Predictive response biomarker discovery method
By identifying TCR clonotypes and gene expressions before and after T cell treatment, predictive response biomarkers are generated, and patients who are beneficial to the treatment are selected to target T cell immunomodulation therapy, the existing treatment methods are solved for most patients. The problem of ineffectiveness improves the targetedness and effectiveness of treatment.
Patent Information
- Application Number
- CN202380029565.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-03-28
- Filing Date
- 2023-03-27
- Publication Date
- 2024-11-08
AI Technical Summary
Existing treatments are ineffective for most patients, especially expensive targeted therapy options, potentially harmful to some patients and may lead to the waste of other treatment opportunities, requiring the identification of appropriate patient populations to maximize treatment success rates.
Prognostic response biomarkers (PRBs) were generated by identifying T cell TCR cloning types before and after treatment with anti-PD-1, anti-PD-L1, anti-TNFR2 and anti-OX40 agonists. To determine whether the patient is capable of responding to treatment, use logistic regression coefficients to weight gene expression status to select patients who are beneficial to the treatment for treatment.
Improve the targeted and effective treatment, reduce the economic and health risks of ineffective treatment, and ensure the success rate of treatment.
Smart Images

Figure CN118922531A_ABST
Abstract
Description
[0001] Citation to Related Applications
[0002] This application claims priority and the benefit of the filing date of U.S. Provisional Patent Application No. 63 / 324,316, filed on March 28, 2022.
[0003] This application also relates to (but does not claim priority to) international patent application PCT / US2021 / 037916 filed on June 17, 2021, which claims the benefit of the filing dates of U.S. Provisional Patent Application No. 63 / 041,440 filed on June 19, 2020; U.S. Provisional Patent Application No. 63 / 079,507 filed on September 17, 2020; and U.S. Provisional Patent Application No. 63 / 172,904 filed on April 9, 2021.
[0004] The entire contents of each of the above-mentioned applications (including any sequences and figures in the listings) are incorporated herein by reference. Background Art
[0005] Most therapeutic interventions for disease indications such as cancer, autoimmune diseases or neurodegenerative diseases are only effective in a subset of patients, while being ineffective at best, if not harmful, in the remainder. Treating patients who may not benefit from a treatment is not only financially burdensome (especially for certain expensive treatment options involving proprietary innovative monoclonal antibodies), but can also be potentially harmful to the patients being treated, not to mention the lost opportunity to treat these patients with other alternative treatments that may be effective.
[0006] Therefore, there is a need to identify the appropriate patient population for a particular treatment regimen in order to maximize the chances of successful treatment. Summary of the invention
[0007] One aspect of the present invention provides a method for identifying a predictive response biomarker (PRB) for treating cancer (e.g., renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL) or uterine cancer (UC)) using a T cell-targeted immunomodulatory therapy (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3) and / or an anti-OX40 agonist antibody (such as HFB10)), the method comprising the following steps: (a) identifying, in a pre-treatment sample of the cancer, a TCR (T cell receptor) clonal type of a T cell clonally expanded in a matched post-treatment sample of the cancer treated by the T cell-targeted immunomodulatory therapy. (a) generating a list of genes that are differentially upregulated and / or downregulated (e.g., log-fold >1, or upregulated by 10-fold or downregulated by 10-fold) in the responsive T cells in the pre-treatment sample compared to the corresponding genes in the unresponsive T cells in the pre-treatment sample to produce a PRB; and (c) optionally, weighting each gene in the PRB using a logistic regression coefficient obtained by fitting each gene as a predictor to the expansion status and / or clinical response status as an outcome.
[0008] In certain embodiments, step (a) comprises: (1) contacting an ex vivo culture of an untreated sample of cancer with an immunomodulatory compound targeting T cells (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3) and / or an anti-OX40 agonist antibody (such as HFB10)) under conditions and for a period of time sufficient to allow the immunomodulatory effect of the compound on the T cell population in the ex vivo culture to manifest; (2) encapsulating single T cells isolated, purified or enriched from the ex vivo culture into picoliter droplets for single cell profiling of functional properties; ng), thereby dividing each encapsulated single T cell into a first responding T cell pool and a second non-responding T cell pool based on the presence or absence of the functional property, respectively; (3) determining the TCR clonotype of each encapsulated single T cell in the first responding T cell pool and the second non-responding T cell pool, thereby identifying the TCR clonotype of the responding T cell as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample; (4) encapsulating the single T cells isolated, purified or enriched from the pre-treatment sample into picoliter droplets to identify the responding T cell having the same TCR clonotype as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample.
[0009] In certain embodiments, step (a) comprises: (1) obtaining the pre-treatment sample and the matched post-treatment sample of the disease (e.g., cancer) from a suitable donor, wherein the matched post-treatment sample has been contacted with the compound under conditions and for a period of time sufficient to allow the immunomodulatory effect of the compound on the T cell population in the post-treatment sample to manifest; (2) encapsulating the single T cells isolated, purified or enriched from the post-treatment sample into picoliter droplets for single-cell characterization of functional properties, thereby characterizing each encapsulated single T cell based on the presence or absence of the functional property, respectively. (3) determining the TCR clonotype of each encapsulated single T cell in the first responding T cell library and the second non-responding T cell library, thereby identifying the TCR clonotype of the responding T cell as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample; (4) encapsulating the single T cells isolated, purified or enriched from the pre-treatment sample into picoliter droplets to identify the responding T cells having the same TCR clonotype as the TCR clonotype of the clonally expanded T cells in the matched post-treatment sample.
[0010] In certain embodiments, step (b) comprises: (5) using whole transcriptome single-cell RNA sequencing (scRNA-seq) to identify the list of genes upregulated in the responsive T cells in the pre-treatment sample, wherein each gene in the PRB has a log2 fold change > 0.2 and is expressed in < 40% of T cells with a TCR clonotype that is not clonally expanded in the matched post-treatment sample.
[0011] In certain embodiments, the scRNA-seq is performed using a gene panel specifically designed for the mechanism of action of the compound or a predetermined gene panel designed to assess immune modulation.
[0012] In certain embodiments, the pre-treatment sample is a blood sample.
[0013] In certain embodiments, the cancer is renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL), or uterine cancer (UC).
[0014] In certain embodiments, the ex vivo culture is freshly isolated from cancer tissue.
[0015] In certain embodiments, the ex vivo culture is established from stored (eg, frozen) cancer tissue.
[0016] In certain embodiments, the ex vivo culture is a single cell suspension.
[0017] In certain embodiments, the ex vivo culture is an adherent culture.
[0018] In certain embodiments, the period of time is about 12 hours to 24 hours, 24 hours to 36 hours, 36 hours to 48 hours, up to 3 days, up to 4 days, or up to 5 days.
[0019] In certain embodiments, the conditions include: one concentration within a range of concentrations of the compound, and / or with or without combination with a second therapeutic agent.
[0020] In certain embodiments, the second therapeutic agent comprises a cytokine (e.g., IL-2, TNF), TCR stimulation (e.g., via CD3 cross-linking), and / or co-stimulation (e.g., CTLA, B7, CD28, etc.).
[0021] In certain embodiments, the compound is an immunomodulatory antibody.
[0022] In certain embodiments, step (2) is performed by a microfluidic device or a droplet chip device (eg, a device capable of performing phenotypic / genotypic assays).
[0023] In certain embodiments, by isolating, purifying or enriching for a specific T cell subset or population (e.g., CD4 + or CD8 + T cells) to separate, purify or enrich the single T cells.
[0024] In certain embodiments, the single T cells are isolated, purified or enriched by isolating, purifying or enriching extensively for T cells.
[0025] In certain embodiments, the functional properties include: IFN-γ secretion, secretion of markers associated with T cell response or activation, and / or upregulation of activation markers.
[0026] In certain embodiments, the gene set includes proliferation-related genes (such as Ki67); stem-like feature-related genes (such as TCF7); T cell activation-related genes (such as CD25, granzyme B, perforin, CD28, TNF, IL2, IFNG, 4-1BB, CD38, CD69, OX-40, GITR, IL4, IL6); T cell exhaustion-related genes (CD39, CTLA-4, EOMES, PD-1, TIGIT, 2B4, LAG-3, T-bet, TIM3, TOX, CD160, IL-10); migration-related genes (such as CD31, CD103, CXCR5, CXCR4, CCR7, CCR3, CCR4, CCR8, CCR5, CXCR3); and combinations thereof.
[0027] In certain embodiments, the gene panel further comprises CD4, CD8, FOXP3, CD62L, CD44, CD127, CD27, IL33R and / or PTPRC.
[0028] In certain embodiments, the gene set includes CST7 (optionally with a weight of 10.8), GZMA (optionally with a weight of 10.7), GZMB (optionally with a weight of 8.2), GZMK (optionally with a weight of 11.3), HAVCR2 (optionally with a weight of 10.7), IFNG (optionally with a weight of 10.0), LAG3 (optionally with a weight of 11.2), NKG7 (optionally with a weight of 9.9), PRF1 (optionally with a weight of 9.9), and TMEM155 (optionally with a weight of 10.8). The optional weights of the individual genes in the PRB gene set typically fall within the range of 10±2 (or 10±1), and can vary within ±10%-20% of the preferred value without significant loss of predictive accuracy.
[0029] In certain embodiments, in step (2), the functional property is used to physically separate the responder T cells and non-responder T cells based on the presence or absence of the functional property using a microfluidic device or a droplet chip device.
[0030] In certain embodiments, in step (2), the functional property is an activation marker or a response marker (either or both of which can be determined, for example, by single-cell RNA sequencing), and wherein the responder T cells and the non-responder T cells are not physically separated, but are actually distinguished based on the expression or lack of expression of the activation marker or response marker.
[0031] In certain embodiments, the method further comprises characterizing the pre-treatment sample and / or the post-treatment sample to characterize the sample, for example, in terms of cell type composition (eg, via flow cytometry) and target expression on T cells.
[0032] Another aspect of the present invention provides a method for selecting a patient for treatment of cancer (e.g., renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL) or uterine cancer (UC)) using a T cell-targeting immunomodulatory compound (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3) and / or an anti-OX40 agonist antibody (such as HFB10)), the method comprising: obtaining a clinical response score for the patient by evaluating the expression status and / or expression level of a gene in a predictive response biomarker (PRB) for cancer obtained by the method of the present invention to determine whether the clinical response score of the patient exceeds a predetermined threshold clinical response score; wherein a patient whose clinical response score exceeds the threshold score is identified as benefiting from the treatment and is selected for the treatment, and / or a patient whose clinical response score is below a second threshold score is identified as not benefiting from the treatment and is not selected for the treatment.
[0033] Another aspect of the present invention provides a method for treating a patient's cancer (e.g., renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL) or uterine cancer (UC)) using an immunomodulatory compound targeting T cells (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3) and / or an anti-OX40 agonist antibody (such as HFB10)), the method comprising: (i) obtaining a clinical response score for the patient by evaluating the expression status and / or expression level of a gene in a predictive response biomarker (PRB) for a disease (e.g., cancer) obtained by the method of the present invention to determine whether the clinical response score of the patient exceeds a predetermined threshold clinical response score; wherein a patient whose clinical response score exceeds the threshold score is identified as being beneficial to the treatment and is selected for the treatment; and, (ii) administering a therapeutically effective amount of the compound to the patient identified as being beneficial to the treatment and selected for the treatment in (i).
[0034] In certain embodiments, the cancer is renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL), or uterine cancer (UC).
[0035] In certain embodiments, immunomodulatory compounds targeting T cells include anti-PD-1 antagonist antibodies, anti-PD-L1 antagonist antibodies, anti-TNFR2 agonist antibodies (such as HFB3), and / or anti-OX40 agonist antibodies (such as HFB10).
[0036] In certain embodiments, the immunomodulatory compound targeting T cells is an anti-PD-1 antagonist antibody or an anti-PD-L1 antagonist antibody, and wherein the PRB includes CST7 (optionally weighted at 10.8), GZMA (optionally weighted at 10.7), GZMB (optionally weighted at 8.2), GZMK (optionally weighted at 11.3), HAVCR2 (optionally weighted at 10.7), IFNG (optionally weighted at 10.0), LAG3 (optionally weighted at 11.2), NKG7 (optionally weighted at 9.9), PRF1 (optionally weighted at 9.9), and TMEM155 (optionally weighted at 10.8). The optional weights of individual genes in the PRB gene set typically fall within the range of 10±2 (or 10±1), and can vary within ±10%-20% of the preferred value without losing significant predictive accuracy.
[0037] In certain embodiments, the anti-PD-1 antagonist antibody comprises pembrolizumab, nivolumab, cemiplizumab, dotalimumab, spartalizumab, carrelizumab, sintilimab, tislelizumab, toripalimab, INCMGA00012, JTX-4014, AMP-224, or AMP-514; and / or the anti-PD-L1 antagonist antibody comprises atezolizumab, avelumab, durvalumab, or KN035.
