed for patients in different clinically or single-cell-defined subgroups. To determine whether patient subgroups significantly deviated from the remaining patient survival rates, or from the survival rates of similar SCP groups or other patients in the same clinical classification, both the log-rank test and Coxph models were employed. To identify significant associations between specific community or single-cell types and patient risk and to examine hazard ratios, log-transformed community or single-cell densities, along with clinical subgroup classification and ranking, were applied to the Coxph survival models. Nested Coxph models were compared using likelihood ratio tests (R package anova.coxph) to assess whether additional variables improved the survival model.

[0162] Example 1: Using imaging mass cytometry, we simultaneously quantified 35 biomarkers and generated 720 high-dimensional immunohistochemical pathology images of tumor tissue from 352 breast cancer patients with available long-term survival data. Spatial single-cell analysis distinguished the phenotype, organization, and heterogeneity of single cells in the tumor and stroma, enabling the categorization of breast cancer cellular structures based on cellular composition and histological organization. This analysis revealed multicellular signatures of the tumor microenvironment and novel breast cancer subgroups associated with distinct clinical outcomes. Thus, spatially resolved single-cell analysis can characterize intratumor phenotypic heterogeneity in a disease-relevant manner that can inform patient-specific diagnostics.

[0163] Spatially resolved single-cell phenotypes To comprehensively quantify the cellular heterogeneity and spatial organization of breast cancer tissues, we designed a breast cancer tissue-specific imaging mass cytometry (IMC) panel to image 281 tumor samples representing all clinical subtypes and pathological grades. IMC combines immunohistochemical staining using metal isotope-labeled antibodies with laser ablation and mass spectrometry-based detection to generate high-dimensional images (Giesen et al., Nat. Methods 11, 417-422 (2014)). This panel of 35 antibodies simultaneously quantified clinically established breast cancer targets estrogen receptor (ER), progesterone receptor (PR), and HER2; the proliferation marker Ki-67; markers of epithelial, mesenchymal, immune, and endothelial cell lineages; and targets providing insights into signaling pathways, oncogenes, and epigenetics (Figure 16). IMC produces images comparable to immunofluorescence or immunohistochemistry and is capable of highly multiplexed staining (Figures 1 and 2).

[0164] Images were segmented into single cells, tumor regions, and stromal regions using a random forest pixel classifier (Ilastik) and CellProfiler (Bodenmiller, 2016). We identified 855,668 cells in 381 images (289 tumor, 87 healthy breast, and 5 liver control images) and quantified their marker expression and spatial characteristics (Figure 3). Clustering using PhenoGraph (Levine et al., 2015) identified 59 diverse tumor cell phenotypes, including clusters of endothelial cells, T and B cells, macrophages, and stromal cell populations (hereafter referred to as phenotypes) (Figures 4 and 5). Some tumor phenotypes were unique to individual patients (Figure 4). To identify common cellular subtypes within this diversity, hierarchical clustering of tumor single-cell phenotypes defined by PhenoGraph defined 14 tumor cell metaclusters (Figure 6).

[0165] Tumors of all clinical subtypes contained similar densities of fibroblasts, endothelial cells, and immune cell populations, but cytokeratins, hormone receptors, and HER2 expression were enriched in tumor cell populations with variability depending on the clinical subtype (Figures 5 and 7a). Across all patients, immune cells were excluded from the tumor mass, but immune cells and fibroblasts infrequently infiltrated the tumor mass and were present in rare HR cells lacking cytokeratins. low / - Cells invaded beyond the tumor-stroma front in some samples (Figs. 5 and 7a). The tumor area contained various luminal HRs identified by combinations of ER, PR, GATA3, E-cadherin, and multiple cytokeratins. + Although epithelial cell phenotypes were included, hormone receptors were also expressed in a few cases (metacluster 26) without cytokeratins (Figures 5 and 7b). Of the luminal cytokeratins (CK7, CK8 / 18, and CK19), only CK7 was associated with a specific subset of luminal tumor cells (metaclusters 19 and 20) (Figure 5). HER2 expression was not a defining feature of the metaclusters and was observed at various levels across multiple phenotypes. The absence of hormone receptor and HER2 receptor expression (a hallmark of triple-negative breast cancer (TNBC)) included metaclusters with elevated levels of Ki-67, p53, EGFR, and the hypoxia marker CAIX (metaclusters 15–17), basal cytokeratins (metacluster 18), and luminal cytokeratins (PG clusters within metaclusters 19 and 22) (Figure 5).

