WO2025083140A1 - Biomarkers for predicting immunotherapy response - Google Patents

Abstract

The present invention relates to methods for predicting whether a human subject having cancer will be responsive to treatment with an immune checkpoint inhibitor. The method comprises measuring the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject, and making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor therapy based on the proliferative fraction profile of the subject. Also provided are methods of treating cancer comprising administering an immune checkpoint inhibitor to a subject that has been determined to be responsive to the immune checkpoint inhibitor based on a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject.

Description

BIOMARKERS FOR PREDICTING IMMUNOTHERAPY RESPONSE

Field of the Invention

The present invention relates to methods for predicting whether a human subject having cancer will be responsive to treatment with an immune checkpoint inhibitor. The present invention also relates to methods of treating cancer.

Background

Triple-negative breast cancer (TNBC), which lacks oestrogen and progesterone hormone receptor and human epidermal growth factor 2 (HER2) expression, is an aggressive subtype of breast cancer which develops in about 15 to 20% of subjects with breast cancer.

Treatment for TNBC typically involves chemotherapy, along with surgery and radiotherapy. However, TNBC is often more aggressive, harder to treat and more likely to recur than cancers which are hormone receptor-positive or HER2 positive. There remains a need for improved therapeutic options for TNBC.

Immunotherapy such as immune checkpoint blockade (ICB) represents an exciting therapeutic option for cancers. In cancer, chronic T cell stimulation leads to dysfunction, owing to interactions between cells expressing immune checkpoint receptors and ligands. ICB prevents this, which can invigorate dysfunctional T-cells. These invigorated T-cells can then interact with target cancer cells to induce cell death. Interactions between proximate specialized cells in distinct activation states underpin the effect of ICB. The efficacy of ICB therefore depends on both the cellular composition and multicellular organization of tumours.

Immunotherapy has transformed the treatment of solid tumours but its best use in breast cancer remains unclear. In TNBC, trials of immune checkpoint blockade (ICB) targeting the interaction between programmed death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1) have shown that some, but not all, subjects benefit (Schmid et al., 2020, Schmid et al., 2022, Rizzo et al., 2022, Loibl et al., 2022, Loibl et al, 2019 and Mittendorf et al., 2020). It remains unclear as which subset of subjects are likely to benefit.

Various clinical trials for cancer therapeutics have utilised the surrogate end point pathological complete response (pCR) for assessment. pCR can be defined as the absence of residual invasive cancer on hematoxylin and eosin evaluation of the complete resected cancer specimen and all sampled regional lymph nodes following completion of neoadjuvant systemic therapy. However, in recent years there has been ongoing debate over the clinical relevance of pCR. Several retrospective analyses including the one used by FDA for accepting pCR as a surrogate (Cortazar et al., 2014), failed to show that an improved rate of pCR translated into an improved event free survival at a trial level. Therefore, there is little correlation between an improved rate of pCR and ultimate survival and response of the subject. It thus remains challenging to identify biomarkers which can identify true responders to immunotherapy in the long term. This is also the case for immunotherapy in TNBC.

There remains a need for a reliable biomarker to identify responders to immunotherapy.

The present invention has been devised in light of the above considerations.

Summary of the Invention

The present inventors have found that the proliferative fraction of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCir cells in a sample from a subject having cancer is an accurate and reliable biomarker to identify long-term responders to immunotherapy. These findings are especially surprising, given that the inventors did not find any correlation between cell density of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells and the subjects response to immunotherapy. The proliferative fraction of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells therefore represents an unexpected and accurate biomarker for responsiveness to immunotherapy.

Accordingly, in a first aspect, the present invention provides a method for predicting whether a human subject having cancer will be responsive to treatment with an immune checkpoint inhibitor, the method comprising: a) measuring the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject, and b) making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor therapy based on the proliferative fraction profile of the subject.

Advantageously, this method ensures that only subjects likely to be responsive are selected for treatment with an immune checkpoint inhibitor therapy. This ensures that immunotherapy is targeted only to those likely to obtain benefit from immunotherapy, which is especially important given that some subjects can exhibit tolerability issues or chronic side effects upon administration. By restricting immunotherapy only to subjects likely to benefit, this also reduces costs of administration.

In some embodiments, the MHCI⁺MHCII⁺ cells are cancer cells, optionally epithelial cells comprising a detectable expression level of one or more of CK5/14, CK/8/18, PanCK and AR.

In some embodiments, step a) comprises measuring the proportion of CD8⁺TCF1⁺ T-cells or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cells or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject. Preferably, step a) comprises measuring the proportion of CD8⁺TCF1⁺ T-cells which comprise a detectable expression level of the proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cell proliferative fraction profile of the subject. In some embodiments, the cancer comprises breast cancer, ovarian cancer, pancreatic cancer, colorectal cancer, lung cancer, pancreatic cancer, colon cancer, gastric cancer, bladder cancer, skin cancer, myeloma, non-Hodgkin lymphoma, prostate cancer, oesophageal cancer, head and neck cancer, endometrial cancer, hepatobiliary cancer, duodenal carcinoma, thyroid carcinoma, or renal cell carcinoma.

In some embodiments, the cancer comprises breast cancer. The cancer may comprise triple-negative breast cancer.

In some embodiments, the immune checkpoint inhibitor therapy comprises atezolizumab.

In some embodiments, the proliferation marker comprises Ki67.

In some embodiments, step b) comprises comparing the proliferative fraction profile of the subject to a plurality of reference proliferative fraction profiles and making a prediction of whether the subject will be responsive or non-responsive to treatment based on the comparison. The plurality of reference proliferative fraction profiles may be separated into at least two quantiles and the proliferative fraction profile of the subject is compared to the quantiles, wherein the subject is predicted to be responsive to treatment when the proliferative fraction profile of the subject fits into the top quantile. In some embodiments, the at least two quantiles comprise at least three quantiles.

Step a) may comprise measuring the proportion of CD8⁺TCF1⁺ T-cells and MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferation marker to obtain a CD8⁺TCF1⁺ T-cells and MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject.

In some embodiments, responsiveness to treatment comprises event-free survival for a time period of at least 18 months, optionally for a time period of at least 24 months.

In some embodiments, the sample comprises tumour tissue.

In some embodiments, the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker is measured using immunohistochemistry (IHC) or imaging mass cytometry (IMC).

The sample may have been obtained from the subject prior to any treatment.

In some embodiments, the subject is predicted to be responsive to the immune checkpoint inhibitor therapy, and wherein the method further comprises the step of administering a therapeutically effective amount of an immune checkpoint inhibitor. The immune checkpoint inhibitor may be administered as part of a combination therapy with chemotherapy, optionally wherein the chemotherapy comprises carboplatin and/or nab-paclitaxel.

In some embodiments, the method comprises concurrent, sequential or separate administration of: a) atezolizumab; and b) carboplatin and/or nab-paclitaxel. In some embodiments, the subject is predicted to be non-responsive to immune checkpoint inhibitor therapy, and wherein the method further comprises administering chemotherapy to the subject in the absence of any immune checkpoint inhibitor therapy.

According to a further aspect, the present invention provides a computer-implemented method for predicting whether a human subject having cancer will be responsive to treatment with an immune checkpoint inhibitor, the method comprising: a) Obtaining proliferative fraction profile data representing the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject, and b) making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor therapy based on the proliferative fraction profile data of the subject.

The present invention also provides an immune checkpoint inhibitor for use in a method of treating cancer, wherein the method comprises: a) measuring the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject; b) making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor based on the proliferative fraction profile of the subject; and c) treating the subject with the immune checkpoint inhibitor if the subject is predicted to be responsive to therapy in b).

According to a further aspect there is provided a method of treating cancer comprising administering an immune checkpoint inhibitor to a subject that has been determined to be responsive to the immune checkpoint inhibitor based on a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject.

Further aspects and embodiments will be apparent to those skilled in the art. All documents mentioned in this text are incorporated herein by reference.

The invention includes the combination of the aspects and preferred features described except where such a combination is clearly impermissible or expressly avoided.

Summary of the Figures

Embodiments and experiments illustrating the principles of the invention will now be discussed with reference to the accompanying figures in which: Figure 1 . Flowchart of longitudinal tumour sampling from the NEOTRIP randomised clinical trial for high parameter imaging.

Figure 2. Semi-supervised workflow for distinguishing epithelial and TME cells from multiplexed images.

Figure 3. Spatial predictors of immunotherapy response at baseline, a, Schematic illustrating cell phenotype density calculation, b, Odds ratios for associations between cell density and pCR for TME cell phenotypes, c, Boxplot of PD-L1 + IDO+ APC density across treatment arms and response.

Boxes show 25th, 50th and 75th centiles; whiskers indicate 75th centile plus 1 .5 x inter-quartile range and 25th centile less 1 .5 x inter-quartile range; points beyond whiskers are outliers. *P < 0.05, ***P < 0.001 , based on two-sided Wilcoxon tests, d, Representative image of a tumour from a responder treated with immunotherapy with high PD-L1 + IDO+ APC density.

Figure 4. Schematic illustrating principles of homotypic and heterotypic cell-cell interaction metrics.

Figure 5. a, Odds ratios for associations between heterotypic epithelial-to-TME cell phenotypes and pCR. Odds ratios are derived from univariate logistic regression: circles represent point estimates and whiskers indicate 95% confidence intervals. Depicted P values are derived from a term for interaction between the predictor and treatment in logistic regression models (including separate terms for the predictor and treatment). Asterisks indicate associations with an FDR < 0.1 by the Benjamini-Hochberg method, b, Bar charts of the proportion of tumours achieving pCR in patients with no selected epithelial— TME interactions per arm (0) or increasing tertiles of epithelial-TME interactions per arm (T1-T3). Numbers on bars are absolute numbers of patients in each category.

Figure 6. Odds ratios for associations between TME to epithelial cells (a) and TME to TME cells (b).

Figure 7. a, Boxplots of mean expression levels (per tumour) of activation markers for T cells in contact or not in contact with cancer cells at baseline, b, Proportion of T cells positive for Ki67 in contact or not in contact with tumour cells at baseline.

Figure 8. Proliferative fractions of cancer and TME cell phenotypes predict response to immunotherapy, a, Schematic illustrating calculation of cell phenotype-specific proliferative fractions (proportion of Ki67+ cells). b,c, Odds ratios for associations between proliferative fraction and pCR for epithelial (b) and TME (c) cell phenotypes. Odds ratios are derived from univariate logistic regression: circles represent point estimates and whiskers indicate 95% confidence intervals. Depicted P values are derived from a term for interaction between the predictor and treatment in logistic regression models (including separate terms for the predictor and treatment) and have not been adjusted for multiple tests. Asterisks indicate associations with an FDR < 0.1 by the Benjamini-Hochberg method, d, Bar charts of the proportion of tumours achieving pCR in patients with no Ki67+ cells of the selected phenotype per arm (0) or increasing proportion of Ki67+ cells per arm, as quantified by tertiles (T1-T3). Absolute numbers of patients in each category are depicted inside bar charts, e, Relationship between proliferative fraction of CD8+ TCF1+ T cells and MHCl&llhi cells. Spearman rank correlation coefficient. Shaded area represents 95% confidence interval of the loess regression line.