[0038] It should be understood that any embodiment of the invention, including those described only in the examples or claims, may be combined with any one or more embodiments of the invention, unless expressly disclaimed or inappropriate. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 is a non-limiting schematic diagram showing one embodiment of the present invention.
[0040] Figure 2 is a non-limiting schematic diagram showing another embodiment of the present invention.
[0041] Figure 3 is an exemplary list of genes in a predetermined gene panel that can be used to characterize T cell responses to treatment with immunomodulators, such as OX-40 agonists. The gene panel approach captures distinct T cell phenotypes while facilitating high throughput screening.
[0042] Figure 4 This is a volcano plot showing the expansion of CD8 + T cells with unexpanded CD8 + Genes that are differentially expressed (significantly increased or significantly decreased) in T cells.
[0043] Figure 5A In the atezolizumab arm of the POPLAR trial, survival analysis revealed that 47 patients with high CD8 T cell expansion markers (Sig Hi ) patients responded lower than the 46-position marker (Sig Lo Based on the patients' marker scores, the patients were divided into two groups using the median cutoff value.
[0044] Figure 5B In the atezolizumab arm of the IMvigor210 trial, survival analysis revealed that 174 patients with high CD8 T cell expansion markers (Sig Hi ) of patients with a lower response than the 174-position marker (Sig Lo Based on the patients' marker scores, the patients were divided into two groups using the median cutoff value.
[0045] Figure 6 Show the classification of expanded and unexpanded T cell subsets from patients treated with ICB. 12,551 T cells from BCC (a, c, e) and 14,522 T cells from SCC (b, d, f) were clustered and visualized based on the similarity of their gene expression profiles using a dimensionality reduction algorithm. The association of single cells with CD8+, CD4+ or regulatory T cell subsets was identified based on marker gene expression (a, b), and the association of single cells with patient identifiers was indicated (c, d). T cells that were amplified in response to anti-PD-L1 treatment were identified based on scTCR-seq data and marked in red. All other T cells are displayed in gray (e, f).
[0046] Figure 7 Differentially expressed genes that distinguish T cells that will expand after PD-1 blockade based on their pre-treatment gene expression profiles are shown. Samples were divided into BCC (a, b) and SCC (c, d) and CD8+ (a, c) and CD4+ T cells (b, d). In order to select the most clearly defined differentially expressed genes, we removed genes with high background expression, exceeding the fold change and statistical significance thresholds (gray lines). The resulting list of differentially expressed genes is named and highlighted in color, where blue means downregulation of the amplified fraction and red means upregulation of the amplified fraction.
[0047] Figure 8 A graphical representation of the score values of the identified predictive gene expression markers as an overlay on the single cell data visualization is shown. Two different markers were identified, a CD8+ T cell marker in BCC patient samples that was a positive predictor of expansion (a) and a CD4+ T cell marker in SCC patient samples that was a negative predictor of expansion (b). The marker score values are relative, unitless and color-coded, with score values shown in red associated with a high likelihood of expansion and blue indicating a lower likelihood of expansion.
[0048] Fig. 9 Survival analysis of clinical trial data using predictive markers is shown.The Kaplan-Meier graph of progression-free survival (PFS) of the patient treated with atezolizumab in each clinical trial is shown, wherein the patient is divided by its predictive marker score of the median value of all patients in the corresponding clinical trial higher than (Sig+) or lower than (Sig-).The observed results of the review are marked with a plus sign.The hazard ratio (*pval<0.05, **pval<0.01, ***pval<0.001) from the Cox proportional hazard model analysis is shown.In b, based on the two marks shown in a, the patient is divided into four groups (low-low, low-high, high-low and high-high), wherein the CD8 marker is first indicated.The hazard ratio and significance relative to the low-low group are indicated.
[0049] Fig.10 The results of the logistic regression model training for deriving gene expression markers are shown, and these results can best predict the T cell expansion after PD-1 blockade. The figure shows the results of training a logistic regression model to predict the T cell expansion after PD-1 blockade at the single cell level. Based on analyzing differential gene expression, independent markers have been derived for each T cell subset. For each subgroup, the threshold of the differential gene expression fold change is changed to study whether a more stringent cutoff value will affect the prediction performance. The prediction performance is measured by accuracy (the correct prediction ratio of true positive and true negative in the total number of cases), sensitivity (true positive rate) and specificity (true negative rate).
[0050] Fig.11 Comparison of hazard ratios between the experimental and comparator arms in biomarker positive (BM+) and negative (BM-) populations is shown. Hazard ratios and two-sided statistical tests from Cox proportional hazard models for both groups of patients are shown, with significance indicated (*pval<0.05, **pval<0.01, ***pval<0.001). pval is 0.06.
[0051] Fig.12 T cell subsets from pre-treatment BCC and SCC samples are shown, identifying the patient's source and expansion status. Based on the similarity of their gene expression profiles, cells were clustered and visualized using a dimensionality reduction algorithm. The association of single cells with patient identifiers is indicated (ad). T cells that have expanded in response to anti-PD-L1 therapy have been identified based on scTCR-seq data and are marked in red, with all other T cells displayed in gray (eh).
[0052] Fig.13 The workflow for inferring predictive gene expression signatures from single-cell data is shown in the flow chart, which illustrates the steps in the workflow for inferring predictive gene expression signatures from scRNA-seq data.
[0053] Fig.14 A workflow for applying predictive gene expression signatures to bulk RNA-seq data from clinical trials is shown in a flowchart illustrating the steps in a workflow for applying predictive gene expression signatures to bulk RNA-seq data from clinical trials.
[0054] Fig.15 Gene expression of FOXP3 (a) and GZMA (b) contained in the CD4- non-expansion marker identified from CD4+ T cells in SCC patient samples is shown. The expression intensity of each gene is indicated by a color scale. Clusters identified as Tregs are indicated by circular dotted lines.
[0055] Fig.16 Comparison of hazard ratios between biomarker positive and negative groups is shown. Hazard ratios and two-sided statistical tests from Cox proportional hazards models for both groups of patients are shown, with significance indicated (*pval<0.05, **pval<0.01, ***pval<0.001).
[0056] Fig.17 is an illustrative overview of the experimental procedures in Example 3.
[0057] Fig.18 Figure 2 is a typical computational analysis workflow diagram showing the general steps involved in obtaining a PRB and validating it against clinical data.
[0058] Figures 19A-19D We show that single-cell characterization of T cells from ex vivo cultures captures T cell biology that is relevant to the TME and specific to therapy. Fig.19A Shown are UMAP embeddings of scRNA-seq transcriptomes of 156,522 cells treated in ex vivo culture with isotype control, αCD3, αPD1, HFB200301, or HFB301001. Cell type labels based on transfer learning from a scRNA-seq tumor microenvironment reference dataset of 362,490 cells across 10 tumor indications (not shown). Fig.19B 54,215 T cells (from Fig.19A UMAP embedding of scRNA-seq transcriptomes of immediate subpopulations of cells. Identification of cell types. Fig.19C Shows that Fig.19B Identification of tumor indications using the same UMAP embeddings. Fig.19D In the figure, T cells from each treatment are colored red to show their relative distribution upon UMAP embedding.
[0059] Figure 20A-20C Based on the characterization of T cells responding to treatment ex vivo, the responding cells were characterized by activation markers CD8 + T cells and consistent with the large TCR clonotype. Fig. 20A Shows Fig.19B UMAP embedding of the scRNA-seq transcriptome of T cells as seen in Figure 2. Leiden clusters were identified. Clusters 3, 4, 10, 14, 16, and 18 correspond to activated T cells and are circled in red. Fig. 20Bis a box plot showing the expression level of activation gene signatures for each Leiden cluster, calculated as activation score. The activation score is a composite of published markers defining primary human T cells activated by αCD3 and αCD28 in culture (Szabo et al.). Clusters with median activation scores above cluster 18 (a group of αCD3 activated T cells) are considered activated and are marked with an asterisk. Fig. 20C is with Fig. 20A The same UMAP embedding with T cells from different sized clonotypes in red to show their relative distribution in the UMAP embedding. Clonotype size is expressed as a percentage of the total number of T cells with TCR reads from the starting sample. Activated T cell clusters are circled in red and coincide with large TCR clonotypes.
[0060] Fig.21 The relative distribution of responsive (R) and non-responsive (NR) T cells to αPD1, HFB3, and HFB10 (red) upon UMAP embedding is shown. Fig.19B UMAP embedding of T cells as seen in αPD1 treated T cells: R=451, NR=240. HFB3 treated T cells: R=735, NR=355. HFB10 treated T cells: R=740, NR=291.
[0061] Figure 22A-22B Predictive response biomarkers (PRBs) for response to αPD1, HFB3, and HFB10 are shown. Fig.22A Depicted are heat maps showing log-normalized and z-scored expression of genes from αPD1, HFB3, and HFB10 biomarkers (columns) generated from differential gene expression analysis of R cells versus NR cells (rows). Fig. 22B It is a Venn diagram showing the number of genes shared by the three PRBs. DETAILED DESCRIPTION
[0062] In one aspect, the invention described herein provides a method of predicting clinical response to a treatment such as an anti-cancer T cell-based immunotherapy based on checkpoint blockade (e.g., anti-PD-L1 or anti-PD-1 therapy) based on the differences in gene expression profiles between T cell clonotypes that ultimately expand in a tumor following the immunotherapy and T cell clonotypes that do not ultimately expand in the tumor following the immunotherapy.
[0063] A notable feature of the present invention is the identification of sets of gene signatures or predictive response biomarkers (PRBs) in populations of untreated effector cells, such as T- / B-cells, based in part on single cell profiling.
[0064] One advantage of single-cell feature analysis is that single-cell resolution improves statistical power by exploiting heterogeneity within patient samples, which typically contain at least millions (if not tens of millions) of heterogeneous cells, including disease (e.g., cancer) cells, immune cells (T cells, B cells, NK cells, macrophages, neutrophils, etc.). The complexity of disease tissue and the microenvironment of the disease (such as the tumor microenvironment or TME) are at least partially maintained during treatment; at the same time, the specific response in a single target cell such as an immune cell that is a target of an immunomodulatory drug (e.g., an immune checkpoint inhibitor) can be evaluated separately.
[0065] A direct benefit of increased statistical power is a significant reduction in the number of patients required to detect PRB.
[0066] In certain embodiments, samples suitable for the methods of the present invention (i.e., for identifying PRB and treating patients based on PRB) can be readily available body fluids, such as blood samples, which further enhances the ease with which the methods of the present invention can be performed. In addition, the transport of target cells (e.g., immune cells) between the TME and PBMC compartments makes it possible that blood-based biomarker screening can enhance clinical success.
[0067] Another distinctive feature of the present invention is the identification of PRBs from untreated samples, which facilitates the discovery of biomarkers that would not otherwise be identified based on traditional methods, which typically rely on differences in expression or activity in treated samples compared to controls.
[0068] With respect to the use of PRBs identified using the methods of the present invention, based on the results of single cell analysis, the identified PRBs can be implemented in patient stratification using more readily available techniques such as immunohistochemistry (IHC), bulk qPCR or gene panels.
[0069] The invention described herein is particularly suitable for identifying PRBs in immune cells, such as T cells or B cells, but the same process can be easily adapted to other target cells or effector cells for therapy.
[0070] Therefore, in one aspect, the present invention provides a method for identifying predictive response biomarkers (PRBs) for treating a disease or condition (e.g., cancer) using immunomodulatory therapy targeting T / B cells, the method comprising the following steps: (a) in a pre-treatment sample of the disease (e.g., cancer, autoimmune disease, or neurodegenerative disease), identifying responsive T / B cells having the same TCR (T cell receptor) / BCR (B cell receptor) clonotype as the TCR (T cell receptor) / BCR (B cell receptor) clonotype of clonally expanded T / B cells in a matched post-treatment sample of the disease (e.g., cancer, autoimmune disease, or neurodegenerative disease), wherein the clonally expanded T / B cells in the matched post-treatment sample are clonally expanded after treatment; and, (b) generating a list of genes that are upregulated and / or downregulated in the responsive T / B cells in the pre-treatment sample to produce a PRB; (c) optionally, weighting each gene in the PRB using a logistic regression coefficient obtained by fitting each of the genes as a predictor to an amplification status and / or a clinical response status as an outcome.
[0071] As used herein, "predictive response biomarkers (PRBs)" include genes whose expression or lack of expression can be used to predict whether a target cell (such as a T cell or B cell in an immunomodulatory therapy) or a patient containing the target cell will respond to a specific treatment. A PRB can consist of one gene, or can include a collection of genes, which can be collectively referred to as a "gene signature."
[0072] The disease or condition may include any disease or condition that can be treated by immunomodulatory therapy such as a small molecule compound or antibody (e.g., anti-PD-1Ab or anti-PD-L1 Ab). For example, the disease may be cancer, such as those in the embodiments, including BCC, SCC, NSCLC, RCC, metastatic urothelial carcinoma, etc.