[0166] Multicellular breast cancer structure We evaluated patterns of multicellular structure in breast tumor tissue based on these single-cell phenotypes. Tissue function is realized by multicellular units or communities, consisting of higher-order interactions rather than pairwise interactions between one or more cell phenotypes. We identified communities by first constructing a topological neighbor-cell interaction network and then applying a graph-based community detection approach (Blondel, 2008) using the Louvain algorithm. Community detection applied only to tumor cells identified dense epithelial patches of various sizes, termed tumor communities (TCs); when applied to all cells, microenvironment communities (MCs), containing tumor and stromal cell components, were identified (Figure 8). Using PhenoGraph, we grouped multicellular communities according to community size and tumor cell phenotype (Figure 9, TCs) or grouped all cells agnostic to tumor cell type (Figure 10, MCs). Tumor communities were separated based on absolute cell numbers, with single-cell metaclusters dominating (Figure 9) (TCs 4, 7, and 18; Supplementary Image). Some microenvironment communities consisted of fibroblasts interacting with various tumor cells (MCs 2, 5, and 8), while others displayed sparse stromal content (MCs 14, 17, 18, 20, 21, and 22). Others were enriched for T cells (MCs 19, 25, and 30), macrophages (MCs 27), a large network of T and B cells (MCs 1), or endothelial cells (MCs 13, 6, 30, 25, and 7) (Figure 10). Fibroblast-enriched communities had fewer interacting immune cells, consistent with fibroblasts' known role as agents of tumor desmoplasia and immune exclusion.

[0167] Single-cell pathological subgroups are associated with clinical outcomes The organization of single cells into communities contributes to the histological architecture of breast cancer. Therefore, we investigated tumor subtypes with different clinical outcomes. Cells derived from multicellular metaclusters were detected in all clinically defined breast cancer subtypes (Figure 7), supporting the conclusion that more general pathological classifications do not adequately resolve inter- and intrapatient cellular heterogeneity. To determine whether single-cell pathological landscapes can classify patients with higher resolution than classical histology-based clinical subtypes, we used unsupervised clustering to group patient tumors based on tumor cell metacluster composition and identified 18 single-cell pathological (SCP) subgroups that separate classical clinical subtypes (Figures 11a and 12). SCP subgroups contained varying proportions of epithelial tumor communities (Figure 11b), and individual SCP subgroups had different clinical outcomes when compared with all other patients, SCP subgroups of the same clinical classification, and other SCP subgroups containing similar cellular metaclusters but with different structures (Figures 12 and 13, Tables 2 and 3).