Figure 9. Cell phenotypes predictive of immunotherapy response early on-treatment, a, Odds ratios for associations between cell density and pCR for epithelial cell phenotypes, b, Boxplot of CD8+ GZMB+ T epithelial cell density across treatment arms and response, c, Odds ratios for associations between cell density and pCR for epithelial cell phenotypes. For a and c, odds ratios are derived from univariate logistic regression: circles represent point estimates and whiskers indicate 95% confidence intervals. Depicted P values are derived from a term for interaction between the predictor and treatment in logistic regression models (including separate terms for the predictor and treatment) and have not been adjusted for multiple tests. Asterisks indicate associations with an FDR < 0.1 by the Benjamini-Hochberg method, d, Boxplot of CD15+ cell density across treatment arms and response. For b and d, boxes show 25th, 50th and 75th centiles; whiskers indicate 75th centile plus 1.5 x inter-quartile range and 25th centile less 1 .5 x inter-quartile range; points beyond whiskers are outliers. ***P < 0.001 , based on two-sided Wilcoxon tests.

Figure 10. Multivariate modelling to predict ICB response, a, Analytical workflow for predictive modelling using multitiered multiplexed imaging data. Three models were trained using (1) baseline variables alone, (2) on-treatment variables alone or (3) combining baseline and on-treatment variables, b, AUC statistics for prediction probabilities derived from multivariate regularized logistic regression models to predict pCR. AUCs were computed using random held-out test data repeated 100 times, as described in a; circles are mean AUCs, error bars are 95% confidence intervals, c, Diagram illustrating variable importance analysis including all predictors described in a. d,e, Boxplots depicting baseline (d) and on-treatment (e) predictors of immunotherapy response (associated with a Pbinomial < 0.01) ranked by importance in the model. On the right are heatmaps showing scaled mean values by response. Pbinomial indicates P values derived from variable importance analysis illustrated in c. For all boxplots, boxes show 25th, 50th and 75th centiles; whiskers indicate 75^th centile plus 1 .5 x inter-quartile range and 25th centile less 1 .5 x interquartile range; points beyond whiskers are outliers. Dn, density; Het, heterotypic interactions; Hom, homotypic interactions; RF, random forest. Figure 11. K-M curves of EFS in two arms of therapy (with or without atezolizumab). a. K-M curve for

CD8+TCF1+Ki67+ low and b. K-M curve for CD8+TCF1+Ki67+ high

Detailed Description of the Invention

For the avoidance of any doubt, any theoretical explanations provided herein are provided for the purposes of improving the understanding of a reader. The inventors do not wish to be bound by any of these theoretical explanations.

Any section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.

Throughout this specification, including the claims which follow, unless the context requires otherwise, the word “comprise” and “include”, and variations such as “comprises”, “comprising”, and “including” will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

It must be noted that, as used in the specification and the appended claims, the singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from “about” one particular value, and/or to “about” another particular value. When such a range is expressed, another embodiment includes from the one particular value and/or to the other particular value. Similarly, when values are expressed as approximations, by the use of the antecedent “about,” it will be understood that the particular value forms another embodiment. The term “about” in relation to a numerical value is optional and means for example +/- 10%.

In the context of the present invention, the term “CD8⁺TCF1⁺ cell” or “CD8⁺TCF1⁺ T-cell” will be understood to refer to a T cell which comprises a detectable expression level of CD8 and TCF1. The cell will therefore be understood to be dual positive for CD8 and TCF1 .

Likewise, in the context of the present invention, the term “MHCI⁺MHCir cell” will be understood to refer to a cell which comprises a detectable expression level of MHCI and MHCII. The cell will therefore be understood to be dual positive for MHCI and MHCII.

The MHCTMHCir cell may be a cancer cell. Preferably, the MHCI⁺MHCH⁺ cell is an epithelial cell. Various markers of epithelial identity are known to the skilled person. These include, but are not necessarily limited to CK5/14, CK8/18, PanCK and AR. Thus, in some embodiments, the MHCI⁺MHCH⁺ cell comprises a detectable expression level of one or more of CK5/14, CK8/18, PanCK and AR. In some embodiments, the MHCI⁺MHCH⁺ cell comprises a detectable expression level of CK5/14. In some embodiments, the MHCI⁺MHCH⁺ cell comprises a detectable expression level of CK8/18. In some embodiments, the MHCI⁺MHCH⁺ cell comprises a detectable expression level of PanCK. In some embodiments, the MHCI⁺MHCH⁺ cell comprises a detectable expression level of AR.

In the context of the present invention, a detectable expression level will be understood to refer to a detectable amount of mRNA and/or protein of the marker (for example, the proliferative marker, CD8, TCF1 , MHCI or MHCII, depending on the context). In some embodiments, the detectable expression level comprises a detectable amount of marker mRNA. In some embodiments, the detectable expression level comprises a detectable amount of marker protein. In other embodiments, the detectable expression level comprises a detectable amount of marker protein and mRNA. Methods for measuring the presence of mRNA and/or protein are known in the art and discussed in more detail below. The mRNA and/or protein may be present in any amount. Amounts of the mRNA and/or protein are discussed in more detail below. In the context of the present invention, the terms “amount” and “level” are interchangeable.

In some embodiments, a detectable expression level comprises a detectable expression level above a threshold value. A threshold value may be identified as a quantile threshold.

Various methods are known in the art to detect proteins including, for example, western blots, flow cytometry and ELISAs. Normalisation of protein expression may be to a housekeeping protein product. The skilled person will be aware of suitable housekeeping genes and products. The expression and/or detectable amount of protein may comprise a normalised value.

The expression and/or detectable amount of protein may be detected using flow cytometry. Normalisation of protein expression may be to an isotype control antibody. The expression and/or detectable amount of protein detected using flow cytometry may be quantified by population shift and/or mean fluorescence intensity.

The expression and/or detectable amount of mRNA can be detected using RNA-seq methods. Various RNA-seq methods are commercially available and known to those skilled in the art. RNA-seq methods function by mapping the number of RNA reads aligned to each gene under each biological condition, to obtain a read count. The reads can then be normalised to provide a normalised read count.

In some embodiments, the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker is measured using imaging mass cytometry (IMC) or immunohistochemistry (IHC).

In some embodiments, the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker is measured using imaging mass cytometry (IMC).

As the skilled person will appreciate, imaging mass cytometry is an imaging method which utilises cytometry by time-of-flight. In contrast to other imaging methods, imaging mass cytometry utilises metal- tagged antibodies rather than fluorochromes.

In other embodiments, the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker is measured using immunohistochemistry (IHC). As the skilled person will appreciate, immunohistochemistry is a process where proteins in a tissue section are labelled by antibodies specific for the proteins. In the context of the present invention, the antibodies may comprise antibodies specific for the proliferation marker, CD8 and TCF1. In some embodiments, the antibodies comprise antibodies specific for the proliferation marker, MHCI and MHCII. In some embodiments, the antibodies comprise antibodies specific for the proliferation marker, CD8, TCF1 , MHCI and MHCII. Various methods and reagents for performing IHC are known and commercially available to those skilled in the art.

Various proliferation markers are suitable for use in the present invention. The skilled person will be aware of various proliferation markers. Exemplary proliferation markers may include, but not necessarily be limited to Ki67, proliferating cell nuclear antigen (PCNA) and minichromosome maintenance (MCM) proteins. In some embodiments, the proliferation marker comprises Ki67. The inventors have found that the proliferative fraction of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells in a sample from a subject having cancer is predictive of the subject’s response to immunotherapy, specifically to an immune checkpoint inhibitor. This is especially surprising given that the density of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells in the sample is not indicative of the subject’s response to immunotherapy; it is the proportion of these cells which is proliferating, as identified by expression of the proliferative marker.

By “immune checkpoint inhibitor” or “immune checkpoint inhibitor therapy”, this will be understood to refer to therapies which boost anti-cancer response by specifically targeting immunologic receptors on the surface of T-cells. As used herein, “immune checkpoint inhibitor” may be used interchangeably with “immune checkpoint blockade”.

Immune checkpoint inhibitors known in the art include anti-CTLA-4, anti-PD-1 and anti-PD-L1 molecules. Thus, in some embodiments, the immune checkpoint inhibitor therapy comprises an anti-CTLA-4, anti- PD-1 or anti-PD-L1 molecule. In some embodiments, the immune checkpoint inhibitor comprises an anti- PD-1 or anti-PD-L1 molecule. The molecule may comprise an antibody. In some embodiments, the immune checkpoint inhibitor therapy comprises an anti-PD-L1 molecule, for example an anti-PD-L1 antibody. The anti-PD-L1 antibody may comprise an anti-PD-L1 monoclonal antibody.

An exemplary anti-CTLA-4 antibody is ipilimumab.

Various anti-PD-1 antibodies are known to those skilled in the art. These include nivolumab, pembrolizumab and cemiplimab.

Likewise, various anti-PD-L1 antibodies are known to those skilled in the art. These include atezolizumab, avelumab and durvalumab.

In some embodiments, the immune checkpoint inhibitor therapy comprises atezolizumab. Atezolizumab is a known anti-PD-L1 monoclonal antibody.

The cancer may comprise any cancer which has been found to be associated with expression of PD-1 within the tumour microenvironment (TME), for example expression of PD-1 on activated monocytes, dendritic cells (DCs), natural killer (NK) cells, T-cells and/or B-cells.

In some embodiments the cancer comprises breast cancer, ovarian cancer, pancreatic cancer, colorectal cancer, lung cancer, pancreatic cancer, colon cancer, gastric cancer, bladder cancer, skin cancer, myeloma, non-Hodgkin lymphoma, prostate cancer, oesophageal cancer, head and neck cancer, endometrial cancer, hepatobiliary cancer, duodenal carcinoma, thyroid carcinoma, or renal cell carcinoma. In some embodiments, the cancer comprises colon, breast, ovarian, lung, skin head and neck or pancreatic cancer. The cancer may comprise lung cancer, skin cancer, head and neck cancer or breast cancer. The lung cancer may comprise non-small-cell lung cancer (NSCLC). The skin cancer may comprise Merkel cell carcinoma (MCC) or melanoma. The head and neck cancer may comprise head and neck squamous cell carcinoma (HNSCC).