[0073] As used herein, "clonotype" refers to the specific pairing of BCR or TCR chains on B cells and T cells, respectively, because each mature B cell and T cell expresses a BCR or TCR on their surface that can potentially recognize an antigen. Such B cells and T cells can be cloned and expanded under the correct conditions, and after binding to the antigen, all B cells and T cells that are the offspring of such clone expansion share the same BCR / TCR pairing, which distinguishes them from the BCR / TCR on other B / T cells.
[0074] "Logistic regression" is a statistical model that, in its basic form, uses a logistic function to model a binary dependent variable. In regression analysis, logistic regression estimates the parameters of a logistic model, which is a form of binary regression. Mathematically, a binary logistic model has a dependent variable with two possible values labeled "0" and "1".
[0075] In certain embodiments, step (a) comprises: (1) generating the matched post-treatment sample by contacting an ex vivo culture of an untreated sample of a disease (e.g., cancer) with an immunomodulatory compound targeting T cells under conditions and for a period of time sufficient to allow the immunomodulatory effect of the compound on the T cell population in the ex vivo culture to manifest; (2) encapsulating single T cells isolated, purified, or enriched from the ex vivo culture into picoliter droplets for single-cell characterization of functional properties, thereby classifying each encapsulated single T cell as either a single T cell or a single T cell based on the presence or absence of the functional property, respectively. (3) determining the TCR clonotype of each encapsulated single T cell in the first responding T cell library and the second non-responding T cell library, thereby identifying the TCR clonotype of the responding T cell as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample; (4) encapsulating the single T cells isolated, purified or enriched from the pre-treatment sample into picoliter droplets to identify the responding T cells having the same TCR clonotype as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample.
[0076] The T cell population may include all T cells or T cell subsets, such as helper CD4 + T cells (including Th1, Th2, Th9, Th17 and Tfh cells), cytotoxic CD8 + T cells (Tc cells, CTL, T killer cells, killer T cells), memory T cells (T CM Cells, T EM Cells and T EMRA Cells, T RM cells and virtual memory T cells), regulatory CD4 + T cells (FOXP3 + Treg cells and FOXP3 - Treg cells), natural killer T cells (NKT), γδ T cells, all of which can be selected based on characteristic cell surface markers, secreted cytokines and / or key transcription factors.
[0077] For the same disease or indication, different target cells (e.g., T cell subsets) can produce different PRBs.
[0078] In certain other embodiments, step (a) comprises: (1) obtaining the pre-treatment sample and the matched post-treatment sample of the disease (e.g., cancer) from a suitable donor, wherein the matched post-treatment sample has been contacted with the compound under conditions and for a period of time sufficient to allow the immunomodulatory effect of the compound on the T cell population in the post-treatment sample to manifest; (2) encapsulating the single T cells isolated, purified or enriched from the post-treatment sample into picoliter droplets for single-cell characterization of functional properties, thereby characterizing each encapsulated single T cell based on the presence or absence of the functional property, respectively. The cells are divided into a first responding T cell library and a second non-responding T cell library; (3) determining the TCR clonotype of each encapsulated single T cell in the first responding T cell library and the second non-responding T cell library, thereby identifying the TCR clonotype of the responding T cell as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample; (4) encapsulating the single T cells isolated, purified or enriched from the pre-treatment sample into picoliter droplets to identify the responding T cells having the same TCR clonotype as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample.
[0079] Many microfluidic devices can be used to generate picoliter droplets to encapsulate cells for single-cell characterization. See, e.g., Gerard et al., Nat Biotechnol 38:715-721, 2020 (incorporated by reference).
[0080] In certain embodiments, step (b) comprises: (5) using (whole transcriptome) single cell RNA sequencing (scRNA-seq) to identify the list of genes upregulated in the responsive T cells in the pre-treatment sample, wherein each gene in the PRB has a log2 fold change > 0.2 and is expressed in < 40% of T cells with a TCR clonotype that is not clonally expanded in the matched post-treatment sample.
[0081] Single-cell transcriptome sequencing (or scRNA-seq for short) provides expression profiles of individual cells. Compared to standard methods for analyzing RNA expression from large cell populations, such as microarrays and bulk RNA-seq analysis (which can obscure key differences between individual cells within these populations), scRNA-seq can identify gene expression patterns through gene cluster analysis. See Eberwine et al., Nature Methods. 11(1): 25–27, 2014 (incorporated by reference).
[0082] In certain embodiments, PRB is associated with a positive treatment outcome (e.g., the presence of PRB predicts that treatment is favorable or successful). For example, in the Examples section below, it is shown that activated CD8+ T cell-associated PRB predicted positive clinical outcomes (eg, longer progression-free survival).
[0083] In certain embodiments, PRB is associated with a negative treatment outcome (e.g., the presence of PRB predicts that treatment will be unfavorable or unsuccessful). For example, in the Examples section below, it is shown that activated CD4 + T cell-associated PRB predicted negative clinical outcomes (eg, shorter progression-free survival).
[0084] In certain embodiments, scRNA-seq is performed using a gene panel specifically designed for the mechanism of action of the compound or a predetermined gene panel designed to assess immune modulation.
[0085] In certain embodiments, the gene panel includes genes indicative of T cell activation, exhaustion, migration and / or proliferation.
[0086] In certain embodiments, the pre-treatment sample is a blood sample.
[0087] In certain embodiments, the disease is cancer, such as a solid tumor.
[0088] In certain embodiments, the ex vivo culture is freshly isolated from diseased tissue.
[0089] In certain embodiments, the ex vivo culture is established from stored (eg, frozen) diseased tissue.
[0090] In certain embodiments, the ex vivo culture is a single cell suspension.
[0091] In certain embodiments, the ex vivo culture is an adherent culture.
[0092] In certain embodiments, the period of time is about 12 hours to 24 hours, 24 hours to 36 hours, 36 hours to 48 hours, up to 3 days, up to 4 days, or up to 5 days.
[0093] In certain embodiments, the conditions include: one concentration within a range of concentrations of the compound, and / or with or without combination with a second therapeutic agent.
[0094] In certain embodiments, the second therapeutic agent comprises a cytokine (e.g., IL-2, TNF), TCR stimulation (e.g., via CD3 cross-linking), and / or co-stimulation (e.g., CTLA, B7, CD28, etc.).
[0095] In certain embodiments, the compound is an immunomodulatory antibody.
[0096] In certain embodiments, step (2) is performed by a microfluidic device.
[0097] In certain embodiments, by isolating, purifying or enriching for a specific T cell subset or population (e.g., CD4 + or CD8 + T cells) to isolate, purify or enrich single T cells.
[0098] In certain embodiments, single T cells are isolated, purified or enriched by isolating, purifying or enriching extensively for T cells.
[0099] In certain embodiments, the functional property includes secretion of cytokines (eg, IFN-γ) and / or upregulation of activation markers.
[0100] In certain embodiments, the gene set includes proliferation-related genes (such as Ki67); stem-like feature-related genes (such as TCF7); T cell activation-related genes (such as CD25, granzyme B, perforin, CD28, TNF, IL2, IFNG, 4-1BB, CD38, CD69, OX-40, GITR, IL4, IL6); T cell exhaustion-related genes (CD39, CTLA-4, EOMES, PD-1, TIGIT, 2B4, LAG-3, T-bet, TIM3, TOX, CD160, IL-10); migration-related genes (such as CD31, CD103, CXCR5, CXCR4, CCR7, CCR3, CCR4, CCR8, CCR5, CXCR3); and combinations thereof.
[0101] In certain embodiments, the gene panel further comprises CD4, CD8, FOXP3, CD62L, CD44, CD127, CD27, IL33R and / or PTPRC.
[0102] In certain embodiments, in step (2), the functional property is used to physically separate the responder T cells from the non-responder T cells using a microfluidic device based on the presence or absence of the functional property.
[0103] In certain embodiments, in step (2), the functional property is an activation marker or a response marker (either or both of which can be determined, for example, by single-cell RNA sequencing), and wherein the responder T cells and the non-responder T cells are not physically separated, but are actually distinguished based on the expression or lack of expression of the activation marker or response marker.
[0104] In certain embodiments, the method further comprises characterizing the pre-treatment sample and / or the post-treatment sample to characterize the sample, for example, in terms of cell type composition (eg, via flow cytometry) and target expression on T cells.
[0105] In view of Figure 1 and Figure 2 The invention described herein may be better illustrated by the following description of non-limiting, exemplary embodiments depicted in the schematic diagrams of FIG.
[0106] Figure 1 An exemplary embodiment is described, wherein fresh, untreated patient samples of a disease (such as cancer) are obtained for the purpose of identifying predictive response biomarkers (PRBs), which can be used to stratify patients for a specific treatment to ensure the maximum chance of therapeutic efficacy. This example focuses on T cells as target cells for potential immunomodulatory drugs (e.g., anti-PD-1 or anti-PD-L1 antibodies), but the same method can be applied to other cells, such as B cells. For simplicity, the description focuses on T cells, which are not restrictive.
[0107] Specifically, Figure 1 Exemplary procedures are described that can be used to identify predictive response biomarkers (PRBs) for immunomodulatory drugs targeting T cells, such as agonist antibodies specific for immunomodulatory receptors expressed on the surface of target T cells (e.g., OX40), using ex vivo cell culture, single cell screening, and matched TCR clonal typing and phenotyping.
[0108] Prior to entering the experimental workflow (AF), the process begins with the collection of tissue from cancer patients with accessible solid tumor lesions, such as by resection or biopsy.
[0109] In (A) preparation of ex vivo cultures: Fresh tissue is dissociated into a single cell suspension, which can optionally be further characterized using cell composition profiling, such as by flow cytometry, to characterize the cell type composition of the sample. Target expression on T cells (e.g., OX40 receptor expression) can also be assessed at this stage. Several aliquots of the sample can optionally be frozen and stored for future repeat experiments. The remaining dissociated cells can then be distributed under one or more experimental conditions.
[0110] (B) more than one experimental condition can be tested sequentially or in parallel. Experimental conditions can include control or untreated conditions (NT), for example, by using isotype-matched controls for any antibody-based treatment conditions. Experimental conditions can also include treatment conditions (Tx) using immunomodulatory drugs, such as immunomodulatory antibodies. This can include different concentrations alone or in combination with other drugs, or supplementary conditions related to T cells, such as cytokines (e.g., IL-2, TNF), TCR stimulation (e.g., using anti-CD3 cross-linking) or co-stimulation (e.g., CD28 stimulation). Regardless of the specific conditions used for specific treatment, the ex vivo cell culture is incubated with the selected treatment conditions for a sufficient time so that the treatment acts on the target cells (T cells), and then the activity readings on the single cells are evaluated. For example, for gene expression or early activation markers, the typical incubation time can be in the range of 12-48 hours, but cells can be cultured up to 3-5 days to make typical phenotypic readings appear.
[0111] (C) T cell enrichment: After sufficient time for immunomodulatory drugs (e.g., anti-OX40 agonist antibodies) to induce activity on target cells (T cells) via their mechanism of action, for each experimental condition, target cells are enriched from the heterogeneous cell population in ex vivo cell culture for further characterization at the single cell level. Any and all T cells in the cells can be broadly enriched, or specific T cell subsets, such as CD4 + or CD8 + T cells. Subpopulations can be analyzed individually and the results can be considered together. Regardless of the enrichment chosen, the enriched T cells then enter single cell characterization analysis.
[0112] (D) For each treatment (Tx) condition, a single enriched target cell (T cell) can be encapsulated in a picoliter droplet using a microfluidic device and one or more functional properties of each target cell, such as IFN-γ secretion or upregulation of activation markers, can be determined. Based on the presence or absence of such functional properties, the target cells (T cells) can be physically or virtually sorted into a first responder pool (1) (responding Tx cells) and a second non-responder pool (2) (non-responding Tx cells) for a specific treatment condition. At the same time, the untreated (NT) condition sample can be similarly enriched with the same target cells (e.g., T cells) and similarly encapsulated in picoliter droplets using a microfluidic device (3).
[0113] (E) TCR clonal typing and / or single cell RNA sequencing (e.g., using gene panels): Each single target cell (T cell) under Tx conditions (including physically or virtually sorted fractions (1) and (2) and NT conditions (3)) can now be subjected to TCR clonal typing to identify their respective association with clonotypes. At the same time, each single target cell (T cell), especially those under NT conditions (3), can be characterized using single cell RNA sequencing, optionally by using a gene panel specifically designed for the respective mechanism of action of the drug under study (such as an anti-OX40 agonist antibody). Figure 3 An exemplary gene panel for OX40 agonist antibodies as immunomodulatory drugs is shown in . Thus, each TCR clonotype under NT conditions (3) is associated with an expression profile in terms of the gene panel (or with a genome-wide expression profile if a gene panel is not used).