[0168] HR + Clinically defined tumors are classified into two groups: tumors strongly enriched in cells that highly express hormone receptors (SCPs 1-5, 12) and tumors with a small number of HR receptors surrounded by many cells that express only low levels of or lack HR receptors. hi / + These are currently not clinically classified (Fig. 11a, 7, 12). + / HR hi SCP1, which contains predominantly tumor cells, was only associated with patients who did not die of disease (succumb). Conversely, SCP1, which contains the same cellular metaclusters but has a different structure, is a smaller community, and is associated with CK low / HR low SCP3, which has a relatively high proportion of cells in metaclusters 22 and 25, is a CK low / HR low CK cells were predominantly involved in SCP6 and SCP9, and were also associated with poor prognosis (Figs. 11a, 13c, 13e, and 14). + / HR + SCP2 containing cells are HR+ / HER2 + Significantly enriched in clinical subtypes, otherwise CK low / HR low SCP11 and SCP12 were dominated by metacluster 22 (Figs. 11a, 7, and 12). SCP11 and SCP12 were dominated by metaclusters 20 and 19 (CK7). + SCP11 was characterized by clinically assigned HR - / HER2 + Overlapping with tumor type, this clinical subtype is usually associated with poor outcomes (Coates, AS, et al. Ann, Oncol. 26, 1533-1546 (2015)), but patients with SCP11 had significantly better outcomes than other patients in this cohort. + / HER2 - A small number of CK7s are assigned to + SCP12 patients did not survive long-term (Figures 11a, 12, 13a, 13c, 13e, and 14). Tumors from high-risk TNBC patients contained distinct cell types, including cells with cytokeratin expression, indicative of luminal cell rather than myoepithelial cell origin (Figures 4, 5, 11a, and 12). Absence of luminal epithelial markers, hypoxia, p53 + / EGFR + The TNBC phenotype, with high levels of basal, or proliferation markers, distinguished SCPs 13, 14, 15, and 17, which were associated with poor outcomes (Figures 4, 5, 11a, and 13). SCP 16 tumors were associated with p53 + and expressed apoptotic markers, and interestingly, patients with tumors in this group did not die from the disease, despite being clinically classified as TNBC (Fig. 13 and Fig. 14f).

[0169] By mapping the spatial cellular organization of these tumors, we observed varying architecture and cell density, as well as the relationship between cellular phenotype and tissue organization (Figure 11). Heterogeneous tumors consisted of multiple phenotypically pure communities, represented by many bands on the heatmap, whereas homogeneous tumors, organized into a single epithelial layer or similar communities of different sizes, had only a few clustered bands (Figure 11b). While most tumors were dominated by a single tumor cell metacluster and a few community types, tumors in SCP8 and some tumors in SCP10 were unusually heterogeneous, consisting of multiple epithelial cell metaclusters of similar proportions localized in spatially distinct communities (Figure 11b). Patients with such heterogeneous tumors in SCP8 had very poor outcomes. Overall, intratumor phenotypic heterogeneity was spatially segregated into distinct tumor communities, in contrast to the heterogeneous tumor mass, and patients with tumors with greater spatial phenotypic heterogeneity had poorer outcomes.

[0170] Compared with clinically defined subtypes, the SCP grouping improved the ability to predict patient overall survival using Cox proportional hazards modeling (Figure 18). To identify characteristics associated with patient risk not captured by clinical grading and classification, we investigated the contribution of epithelial and stromal single cells and communities to this model. While single-cell phenotypes or cell metaclusters were rarely independently associated with outcome (not shown), spatially defined cell communities were (Figure 15). For certain cell types, larger tumor cell communities were associated with better outcomes, while smaller similar network sizes were associated with poorer outcomes (Figures 9 and 15; TC 12 vs. 13, 17 vs. 23, 5 vs. 15). Furthermore, the microenvironment community MC6, characterized by angiogenesis with T cell involvement, had a lower HR compared to other subgroups. +Despite being more prevalent in certain clinical subgroups, MC19 with high T cell infiltration and MC27 with macrophage enrichment were significantly associated with a higher risk of death (Figures 10 and 15). On the other hand, high T cell infiltration (MC19) and macrophage enrichment (MC27) were significantly associated with favorable patient outcomes, despite inflammation being more prevalent in high-risk TNBC tumors than in other clinical subgroups (Figures 10 and 15). The tumor type and tumor and stromal architecture defined by the SCP provided prognostic information beyond current clinical classifications. Tumor morphology, drug sensitivity, drug resistance, and assignment or exclusion of tumors as significantly high or low risk are all potentially useful pieces of information that can be derived from the SCP in the clinical context. This information is summarized in Figure 19.