In some embodiments, the cancer comprises breast cancer. The cancer may comprise triple-negative breast cancer. As used herein, “triple-negative breast cancer” will be understood to refer to ER; HER2- primary breast cancer.

The proliferative fraction profile may comprise a percentage of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCH⁺ cells which comprise a detectable expression level of the proliferative marker. Optionally, when the proliferative fraction profile comprises a percentage of at least 0.1 % of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferative marker, the subject may be predicted to be responsive to treatment with an immune checkpoint inhibitor therapy. In some embodiments, when the proliferative fraction profile comprises a percentage of at 1 %, optionally at least 5% of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferative marker, the subject is predicted to be responsive to treatment with an immune checkpoint inhibitor therapy. When the proliferative fraction profile comprises a percentage of at least 10% of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferative marker, the subject may be predicted to be responsive to treatment with an immune checkpoint inhibitor therapy. In some embodiments, when the proliferative fraction profile comprises a percentage of at least 20% of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferative marker, the subject is predicted to be responsive to treatment with an immune checkpoint inhibitor therapy. In some embodiments, when the proliferative fraction profile comprises a percentage of at least 30% of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferative marker, the subject is predicted to be responsive to treatment with an immune checkpoint inhibitor therapy. In some embodiments, when the proliferative fraction profile comprises a percentage of at least 40% of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferative marker, the subject is predicted to be responsive to treatment with an immune checkpoint inhibitor therapy. In some embodiments, when the proliferative fraction profile comprises a percentage of at least 50% of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferative marker, the subject is predicted to be responsive to treatment with an immune checkpoint inhibitor therapy.

In some embodiments, step b) comprises comparing the proliferative fraction profile of the subject to a plurality of reference proliferative fraction profiles and making a prediction of whether the subject will be responsive or non-responsive to treatment based on the comparison. The reference proliferative fraction profiles may comprise the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker obtained or derived from a sample of reference subjects. In some embodiments, the reference subject is known not to have cancer. In other embodiments, the reference subject is a subject known to have and having already been treated for cancer. In such embodiments, the sample from the reference subject may be a baseline sample, i.e. a sample obtained prior to the subject starting treatment.

The plurality of reference proliferative fraction profiles may be separated into at least two quantiles and the proliferative fraction profile of the subject is compared to the quantiles, wherein the subject is predicted to be responsive to treatment when the proliferative fraction profile of the subject fits into the top quantile. In some embodiments, the at least two quantiles comprise at least three quantiles.

Where there are two quantiles, it will be appreciated that there will be an upper and lower quantile, which may otherwise be referred to as a top and bottom quantile, respectively. The upper quantile defines the top 50% of proliferative fraction profiles in a plurality of reference proliferative fraction profiles, and the lower quantile defines the bottom 50% of proliferative fraction profiles in a plurality of reference proliferative fraction profiles.

In some embodiments, the at least two quantiles comprise three quantiles. In such embodiments, it will be appreciated that the three quantiles comprise a lower quantile, a middle quantile and an upper quantile. The lower quantile defines the bottom 33.33% of proliferative fraction profiles in a plurality of reference proliferative fraction profiles, the middle quantile defines the middle 33.33% of proliferative fraction profiles in a plurality of reference fraction profiles and the upper quantiles defines the top 33.33% of proliferative fraction profiles in a plurality of reference fraction profiles.

In embodiments comprising three quantiles, each quantile may otherwise be referred to as a fertile.

Therefore, when the proliferative fraction profile of the subject fits into the top (which may otherwise be referred to as “upper” quantile), this may be indicative of the subject having a proliferative fraction which is higher than at least 50%, optionally at least 63% of reference subjects and is therefore predictive of the subject being responsive to immune checkpoint inhibitor therapy.

The plurality of reference proliferative fraction profiles may comprise at least 10 reference proliferative fraction profiles, at least 20 reference proliferative fraction profiles, at least 30 reference proliferative fraction profiles, at least 40 reference proliferative fraction profiles, at least 50 reference proliferative fraction profiles or at least 100 reference proliferative fraction profiles.

In some embodiments, the plurality of reference proliferative fraction profiles comprises at least 150 or at least 200 reference proliferative fraction profiles.

Step a) may comprise measuring the proportion of CD8⁺TCF1⁺ T-cells and MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferation marker to obtain a CD8⁺TCF1⁺ T-cells and MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject.

In some embodiments, responsiveness to treatment comprises event-free survival for a time period of at least 18 months, optionally for a time period of at least 24 months. In some embodiments, responsiveness to treatment comprises event-free survival for a time period of at least 36 months, optionally at least 48 months and further optionally at least 60 months. In some embodiments, responsiveness to treatment comprises event-free survival for a time period of at least 10 years. In the context of the present invention, event free survival will be understood to refer the time from treatment of the subject to the first date of disease progression, disease recurrence (local, regional, distant, invasive, other invasive cancers) or death due to any cause, including unknown causes.

In other embodiments, responsiveness to treatment comprises overall survival for a time period of at least 18 months, optionally for a time period of at least 24 months. In some embodiments, responsiveness to treatment comprises overall survival for a time period of at least 36 months, optionally at least 48 months and further optionally at least 60 months. In some embodiments, responsiveness to treatment comprises overall survival for a time period of at least 10 years. “Overall survival (“OS”), as used herein, will be understood to refer to the time from diagnosis of the subject or the start of treatment of the subject to the time of death.

The human subject may be female. The human subject may be at least 18 years old, optionally at least 30 years old, further optionally at least 40 years old. In some embodiments, the human subject is at least 50 years old.

The sample may be a cell or tissue sample (e.g. a biopsy) or a biological fluid. In particular, the sample may comprise tumour tissue, for example a breast tumour (primary or secondary). The sample will generally comprise cells. In some embodiments, the sample may comprise blood or plasma.

The sample may be one which has been freshly obtained from the subject or may be one which has been processed and/or stored prior to making a determination (e.g. frozen, fixed or subjected to one or more purification, enrichment or extractions steps). In embodiments, the sample is a fixed tumour tissue sample (such as e.g. a formalin-fixed paraffin-embedded (FFPE) tissue sample), or a frozen tumour tissue sample (such as e.g. a fresh frozen (FF) tissue sample). The preferred sample type according to the present invention is a FFPE tissue sample, as this type of sample is widely available. Indeed, FFPE tissue samples are commonly obtained in clinical settings, for example for histopathological diagnosis.

The sample may have been obtained from the subject prior to any treatment. The inventors have advantageously found that the proliferative fractions obtained from samples obtained prior to any treatment have increased predictive value compared to proliferative fractions obtained from samples obtained later. In some embodiments, the sample was obtained from the subject prior to surgery and/or chemotherapy. In other embodiments, the sample was obtained from the subject during or after chemotherapy. In some embodiments, the sample was obtained from the subject prior to radiotherapy, optionally prior to radiotherapy and chemotherapy.

In some embodiments, the sample comprises a plurality of samples obtained from the subject at different time points. For example, the sample may comprise at least two samples, comprising a first sample obtained from the subject prior to treatment starting and a second sample obtained from the subject during treatment.

In some embodiments, the subject is predicted to be responsive to the immune checkpoint inhibitor therapy, and wherein the method further comprises the step of administering a therapeutically effective amount of an immune checkpoint inhibitor. In such embodiments, the method may further comprise steps of administering a therapeutically effective amount of an immune checkpoint inhibitor and radiotherapy. The immune checkpoint inhibitor and radiotherapy may be administered concurrently, sequentially or separately. Alternatively, the method may further comprise steps of administering a therapeutically effective amount of an immune checkpoint inhibitor and surgery to excise the tumour.

The immune checkpoint inhibitor may be administered with a further anti-cancer therapy and/or surgery to excise the tumour. Exemplary anti-cancer therapies include, but are not necessarily limited to cyclin inhibitors, endocrine therapy, tyrosine kinase inhibitors, TORC1/2 active drugs, chemotherapy and radiotherapy. The further anti-cancer therapy may comprise one or more of cyclin inhibitors, endocrine therapy, tyrosine kinase inhibitors, TORC1/2 active drugs, chemotherapy, radiotherapy and surgery to excise the tumour.

The immune checkpoint inhibitor may be administered as part of a combination therapy with chemotherapy, optionally wherein the chemotherapy comprises carboplatin and/or nab-paclitaxel.

In some embodiments, the immune checkpoint inhibitor is administered as part of a combination therapy with chemotherapy prior to or after surgery to excise the tumour.

In some embodiments, the immune checkpoint inhibitor is administered as part of a combination therapy with one or more of cyclin inhibitors, endocrine therapy, tyrosine kinase inhibitors and TORC1/2 active drugs prior to or after surgery to excise the tumour.

Any suitable immune checkpoint inhibitor may be administered. The immune checkpoint inhibitor may be as described herein. For example, the immune checkpoint inhibitor may comprise an anti-CTLA-4, anti- PD-1 or anti-PD-L1 molecule. In some embodiments, the immune checkpoint inhibitor comprises an anti- PD-1 or anti-PD-L1 molecule. In some embodiments, the immune checkpoint inhibitor therapy comprises an anti-PD-L1 molecule, for example an anti-PD-L1 antibody. The anti-PD-L1 antibody may comprise an anti-PD-L1 monoclonal antibody. The immune checkpoint inhibitor may comprise atezolizumab.

In some embodiments, the immune checkpoint inhibitor is administered concurrently with chemotherapy, for example, carboplatin and/or nab-paclitaxel.

In some embodiments, the method comprises concurrent, sequential or separate administration of: a) atezolizumab; and b) carboplatin and/or nab-paclitaxel.

In some embodiments, the method comprises concurrent administration of atezolizumab, carboplatin and nab-paclitaxel.

The immune checkpoint inhibitor may be administered every three weeks. In some embodiments, the immune checkpoint inhibitor is administered for eight cycles.

In some embodiments, the subject is predicted to be non-responsive to immune checkpoint inhibitor therapy, and wherein the method further comprises administering chemotherapy to the subject in the absence of any immune checkpoint inhibitor therapy. The chemotherapy may be any suitable chemotherapy. For example, the chemotherapy may comprise carboplatin and/or nab-paclitaxel.

When the subject is predicted to be non-responsive to immune checkpoint inhibitor therapy, the method may further comprise administering radiotherapy to the subject in the absence of any immune checkpoint inhibitor therapy. Alternatively, in embodiments where the subject is predicted to be non-responsive to immune checkpoint inhibitor therapy, the method may further comprise surgically removing the tumour in the absence of any immune checkpoint inhibitor therapy.