[0114] (F) Data analysis. Solid tumors are known to contain clonally expanded T cells that are specifically associated with anti-tumor antigens, and it is believed that these expanded T cell clonotypes are important cells responsible for anti-tumor activity. However, such T cells are often found to be suppressed in the tumor microenvironment, which may be due to checkpoint inhibition through, for example, the PD-1 / PD-L1 pathway, which is due to the presence of PD-L1 expressed by tumor cells. Therefore, immune checkpoint blockers (such as anti-PD-1 antibodies or anti-PD-L1 antibodies) or other immunomodulatory drugs (such as anti-OX40 agonist antibodies) can be used to alleviate such inhibition to re-stimulate dormant T cells.
[0115] While not wishing to be bound by any particular theory, it is believed that for the purpose of identifying PRBs, the T cell clonotypes expanded in the responder fraction (1) (consisting of Figure 1 T1-T3 in Figure 2 are important drivers of anti-tumor activity as a result of drug treatment. However, on the other hand, there will be non-responding T cell clonotypes (described by Figure 1 By comparing the clonotypes in fraction (1) and fraction (2), it is possible to identify / assign clonotypes that ultimately do respond to drug treatment (such as those in T1-T3) and clonotypes that ultimately do not respond to drug treatment (such as those in T4-T7).
[0116] Importantly, since all T cells of all clonotypes are randomly distributed into the Tx and NT experimental conditions in (A), important responding clones will be expanded and present multiple times in the original sample, including in the never treated sample under the NT condition (3). Responders in NT (3) have the same clonotypes ultimately identified as responders in Tx (1), with the difference that such responders in NT (3) have not been previously exposed to the treatment condition Tx. The other clonotypes in NT (3) are presumably those of non-responders.
[0117] Thus, according to the method of the invention, responding and non-responding clonotypes under NT conditions (3) can be identified even if no T cells under NT conditions were previously exposed to the therapeutic condition Tx.
[0118] By comparing the gene expression profiles between cells with clonotypes associated with responders to drug treatment (illustrated by T1-T3) and cells with clonotypes associated with non-responders to drug treatment (illustrated by T4-T7) in (3), markers specific to responding T cell clonotypes can be identified as PRBs (predictive response markers) to drug treatment before they are treated with the therapeutic condition Tx.
[0119] In certain embodiments, the PRB includes one or more genes whose expression is significantly increased or significantly decreased in responding (e.g., T) cells compared to non-responding (e.g., T) cells. Figure 4 .
[0120] In certain embodiments, the log2 fold change of one or more (e.g., substantially all or all) genes in a PRB is >0.2, >0.3, >0.4, >0.5, >0.6, >0.7, >0.8, >0.9, >1 (2-fold) or more.
[0121] In certain embodiments, one or more (e.g., substantially all or all) genes in the PRB are expressed in <5%, 10%, 20%, 30%, <40%, <50%, <60% of T / B cells having a TCR / BCR clonotype that is not clonally expanded in a matched post-treatment sample.
[0122] In certain embodiments, all genes in the PRB have a log2 fold change > 0.2 and are expressed in < 40% of T / B cells with TCR / BCR clonotypes that are not clonally expanded in matched post-treatment samples.
[0123] In certain embodiments, the gene in PRB is weighted separately based on the individual prediction value of gene.For example, some differentially expressed (for example, based on threshold log2 multiple values, and the prevalence of expression % value in responder vs. non-responder, significantly increased or decreased in responder compared with non-responder) genes can be strongly related to desired biological function or outcome (such as T cell amplification, or patient responsiveness to treatment). Such genes carry high weights (for example, close to 1 in a scale of 0 to 1, 0 is no weight, 1 is the maximum weight). On the contrary, some genes can have a relatively low correlation with desired biological function or outcome, so that the expression of such genes increases only with, for example, 30% of favorable biological outcomes. Such genes, although still part of PRB, can be given a lower weight (for example, 0.3 in a scale of 0-1). According to this weight distribution method, the gene whose expression increases and is anti-correlated with desired biological function or outcome is given a negative weight (for example, -0.6). Such a PRB having each or most of its genes weighted according to their respective predictive values may be more accurate for prediction compared to a corresponding PRB in which each gene is given equal weight, regardless of the predictive value of individual genes for a specific biological outcome.
[0124] The weights are assigned to the genes in the PRB and can be performed as part of a machine learning process, such as a machine learning based on a logistic regression classifier. In this embodiment, a training data set can be provided to include the characteristics of the genes in the PRB, and / or their respective expression levels (as absolute values in responders, or relative values compared to non-responders), and associated biological outcomes (e.g., clonal expansion or non-clonal expansion of T / B cells with / without such increased / reduced gene expression, responsiveness or non-responsiveness of the patient to treatment, etc.). In view of the many (e.g., tens of thousands, if not hundreds of thousands) available T / B cells that can provide such training data sets, the weighting of the individual genes in the PRB and the predictive efficacy of the PRB can be highly accurate and statistically powerful.
[0125] Alternatively or additionally, cells obtained from matched blood samples from the same donor can be assayed using the same process. In the resulting data (4), clonotypes that ultimately do respond to drug treatment from the earlier step (illustrated by T1) can be identified again. Based on corresponding gene expression signatures in the blood (similar to the NT condition (3) described above), PRBs can be identified based on blood samples, which enables the method of the invention to be implemented more practically than corresponding methods using tissue samples (such as biopsies) from diseased tissues (such as tumors).
[0126] As described above, fractions (1) and (2) can be physically separated based on predetermined functional properties by using a microfluidic device with sorting capabilities. For example, the functional properties can be the expression of cytokines such as IL-12 or IFN-γ in T cells, or the secretion of antibodies by B cells, or the expression of cell surface markers, all of which can be labeled with fluorescent signals or dyes that can be used for sorting.
[0127] Alternatively, physical sorting is not required if, for example, such expressed markers are not easily detectable or labeled for physical sorting. In this case, the responding fraction (1) and the non-responding fraction (2) can be virtually sorted based on the expression of certain genes or sets of genes in responders, and the lack of such expression or differences in the expression levels of such genes in non-responders. This can be achieved by using the same set of genes under NT conditions (3), or by using different sets of genes designed to optimally capture differences in gene expression between fractions (1) and (2).
[0128] In another embodiment, instead of using fresh disease tissue that has not been previously treated with a therapy (e.g., an immunomodulatory drug targeting T cells) to test a therapy in ex vivo culture, a treated (post-treatment) sample and a matched untreated (pre-treatment) sample can be similarly used without using ex vivo culture. This is particularly useful in cases where stored treated and matched untreated samples are available. Figure 2 is a description of this exemplary embodiment of identifying predictive response biomarkers (PRBs) for immunomodulatory drugs targeting T cells, such as anti-OX40 agonist antibodies, using pre- and post-treatment patient samples, single cell screening, and matched TCR clonal typing and phenotyping.
[0129] Similarly, the process starts with matched pre- and post-treatment tissues collected from patients with accessible lesions (such as solid tumors). Likewise, samples may be obtained using any conventional means such as resection or biopsy before they enter the experimental workflow (AC).
[0130] (A) Preparation of cells: Pre- and post-treatment tissues can be dissociated into single cell suspensions. Several aliquots of these samples can be frozen and stored for future replicates. Cells from each sample can be enriched for broad T cells or specific T cell subsets (e.g., CD4 + or CD8 +T cells). The enriched T cells can then proceed to the next single cell characterization step by using, for example, a microfluidic device that encapsulates a single target cell (e.g., T cell) into a picoliter droplet. As previously described, the enriched T cells can be virtually or physically sorted based on the presence or absence of a functional characteristic to produce a responder pool (1) and a non-responder pool (2). The pre-treatment sample produces the untreated condition NT (3).
[0131] (B) TCR clonal profiling and / or single-cell RNA sequencing (e.g., using gene panels): Cells from each sample can be characterized to identify the TCR clonotype of each single T cell in order to identify their respective associations with the clonotype. Additionally, cells from the samples, particularly cells from pre-treatment samples (NT(3) conditions), can be characterized using single-cell RNA sequencing, by using whole transcriptome sequencing, or by using more focused gene panels specifically designed for the respective mechanism of action of the drug under investigation.
[0132] (C) Data Analysis. As previously described, based on data from post-treatment samples, responder T cells (1) or non-responder T cells (2) can be identified based on the presence or absence of functional properties, such as their gene expression patterns, for example, expression of one or more activation markers (e.g., IL-2RA), cytokines (e.g., IFN-γ), and / or by increased clonal expansion compared to clonal typing of pre-treatment samples.
[0133] Solid tumors contain clonally expanded T cells that are specifically associated with anti-tumor antigens, and it is believed that these expanded T cell clonotypes are important cells responsible for anti-tumor activity. Therefore, for the purpose of the PRB identification method, it is believed that the expanded clonotypes in the responding fraction (1) (illustrated by T1-T3) are important drivers of anti-tumor activity as a result of drug treatment. On the other hand, non-responsive T cell clonotypes (illustrated by T4-T7) will be collected in the non-responsive fraction (2).
[0134] Importantly, since all T cells of all clonotypes are initially randomly distributed into the Tx (post-treatment) and NT (matched pre-treatment) experimental conditions in (A), and the important responding clones expand and are present multiple times in the original sample, responding and non-responding clonotypes can be identified from the pre-treatment sample (3). By comparing the gene expression profiles between the clonotypes associated with responders to drug treatment (illustrated by T1-T3) and the clonotypes associated with non-responders to drug treatment (illustrated by T4-T7) in (3), markers (PRBs) specific for responding T cell clonotypes can be identified before responders are subjected to the treatment conditions. Thus, according to the method of the present invention, the identified markers constitute predictive response biomarkers (PRBs) to drug treatment.
[0135] Alternatively or additionally, cells obtained from matched blood samples from the same donor can be assayed using the same process ( Figure 2 ). In the resulting data (4), clonotypes that did respond to the drug treatment from the earlier step can be identified again (illustrated by T1). Based on the corresponding gene expression signatures in the blood, it can be confirmed that the identified response signatures (PRBs) can be based on blood samples, which enables a more practical implementation of the subject PRBs for drug treatment compared to using tissue samples from the tumor.
[0136] Although the above description uses T cell stimulation by immunomodulatory drugs as an illustrative example, it can be easily envisioned that the methods of the present invention can be applied to other immune cells, including B cells and innate immune system effector cells, and can be applied to other treatment conditions, such as combinations of different drugs with different mechanisms of action, and / or different dosing regimens and treatment schedules.
[0137] Therefore, in a more general sense, the invention described herein provides a method for identifying PRBs for a particular treatment in pre-treatment samples from patients, and a method for using the PRBs so identified to predict treatment outcomes in individual patients, so that patient populations can be appropriately and effectively stratified for any particular treatment.
[0138] The method of the present invention utilizes a single cell-based method, which starts with collecting material / samples from the patient's disease site, for example, from a solid tumor tissue of a cancer patient. Optionally, matching PBMCs from the same patient can also be collected to facilitate the identification of potential blood-based biomarkers. Next, the tissue can be processed for in vitro culture and treated with one or more experimental conditions / drugs in parallel or sequentially. At this stage, several pre-tests can be optionally performed to better characterize the sample. For example, flow cytometry can be used to determine tissue composition to characterize the resident cell types (e.g., T cells in tumor tissue) in the tissue. Similarly, drug target expression on target cells can be quantified, and experimental conditions can be optimized in this pre-test step. After the in vitro culture of the tissue sample has been treated with experimental conditions / drugs, and after enough time has passed to induce any drug effect, the cell bank is harvested. At this point, conventional / mass analysis can be used to characterize the treatment outcome, such as flow cytometry, qRT-PCR or ELISA. Next, the target cells (such as T or B cells) are purified and enter the single cell analysis channel. At this stage, all experimental conditions are characterized at the single cell level to obtain readings that characterize therapeutic activity and to characterize the cell state under control conditions. For example, a microfluidic system can be used to screen functional activity readings for treatment, such as activation markers or secreted soluble factors such as cytokines. In addition, sequencing-based readings such as TCR or BCR clonal typing and gene expression readings can be used to characterize the activity and properties of cells. Finally, single cell data obtained from different experimental conditions and readings and data from conventional analysis enter computational analysis to identify predictive response biomarkers (PRBs) based on differences observed in treatment conditions compared to controls.
[0139] Another aspect of the present invention provides a method for selecting patients for treatment of a disease (e.g., cancer) using a T-cell targeting immunomodulatory compound, the method comprising: obtaining a clinical response score for the patient by evaluating the expression status and / or expression level of genes in a predictive response biomarker (PRB) for the disease (e.g., cancer) obtained by the method of the invention described herein to determine whether the clinical response score of the patient exceeds a predetermined threshold clinical response score; wherein patients whose clinical response scores exceed the threshold score are identified as being beneficial to the treatment and are selected for the treatment; and / or patients whose clinical response scores are below a second threshold score are identified as being unsuitable for the treatment and are not selected for the treatment.
[0140] Yet another aspect of the present invention provides a method for treating a disease (e.g., cancer) in a patient using an immunomodulatory compound targeting T cells, the method comprising: (i) obtaining a clinical response score for the patient by evaluating the expression status and / or expression level of genes in a predictive response biomarker (PRB) for the disease (e.g., cancer) obtained by any one of the methods of the present invention to determine whether the clinical response score of the patient exceeds a predetermined threshold clinical response score; wherein a patient whose clinical response score exceeds the threshold score is identified as being beneficial to the treatment and is selected for the treatment; and, (ii) administering a therapeutically effective amount of the compound to the patient identified in (i) as being beneficial to the treatment and selected for the treatment.