[0171] In summary, the present invention provides a systematic, multidimensional histology of breast cancer, enabling the generation of detailed spatial maps of disease-related single-cell phenotypes and cellular communities. This single-cell pathology allows for better differentiation of patients with different clinical outcomes than clinical subtyping strategies currently available in the art. The examples herein demonstrate how analysis of the multicellular architecture revealed that tumor phenotypic heterogeneity is spatially localized to distinct regions or foci. Furthermore, this multicellular architecture provided information relevant to patient outcomes that was superior to single-cell data alone. Co-occurring breast cancer phenotypes were identified, and phenotypic and spatial heterogeneity differed among clinically established subtypes. Thus, multicellular spatial information is medically relevant and provides a basis for understanding how spatial phenotypic tissue features influence patient disease progression.

Claims

1. 1. A method for characterizing a tumor in a breast cancer patient, comprising: The following steps: a. providing a cancer tissue sample obtained from a patient; b. labeling the cancer tissue sample with a plurality of molecular probes, each probe specific for a biomolecule, each of the molecular probes characterized by a detectable marker, and the biomolecule is: i. epithelial cadherin (E-cadherin), ii. cytokeratin (CK) 18 and / or 19; iii. CK7, iv. estrogen receptor (ER) and / or progesterone receptor (PR), v. a marker of cell proliferation selected from Ki-67 and PCNA; vi. CK5 and / or p63 and / or CK14, vii. p. 53, viii. a hormone receptor (HR) selected from an estrogen receptor (ER) and a progesterone receptor (PR); ix. a marker of apoptosis selected from cleaved form of poly ADP-ribose polymerase (cPARP) and cleaved form of caspase 3 (cC3); x. epidermal growth factor receptor (EGFR), xi. carbonic anhydrase (CAIX); and xii. DNA intercalating dyes; the step selected from the list comprising or consisting of: c. In the reading step, obtaining information regarding the expression of each of the plurality of biomolecules at single-cell resolution; d. In a cell assignment step, assigning a cellular identity (CI) to each single cell in the labeled tissue sample based on the expression of the plurality of biomolecules, wherein the cellular identity is selected from the following list: CI1. CIAX ｈｉ , EGFR-, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)+; CI2.p53 ｈｉ , (cC3 and / or cPARP)+, (ER and / or PR)-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-; CI3. (Ki-67 and / or PCNA)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI4.p53 ｈｉ , EGFR+, CIAX ｈｉ , (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI5. (CK5 and / or p63 and / or CK14)+, CK7-, (CK18 and / or CK19)-, (ER and / or PR)-; CI6. E-cadherin ｈｉ , (CK18 and / or CK19) ｈｉ , CK7+, (ER and / or PR)+, (CK5 and / or p63 and / or CK14)-; CI7. CK7+, (CK18 and / or CK19)+, (ER and / or PR)-, (CK5 and / or p63 and / or CK14)-; CI8. E-cadherin-, CK7-, (CK18 and / or CK19)-, (CK5 and / or p63 and / or CK14)-, (ER and / or PR)-; CI9. (E-cadherin ｌｏ or E-cadherin-), ((CK18 and / or CK19)- or (CK18 and / or CK19) ｌｏ ), (ER and / or PR) ｌｏ , (CK5 and / or p63 and / or CK14)-; CI10. ((CK18 and / or CK19) ｈｉ or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR) ｈｉ , CK5 and / or p63 and / or CK14)-; CI11. ((CK18 and / or CK19) ｈｉ or (CK18 and / or CK19)+), E-cadherin+, (ER and / or PR)+, CK5 and / or p63 and / or CK14)-; CI12. (E-cadherin ｌｏ or E-cadherin+), ((ER and / or PR) ｌｏ or (ER and / or PR)-), (CK5 and / or p63 and / or CK14)-; CI13. p53 ｈｉ , EGFR+, (ER and / or PR) ｈｉ , ((CK5 and / or p63 and / or CK14) ｌｏ and / or (CK5 and / or p63 and / or CK14)-); CI14. CK7+, (CK18 and / or CK19)+, (CK5 and / or p63 and / or CK14)+, (ER and / or PR)- the step of assigning the cells according to the expression of a biomolecule identified by a marker according to e. In the pathology group assignment step, the cancer tissue samples are assigned to single