In embodiments where the subject is predicted to be non-responsive to immune checkpoint inhibitor therapy, it will be appreciated that any combination of treatment may be envisaged (for example chemotherapy, surgery, radiotherapy, cyclin inhibitors, endocrine therapy, tyrosine kinase inhibitors, TORC1/2 active drugs or any combination thereof), provided that an immune checkpoint inhibitor is not administered to the subject.

According to a further aspect, the present invention provides a computer-implemented method for predicting whether a human subject having cancer will be responsive to treatment with an immune checkpoint inhibitor, the method comprising: a) Obtaining proliferative fraction profile data representing the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject, and b) making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor therapy based on the proliferative fraction profile data of the subject.

It will be appreciated that features of the first aspect may equally apply to the computer-implemented method aspect. For example, the subject, proliferative fraction profile and/or prediction may be as defined in relation to the first aspect.

As described herein, the sample obtained from the subject may comprise a tumour tissue sample. In some embodiments, the sample was obtained from the subject prior to the subject having any treatment. In some embodiments, the sample comprises a plurality of samples obtained from the subject at different time points. For example, the sample may comprise at least two samples, comprising a first sample obtained from the subject prior to treatment starting and a second sample obtained from the subject during treatment.

Step b) may comprise providing the proliferative fraction profile data as input to a machine learning model comprising a classifier trained on a training data set comprising reference proliferative fraction profile data from at least one reference subject. The reference proliferative fraction profile data and reference subject may be as defined in the above aspects.

The machine learning model may comprise one or more of a decision tree, a logistic regression model, an artificial neural network or a support vector machine (SVM). In some embodiments, the machine learning model comprises one or more of a decision tree and a logistic regression model. An exemplary decision tree may comprise a random forest classifier.

In some embodiments, step b) comprises providing the proliferative fraction profile data as input to a machine learning model comprising a classifier trained on a training data set comprising reference proliferative fraction profile data from at least one reference subject and receiving an output from the machine learning model, the output indicative of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor therapy.

The output may comprise a probabilistic score for whether the subject will be responsive or non- responsive to treatment with an immune checkpoint inhibitor therapy.

The present invention also provides an immune checkpoint inhibitor for use in a method of treating cancer, wherein the method comprises: a) measuring the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject; b) making a prediction of whether the subject will be responsive or non-responsive to therapy with an immune checkpoint inhibitor based on the proliferative fraction profile of the subject; and c) treating the subject with the immune checkpoint inhibitor if the subject is predicted to be responsive to therapy in b).

The immune checkpoint inhibitor may be as defined in relation to the first aspect. For example, the immune checkpoint inhibitor may comprise an anti-CTLA-4, anti-PD-1 or anti-PD-L1 molecule. The immune checkpoint inhibitor may comprise atezolizumab.

Step c) may comprise treating the subject with the immune checkpoint inhibitor and chemotherapy if the subject is predicted to be responsive to therapy in b). The chemotherapy may be as defined in relation to any of the above aspects.

The subject may be predicted to be responsive to therapy with an immune checkpoint inhibitor when the proliferative fraction profile of the subject comprises a percentage of at least 10%, at least 20%, at least 30%, at least 40% or at least 50% of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells, which comprise a detectable expression level of the proliferative marker.

According to a further aspect there is provided a method of treating cancer comprising administering an immune checkpoint inhibitor to a subject that has been determined to be responsive to the immune checkpoint inhibitor based on a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject. Administration of the immune checkpoint inhibitor may be as defined in any of the above aspects. For example, the immune checkpoint inhibitor may comprise atezolizumab. The immune checkpoint inhibitor may be administered with chemotherapy.

While the invention has been described in conjunction with the exemplary embodiments described above, many equivalent modifications and variations will be apparent to those skilled in the art when given this disclosure. Accordingly, the exemplary embodiments of the invention set forth above are considered to be illustrative and not limiting. Various changes to the described embodiments may be made without departing from the spirit and scope of the invention.

Examples

EXAMPLE 1: Identification of pCR predictors in atezolizumab treatment of breast cancer

Interactions between proximate specialized cells in distinct activation states underpin the effect of ICB. In cancer, chronic T cell stimulation leads to dysfunction, owing to interactions between cells expressing immune checkpoint receptors and ligands. ICB prevents this to invigorate dysfunctional T cells and these invigorated T cells must then interact with target cancer cells to induce cell death. The efficacy of ICB therefore depends on both the cellular composition and multicellular organization of tumours because they orchestrate these interactions. Breast tumours are heterocellular ecosystems of cancer and tumour microenvironment (TME) cells that self-organize as distinct, recurring multicellular structures. Despite this, the relationship between phenotypic spatial organization of tumours and ICB response has been little explored.

Although multicellular organization pre-treatment may indicate whether the immune response can be augmented by ICB, how ICB remodels tissue structure to achieve this remains obscure. Serial tumour sampling before, during and after treatment could uncover treatment-induced remodelling but is challenging in routine clinical practice. This may explain why the relationship between tissue dynamics during treatment and response is unknown. To characterize the relationship between tissue structure, its dynamics on therapy and immunotherapy response in triple-negative breast cancer (TNBC), imaging mass cytometry (IMC) was used to precisely quantify the phenotype, activation state and spatial location of cells in tumours sampled at three timepoints from patients enrolled in a randomized trial of neoadjuvant immunotherapy.

MATERIALS AND METHODS

Study design and prospective tissue collection

Breast tumour samples were obtained from patients enrolled in a multicentre, randomized, open-label, phase III clinical triall 4 (NeoTRIPaPDLl or NeoTRIP; NCT02620280). NeoTRIP was a neoadjuvant immunotherapy trial of early high-risk TNBC in which 280 patients were randomized to receive neoadjuvant carboplatin and nab-paclitaxel on days 1 and 8, with or without atezolizumab (anti-PD-L1) on day 1. This treatment was given every 3 weeks (one cycle), for a total of eight cycles. Tumours were subsequently surgically excised and, if the responsible clinician opted to do so, an additional four cycles of a nth ra cy clines were given. Patients with treatment-naive, early high-risk TNBC were eligible. Tumour receptor and PD-L1 status (by SP142, Ventana Medical Systems) were determined by central pathology review. Tumour infiltrating lymphocytes were also assessed using established methods. The study protocol was approved at each participating institution; all patients provided written, informed consent. Core biopsies for research were obtained at baseline and after one cycle of therapy (first day of second treatment cycle; on-treatment). Following the full course of therapy, tumours were surgically removed (post-treatment). Tissue microarrays (TMAs) were constructed for surgical excisions only. Regions of tumour, tumour-TME interface and adjacent stroma were annotated by a breast pathologist on corresponding H&E slides to guide TMA construction. Cores of 1 mm in diameter in the identified regions were removed and processed as TMAs.

RNA sequencing and tumour molecular subtypinq

Gene expression data were generated for the biopsy and surgical specimens using exome capture-based RNA sequencing on total RNA samples derived from 5 pm tumour sections. Briefly, exome-enriched complementary DNA libraries were constructed according to the manufacturer’s instructions (TruSeq RNA Exome, Illumina). Pools of 48 libraries were sequenced on a NextSeq 500 or NextSeq 2000 sequencing system (Illumina) with a high-output reagent kit for 75 base pair paired-end reads, with a mean of 10 million paired-end reads per sample per run. Pools were sequenced across replicate runs to achieve over 40 million paired-end reads per sample. Base call files from each sequencing run were converted to fastq format using bcl2fastq conversion software v.2.20, replicate fastq files for each sample were merged and files were aligned to the Ensembl GRCh37 Homo sapiens reference using STAR v.2.5.2 (Dobin et al., 2013). Transcript assembly and expression analysis were performed on each sample with cufflinks v.2.2.1 (Trapnell et al., 2012), resulting in fragments per kilobase million (FPKM) values for each transcript in the genes of interest. The TNBC subtypes were determined using the minimal 101 -gene TNBCtype. Briefly, gene expression for each sample for 101 genes was extracted from the whole transcriptome data and compared with five centroids representing each of the five subtypes by Pearson correlation. The sample was assigned to the subtype with the highest correlation. If no correlation was above 0.195, the subtype was not determined.

Multiplexed imaging antibody panel

Candidate commercial antibodies intended for use in IMC were first validated by immunofluorescence using tonsil and breast cancer tissue to confirm optimal staining intensity, specificity and signal-to-noise ratio. Antibodies that passed validation by immunofluorescence were conjugated to metal isotopes and validated using IMC to ensure preservation of staining specificity and intensity. Sensitivity and specificity were further validated in multiplexed IMC experiments to ensure appropriate patterns of marker colocalization. Finally, optimal concentrations of all metal-conjugated antibodies were determined by visual inspection of IMC images in both tonsil and breast cancer tissue. Antibody conjugation

Indium, yttrium and lanthanide metals were conjugated to antibodies according to the manufacturer’s instructions (Maxpar X8 Multi-Metal Antibody Labelling Kit, Fluidigm). Platinum isotopes were conjugated directly to the reduced antibody without the polymer. Conjugation of bismuth to the antibody required substitution of the L buffer from the Maxpar X8 labelling kit with 5% nitric acid (HNO3) during loading of the metal onto the polymer and deionized water (MilliQ) during the washes. Metal-tagged antibodies were stored in a Candor Antibody Stabilizer (Candor Biosciences) at 4 °C.

Tissue labelling

FFPE slides were dewaxed in xylene and rehydrated in an alcohol gradient. Tissue underwent antigen retrieval (Tris pH 9.0, 95 °C for 30 min) before blocking with 3% BSA in TBS for 1 h. Slides were incubated with unconjugated primary antibodies (PD-L1 clone SP142, PD-1 clone NAT105) overnight at 4 °C, then with metal-conjugated secondary anti-mouse and anti-rabbit antibodies for 3 h at room temperature. Next, slides were incubated with the remainder of the metal-tagged antibodies overnight at 4 °C, then with 0.5 pM iridium for DNA detection (Fluidigm, 201192B) for 30 min. Slides were washed with TBS 0.1 % Tween between each labelling step, and air-dried following the final incubation.

ROIs and IMC

Two sequential sections of FFPE tissue were prepared from core biopsy (baseline and on-treatment) and TMA blocks (post-treatment). One was stained with H&E using an autostainer (Leica ST5020 Stainer/TS5025 Transfer Station/CV5030 Coverslipper Workstation). H&E slides were scanned using the Leica Aperio AT2 Automated Digital Whole Slide Scanner. For core biopsies, ROIs measuring 500 x 500 pm2 were identified by a breast pathologist for acquisition by IMC using the Aperio eSlideManager web application (Leica Biosystems). Three ROIs were selected for each sample unless the biopsy was too small or, for baseline samples, contained no tumour cells. In on-treatment biopsies, if no invasive cancer was identified, regions of tumour bed were instead targeted for IMC. The adjacent section was labelled with antibodies for IMC as described above. ROIs were mapped by manual inspection of annotated H&E images and subjected to IMC (Fluidigm): tissue was raster laser-ablated at 1 pm resolution, then ablated tissue aerosol was ionized using inductively coupled plasma, and resulting isotopic ion reporters quantified using time-of-flight mass spectrometry to infer protein abundance.