[0141] Example
[0142] Example 1
[0143] The examples described herein provide an illustration of one non-limiting embodiment of the present invention and are in no way limiting.
[0144] In a previous study, 4 cancer patients with SCC (squamous cell carcinoma) and 11 patients with BCC (basal cell carcinoma) were subjected to anti-PD-1 antibody therapy. For SCC patients, about an average of 31 days after treatment, and for BCC patients, about an average of 54 days after treatment, site-matched normal samples or untreated samples were obtained from the patients, and more than 26,000 T cells (TIL, tumor infiltrating lymphocytes) isolated from these samples were subjected to single-cell feature analysis. Specifically, single-cell RNA sequencing (scRNA-seq) and single-cell TCR sequencing (scTCR-seq) were performed on the isolated T cells from each sample to allow identification of TCR clonal types that have been amplified after treatment. The conclusion is that the amplified T cell clones are mainly composed of novel clonal types that migrate from outside the tumor. This study did not examine any amplified T cell clones derived from within the tumor.
[0145] Therefore, the question arises as to the extent to which gene expression of expanded TCR clonotypes differs from that of non-expanded TCR clonotypes in tumors prior to checkpoint blockade by anti-PD-1 antibody treatment. Furthermore, the extent to which these gene expression differences can be used to predict clinical response to anti-PD-1 or anti-PD-L1 therapy.
[0146] This example demonstrates that using TCR clonotype as a matching condition, T cells that expand after treatment can be identified in pre-treatment samples.
[0147] In general, for immunomodulatory drugs that target T cells (i.e., modulate T cell activity), T cells from BCC and SCC pre-treatment samples are enriched for specific T cell markers. In this example, T cells were further divided into CD8 + T cells and CD4 from BCC / SCC pre-treatment samples + T cells. These T cell libraries are subjected to the methods of the invention to identify genes that are upregulated in clonally expanded CD8 / CD4 T cells. Gene expression filters are then applied to identify a list of upregulated genes (log2 fold change > 0.2) that are expressed in a low percentage (e.g., in this case < 40%) of unamplified T cells. This list constitutes a gene signature or predictive response biomarker (PRB) for treatment. See Figure 4 .
[0148] The genes of the gene markers are then used as predictors and logistic regression is performed with the T cell expansion status as the outcome. The coefficients obtained in this way are used as weights for each gene for future predictions. If the upregulation of a gene is closely related to the clonal expansion of T cells, then the gene has a higher predictive value or predictive efficacy than another gene that is less closely related to the clonal expansion of T cells.
[0149] Next, a large amount of RNA-seq data was obtained from several anti-PD-L1 (atezolizumab) clinical trials, namely IMvigor210 (treatment of metastatic urothelial carcinoma, n = 348), IMmotion150 (treatment of renal cell carcinoma or RCC, n = 86 for atezolizumab alone, n = 88 for atezolizumab + bevacizumab (which is an anti-angiogenic drug designed to block VEGF) and POPLAR (treatment of NSCLC, n = 93). From these studies, a large amount of RNA-seq data was collected from pre-treatment tumor samples, and progression-free survival (PFS) data was collected as a treatment endpoint.
[0150] With these data, PRB (ie, marker of T cell expansion) can be correlated with actual clinical response or PFS data.
[0151] Specifically, the weight of each gene signature from the scRNA-seq data was applied to the corresponding gene in the bulk RNA-seq data of the clinical trial to calculate the score for each patient as the weighted sum of the genes from the PRB / gene signature. The median score of the patients was taken, and half of the patients with scores above the median were designated as Sig 高 group, and the other half of the patients with scores below the median were designated as Sig 低 The censored PFS data were then fitted to a Cox proportional hazards model to obtain the Sig 高 Group and Sig低 Hazard ratios between groups, and Kaplan-Meier (KM) plots for progression-free survival (PFS). Here, HR is defined as the difference between a group (e.g., Sig 高 group) / the risk of an outcome (e.g., death) in another group (e.g., Sig 低 The risk of an outcome (e.g., death) occurring in either group (eg,
[0152] The results showed that a single CD8 T cell expansion marker (PRB) predicted a positive clinical response in different trials. Specifically, in the IMvigor210 trial of 348 patients, CD8 + PRB is applied to large amounts of RNA data from patients before treatment. 高 Group and Sig 低 The hazard ratio (HR) of patients in the group was 0.7 (p<0.0029), indicating that Sig 高 Patients in the group with Sig 低 Patients in the 2-week follow-up group had a significantly better chance of survival than those in the 2-week follow-up group. Figure 5B Similarly, in the 93-patient POPLAR trial, Sig 高 Group and Sig 低 The HR for the group was 0.38 (p=7.5e-05), indicating a statistically significant production advantage. Figure 5A The data demonstrate that CD8 + The T cell expansion marker correlated with a positive clinical response and the predictive value of this marker was superior to that of published markers (see comparison below).
[0153] CD8-amplification marker Wu et al., Nature (2020)* POPLAR HR = 0.38 HR = 0.45 IMvigor210 HR = 0.70 HR = 0.72
[0154] *Wu et al., Nature 579:274-278 (2020).
[0155] Similar results were obtained in the IMmotion 150 trial (atezolizumab alone) of 86 patients, with an HR ratio of 0.87, but the p value (p = 0.58) did not support statistical significance. In the same trial with atezolizumab + bevacizumab (n = 88 patients), the HR was 0.68 (p = 0.12).
[0156] On the other hand, the following data show that CD4 T cell expansion markers predict negative clinical responses, supporting another embodiment of the present invention, wherein CD4 + Clonal expansion of T cells appears to be detrimental to survival in this treatment.
[0157] Specifically, in the IMvigor210 trial, the use of +T cell-acquired PRB (GZMA) for Sig 高 Patients produced a HR of 1.21 (p = 0.11), indicating that 低 The survival rate of patients was slightly lower.
[0158] In the POPLAR trial using the GGMA gene in the PRB, the HR was 1.44 (p=0.11).
[0159] In the IMmotion150 trial of atezolizumab alone, PRB(GGMA) had an effect on Sig 高 Patients yielded a HR of 1.59 (p=0.059). With atezolizumab and bevacizumab, PRB (GGMA) yielded a HR of 1.86 (p=0.013).
[0160] In summary, the above data show that genes upregulated in CD8 or CD4 T cells expanded after anti-PD1 treatment can predict positive or negative clinical responses, respectively. Gene signatures or PRBs derived from one cancer (e.g., BCC and SCC datasets) are applicable to other tumor types (RCC, NSCLC, metastatic urothelial carcinoma). Therefore, combined clonal and phenotypic analysis can be used to identify predictive response markers associated with patient outcomes. The methods described herein provide an unbiased systematic method for discovering predictive response biomarkers.
[0161] References
[0162] 1.clinicaltrials.gov / ct2 / show / NCT02108652, “A Study of Atezolizumab inParticipants With Locally Advanced or Metastatic Urothelial Bladder Cancer (Cohort 2).”
[0163] 2. Balar et al., Atezolizumab as first-line treatment in cisplatin-ineligible patients with locally advanced and metastatic urothelialcarcinoma: a single-arm, multicentre, phase 2trial. The Lancet, 389(10064):67-76, 2017.
[0164] 3.clinicaltrials.gov / ct2 / show / NCT01903993, “A Randomized Phase 2Study of Atezolizumab(an Engineered Anti-PDL1 Antibody)Compared With Docetaxel inParticipants With Locally Advanced or Metastatic Non-Small Cell Lung CancerWho Have Failed Platinum Therapy–'POPLAR'.”
[0165] 4. Weinstock et al., US Food and Drug Administration Approval Summary: Atezolizumab for Metastatic Non-Small Cell Lung Cancer. Clin Cancer Res. 23(16): OF1-OF6, 2017.
[0166] Example 2
[0167] Single-cell immune profiling using clonotypic barcoding identified biomarker signatures predictive of response to immune checkpoint blockade.
[0168] There is an urgent need to identify predictive biomarkers of patient treatment response to increase the probability of success of existing and novel experimental therapies. Single-cell profiling has provided novel biological insights into drug response in the tumor microenvironment, but its potential for biomarker discovery has not yet been explored for therapeutic purposes. This example describes a novel method for discovering predictive response biomarkers from single-cell data, which uses the T cell receptor sequence inherent to each T cell to match clonotypes between pre- and post-treatment tumor samples. Therefore, for immune checkpoint blockade, predictive gene expression markers were identified from a small patient cohort, and their predictive performance was validated using data from three larger clinical studies.
[0169] The results presented herein demonstrate that the application of clonal typing and genomic characterization is a novel approach for biomarker identification using single-cell technology. This approach leads to an increased probability of success and reduced clinical trial size, and thus significantly impacts the future clinical development of novel immunomodulatory therapeutics.
[0170] To investigate which T cells respond to ICB by clonal expansion, data from 14 skin cancer patients were used, each with site-matched tumor biopsies from before and after anti-PD-1 treatment. Each biopsy was characterized using single-cell RNA sequencing (scRNA-seq) for gene expression and single-cell TCR sequencing (scTCR-seq) for clonal analysis. Samples from basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) were separated for further analysis. Clustering and dimensionality reduction algorithms were then used to identify cell populations sharing common features based on their gene expression profiles, and clusters separated by T cell subsets and patient-specific identifiers were found. See Figure 6 Quantify clonal expansion triggered by ICB by identifying matched T cell clonotypes between pre- and post-treatment samples to calculate the fold increase in clonotype size induced by treatment. Figure 6 . In the BCC data, 2.02% of all CD8+ T cells, 0.63% of all conventional CD4+ T cells, and 0.99% of all FOXP3+CD4+ Tregs expanded after ICB, with mean fold increases of 3.82, 2.75, and 4.54, respectively. While CD8+ T cells primarily responded in BCC, there was a more extensive T cell response in SCC, with 5.14% of all CD8+ T cells, 5.04% of all CD4+ T cells, and 2.70% of all Tregs expanding after ICB, with mean fold increases of 4.88, 5.26, and 4.06, respectively. In both datasets, the absolute number of Treg clonotypes that responded to ICB with clonal expansion was limited. Therefore, Tregs and conventional CD4+ T cells were combined for subsequent analyses. To increase resolution, clonal expansion analysis was repeated for each major T cell subset, and in BCC, the majority of responding CD8+ T cells were from a single patient. See. Fig.12 . However, this patient still had only 6.74% of individual CD8+ T cells identified as responding. In the SCC data, responding cells were more dispersed among patients. In summary, based on scTCR-seq clonotype characterization in BCC and SCC cohorts, CD8+ and CD4+ T cell clonotypes that expanded in response to ICB were identified.
[0171] Next, we investigated the possibility of using pre-treatment gene expression profiles to distinguish T cells that responded to ICB with clonal expansion from those that did not. Using the CB approach, TCR clonotypes in pre-treatment samples were grouped into a fraction that responded and a fraction that did not respond after ICB. Fig.12. Genes differentially expressed in the responder cell pool were then identified using their scRNA-seq gene expression profiles. To identify differentially expressed genes, the gene expression fold change between the two fractions was calculated for each gene, and several filtering steps were used to ensure that the identified genes represented clear (not affected by the sparseness of single-cell data or high baseline expression), meaningful (with appreciable log fold changes), and statistically robust signals (significant based on a rank-sum test corrected for multiple testing). See Fig.13 In BCC patient samples, eight genes were found to be upregulated (IFNG, ADGRE2, TNF, FGL2, RHOB, ADRB2, NFKBIZ, STAT1) and nine genes were found to be downregulated (TRAT1, MRPL18, PEBP1, CMTM7, CCR7, CYSTM1, AHSA1, CMSS1) in CD8+ T cells relative to clonotypes that did not expand after ICB. Figure 7 In contrast, no genes were differentially expressed in CD4+ T cells. Figure 7 This may be due to the low frequency of expanded cells in this subset (0.7%). In SCC samples, no genes were found to be differentially expressed in CD8+ T cells, but five genes (LAIR2, ACP5, IL1R2, GZMA, FOXP3) and two genes (G0S2, BAG3) that were downregulated and upregulated in CD4+ T cells relative to clonotypes that did not expand after ICB were identified. See Figure 7 Overall, two distinct sets of differentially expressed genes at the single-cell level were identified, one associated with CD8+ T cells in BCC and the other with CD4+ T cells in SCC, and these gene sets were associated with clonal expansion in response to ICB.