cell pathology (SCP) patient groups according to the proportion of each cell identity contained in the samples assigned in the cell assignment step, and the list of SCP patient groups is: SCP1. (>)70% of single cells are CI10; SCP 2. (>)70% of single cells are CI11; SCP 3. ≤70% of single cells are CI10; SCP 4. (>)70% single cells for CI12; SCP 5. ≤70% of single cells are CI12; SCP 6. (>)80% of single cells are CI9; SCP 7. (>)80% of single cells are CI8; SCP 8. (≦) 70% or less of the single cells are CI9; or CI10; or CI12; SCP 9. (>)60% of single cells are CI9; SCP 10. (>) 70% of single cells are CI9; or CI10; or CI12; SCP 11. (>)60% of single cells are CI7; SCP 12. (>)70% of single cells are CI6; SCP 13. (>)50% of single cells are CI5; SCP 14. (>)60% of single cells are CI3; SCP 15. (>)70% of single cells are CI4; SCP 16. (>)50% of single cells are CI2; SCP 17. >50% of single cells are CI1; SCP 18. (>)90% of single cells are CI 14; the process comprising or consisting of Including, In the SCP patient group assignment step, patients are assigned to a predicted outcome group: - SCP1, high probability of a good outcome and: - selective estrogen receptor modulator (SERM) antineoplastic agents; - selective estrogen receptor degrader (SERD) anti-cancer drugs; aromatase inhibitor antineoplastic agents; and / or ・PI3K pathway inhibitors likely to be susceptible to; - SCP2, as follows: anti-angiogenic antineoplastic agents; and / or ・HER2-targeted anti-cancer drugs likely to be susceptible to; - SCP3, high probability of poor outcome, with: - anthracycline anti-cancer drugs; -mitotic inhibitors; Anti-cancer platinum complexes; Alkylating antineoplastic agents; - antimetabolite antineoplastic agents; Selective SERM antineoplastic drugs; SERD antineoplastic drugs; aromatase inhibitors; and / or ・PI3K pathway inhibitors likely to be susceptible to; - High possibility of lack of sensitivity to anti-cancer drugs that target SCP4 and ER; - SCP5, as follows: - Anti-cancer drug that is an EZH2 methyltransferase inhibitor likely to be susceptible to; - SCP 6, high likelihood of poor outcome, with: ・HER2-targeted anti-cancer drugs likely to be susceptible to; - SCP7, as follows: Anti-angiogenic anti-cancer drugs likely to be insensitive to and the following: ・HER2-targeted anti-cancer drugs likely to be susceptible to; - SCP8, high probability of adverse outcome; - SCP9, as follows: ・HER2-targeted anti-cancer drugs likely to be susceptible to; - SCP 10, including: ・HER2-targeted anti-cancer drugs likely to be susceptible to; - SCP11, high probability of a good outcome and: HER2-targeted anti-cancer drugs, and / or ・PI3K pathway inhibitors likely to be susceptible to; - SCP12, high likelihood of poor outcome, with: - Anti-cancer drug that is an EZH2 methyltransferase inhibitor likely to be susceptible to; - SCP 13, as follows: ・HER2-targeted anti-cancer drugs likely to be susceptible to; - SCP14, high likelihood of adverse outcome and: - anthracycline anti-cancer drugs; -mitotic inhibitors; Anti-cancer platinum complexes; alkylating antineoplastic agents; and / or -Antimetabolic anticancer drugs likely to be susceptible to; - SCP 15, as follows: - anthracycline anti-cancer drugs; -mitotic inhibitors; Anti-cancer platinum complexes; Alkylating antineoplastic agents; antimetabolite antineoplastic agents; and / or - Anti-cancer drug that is an EGFR bioactivity inhibitor likely to be susceptible to; - SCP16, high probability of good outcome, with: - anthracycline anti-cancer drugs; -mitotic inhibitors; antineoplastic platinum complexes; or -Antimetabolic anticancer drugs likely to be susceptible to; - SCP17, high likelihood of adverse outcome and: ・Quinone alkylating anticancer drugs likely to be susceptible to; - High possibility of lack of sensitivity to anti-cancer drugs targeting SCP18 and ER, including the following: ・PI3K pathway inhibitors likely to be susceptible to; - cellular community identity (CCI) 1, CCI3, CCI5, CCI8, CCI9, or CCI10: high probability of good outcome; - CCI2, CCI4, CCI6, or CCI7: High probability of poor outcome The method according to claim 1,