Spillover compensation

In mass cytometry applications such as IMC, signal from one channel may spill over to another channel due to trace amounts of contaminating isotopes in metal stock solutions. To account for this, all metal- conjugated antibodies in our panel were spotted separately onto glass slides and dried. Quantification by IMC of all metal isotopes in the panel was then conducted for each dried antibody spot on the slide.

A ‘spillover matrix’ quantifying crosstalk was generated using the Bioconductor CATALYST41 package and subsequently used to correct single-cell measurements.

Image processing, epithelial masks and single-cell measurements

Raw txt file data were converted into multistack image tiff files using existing software (Zanotelli et al., 2017) . To identify regions of contiguous epithelium, we labelled pixels as epithelial based on their expression of cytokeratins and used a random-forest pixel classifier (llastik, Berg et al., 2019) to assign all remaining pixels a probability of belonging to an epithelial region. Probability maps were saved as a red, green and blue (RGB) tiff file, and epithelial regions segmented to generate image masks using standard segmentation tools. For segmentation of single cells, ‘salt and pepper’ noise was removed using a median filter and relevant channels rescaled per image to lie between zero and one. A two-channel image was passed to the Mesmer deep learning single-cell segmentation model (Greenwald et al., 2021) : a nuclear channel (sum of Histone H3 and I r191) and a cytoplasmic channel (sum of panCK and CK5). Whole-cell image masks were used for downstream measurements (single-cell proteomic profiles and size) using CellProfiler44 .

Nuclei were mapped to whole-cell regions, and whole cells mapped to epithelial masks. To be considered ‘related’ to an epithelial mask, at least 30% of the pixels from a whole cell had to overlap with the epithelial region. Before taking measurements, multistack tiff files were filtered for single hot pixels. Single-cell proteomic measurements were taken by computing the mean ion count for each segmented whole cell; these data were spillover corrected using the spillover matrix described above with a nonnegative least squares linear model implemented in CATALYST (Chevrier et al., 2018). Small objects (cells with area less than 31 pm2 ) were excluded from analyses. We left one platinum isotopic channel empty for detection of carboplatin. Carboplatin signal detected in other platinum isotopes, for which conjugated antibodies were included (Vimentin and Calponin), was corrected by fitting a linear model to all cells: log-transformed cell expression was predicted using the carboplatin isotope, and the resulting model residuals were taken as corrected values.

Image curation and cell phenotyping

Cell phenotypes were assigned by semi-supervised clustering. Cells were first classified as epithelial or TME using multiple classification methods, in which the best performing method for each image was manually selected by visual inspection of tissue morphology and cytokeratin expression. The classification methods are listed below. A two-component Gaussian mixture model was fit to the log- transformed sum of all cytokeratins (panCK, CK8/18, CK5/14) to distinguish cells as positive or negative for cytokeratin. Cells related to epithelial masks based on a 30% area overlap were deemed ‘mask- positive’. All images were annotated with mask-positive, cytokeratin-positive or double-positive cells (those that were positive by both criteria). All images were inspected, using tissue morphology as the standard, to determine which method best captured epithelial cells. Most epithelial cells were accurately classified by this approach, but some infiltrating leukocytes were misclassified. To capture infiltrating leukocytes, cells that were mask-positive and cells that were cytokeratin-positive were subclustered using a combination of key epithelial (panCK, CK8/18, CK5/14, AR, GATA3) and immune markers (CD3, CD4, CD8, CD68, CD163, CD11c). Inspection of average expression profiles per cluster was used to identify infiltrating leukocytes. This process was repeated until satisfactory results (determined by inspection of images annotated with cell phenotypes) were obtained. Tumours poorly classified by these approaches (often owing to low cytokeratin expression) were subjected to unsupervised clustering by Phenograph (Levine et al., 2015) per image. Every marker (except for DNA, H3, Carboplatin, c-PARP, CD68, Calponin and Caveolin) was used for clustering, and values were rescaled to lie between zero and one per image. Clusters were categorized as either epithelial or TME by manual inspection of annotated images. Cell phenotypes were derived separately for epithelial and TME cells. Only proteins known to be expressed by epithelial or TME cells based on previous knowledge or manual inspection were included. The proliferation marker Ki67 was excluded from cell clustering. Expression values were clipped at the 99th centile, mean centred and scaled before clustering. Clustering was performed in two steps. First, a selforganizing map (SOM) was created using GigaSOM46 (Kratochvil et al., 2020), then median expression values per SOM node were passed to Phenograph45 and resulting clusters mapped back to single cells. Heatmaps of scaled median expression values were inspected, and clusters lacking meaningful differences merged. Resulting clusters were labelled based on their expression profiles. Cluster validity was further investigated by inspecting images annotated with cluster labels and expression profiles to ensure cell morphology and expression values were concordant with the cluster label. Extensive image curation was conducted under the supervision of a pathologist to identify invasive cancer cells and exclude in situ or normal epithelial cells from clinical correlative analyses. All TME cells were retained for downstream analysis.

Thresholds for marker positivity

We identified thresholds for assigning a cell as ‘positive’ for a given marker by inspecting a random selection of at least 50 images for which cells passing a quantile threshold (calculated using all data) were highlighted. This procedure was repeated at differing quantile thresholds until the value that most closely aligned with marker positivity was identified.

Differential cell phenotype abundance analysis

We used generalized linear models under a binomial distribution with logit link function to determine whether abundance of a given cell phenotype differed between categorical groups47 (PD-L1 status and tumour transcriptional subtype). Cell phenotype proportion per tumour was taken as the response variable and predicted by the binary predictor. The precision of proportion estimates varied substantially between tumours because the total number of cells sampled was also highly variable. When more cells were sampled, proportion estimates were more precise. To account for this variable precision, generalized linear models were weighted by the total number of cells. Proportion values were computed separately by epithelial or TME compartments. The same approach was taken to investigate the relationship between stromal infiltrating lymphocytes and cell phenotype proportions, fitting stromal infiltrating lymphocytes as a continuous predictor. Model coefficients were exponentiated, Iog2- transformed and reported as Iog2 odds ratios. P values were adjusted for multiple testing by the Benjamini-Hochberg method.

Cell densities

Cell phenotype densities were calculated by dividing the number of total cells obtained per biopsy by the total area of the tissue acquired (per mm2 ; Fig. 3a). As tissue did not cover the entirety of all ROIs, the convex hull method was used to draw a ‘tissue’ area based on the existence of all segmented cells.

Cell-cell interaction metrics

Cells were defined as participating in an interaction if their whole-cell masks were in direct contact (that is, their pixels were contiguous. Taking direct contact as the criterion, we computed interactions for all cells using CellProfiler. Cell phenotypes were mapped to cell-cell interaction maps. For each tumour, four ‘flavours’ of cell-cell interaction metric were computed (epithelial homotypic (all epithelial cells interacting with each of the 17 epithelial phenotypes), epithelial heterotypic (all epithelial cells interacting with each of the 20 TME phenotypes), TME homotypic (all TME cells interacting with each of the 20 TME phenotypes) and TME heterotypic (all TME cells interacting with each of the 17 epithelial phenotypes)). The homotypic interactions for an epithelial cell phenotype of interest, for example, were computed as the total number of interactions between that phenotype and all other epithelial cells (regardless of phenotype), divided by the total number of cells in the tumour sample (epithelial and TME cells combined). By contrast, the epithelial heterotypic interactions for a TME phenotype of interest were computed as the number of epithelial-TME interactions (with that TME phenotype) divided by the total number of cells. The definitions for TME- centric interactions were the same but computed from the perspective of the TME (heterotypic interactions were with different epithelial cell phenotypes).

Proliferative fractions

The proportion of cells positive for the proliferation marker Ki67 was computed per cell phenotype per tumour per timepoint (Fig. 4). When a cell phenotype was absent, its corresponding proliferative fraction was also zero. Ki67 status per cell was determined using the method described above to find a suitable threshold for positivity.

Associations with immunotherapy response

We used pCR as a response end point and fitted univariate logistic regression models to test for associations between response and tissue features. Plots illustrating estimates of association between tissue features and pCR depict two odds ratios (and 95% confidence intervals) per predictor: one for each treatment arm, resulting from a univariate logistic regression model restricted to the relevant (C or C&l) study population. Adjacent to these two odds ratios, Rvalues for statistical interaction (Pinteraction, are depicted: these were derived from trivariate logistic regression models that included the feature of interest, treatment (C or C&l) and a term for statistical interaction between the feature of interest and treatment. Pinteraction values corresponded to the statistical term for interaction computed in these models. All predictors (cell densities, cell-cell interactions and proliferative fractions) were square-root- transformed and modelled as continuous. Model coefficients and 95% confidence intervals were exponentiated and reported as odds ratios. When necessary, predictor values were multiplied (proliferative fractions by 10, and cell-cell interactions by 100) so that more interpretable odds ratios could be derived. All clinical correlative analyses were limited to the per-protocol patient population (n = 258; that is, patients who were treated according to the entire trial protocol). To account for multiple testing and to evaluate the likelihood of false positives among significant associations, we computed the FDR using the Benjamini-Hochberg method.

Differential T cell activation

To compare the differential activation state between T cells in contact with tumour cells versus those not in contact, we computed the mean expression level (of activation markers TOX, PD-1 , GZMB, 0X40, ICOS) per tumour per timepoint for each group of T cells (in contact and not) and compared the resulting distributions using two-sided Wilcoxon tests. The same method, taking per-tu mo ur averages, was deployed for comparison of the proportion of Ki67+ cells. The same analysis was conducted when comparing proliferating versus non-proliferating CD8+ TCF1+ T cells.

Immunotherapy-induced tissue dynamics

We compared the cellular composition of tumours through treatment to identify changes that characterized sensitivity and resistance to immunotherapy. We plotted the mean proportion of each cell phenotype (computed separately for epithelial and TME compartments) across timepoints, treatments and response. Means were Z-scored per phenotype and illustrated as trend plots. Significant differences between treatments were illustrated as boxplots and tested using a two-sided Wilcoxon test. Multivariate modelling and variable importance

We fitted regularized logistic regression models (using the R package glmnet) to determine the discriminatory performance of tissue features taken in aggregate to predict pCR. We derived three distinct sets of features for each tumour, separating epithelial and TME compartments: cell phenotype densities; cell interaction metrics as described above; proliferative fractions of cell phenotypes.