[0172] Given the set of differentially expressed genes in CD4+ and CD8+ T cells, it was important to discover which single-cell gene expression signature best predicted the response to ICB, which could later be applied to bulk gene expression data to predict patient response to ICB in clinical trials. Bulk RNA-seq data represent not only tumor-infiltrating T cells, but also a mixture of all cells in the TME, including immune cells, stromal cells, and malignant cells. To obtain a T-cell-specific predictive signature, the list of differentially expressed genes was restricted to genes whose expression was more restricted to T-cell subsets by calculating their correlation with T-cell lineage markers. To obtain a gene expression signature that best predicted T-cell expansion after ICB at the single-cell level, a logistic regression model was trained on the remaining differentially expressed genes. See Fig.13 We also investigated whether a more stringent cutoff for differential expression, which resulted in a reduced number of genes in the signature, would affect the predictive performance of the logistic regression model. Fig.10 For the CD8+ T cell expansion marker, the most parsimonious model with only one gene, IFNG, had the highest accuracy and specificity as a positive predictor of expansion. Figure 8 Comparison Fig.12 For CD4+ T cell signatures, the best accuracy and specificity were found for a model with two genes, GGMA and FOXP3; CD4+ T cell transcription of these two genes was a negative predictor of expansion (i.e., before treatment, these transcripts were downregulated in clonotypes that responded vs. non-responded after ICB). Figure 8 Comparison Fig.12 .
[0173] In summary, gene expression signatures were identified that predicted response to ICB at the single-cell level: expression of one signature was a positive predictor of CD8+ T cell expansion (CD8-expanded signature), whereas expression of the other signature was a negative predictor of CD4+ T cell expansion (CD4-non-expanded signature).
[0174] To study the effects of single-cell data ( Figure 8 ) can predict ICB treatment outcomes. We applied a CD8-amplified signature and a CD4-non-amplified signature to data from three phase II clinical trials of the anti-PD-L1 checkpoint inhibitor (atezolizumab) for which bulk RNA-seq on pretreatment patient samples was available: the POPLAR study in locally advanced or metastatic non-small cell lung cancer16 (NCT01903993), the IMvigor210 study in locally advanced or metastatic urothelial bladder cancer21 (NCT02951767, NCT02108652), and the IMmotion150 study in advanced renal cell carcinoma22 (NCT01984242). For each study, a signature score was generated for each patient, a median cutoff was used to distinguish patients with high from those with low signatures, and an investigation was performed to determine whether the signature could predict treatment outcomes using survival analysis. See Fig.14 For the CD8-amplification marker, patients with a high marker tend to have a longer progression-free survival (PFS). Fig. 9 For CD4-non-amplified markers, patients with high markers also tended to experience longer PFS. Fig. 9 This was surprising, as a high signature means that the CD4+ T cells in their tumors would have a lower probability of expanding after ICB. Expression of FOXP3 and GGMA (which are genes contributing to this signature) was found in different clusters of conventional CD4+ T cells, however, there was also some overlap in the Treg clusters. See Fig.15Therefore, the potential predictive importance of this non-amplified marker originating from within the Treg subset cannot be excluded. While both markers were significantly (p-value < 0.05) enriched for responding patients in IMvigor210 and POPLAR, only patients with only the CD4- non-amplified marker in the IMmotion150 trial reached significance in the combined arm (atezolizumab + bevacizumab).
[0175] Next, we investigated whether the combination of two markers identified could further improve treatment outcome prediction. Patients were divided into four groups, and the CD8 marker was first labeled: both markers were above the median score (Pos-Pos), one marker was above the median score (Pos-Neg, Neg-Pos), or both markers were below the median score (Neg-Neg). The Pos-Pos combination signature could further improve prediction in the combined arm of IMvigor210 and IMmotion150. Fig. 9 For POPLAR, all groups positive for at least one of the two markers had significantly improved outcomes relative to the low-low group. Fig. 9 Based on these observations, a combination signature was derived that took into account both markers. When compared to other published markers and gene expression markers of interest, including CD274 (PD-L1), PDCD1 (PD-1), CD8A, CD4, and FOXP3, the combination signature consistently achieved similar or superior hazard ratios compared to the other markers or markers of interest. Fig.16 . Notably, no marker provided a significantly reduced hazard ratio (HR) for the IMmotion150 atezolizumab monotherapy arm. Overall, gene expression signatures derived from pre-treatment scRNA-seq data from a small number of patients can predict outcomes in a larger cohort of ICB-treated patients when applied to pre-treatment bulk RNA-seq data. Furthermore, combination signatures yielded further improved predictions depending on tumor type.
[0176] To further demonstrate how predictive signatures derived from single-cell data can impact the clinical development of novel experimental drugs, patient stratification using our combined signature was analyzed to determine whether it could improve the probability of achieving significantly improved PFS in clinical trials. To this end, subsets of patient data from the POPLAR and IMmotion150 studies were used, for which pre-treatment extensive RNA-seq data were available for the experimental and standard of care comparator arms. Patients treated with ICB in each trial were divided into biomarker-positive (BM+) and biomarker-negative (BM-) groups using the combined signature. For each trial and biomarker group, survival analyses between the comparator and experimental arms were performed, and the results were compared with outcomes in the unstratified patient population. See Fig.11. In the unstratified setting, the experimental arms did not significantly improve HR in any of the three analyses. After stratifying patients based on the combined signature, there was an overall reduction in HR in the BM+ group compared to unstratified patients. The IMmotion150 trial combined arm reached statistical significance, suggesting that the combination of atezolizumab and bevacizumab is more beneficial than the multi-targeted receptor tyrosine kinase (RTK) inhibitor sunitinib in this patient population. In contrast, increased HR was found in the BM- group compared to the unstratified group. Significance was reached in the BM- group of POPLAR, suggesting that atezolizumab was less beneficial than the chemotherapy docetaxel in this patient group. Correspondingly, in the BM+ group of POPLAR, there was a strong trend towards reduced HR compared to the unstratified group; however, it did not formally reach significance (p=0.06). We also compared our combined signature with previously published signatures and other biomarkers of interest. See Fig.11 . This signature consistently showed better or similar performance than other relevant markers. Notably, no marker was able to provide a significantly reduced HR for the IMmotion150 atezolizumab monotherapy arm. Overall, it was demonstrated that gene expression signatures derived from single-cell data from a small number of patients can prospectively identify patient populations with a greater chance of benefiting from experimental treatments compared to patients in an unstratified patient population.
[0177] Using single cell data from patients receiving ICB before and after treatment, this example demonstrates a novel method for identifying predictive gene expression markers that can be used as biomarkers to stratify patients to improve clinical trial outcomes. The clonal barcoding (CB) method described herein uses the TCR sequence inherent to each T cell to match the clonal type between tumor samples before and after treatment to identify T cells that are amplified in response to treatment. The CB method gives priority to gene expression patterns from amplified and recurring T cell clonal types detected in the TME for predicting treatment outcomes. The CB method can also be applied to predictive biomarkers for identifying therapeutic agents targeting pathogenic B cells in autoimmune diseases.
[0178] Single cell feature analysis enables the CB method to exploit heterogeneity at the single cell level to derive predictive gene expression markers. Compared with conventional methods that require a large number of patient samples in order to compare different groups of patients, the T cell group that can contact the pre- and post-treatment samples from the same patient provides an excellent way to centrally analyze and identify the most critical predictive parameters. Therefore, this method allows unbiased searches for predictive markers from smaller patient cohorts (such as those available from early clinical studies). The markers found using the CB method are surprisingly simple and easy to interpret and practically implement clinically (i.e., immunohistochemistry or in situ hybridization), but otherwise will be missed as part of a more complex marker. Although the single cell data used in this study are still limited in the number of matched T cell clonotypes before and after treatment, the generation of scRNA-seq data sets with a larger number of cells that have been characterized will result in more refined markers, which more clearly link T cell biology to the patient's response to immunotherapy. In addition, more optimized and standardized single cell data generation and analysis will further enrich the data set and provide more meaningful analysis.
[0179] Mechanistically, the identification of IFNG as the best predictor of CD8+ T cell clonal expansion in BCC patients treated with ICB indicates the presence of functional tumor-reactive cells that actively recognize cognate tumor antigens but whose expansion is inhibited by the PD-1 / PD-L1 pathway. Therefore, it is surprising that CD8+ T cell markers were not identified in SCC patients despite a larger proportion of pre-treatment CD8+ T cell clonal expansion (2.02% and 5.14% of CD8+ T cell clones in BCC and SCC, respectively). The T cell clonal response in SCC was more extensive than in BCC, with a larger proportion of conventional CD4+ T cells and Tregs expanding in SCC, which may indicate different ICB mechanisms of action between these cancer indications. Although CD8+ T cell expansion in BCC may have been a direct consequence of ICB, it may have been an indirect consequence in SCC, potentially caused by CD4+ T cell help, reduced regulatory T cell suppression, or anti-tumor activity from immune cell subsets (such as NK cells) not captured in our CB approach.
[0180] Interpretation of CD4-non-expansion markers may be more complicated. FOXP3 and GZMA do not necessarily correspond to the same cell subsets, and GZMA itself still has predictive efficacy outside of the Treg cluster. Although the transcription factor FOXP3 is a characteristic of Tregs, FOXP3 transcription is also expressed in conventional T cells after activation, and in fact FOXP3 was found within the conventional CD4 T cell cluster in the SCC scRNA-seq data. GGMA, which encodes the cytolytic protein granzyme A and is associated with effector T cells, has previously been identified as a positive prognostic or predictive biomarker for ICB treatment, but this is generally attributed to CD8+ T cells. However, since there is some overlap between FOXP3 and GGMA within the Treg cluster, this Treg subset may have potential predictive importance. Granzyme A produced by activated Tregs has been reported to make these cells more sensitive to self-inflicted apoptosis.
[0181] Furthermore, it was demonstrated that the predictive gene expression signatures identified using the CB approach can be validated in larger clinical trials. Both the CD8-amplified signature as well as the CD4-non-amplified signature can predict favorable outcomes with treatment, and the combination of the two signatures can produce further improved results. It is interesting to note that the predictive signatures were derived based on data from skin cancer patients, but they still showed predictiveness in a larger cohort of three other solid cancer types. However, the optimal combination of markers depends on the patient population being treated.
[0182] Finally, it is demonstrated how predictive markers can be applied to patient stratification to increase the probability of success in clinical trials. Although these markers consistently perform better or similarly to other relevant markers, PD-1 expression alone is also a strong predictive biomarker. This is not unexpected based on the mechanistic understanding of ICB. Importantly, the power of the CB approach lies in its ability to identify biomarker signatures in an unbiased manner that performs similarly well. This may be most relevant for identifying biomarkers for novel experimental therapies, where the mechanisms may be less understood or more complex. Therefore, the proposed CB method is a promising new use of single-cell data for biomarker identification that will impact the future clinical development of novel immunomodulatory therapeutics.
[0183] method
[0184] Identification of clonal expansion based on scTCR-seq data
[0185] TCR clonotype signatures of individual T cells were obtained from Yost et al. Clonotypes shared by pre- and post-treatment T cells within the same patient were identified based on their matching TCR CDR3 sequences. Clonotypes whose post-treatment clonotype size (number of T cells sharing the same TCR CDR3 sequence) was greater than the pre-treatment clonotype size were considered to have expanded in response to ICB. For patient su013 in the SCC dataset, 4488 T cells were characterized in the pre-treatment sample, but only 69 T cells were characterized in the post-treatment sample; this imbalance in T cell counts partially explains why no clonotypes in patient su013 were considered to have undergone expansion after treatment.
[0186] Clustering and dimensionality reduction of scRNA-seq data
[0187] Clustering of scRNA-seq data from all BCC cell types was performed as previously described using Seurat (version 3.1.1). Based on this first round of clustering and the expression of canonical T cell markers (CD8A, CD4, FOXP3), we identified T cell clusters. T cells specifically from pre-treatment samples were extracted from the complete dataset and then clustered. In this round of clustering, the default variable gene selection criteria in Seurat were used. These variable genes were used for principal component analysis (PCA). The first 20 principal components were used for clustering based on shared nearest neighbors, k=20 and resolution=3. The same principal components were used for visualization of UMAP projections with minimum distance=0.1 and number of neighbors=20. The SCC scRNA-seq dataset is already T cell specific. After only pre-treatment T cells were extracted from this dataset, clustering was performed using Seurat. These pre-treatment T cells were clustered in the same manner as BCC pre-treatment T cells. After pre-treatment T cell clustering, each cluster was annotated as CD8+ T cells, conventional CD4+ T cells, or Tregs based on the expression of CD8A, CD4, or FOXP3, respectively.
[0188] Differential gene expression analysis
[0189] Using only pre-treatment scRNA-seq data, differential gene expression analysis was performed to determine the genes significantly upregulated or downregulated in the T cell population of the amplified T cell population versus the unamplified T cell population. Several filtering steps were performed to remove genes that were unlikely to be biologically relevant from the test. The test was limited to genes detected in at least 10% of cells in any of the two populations compared. Non-coding genes defined by HGNC were removed from the test. Human T cell receptor α variable (TRAV) gene, human T cell receptor β variable (TRBV) gene and human leukocyte antigen (HLA) gene were also removed from the test. For genes that passed the filter, a Wilcoxon rank sum test was performed to detect differential gene expression between the amplified T cell population and the unamplified T cell population. The obtained p value was adjusted using the Benjamini-Hochberg correction to control the false discovery rate. Differentially expressed genes were defined as those with log2 fold change>0.6 and adjusted p value<0.01. In addition, for genes to be considered significantly upregulated in the expanded T cell population, they need to be expressed in less than 30% of the non-expanded T cell population. Similarly, for genes to be considered significantly downregulated in the expanded T cell population, they need to be expressed in less than 30% of the expanded T cell population.