2. The method of claim 1 , wherein obtaining information about the average expression of a plurality of biomolecules comprises constructing an image of the cancer tissue sample.

3. 3. The method for characterizing tumors in breast cancer patients according to claim 1 or 2, wherein the method for obtaining information about the average expression of the plurality of biomolecules is imaging mass cytometry at subcellular resolution, in particular at a resolution of (≦) 5 μm or less, or even at a resolution of (≦) 1 μm or less.

4. The method comprises the following steps: In the labeling step: xiii. CD3 or CD90; xiv. CD20 or CD19; xiv. CD68; xv. CD44 and / or CD45; xvi. fibronectin; xvii. vimentin, and xviii. CD31 and / or von Willebrand factor (vWF) and / or CD34; the step of comprising an additional marker selected from: In the cell allocation step: CI15. CD44+, CD45+, (CD3 or CD90)+, fibronectin-, E-cadherin-, (CK5 and / or p63 and / or CK14) ｌｏ or (CK5 and / or p63 and / or CK14)-); CI16. CD20+, (fibronectin ｌｏ or fibronectin-), ((E-cadherin ｌｏ or E-cadherin-), ((CK5 and / or p63 and / or CK14) ｌｏ or (CK5 and / or p63 and / or CK14)-); CI17. (CD3 or CD90)+, (CD20 or CD19)+; CI18. CD68+; CI19. Vimentin+, (CD34 and / or VWF and / or CD31)+; CI20. Vimentin-, (Fibronectin+ or Fibronectin ｈｉ ), (CD3 or CD90)-, (CD20 or CD19)-, CD45-, CD44-; the steps comprising a further cell identity selected from 4. A method for characterizing a tumor in a breast cancer patient according to any one of claims 1 to 3, comprising:

5. 5. The method for characterizing a tumor in a breast cancer patient according to any one of claims 2 to 4, wherein the single cell is a fragment of an image of the cancer tissue sample, in particular a fragment consisting of pixels within an area where membrane-bound molecules surround a single nucleus.