A total of 148 variables were derived for single timepoints and 296 when baseline and on-treatment timepoints were combined. Only variables with more than six unique values across samples were retained. We further reduced the feature space for multivariate models by identifying groups of highly correlated variables (Spearman rank correlation > 0.95) and selecting one representative variable at random. To identify highly correlated groups, we first built a graph of variables with at least one correlation of greater than 0.95 (edges were weighted by the correlation coefficient) and used Louvain clustering to discretise subgraphs representing groups of highly correlated variables. Next, data were randomly split into training (75%) and test (25%) sets, for which the proportion of responders was approximately balanced between the two. Regularized logistic regression models were fitted to the training set (using cross validation to identify the minimal shrinkage factorlambda) and predictions made using the test data. An AUC statistic was computed using the prediction probabilities in the test data. To estimate the precision of AUC values, and to derive 95% confidence intervals, the random split procedure, model fitting and testing were repeated 100 times. This whole process was conducted separately by treatment arm, by timepoint (baseline and on-treatment) and for a combined model (data from both baseline and on-treatment timepoints). To determine which predictors were most important in driving predictions we used an established feature selection algorithm (implemented in the R package Boruta)₂o. The principle of this method is repeated comparison of true values with randomly shuffled features to identify which outperform random data more often than would occur by chance. Briefly, all predictors are randomly shuffled, doubling the original predictor set (the original plus the shuffled data), and a random-forest classifier fitted to determine the importance of all predictors in the doubled dataset (importance is calculated by replacing a feature with its randomly permuted equivalent and computing the resulting percentage of misclassified tumours, and then scaled by dividing by the standard error from all misclassification rates). The maximum feature importance achieved among all the randomly shuffled predictors is the threshold for a true feature to then be deemed important (on the basis that an important predictor must outperform random equivalents). This process was repeated 1 ,000 times to generate a binomial distribution of the number of times a given feature was regarded as important, and a final set of important variables identified based on a threshold of P< 0.01 . Taking only those features that outperformed randomly shuffled data more often than expected by chance (at Bonferroni-corrected P< 0.01), we plotted the distribution of their importance values (the scaled percentage misclassification rate) across all 1 ,000 runs to rank their importance. RESULTS

Longitudinal multiplexed imaging of TNBC

IMC was used to profile the expression of 43 proteins at subcellular resolution in tumour samples of formalin-fixed paraffin-embedded (FFPE) tissue collected from patients with TNBC enrolled in the

NeoTRIP randomized controlled trial (Fig. 1). NeoTRIP was a trial of early TNBC that compared neoadjuvantchemotherapy (carboplatin and nab-paclitaxel) with chemotherapy plus anti-PD-L1 immunotherapy (carboplatin, nab-paclitaxel and atezolizumab) by 1 :1 randomization. We collected FFPE tissues at three timepoints for IMC (n = 279 patients): pre-treatment (baseline, n = 243), on the first day of the second treatment cycle (on-treatment, n = 207) and at surgical excision of the tumour bed following treatment (post-treatment, n = 210).

IMC uses laser ablation and time-of-flight mass spectrometry to detect antibodies conjugated to rare earth metal reporters to infer protein abundance at subcellular resolution. We labelled tissues using a 43-plex IMC assay to precisely characterize the TME and key cancer cell phenotypes. In addition, we mapped carboplatin directly in situ by detecting platinum and found that levels in on-treatment and post-treatment samples were far greater than baseline, with much of the drug accumulating in macrophages. Our approach successfully generated 1 ,855 high-plex tissue images from both biopsies and excisions using FFPE samples collected prospectively as part of a randomized trial.

Diverse cell phenotypes in TNBC

To precisely characterize cell phenotypes in situ, we segmented single cells using deep learning and derived proteomic profiles. For cell phenotyping, we separated epithelial (cancer cells) and TME cells using several methods (Fig. 2) and, taking cell morphology as the standard, selected the best performing. To discover salient cell phenotypes, we clustered single cells, limiting the proteins used for clustering to those relevant to epithelial or TME cells. Cell clustering resulted in 17 epithelial and 20 TME phenotypes. Epithelial cells were distinguished by markers of lineage, activation state and immunoregulation. Among TME cells, cytotoxic and helper T cells separated according to expression of PD-1 and TCF1 (encoded by TCF7) which identifies stem-like T-cells. We also identified regulatory T (Treg) cells defined by FOXP3 expression and activated cytotoxic T cells with high granzyme B expression (CD8+ GZMB+ T). There were two cell phenotypes positive for PD-L1 ; these were both CD11c+ antigen presenting cells (APCs), of which one was IDO+. We also identified B cells, plasma cells, macrophages, dendritic cells, neutrophils, endothelial cells and three fibroblast phenotypes. We next correlated the cellular compositions of different images from the same tumour to estimate the extent of spatial heterogeneity. Correlations between images from the same tumour varied between compartment and timepoint (from an average of 0.65 for TME cells at baseline, to 0.76 for TME cells post-treatment;). We also examined the variance per cell phenotype and found that it was generally low, with the highest being among basal epithelial cells. To determine the relevance of cell phenotypes to established tumour pathology, we investigated the relationship between cell phenotype proportions and relevant clinical features. In comparison with clinical

PD-L1 status assessed by centralized pathology review, we confirmed that PD-L1 expression by IMC was greater in PD-L1 -positive tumours. Similarly, the cell phenotypes most enriched among PD-L1 -positive tumours were characterized by the highest expression of PD-L1 . All lymphoid cell phenotypes were also positively associated with PD-L1 status at baseline. Stromal infiltrating lymphocytes were also positively correlated with all lymphoid cells and PD-L1+ APCs. We also investigated whether the abundance of cell phenotypes significantly differed between established transcriptomic subtypes of TNBC17. Both epithelial and TME cell phenotypes differed between tumour subtypes. The luminal androgen receptor subtype, for example, was characterized by the highest proportion of epithelial cells positive for androgen receptor (AR+ LAR), and the mesenchymal tumour subtype contained the greatest proportion of all three stromal cell phenotypes.

Cancer-immune interactions predict ICB response

It was decided to investigate if the tissue structure of treatment-naive tumours is a determinant of response to immunotherapy. In neoadjuvant trials such as NeoTRIP (in which primary treatment is given before surgical excision of the tumour) a surrogate end point, pathological complete response (pCR) is used to determine responsiveness. pCR is defined as the absence of invasive cancer cells after treatment and can be used to identify responders before time-to-event follow-up has matured. To evaluate the link between tissue structure and response, logistic regression models were fitted to predict pCR separately by randomization arm and included a term for statistical interaction (Pinteraction) to test whether the association between a given feature and response significantly differed by treatment. To estimate the impact of multiple testing, we also computed the false discovery rate (FDR). We first asked whether the densities of different epithelial and TME cell phenotypes differed in their capacity to predict response (Fig. 3a and b). Only PDL1+ IDO+ APC density predicted ICB (but not chemotherapy) response, although this was associated with an elevated FDR (Pinteraction = 0.01 , FDR = 0.3; Fig. 3c and d).

Because ICB modulates interactions among immune and cancer cells, we explored whether different cell-cell interactions were associated with response to immunotherapy but not chemotherapy. Cells were deemed to be interacting if they were in direct contact (Fig. 4). For all epithelial cells, we computed the number of interactions with each epithelial cell phenotype (homotypic interactions) and with each TME cell phenotype (heterotypic interactions), normalizing by the total number of cells present (Fig. 4). This approach gave greater weight to cells with multiple interactions. The distribution of cancer-TME interactions was a continuum across tumours; hence, we modelled interactions as continuous predictors. Among epithelial-epithelial interactions four cell phenotypes were associated with outcome, but none of the estimates differed significantly between treatment arms. In contrast, associations between eight epithelial-TME interactions and response significantly differed between treatments, with epithelial-CD20+ B (Pinteraction = 0.003, FDR = 0.06) and epithe lial— CD8+ GZMB+ T (Pinteraction = 0.006, FDR = 0.06) cell interactions showing the greatest differential effect (Fig. 5a and b). Repeating this analysis for TME cells did not, however, uncover significant predictors of differential response (Fig. 6a and b). Heterotypic epithelial interactions were only moderately correlated with corresponding cell densities, suggesting they reflect distinct aspects of tumour organization. Because activated T cell-cancer cell interactions predicted ICB response, we investigated whether T cells in contact with cancer cells were functionally distinct from other T cells. We found that contact with a cancer cell was associated with higher expression of key activation markers (TOX and PD-1 for cytotoxic T cells; TOX and 0X40 for T helper cells; Fig. 7a), and that T cells contacting cancer cells were much more likely to be proliferating (Fig. 7b). These findings corroborate the functional significance of cell-cell interactions.

Proliferative fractions predict ICB response

We next computed the proportion of Ki67⁺ cells per phenotype (proliferative fraction) and tested for associations with pCR (Fig. 8a-d). Strikingly, the proliferative fraction of just one cell phenotype was associated with response in the chemotherapy arm (CKIo GATA3+epithelial cells), but when patients were treated with immunotherapy, 12 epithelial and 16 TME cell phenotypes predicted response (Fig. 8 b,c). The proliferative fraction of MHCI&I I hi cells was the strongest predictor of immunotherapy response among epithelial (cancer) cells (Pinteraction = 0.004, FDR = 0.04), whereas the proliferative fraction of CD8+ TCF1 + T cells was the strongest immunotherapy response predictor overall

(Pinteraction = 8 x 10-5, FDR = 0.003). These features (proliferative fractions of MHCI&I Ihi cancer cells and CD8+ TCF1 + T cells) were, however, only moderately correlated (p = 0.46; Fig. 8e). Despite the proliferative fraction of CD8+ TCF1 + T cells being the strongest predictor, neither their overall density nor their interactions with TME cells were associated with response, underscoring that proliferative fractions enrich for cells in distinct activation states. Indeed, proliferating CD8+ TCF1 + T cells showed significantly higher levels of all key activation markers (TOX, PD-1 , GZMB, ICOS, Helios) and were more often in contact with cancer cells and with MHCII+ cells. Pre-treatment proliferative fractions therefore enrich for cells in distinct activation states and identify phenotypes that predict ICB response.