[0190] Generate predictive signatures based on differentially expressed genes
[0191] Differential gene expression analysis produces two lists of differentially expressed genes: one list is from the BCC CD8+ T cell data set, and the other list is from the SCC CD4+ T cell data set. For each list, genes with a fold change>1.5 are used as predictors in the logistic regression classifier that predicts the expansion state of each cell. The classifier calculates the coefficient of each gene, which reflects how strongly the gene expression level of each gene predicts the expansion state of each cell (expansion versus non-expansion). Positive coefficients indicate that cells expressing the gene are more likely to expand. Negative coefficients indicate that cells expressing the gene are more likely not to expand. The process was then repeated with a fold change threshold of 1.75, and then repeated again with a fold change threshold of 2.0; as the fold change threshold increases, the number of genes used in the logistic regression classifier decreases. The coefficients, accuracy, sensitivity, and specificity of each logarithmic regression classifier at each fold change threshold were recorded. For the BCC CD8+ T cell data set, we compared the performance of the logarithmic regression classifier and selected the most parsimonious model that did not encounter a significant decrease in accuracy, sensitivity, and specificity. In this case, the highest performing classifier contained one gene, IFNG, with a positive coefficient. This was defined as the CD8-amplified signature. Using the same classifier selection criteria for the SCC CD4+ T cell dataset, we defined the CD4-non-amplified signature as containing two genes, GGMA and FOXP3, each with a negative coefficient. For each gene in the amplified signature, the absolute value of its log regression coefficient was taken as its weight.
[0192] Predicting patient susceptibility to ICB by applying single-cell-based gene expression signatures to bulk RNA-seq data answer
[0193] The effectiveness of CD8-amplification tags, CD4-non-amplification tags and the combination of these two tags in predicting patient responses was evaluated using a large amount of tumor RNA-seq data from three ICB clinical trials. In POPLAR (NCT01903993), patients with locally advanced or metastatic non-small cell lung cancer were treated with atezolizumab (n=93) or docetaxel (n=100). In IMvigor210 (NCT02951767, NCT02108652), patients with locally advanced or metastatic urothelial bladder cancer (n=354) were treated with atezolizumab alone (n=86) or with both atezolizumab and bevacizumab (n=88) in patients with advanced renal cell carcinoma. For each patient in the clinical trial, the weighted sum of the genes in the marker (CD8+ T cell amplification marker or CD4+ T cell amplification marker) was calculated as the marker score. The median marker score for each clinical trial is determined, and patients with scores above the median are placed in the biomarker-positive group, while the remaining patients in the same trial are placed in the biomarker-negative group. As previously described by Wu et al., survival analysis of the biomarker-positive group and the biomarker-negative group in each clinical trial arm is compared. The Cox proportional hazard model is fitted to the survival data in each arm of each clinical trial; the p-value of the hazard ratio and Wald statistic obtained is reported. The results of these survival analyses are represented by Kaplan-Meier graphs. In order to study the interaction between CD8+ and CD4+ T cell expansion markers in clinical trials, we assign patients in each arm to one of four groups: two markers above the median score (positive-positive), one marker above the median score (positive-negative, negative-positive) or two markers below the median score (negative-negative). Survival analysis and Kaplan-Meier mapping are performed as described above, and the negative-negative group is used as a baseline comparator with each of the other three groups. As a reference, CD274 (PD-L1), PDCD1 (PD-1), CD8A, CD4, and FOXP3 were each treated as their own single unweighted biomarker. Survival analyses comparing biomarker-positive and biomarker-negative groups in each arm of each clinical trial were performed as described above.
[0194] Predicting trial outcomes (experimental arm vs. comparator arm) via predictive markers
[0195] To compare the effectiveness of these T cell expansion markers between the experimental arms (atezolizumab, atezolizumab + bevacizumab) and the comparator arm (sunitinib) in IMmotion150, the CD8+ and CD4+ T cell expansion markers were merged into a single marker, and the corresponding weights of each gene were transferred to the merged marker. First, to establish a baseline, a survival analysis comparing each experimental arm with the comparator arm was performed to detect any pre-stratification differences in survival. Next, based on the merged markers, patients from IMmotion150 were stratified into biomarker-positive and biomarker-negative groups. For each biomarker-stratified group, a survival analysis comparing atezolizumab-treated patients with docetaxel-treated patients was performed as described above. For POPLAR, the same process was used for survival analysis to detect the effectiveness of these T cell proliferation markers between the experimental and comparator arms. The only difference was that the CD8+ and CD4+ T cell expansion markers were not combined, but kept separate; patients with either marker above the median score (positive-positive, positive-negative, negative-positive) were considered to be in the biomarker-positive group, and the remaining patients were considered to be in the biomarker-negative group. This stratification strategy corresponds to previous observations that patients with either expansion marker above the median in POPLAR had more similar generation patterns to each other than those in the negative-negative group. To compare the effectiveness of other markers between the experimental and comparative arms in IMmotion150 and POPLAR, the 19-gene dual expansion signature from Wu et al. was applied, unweighted, as a single marker, while CD274 (PD-L1), PDCD1 (PD-1), CD8A, CD4, and FOXP3 were each treated as their own single unweighted marker.
[0196] References
[0197] 1. Topalian, SLet al. Safety, Activity, and Immune Correlates of Anti–PD-1 Antibody in Cancer. New Engl J Medicine 366, 2443–2454 (2012).
[0198] 2. Rizvi, NA et al. Mutational landscape determines sensitivity to PD-1blockade in non–small cell lung cancer. Science 348,124–128(2015).
[0199] 3.Rooney,M.S.,Shukla,S.A.,Wu,C.J.,Getz,G.&Hacohen,N.Molecular andGenetic Properties of Tumors Associated with Local Immune CytolyticActivity.Cell 160,48–61(2015).
[0200] 4.Miao,D.et al.Genomic correlates of response to immune checkpointblockade in microsatellite-stable solid tumors.Nat Genet 50,1271–1281(2018).
[0201] 5.Kümpers,C.et al.Immune Cell Infiltration of the Primary Tumor,NotPD-L1 Status,Is Associated With Improved Response to Checkpoint Inhibition inMetastatic Melanoma.Frontiers Medicine 6,27(2019).
[0202] 6.Kansy,B.A.et al.PD-1 status in CD8+ T cells associates withsurvival and anti-PD-1 therapeutic outcomes in head and neck cancer.CancerRes 77,canres.3167.2016(2017).
[0203] 7.Gros,A.et al.PD-1 identifies the patient-specific CD8+tumor-reactive repertoire infiltrating human tumors.J Clin Invest 124,2246–2259(2014).
[0204] 8.Kumagai,S.et al.The PD-1 expression balance between effector andregulatory T cells predicts the clinical efficacy of PD-1 blockadetherapies.Nat Immunol 1–13(2020)doi:10.1038 / s41590-020-0769-3.
[0205] 9.Yi,M.et al.Biomarkers for predicting efficacy of PD-1 / PD-L1inhibitors.Mol Cancer 17,129(2018).
[0206] 10.Sharma,P.,Hu-Lieskovan,S.,Wargo,J.A.&Ribas,A.Primary,Adaptive,andAcquired Resistance to Cancer Immunotherapy.Cell 168,707–723(2017).
[0207] 11.Higgs,B.W.et al.Interferon Gamma Messenger RNA Signature in TumorBiopsies Predicts Outcomes in Patients with Non-Small-Cell Lung Carcinoma orUrothelial Cancer Treated with Durvalumab.Clin Cancer Res 24,clincanres.3451.2017(2018).
[0208] 12.Postel-Vinay,S.et al.Challenges of phase 1 clinical trialsevaluating immune checkpoint-targeted antibodies.Ann Oncol 27,214–224(2016).
[0209] 13.Wong,K.M.,Capasso,A.&Eckhardt,S.G.The changing landscape of phaseI trials in oncology.Nat Rev Clin Oncol 13,106–117(2016).
[0210] 14.Chabner,B.A.Approval After Phase I:Ceritinib Runs the Three-MinuteMile.Oncol 19,577–578(2014).
[0211] 15.Huang,A.C.et al.T-cell invigoration to tumour burden ratioassociated with anti-PD-1 response.Nature 545,60–65(2017).
[0212] 16.Fehrenbacher,L.et al.Atezolizumab versus docetaxel for patientswith previously treated non-small-cell lung cancer(POPLAR):a multicentre,open-label,phase 2 randomised controlled trial.Lancet 387,1837–1846(2016).
[0213] 17.Fairfax,B.P.et al.Peripheral CD8+ T cell characteristicsassociated with durable responses to immune checkpoint blockade in patientswith metastatic melanoma.Nat Med 26,193–199(2020).
[0214] 18.Wu,T.D.et al.Peripheral T cell expansion predicts tumourinfiltration and clinical response.Nature(2020)doi:10.1038 / s41586-020-2056-8.
[0215] 19.Sade-Feldman,M.et al.Defining T Cell States Associated withResponse to Checkpoint Immunotherapy in Melanoma.Cell 175,998-1013.e20(2018).
[0216] 20.Yost,K.E.et al.Clonal replacement of tumor-specific T cellsfollowing PD-1 blockade.Nat Med 25,1251–1259(2019).
[0217] 21.Mariathasan,S.et al.TGFβattenuates tumour response to PD-L1blockade by contributing to exclusion of T cells.Nature 554,544–548(2018).
[0218] 22.McDermott,D.F.et al.Clinical activity and molecular correlates ofresponse to atezolizumab alone or in combination with bevacizumab versussunitinib in renal cell carcinoma.Nat Med 24,749–757(2018).
[0219] 23.Baumeister,S.H.,Freeman,G.J.,Dranoff,G.&Sharpe,A.H.CoinhibitoryPathways in Immunotherapy for Cancer.Annu Rev Immunol 34,1–35(2014).
[0220] 24.Kreiter,S.et al.Mutant MHC class II epitopes drive therapeuticimmune responses to cancer.Nature 520,692–696(2015).
[0221] 25.Alspach,E.et al.MHC-II neoantigens shape tumour immunity andresponse to immunotherapy.Nature 574,696–701(2019).
[0222] 26.Francisco,L.M.et al.PD-L1 regulates the development,maintenance,and function of induced regulatory T cells.J Exp Medicine 206,3015–3029(2009).
[0223] 27.Zappasodi,R.et al.Non-conventional Inhibitory CD4+Foxp3–PD-1hi TCells as a Biomarker of Immune Checkpoint Blockade Activity.Cancer Cell 34,691(2018).
[0224] 28.Pesce,S.et al.PD / 1-PD-Ls Checkpoint:Insight on the Potential Roleof NK Cells.Front Immunol 10,1242(2019).
[0225] 29.Subrahmanyam,P.B.et al.Distinct predictive biomarker candidatesfor response to anti-CTLA-4and anti-PD-1immunotherapy in melanoma patients.JImmunother Cancer 6,18(2018).
[0226] 30. Morgan, ME et al. Expression of FOXP3 mRNA is not confined to CD4+CD25+ T regulatory cells in humans. Hum Immunol 66, 13–20 (2005).
[0227] 31. Roufas, C. et al. The Expression and Prognostic Impact of ImmuneCytolytic Activity-Related Markers in Human Malignancies: A ComprehensiveMeta-analysis. Frontiers Oncol 8, 27 (2018).
[0228] 32. Inoue, H. et al. Intratumoral expression levels of PD-L1, GZMA, and HLA-A along with oligoclonal T cell expansion associate with response tonivolumab in metastatic melanoma. Oncoimmunology 5, 00–00 (2016).
[0229] 33.Karreci, ES et al. Human regulatory T cells undergo self-inflicteddamage via granzyme pathways upon activation. Jci Insight 2, e91599 (2017).
[0230] Example 3
[0231] Using ex vivo single-cell profiling combined with TCR clonotype characterization to discover novel targets for T cell-targeted biologics Predictive biomarkers of response
[0232] The discovery of predictive biomarkers of drug response is critical to predicting patients' benefit from novel immunomodulatory therapies. However, their discovery is challenging due to the heterogeneous nature of the tumor immune environment and the lack of effective analysis of drug treatment of these immune cells associated with tumor progression. This is further hampered by the complexity of drug modes of action and the lack of appropriate biological models.
[0233] The invention described herein defines biomarkers of response by subjecting cells from dissociated tissue of human tumor samples to T cell-targeted biologics or a combination thereof, followed by single cell transcriptome profiling and TCR clonotype characterization (supra).