6. The method comprises the following steps: A cell community detection step includes partitioning an image of a cancer tissue sample into multicellular regions, wherein each single cell in the multicellular region is highly interconnected with neighboring cells to provide a cell community; In the cell community assignment step, a cell community identity (CCI) is assigned to each cell community according to the number of cells in the cell community and the proportion of each cell identity (CI) contained therein, and the list of CCIs is: CCI1. Among cells with identities CI1 to CI14, more than (>) 10% of the single cells are CI6 and the average size of the cell community is more than (>) 25 cells; or CCI2. Among cells with identities CI1-CI14, more than (>) 10% of the single cells are CI6 and the average size of the cell community is (<) 25 cells or less; or CCI3. Of cells with identities CI1 to CI14, more than (>) 10% of the single cells are CI2 and the average size of the cell community is more than (>) 25 cells; or CCI4. Of cells with identities CI1 to CI14, more than (>) 10% of the single cells are CI2 and the average size of the cell community is (<) 25 cells or less; or CCI5. Of cells with identities CI1-CI14, >10% of the cells are CI3 and the average size of the cell community is <=25 cells; or CCI 6. More than (>) 5% of the total cells are CI 19, and more than (>) 10% of the cells are CI 20, and more than (>) 3% of the cells are CI 18, and the average size of the cell community is less than (<) 50 cells; or CCI 7. (>) 80% or more of all cells are of any of the CI1-CI15 identities and (<) 10% or less are of CI20, and the average size of the cell community is (<) less than 75 cells; or CCI 8. (>) 20% of all cells are CI15, and / or CI16, and / or CI17, and (<) 40% of cells are any of CI1-CI14 identities, and the average size of the cell community is greater than 75 cells; or CCI 9. More than (>) 5% of all cells have a CI of 18 and the average size of the cell community is more than (>) 25%; or CCI 10. (>) 80% or more of the total cells are of any of the CI1-CI14 identities, and (<) 2% or less of the cells are of CI20, and the average size of the cell community is (>) 115 cells and (<) 125 cells; the process comprising or consisting of 6. A method for characterizing a tumor in a breast cancer patient according to any one of claims 2 to 5, comprising:

7. In the SCP patient group assignment step, patients are assigned to a predicted outcome group according to the SCP classification of the sample in the pathology group assignment step and / or according to the CCI classification of the sample in the cell community assignment step: - SCP1, high probability of a good outcome and: - a SERM drug selected from raloxifene, toremifene and tamoxifen; - a SERD drug selected from fulvestrant, brilandestrant and elacestrant; an aromatase inhibitor antineoplastic agent selected from exemestane, letrozole, vorozole, formestane, fadrozole and anastrozole; and / or a PI3K pathway inhibitor selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235SCP2 likely to be susceptible to; - SCP2, as follows: an anti-angiogenic antineoplastic agent selected from bevacizumab, thalidomide, and lenalidomide; and / or a HER2-targeted anti-cancer agent selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, H2NVAC, HER2.taNK cells, HER2-pulsed dendritic cells, and HER2-targeted T cells. likely to be susceptible to; - SCP3, high probability of poor outcome, with: an anthracycline anti-cancer drug selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; - a mitotic inhibitor-type antineoplastic agent selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; - alkylating antineoplastic agents selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbayine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, tiptepa and trabectedin; antimetabolite antineoplastic agents selected from azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate, and phototrexate; - a SERM drug selected from raloxifene, toremifene and tamoxifen; - a SERD antineoplastic agent selected from fulvestrant, brilandestrant and elacestrant; aromatase inhibitor antineoplastic agents selected from exemestane, letrozole, vorozole, formestane, fadrozole and anastrozole; and / or a PI3K pathway inhibitor selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235 likely to be susceptible to; - High possibility of lack of sensitivity to anti-cancer drugs targeting SCP4 and ER, including the following: SERM anti-cancer drugs; SERD antineoplastic drugs; and / or Aromatase inhibitor anti-cancer drugs likely to be insensitive to - SCP5, as follows: an anti-cancer drug that is an EZH2 methyltransferase inhibitor selected from 3-deazaneplanocin A (DZNep), tazemetostat, EPZ005687, EI1, GSK126, or UNC199; likely to be susceptible to; - SCP 6, high likelihood of poor outcome, with: a HER2-targeting anti-cancer drug selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, and H2NVAC. likely to be susceptible to; - SCP7, as follows: - an anti-angiogenic anti-cancer drug selected from bevacizumab, thalidomide, and lenalidomide likely to be insensitive to and the following: a HER2-targeting anti-cancer drug selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, and H2NVAC. likely to be susceptible to; - SCP8, high probability of adverse outcome; - SCP9, as follows: a HER2-targeting anti-cancer drug selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, and H2NVAC. likely to be susceptible to; - SCP 10, including: a HER2-targeting anti-cancer drug selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, and H2NVAC. likely to be susceptible to; - SCP11, high probability of a good outcome and: - a HER2-targeted antineoplastic agent selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, and H2NVAC; and / or a PI3K pathway inhibitor selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235 likely to be susceptible to; - SCP12, high likelihood of poor outcome, with: an anti-cancer drug that is an EZH2 methyltransferase inhibitor selected from 3-deazaneplanocin A (DZNep), tazemetostat, EPZ005687, EI1, GSK126, and UNC199; likely to be susceptible to; - SCP 13, as follows: a HER2-targeting anti-cancer drug selected from trastuzumab, pertuzumab, SYD985, RC48, A166, HER2ALT-P7, T-DM1, ARX788, KN026, BVAC-B, MT-5111, AVX901, TAS0728, MP0274, MM-302, FS102, and H2NVAC. likely to be susceptible to; - SCP14, high likelihood of adverse outcome and: an anthracycline anti-cancer drug selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; - a mitotic inhibitor-type antineoplastic agent selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; alkylating antineoplastic agents selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbayine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, tiptepa, and trabectedin; and / or antimetabolite antineoplastic agent selected from azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate, and phototrexate likely to be susceptible to; - SCP 15, as follows: an anthracycline anti-cancer drug selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; - a mitotic inhibitor-type antineoplastic agent selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; - alkylating antineoplastic agents selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbayine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, tiptepa and trabectedin; antimetabolite antineoplastic agents selected from azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate, and phototrexate; and / or - an anti-cancer drug which is an EGFR bioactive inhibitor selected from gefetinib, erlotinib, lapatinib, cetuximib, neratinib, osimeratib, panitumamib, vandetanib, necitumumab, and dacomitinib likely to be susceptible to; - SCP16, high probability of good outcome, with: an anthracycline anti-cancer drug selected from daunorubicin, doxorubicin, epirubicin, and idarubicin; - a mitotic inhibitor-type antineoplastic agent selected from cabazitaxel, docetaxel, nab-paclitaxel, paclitaxel, vinblastine, vincristine, and vinorelbine; antineoplastic platinum complexes selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, triplatin, and tetranitrates; an alkylating antineoplastic agent selected from altretamine, bendamustine, busulfan, carboplatin, carmustine, cisplatin, cyclophosphamide, chlorambucil, dacarbayine, ifosfamide, lomustine, mechlorethamine, melphalan, oxaliplatin, temozolomide, tiptepa, or trabectedin; or antimetabolite antineoplastic drug selected from azacitidine, 5-fluorouracil, 6-mercaptopurine, capecitabine, clofarabine, cytarabine, decitabine, floxuridine, fludarabine, gemcitabine, hydroxyurea, hydroxycarbamide, methotrexate, nelarabine, pemetrexed, pentostatin, pralatrexate, or phototrexate likely to be susceptible to; - SCP17, high likelihood of adverse outcome and: - Mitomycin C, a quinone alkylating anticancer drug likely to be susceptible to; - High possibility of lack of sensitivity to anti-cancer drugs targeting SCP18 and ER, including the following: SERM anti-cancer drugs; SERD antineoplastic drugs; and / or Aromatase inhibitor antineoplastic drugs; likely to be insensitive to and: a PI3K pathway inhibitor selected from rapamycin, dactolisib, BGT226, SF1126, PKI-587, and NVPBE235 likely to be susceptible to; CCI1, CCI3, CCI5, CCI8, CCI9, or CCI10: high probability of a good outcome; - CCI2, CCI4, CCI6, or CCI7: High probability of poor outcome 7. A method for characterizing a tumor in a breast cancer patient according to any one of claims 1 to 6, wherein the tumor is assigned to

Images