On-treatment ICB response predictors

Using biopsies taken early on-treatment (first day of second treatment cycle), we investigated the link between on-treatment cell densities, cell-cell interactions and immunotherapy response (Fig. 9a-d).The correlation structure of cell densities and their corresponding cell-cell interaction metrics echoed that pretreatment: heterotypic epithelial (cancer-TME) interactions were moderately correlated with cell densities whereas other cell-cell interaction metrics were highly correlated. Among heterotypic epithelial interactions, only epithelia interactions with CD79a+ Plasma cells were significantly enriched in tumours resistant to immunotherapy but not chemotherapy (Pinteraction = 0.004, FDR = 0.09); the density of CD79a+ Plasma cells was not, however, associated with response (Fig. 9a). Among other TME cell densities, only CD8+ GZMB+ T cells showed significant differential immunotherapy response prediction (Pinteraction = 0.04, FDR = 0.5) and, consistent with their strong correlation, this also held for homotypic CD8+ GZMB+ T interactions (Pinteraction = 0.02, FDR = 0.3) (Fig. 9a, b), but with elevated FDRs. The CD15+ epithelial (cancer) cell phenotype was distinct because it was associated with resistance to immunotherapy (but not chemotherapy) when quantified as a density (Pinteraction = 0.004, FDR = 0.08) or cell-cell interaction metric (heterotypic Pinteraction = 0.003, FDR = 0.05; homotypic

Pinteraction = 0.04). In some ICB-resistant cases, expression of CD15 by cancer cells was characterized by a striking mosaic expression pattern for which clear CD15+ CKIo cells were admixed with CD15

CKhi cells, suggestive of discrete phenotypic state transitions. We also observed foci of CD15+ cancer cells surrounded by CD15+ leukocytes, implicating heterotypic interactions as possible drivers of state transition. Given the predictive value of proliferation in treatment-naive tumours, we asked whether it would perform similarly on-treatment. We found that although the functional significance of proliferation was largely preserved on-treatment, proliferation itself was reduced and proliferating cell fractions were not predictive of response. In conclusion, markers of outcome on-treatment were distinct from those in treatment-naive tumours. Response to immunotherapy was characterized by accumulation of CD8+ GZMB+ T cells, whereas heterotypic CD79a+ Plasma cell interactions and CD15+ cancer cells marked resistant tumours.

Dominant ICB response predictors

In total, we derived 148 tissue features (densities of 37 cell phenotypes; 37 heterotypic, and 37 homotypic cell-cell interactions; 37 proliferative fractions) and found more were predictive of immunotherapy than chemotherapy response (112 versus 26), and more of these were found at baseline than on-treatment (70 versus 42). We therefore asked whether their combined predictive performance would also differ by treatment and timepoint. For each treatment arm, we fit three regularized multivariate logistic regression models: using baseline data, on-treatment data and data from both timepoints (Fig. 10a). Predictive performance was always better among immunotherapy-treated patients, implying that immunotherapy was more dependent on TME activation state and tumour structure (Fig. 10b). Despite finding many more predictors of immunotherapy response in treatment-naive baseline samples, predictive performance was similar for baseline and on-treatment multivariate models (mean receiver-operating characteristic area under the curve (AUC) 0.77 for both). Combining baseline and on-treatment features, however, materially improved predictive performance (mean AUC 0.82), showing that the features measured at each timepoint reflect distinct facets of response dynamics. Overall, multivariate modelling showed that TME activation and tumour structure play a greater role in treatment response when patients receive immunotherapy, and that early on-treatment biopsies improve predictive accuracy and could therefore help guide adaptive treatment strategies. It remained unclear whether response to immunotherapy was driven by the combined effect of many features or just a few. We used an established method to identify the dominant drivers of response (Fig. 10c).This analysis revealed clear immunotherapy response drivers. At baseline, a total of 14 predictors contributed substantively to overall model performance, spanning cancer and TME cells, and mainly comprised cell-cell interactions and proliferative fractions (Fig. 10d). By far the best predictor was the proliferative fraction of CD8+ TCF1+ T cells, followed by the proliferative fraction of MHCl&llhi cancer cells, cancer-B cell interactions and cancer-CD8+ GZMB cell interactions. On-treatment, 12 top predictors were identified (Fig. 10e). The two best predictors of these 12 corresponded to CD8+ GZMB+ T cell abundance (homotypic interactions and density), whereas the next two most important predictors corresponded to CD15+ cancer cell abundance (homotypic interactions and density). In summary, feature importance analysis revealed distinct immunotherapy response drivers at baseline versus on-treatment.

EXAMPLE 2: Treatment across the group with atezolizumab did not result in an improved average EFS

NeoTRIP is a Phase III open label randomized trial that randomly enrolled women with triple negative breast (TNBC) cancer to receive neoadjuvant nab-paclitaxel and carboplatin x 24 weeks w/o (Arm A, n=142) or with (Arm B; n=138) the PD-L1 targeting monoclonal antibody atezolizumab, 1200 mg on day 1 q 3wks for 8 cycles before surgery. After surgery patients were candidate to receive an anthracycline- containing chemotherapy (doxorubicin or epirubicin with cyclophosphamide for 4 cycles) and continued follow up.

Main patient characteristics were fully reported (L. Gianni et al. (2022)). Of note, the status of TNBC was assessed in a central laboratory of pathology to characterize estrogen and progesterone receptor expression, erbB2 status and PD-L1 expression. The primary endpoint of the study is event free survival defined as the time from randomization to the first date of disease progression while on primary therapy or disease recurrence (local, regional, distant, invasive contralateral breast, other invasive cancers) after surgery or death due to any cause, including unknown causes. Among secondary endpoint there are safety and response assessed at surgery to compare rates of pathologic complete response (pCR; ypTO/Tis ypNO).

The analysis showed that addition of atezolizumab to chemotherapy led to a statistically non-significant numerical increase of pCR (48.6 vs 44.4%, respectively; p = 0.48) at cost of minor additional toxicity. In the study specimen of the tumor were collected before starting therapy, after 1 cycle, and at surgery after completing the allocated drug regimen.

An analysis of EFS was conducted at 54 months of median follow up years. The analysis showed that addition of atezolizumab did not result in an improved average EFS (as calculated across the group). The EFS was 70.6% with atezolizumab and 74.9% w/o (HR 1 .076; p=0.76) In a multivariate analysis (see Table 1 below). PD-L1 expression and higher sTILs at baseline were prognostic for better EFS, but not predictive of atezolizumab benefit.

Table 1 : Multi-variate analysis of hazard ratio analysis of EFS. A HR less than 1 indicated a positive association with EFS. A HR greater than 1 indicated a negative association with EFS (i.e., a positive association with disease progression).

EXAMPLE 3: Analysis of predictive factors for EFS in immunotherapy

It was decided to investigate the role of the tumour immune microenvironment in the probability of benefit from the PD-L1 targeting monoclonal antibody atezolizumab in EFS.

Example 1 demonstrates that that the density of interaction between selected activated immune cells and epithelial cancer cells at different locations in the tumour specimen is strongly associated with the probability of achieving pCR with the addition of the ICI. The key aspect of the observation is that the prediction is restricted to immunotherapy treatment, while the same predictors are not linked to the probability of pCR with chemotherapy only.

However, it was not clear if predictors were also linked to the probability of event-free survival. This was particularly in view of the ongoing debate over the clinical relevance of pCR. Previously, the achievement of pCR was considered a surrogate marker of efficacy by the FDA (Prowell and Pazdur 2012). However, several retrospective analyses including the one used by FDA for accepting pCR as a surrogate (Cortazar et al., 2014), failed to show that an improved rate of pCR translated into an improved event free survival at a trial level. In other words, although the achievement of pCR is strongly prognostic at individual level (a subject who has does not have any residual tumour at surgery will do much better than a subject with residual disease at surgery), little correlation existed for the majority of trials between improved pCR and improved EFS in the average subject population. This is also the case for immunotherapy in TNBC.

In unselected TNBC patients the addition of an anti-PD1/anti-PD-L1 agent to chemotherapy increased the pathological Complete Response (pCR) rate by 2.8% to 17%. Whether the heterogeneity of results is due to chance or clinical-biological factors such as the adoption of a different chemotherapy backbone or different patient characteristic is unknown.

Overall, the higher rate of pCR obtained upon addition of ICIs appears of the same order of magnitude or even smaller to that achieved by adding platinum agents to anthracycline-based regimens. However, it is unclear whether pCR is the most appropriate endpoint to measure the role of ICIs. In general, it is known that achieving a pCR has prognostic value for individual patients, but the increase of pCR rate is not a validated surrogate for EFS or overall survival (OS) at trial level. Additional factors could limit the value of surrogacy for pCR with ICIs.

An example of the disconnection between pCR increase and EFS is well shown in the KEYNOTE 522 trial (Bianchini et al., 2022). Overall, in this trial the increase of pCR was only 7.5%. However, at a median follow-up of 39.1 months, a clinically meaningful improvement of EFS was observed (HR 0.63, 95% Cl 0.48-0.82; P = 0.00031), crossing the prespecified interim analysis boundary for significance (P = 0.00517). Notably, EFS benefit was similar in PD-L1 negative (HR 0.48 95% Cl 0.28-0.85) and positive (HR 0.67 95% Cl 0.49-0.92) patients. The trials demonstrated also a 39% reduction of distant recurrence risk (HR 0.61 95% Cl 0.46-0.82), and an encouraging numerical improvement of OS (HR 0.72 95% Cl 0.51-1.02; p=0.0321).

A reinforcement of these positive findings comes from an update of the data from the GeparNuevo trial of neoadjuvant chemotherapy plus either durvalumab or placebo in patients with TNBC. Invasive DFS (HR 0.48 95% Cl 0.24-0.97; p=0.0398), distant EFS (HR 0.31 95% Cl 0.13-0.74; p=0.0078) and OS (HR 0.24 95% Cl 0.08-0.72; p=0.0108) were improved. Of note, as previously mentioned, the survival outcomes in both KEYNOTE-522 and GeparNuevo improved to an extent greater than expected on the basis of pCR rate increase (by 7.5% and 9.2%, respectively), confirming that caution is needed in interpreting pCR results from neoadjuvant trials of ICIs. Consistent with these two trials, a numerical improvement in EFS was also described in the IMpassion031 (Bianchini et al., 2022).

Overall, these results established the role of neoadjuvant ICIs in early high-risk TNBC. Considering the cost of these compounds and the safety profile of ICIs with rare, but sometimes serious and occasionally even fatal irAE, precision immunology and treatment tailoring is a new and urgent unmet need.

It was then decided to test several biomarkers that were identified in Example 1 as being associated with pCR following atezolizumab treatment. The four biomarker profiles identified as having the highest statistical predictive power for predicting pCR in atezolizumab were assessed to determine if there was an association with EFS. The four profiles were:

- CD20B⁺;

CD8⁺ and GZMBT⁺ (i.e., dual expression of CD8 and GZMBT);

MHC I and ll⁺ and Ki67⁺ (i.e., triple expression of MHC I and II and Ki67); and

CD8⁺TCF1⁺ and Ki67⁺ (i.e., triple expression of CD8, TCF1 , and Ki67)

For example, a cell would be considered to meet the fourth profile above if it expressed detectable levels of CD8, TCF1 , and Ki67.

The expression profile of each was obtained from samples obtained from the subjects prior to treatment, and then compared to EFS for the subject at the median follow-up of 39. 1 months.

The expression profile of each of the four biomarkers for each subject was plotted and separated out into three tertiles - an upper fertile for the highest level of expression (fertile 3) and middle fertile (fertile 2) and a bottom tertile for the lowest level of expression (fertile 1). The association between each fertile and EFS was then analysed by calculating a hazard ratio. Results are shown in Table 2.

For the biomarker profiles CD20B+ and CD8+ and GZMBT+, there was no association with EFS in the atezolizumab group (see Table 2). In particular, the highest tertiles of these two profiles did not show any association with better EFS for the atezolizumab group. This was despite the same profiles being associated with pCR.

In contrast, the biomarker profiles MHC I and H⁺ and Ki67⁺ and CD8⁺TCF1⁺ and Ki67⁺ exhibited a positive association between EFS and atezolizumab. This was particularly apparent for the highest tertiles (tertiles 3, Table 2). To further analyse this association, the bottom two tertiles for each of the biomarker profiles MHC I and ll⁺ and Ki67⁺ and CD8⁺TCF1⁺ and Ki67⁺ were combined, to generate a “LOW quantile”. The hazard ratio of the LOW quantile was then compared to the HIGH quantile, which corresponds to the upper fertile of Table 2. Results are shown in Table 3. The HIGH quantile of CD8TCF1 Ki67 had a significant association with increased EFS in atezolizumab groups. This is clearly shown in Figure 11 , which shows that the CD8TCF1 Ki67 high expression is strongly associated with EFS for treatment with atezolizumab. The HIGH quantile of MHCIIhiKi67 demonstrated a trend for increased EFS in atezolizumab groups (analysis performed for the high fertile (HIGH) and the combined low and intermediate tertiles (LOW)).

In summary, we have demonstrated that the baseline biomarker profiles MHC I and H⁺ and Ki67⁺ and CD8⁺TCF1⁺ and Ki67⁺ predicts for EFS in women with TNBC.

References

A number of publications are cited above in order to more fully describe and disclose the invention and the state of the art to which the invention pertains. Full citations for these references are provided below. The entirety of each of these references is incorporated herein.

Berg, S. et al. ilastik: interactive machine learning for (bio)image analysis. Nat. Methods 16, 1226-1232 (2019).

Bianchini et al. Treatment landscape of triple-negative breast cancer - expanded options, evolving needs. Nat Rev Clin Oncol 19(2): 91-113, 2022.

Chevrier, S. et al. Compensation of signal spillover in suspension and imaging mass cytometry. Cell Systems 6, 612-620.e615 (2018).

Cortazar, P. et al., Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. The Lancet. 384:9938, P164-172, 2014

Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15-21 (2013).

Gianni et al. Pathologic complete response (pCR) to neoadjuvant treatment with or without atezolizumab in triple-negative, early high-risk and locally advanced breast cancer: NeoTRIP Michelangelo randomized study. Annals Oncol. 33, 534-543 (2022).

Greenwald, N. F. et al. Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. Nat. Biotechnol. https://doi.org/10.1038/s41587-021- 01094-0 (2021).

Kratochvil, M. et al. GigaSOM.jl: high-performance clustering and visualization of huge cytometry datasets. GigaScience 9, giaa127 (2020).

Levine, J. H. et al. Data-driven phenotypic dissection of AML reveals progenitor-like cells that correlate with prognosis. Cell 162, 184-197 (2015).

Loibl et al. Neoadjuvant durvalumab improves survival in early triple-negative breast cancer independent of pathological complete response. Ann Oncol. 2022 Nov;33(11):1149-1158.

Loibl et al. A randomised phase II study investigating durvalumab in addition to an anthracycline taxane- based neoadjuvant therapy in early triple-negative breast cancer: clinical results and biomarker analysis of GeparNuevo study. Ann Oncol. 2019 Aug 1 ;30(8):1279-1288.

Mittendorf et al. Neoadjuvant atezolizumab in combination with sequential nab-paclitaxel and anthracycline-based chemotherapy versus placebo and chemotherapy in patients with early-stage triplenegative breast cancer (IMpassion031): a randomised, double-blind, phase 3 trial. Lancet. 2020 Oct 10;396(10257):1090-1100.

Prowell T.M. and R Pazdur. Pathological Complete Response and Accelerated Drug Approval in Early Breast Cancer." New England Journal of Medicine 366(26): 2438-2441 , 2012. Rizzo et al. KEYNOTE-522, IMpassion031 and GeparNUEVO: changing the paradigm of neoadjuvant immune checkpoint inhibitors in early triple-negative breast cancer. Future Oncol. 2022 Jun;18(18):2301- 2309.

Schmid et al. Pembrolizumab for Early Triple-Negative Breast Cancer. N Engl J Med. 2020 Feb 27;382(9):810-821.

Schmid et al. Event-free Survival with Pembrolizumab in Early Triple-Negative Breast Cancer. N Engl J Med. 2022 Feb 10;386(6):556-567.

Trapnell, C. et al. Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat. Protoc. 7, 562-578 (2012). Zanotelli, V. R. & Bodenmiller, B. ImcSegmentationPipeline: a pixel classification based multiplexed image segmentation pipeline. GitHub https://github.com/BodenmillerGroup/lmcSegmentationPipeline (2017).

For standard molecular biology techniques, see Sambrook, J., Russel, D.W. Molecular Cloning, A Laboratory Manual. 3 ed. 2001 , Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press

able 2: Hazard ratio analysis to determine predictive power of biomarkers for EFS. A HR less than 1 indicated a positive association with EFS. A HR reater than 1 indicated a negative association with EFS (i.e., a positive association with disease progression).

le 3: Hazard ratio analysis of LOW and HIGH quantiles of biomarker profiles MHC I and H⁺ and Ki67⁺ and CD8⁺TCF1⁺ and Ki67⁺ in relation to ociation with EFS. A HR less than 1 indicated a positive association with EFS. A HR greater than 1 indicated a negative association with EFS (i.e., a itive association with disease progression).

Claims

Claims:

1 . A method for predicting whether a human subject having cancer will be responsive to treatment with an immune checkpoint inhibitor, the method comprising: a) measuring the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject, and b) making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor therapy based on the proliferative fraction profile of the subject.

2. The method of claim 1 , wherein the MHCI⁺MHCII⁺ cells are cancer cells, optionally epithelial cells comprising a detectable expression level of one or more of CK5/14, CK/8/18, PanCK and AR.

3. The method of claim 1 , wherein step a) comprises measuring the proportion of CD8⁺TCF1 ⁺ T-cells which comprise a detectable expression level of the proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cell proliferative fraction profile of the subject.

4. The method of any one of the preceding claims, wherein the cancer comprises breast cancer, ovarian cancer, pancreatic cancer, colorectal cancer, lung cancer, pancreatic cancer, colon cancer, gastric cancer, bladder cancer, skin cancer, myeloma, non-Hodgkin lymphoma, prostate cancer, oesophageal cancer, head and neck cancer, endometrial cancer, hepatobiliary cancer, duodenal carcinoma, thyroid carcinoma, or renal cell carcinoma, preferably wherein the cancer comprises breast cancer.

5. The method of any one of the preceding claims, wherein the cancer comprises triple-negative breast cancer.

6. The method of any one of the preceding claims, wherein the immune checkpoint inhibitor therapy comprises atezolizumab.

7. The method of any one of the preceding claims, wherein the proliferation marker comprises Ki67.

8. The method of any one of the preceding claims, wherein step b) comprises comparing the proliferative fraction profile of the subject to a plurality of reference proliferative fraction profiles and making a prediction of whether the subject will be responsive or non-responsive to treatment based on the comparison.

9. The method of claim 8, wherein the plurality of reference proliferative fraction profiles are separated into at least two quantiles and the proliferative fraction profile of the subject is compared to the quantiles, wherein the subject is predicted to be responsive to treatment when the proliferative fraction profile of the subject fits into the top quantile.

10. The method of claim 9, wherein the at least two quantiles comprise at least three quantiles.

11. The method of any one of the preceding claims, wherein step a) comprises measuring the proportion of CD8⁺TCF1⁺ T-cells and MHCI⁺MHCII⁺ cells which comprise a detectable expression level of the proliferation marker to obtain a CD8⁺TCF1⁺ T-cells and MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject.

12. The method of any one of the preceding claims, wherein responsiveness to treatment comprises event-free survival for a time period of at least 18 months, optionally for a time period of at least 24 months.

13. The method of any one of the preceding claims, wherein the sample comprises tumour tissue.

14. The method of any one of the preceding claims, wherein the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker is measured using immunohistochemistry (IHC) or imaging mass cytometry (IMC).

15. The method of any one of the preceding claims, wherein the sample has been obtained from the subject prior to any treatment.

16. The method of any one of the preceding claims, wherein the subject is predicted to be responsive to the immune checkpoint inhibitor therapy, and wherein the method further comprises the step of administering a therapeutically effective amount of an immune checkpoint inhibitor.

17. The method of claim 16, wherein the immune checkpoint inhibitor is administered as part of a combination therapy with chemotherapy, optionally wherein the chemotherapy comprises carboplatin and/or nab-paclitaxel.

18. The method of claim 17, wherein the method comprises concurrent, sequential or separate administration of: i) atezolizumab; and ii) carboplatin and/or nab-paclitaxel.

19. The method of any one of claims 1 to 15, wherein the subject is predicted to be non-responsive to immune checkpoint inhibitor therapy, and wherein the method further comprises administering chemotherapy to the subject in the absence of any immune checkpoint inhibitor therapy.

20. A computer-implemented method for predicting whether a human subject having cancer will be responsive to treatment with an immune checkpoint inhibitor, the method comprising: a) Obtaining proliferative fraction profile data representing the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject, and b) making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor therapy based on the proliferative fraction profile data of the subject.

21. An immune checkpoint inhibitor for use in a method of treating cancer, wherein the method comprises: a) measuring the proportion of CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells which comprise a detectable expression level of a proliferation marker in a sample obtained from the subject to obtain a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject; b) making a prediction of whether the subject will be responsive or non-responsive to treatment with an immune checkpoint inhibitor based on the proliferative fraction profile of the subject; and c) treating the subject with the immune checkpoint inhibitor if the subject is predicted to be responsive to therapy in b).

22. A method of treating cancer comprising administering an immune checkpoint inhibitor to a subject that has been determined to be responsive to the immune checkpoint inhibitor based on a CD8⁺TCF1⁺ T-cells and/or MHCI⁺MHCII⁺ cells proliferative fraction profile of the subject.