[0234] More specifically, the invention described herein defines biomarkers of response by subjecting dissociated human tumor tissue to drug treatment (particularly biologics targeting T cells or combinations thereof), followed by single cell transcriptome profiling and TCR clonotype characterization. Responding cells are identified as those cells that show a treatment-specific shift in gene expression profiles. For T cells, TCR clonotypes are used to match responding T cells under treatment conditions to their sister clones in the (untreated) baseline state. By comparing the baseline gene expression profiles between T cell clonotypes that respond to treatment and T cell clonotypes that do not respond to treatment, gene expression signatures called PRBs (predictive response biomarkers) that predict drug treatment response are identified.
[0235] A total of 17 tumor samples were obtained from cancer patients and treated ex vivo with two novel biologics currently in clinical development, an OX40 agonist (also referred to herein as "HFB10") and a TNFR2 agonist (also referred to herein as "HFB3") as experimental treatments. Tumor samples were obtained from commercial sources (Discovery Life Sciences (DLS); Cooperative Human Tissue Network (CHTN)). Eight renal cell carcinoma (RCC) tumor samples, seven non-small cell lung cancer (NSCLC) tumor samples, one melanoma (MEL) tumor sample, and one uterine cancer (UC) tumor sample were treated together. These tumors were selected to represent the key indications for the anti-OX40 antibody HFB301001 (a potential best-in-class second-generation OX40 agonist antibody, or "HFB10" for short, see WO2021129872) and the potential first-in-class TNFR2 agonist antibody HFB200301 (or "HFB3" for short, see WO2021141907).
[0236] For PRB biomarker discovery, predictive signatures for response to these drugs were defined using combined scRNA-seq and scTCR-seq data from these samples.
[0237] Specifically, for ex vivo drug treatment, tumors were digested into single-cell suspensions, followed by baseline sample characterization (e.g., immune cell composition and target expression) using flow cytometry. All cells, including tumor cells, were then treated with controls (isotype-matched antibody controls, anti-CD3 Ab, anti-PD-1 Ab) and the above experimental drugs / antibodies (i.e., HFB3 and HFB10) in ex vivo culture for 24 hours. No exogenous TCR cross-linking was used to maintain TCR stimulation provided only by material within the sample. Finally, single-cell RNA and TCR sequencing were used to characterize cells from different treatment conditions. Customized analytical pipelines were applied to identify predictive biomarkers. See Fig.17 .
[0238] Specifically, Fig.18 As generally shown, in the first step, the bioinformatics pipeline is applied to map all reads to transcripts as gene expression readings, and the clonotypes of each T cell are distinguished using TCR-β transcripts. We then apply data cleaning and batch correction methods according to the variational autoencoder framework (Lopez et al., Deep generative modeling for single-cell transcriptomics. Nature Methods 15: 1053-1058, 2018, incorporated herein by reference). Then, we use automatic cell type classification (Lotfollahi et al., Mapping single-cell data to reference atlases by transfer learning. Nature Biotechnology 40: 121–130, 2021, incorporated herein by reference) performed by transfer learning from a reconciled tumor microenvironment scRNA-seq reference data set. The resulting ex vivo dataset contained 156,522 single cells and was visualized using uniform manifold approximation and projection (McInnes et al., UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction, ArXiv e-prints 1802.03426, 2018, incorporated herein by reference; Fig.19A ). We then focused on T cell clustering and obtained a more granular subtype identification ( Fig.19B ). By tumor type ( Fig.19C ) and treatment conditions ( Fig.19D ) Annotate the origin of each single cell.
[0239] As the next step in the process, cells were grouped using unbiased clustering (Xu et al., Probabilistic harmonization and annotation of single-cell transcriptomics data with deep generative models. Molecular Systems Biology 17: 1-21, 2021, incorporated herein by reference; Fig. 20A ). For each single cell, an activation score was calculated, which is a composite of public markers defining primary human T cells activated by αCD3 and αCD28 in culture (Szabo et al., Single-cell transcriptomics of human Tcells reveals tissue and activation signatures in health and disease. Nature Communications 10:4706, 2019, incorporated herein by reference). Specifically, the activation score ( ) was obtained by adding the cytokine score to the proliferation score and then subtracting the resting score. Fig. 20B We then identified responding clusters (#3, #4, #10, #14, #16, #18) based on a threshold defined as the lowest median score of the cluster belonging to the positive control (anti-CD3) treatment condition (cluster #18). As expected, those responding clusters were enriched for cells from the treatment condition (refer to Fig.19D ) and also enriched for expanded clonal types in the absence of anti-CD3 ( Fig. 20C ).
[0240] The corresponding WR (will respond) and WNR (will not respond) T cell clonotypes in the control sample were also determined. The results are shown below:
[0241] R NR WR WN HFB10 1133 10739 591 440 HFB3 1560 12387 597 493 αPD1 360 8456 338 353 CD3 3080 1009 331 24
[0242] R: cells that respond under drug treatment conditions; NR: cells that do not respond under drug treatment conditions; WR: cells that would respond under control / baseline conditions; WNR: cells that would not respond under control / baseline conditions.
[0243] WO2021257890A1 and the methods described above were then used to identify predictive biomarker signatures (PRBs) using a clonotype barcoding approach. In brief, based on the response clusters (#3, #4, #10, #14, #16, #18), T cell clonotypes can be identified in treated samples, and corresponding responding (R) and non-responding (NR) T cells can be identified based on the matching clonotypes in the control samples ( Fig.21 ).
[0244] Predictive gene expression signatures were then defined by identifying differentially expressed genes between responding or non-responding T cells in control samples (Lopez et al.; Fig.22A and Fig. 22B ). The top 30 most differentially expressed genes with a log fold change > 1 were of particular interest. Logarithmic regression weights were calculated based on the z-scored gene expression of each gene. For application to bulk RNA-seq data, genes specifically expressed in T cells were then filtered by requiring a correlation with CD8A of at least 0.4. The final list of PRB genes and their corresponding weights are shown in Table 3-1.
[0245] Table 3-1. Anti-PD-1 10-gene prediction marker genes (PRB) and their corresponding weights calculated by logarithmic regression
[0246]
[0247]
[0248] The ex vivo αPD1 PRB biomarker includes some of the most differentially expressed genes that distinguish WE from WNE cells in breast cancer. This biomarker can serve as a proxy for predicting in vivo clonal expansion.
[0249] To further validate this strategy, single-cell data were generated in an ex vivo system to characterize the response to anti-PD-1 therapy. Among the anti-PD-1 predictive response signatures, as expected, genes involved in the inflammatory response and genes in other co-inhibitory checkpoint pathways were identified.
[0250] Application of the anti-PD-1 predictive response signature to large transcriptomic data from clinical studies using checkpoint inhibitors successfully stratified patients into two groups with significantly different risks of progression.
[0251] Specifically, to validate the predictive markers, the anti-PD-1 predictive markers were applied to data from the atezolizumab clinical trial (Fehrenbacher et al., Lancet 387: 1837-1846, 2016; Mariathasan et al., Nature 554: 544-548, 2018; McDermott et al., Nat Med 24: 749-757, 2018, all incorporated herein by reference). Based on the markers calculated as the weighted sum of all genes, each patient in the data set was stratified into high biomarkers and low biomarkers using the median cutoff point. The hazard ratio and statistical significance from Cox regression were then calculated, and it was found that compared with other publicly available predictive markers (Ayers et al.; US2016 / 0304969 A1; Wu et al.) or other known single genes of interest (CD274, PDCD1, CD8A, CD4, FOXP3), the marker score predicted equally well or better in all four treatment arms of these trials (Table 3-2). As expected, this marker did not stratify patients in the control arms of these studies.
[0252] Table 3-2. αPD1 biomarker derived from ex vivo models predicts ICB-specific patient survival in clinical trials
[0253]
[0254]
[0255] Column A - IMvigor210 atezolizumab; Column B - POPLAR atezolizumab; Column C - IMmotion150 atezolizumab; Column D - IMmotion150 atezolizumab + bevacizumab; Column E - POPLAR docetaxel; Column F - IMmotion150, sunitinib. ***p val < 0.001; **p val < 0.01; *p val < 0.05.
[0256] The data show that for the specific trial arms of IMvigor210, POPLAR and IMmotion150, the subject αPD-1 biomarker (PRB) was used together with other disclosed αPD1 biomarker markers to dichotomize patients by median marker score. The PFS hazard ratio was calculated by fitting the Cox proportional hazard model relative to the PFS survival time examined using the dichotomous variable. The subject PRB appears to be the only one that generated statistically significant results in all 4 trials (see AD column).
[0257] Additional data showed that clusters with a higher proportion of PD1 biomarker+ cells also tended to have a higher proportion of WE cells (data not shown). This result suggests that the αPD1 biomarker corresponds to the WE / WNE state.
[0258] Interestingly, predictive response signatures for two experimental treatments (HFB3 and HFB10) were found to share genes involved in inflammatory pathways with the anti-PD-1 signature but also contained other distinct sets of genes.
[0259] Predictive biomarkers of response to TNFR2 (HFB3) and OX40 (HFB10) agonist antibodies can now be validated in Phase I clinical trials and can be used as a patient stratification approach to enrich for patients who are more likely to respond to monotherapy or combination.
[0260] Collectively, ex vivo single-cell profiling combined with TCR clonotype characterization enables the discovery of predictive response markers (or PRBs) that can inform patient selection strategies for early clinical development of novel therapeutics.
[0261] As a proof of principle, TCR clonotype barcode analysis was used to extract predictive biomarkers (PRBs) for responses to αPD1, HFB200301, and HFB301001. The predictive nature of the αPD1 biomarker was validated using a large number of RNA-seq profiles from the IMvigor210, POPLAR, and IMmotion150 clinical trials. Together, these data demonstrate the effectiveness of the inventive method ("DIS") in combining scRNA-seq and scTCR-seq signature analysis of T cell biology associated with the TME using an ex vivo culture system.
[0262] Therefore, when applied to the large amount of transcriptomic data from TCGA, these same markers can be used to stratify patients by their predicted likelihood of response to single or combined treatments. The HFB200301 and HFB301001 biomarkers will be validated in a Phase I clinical trial.
[0263] References
[0264] Ayers,M.,et al.(2017).IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade.Journal of Clinical Investigation,127,2930–2940.
[0265] Fehrenbacher,L.,et al.(2016).Atezolizumab versus docetaxel forpatients with previously treated non-small-cell lung cancer(POPLAR):Amulticentre,open-label,phase 2randomised controlled trial.The Lancet,387,1837–1846.
[0266] Hifibio Therapeutics,WO2021257890A1(2019).Predictive responsebiomarker discovery process.
[0267] Lopez,R.,et al.(2018).Deep generative modeling for single-celltranscriptomics.Nature Methods,15,1053–1058.
[0268] Lotfollahi,M.,et al.(2021).Mapping single-cell data to referenceatlases by transfer learning.Nature Biotechnology,40,121–130.
[0269] Mariathasan,S.,et al.(2018).TGFβattenuates tumour response to PD-L1blockade by contributing to exclusion of T cells.Nature,554,544–548.
[0270] McDermott,D.F.,et al.(2018).Clinical activity and molecularcorrelates of response to atezolizumab alone or in combination withbevacizumab versus sunitinib in renal cell carcinoma.Nature Medicine,24,749–757.
[0271] McInnes,L,Healy,J,UMAP:Uniform Manifold Approximation and Projectionfor Dimension Reduction,ArXiv e-prints 1802.03426,2018.
[0272] Merck,US2016 / 0304969 A1.IFN-gamma gene signature biomarkers of tumorresponse to PD-1antagonists.
[0273] Szabo,P.A.,et al.(2019).Single-cell transcriptomics of human T cellsreveals tissue and activation signatures in health and disease.NatureCommunications,10,4706.
[0274] Wu,T.D.,et al.(2020).Peripheral T cell expansion predicts tumourinfiltration and clinical response.Nature,579,274–278.
[0275] Xu,C.,et al.(2021).Probabilistic harmonization and annotation ofsingle-cell transcriptomics data with deep generative models.MolecularSystems Biology,17,1-21。
Claims
1. A method for identifying a predictive response biomarker (PRB) for treating cancer (e.g., renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL), or uterine cancer (UC)) using an immunomodulatory therapy targeting T cells (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3), and / or an anti-OX40 agonist antibody (such as HFB10)), the method comprising the following steps: (a) identifying, in a pre-treatment sample of the cancer, responsive T cells having a TCR (T cell receptor) clonotype identical to a TCR clonotype of a clonally expanded T cell in a matched post-treatment sample of the cancer treated with the T cell-targeted immunomodulatory therapy, wherein the clonally expanded T cell in the matched post-treatment sample is clonally expanded after treatment with the T cell-targeted immunomodulatory therapy; and, (b) generating a list of genes that are differentially upregulated and / or downregulated (e.g., log fold>1, or upregulated by 10-fold or downregulated by 10-fold) in the responsive T cells in the pre-treatment sample (compared to the corresponding genes in the unresponsive T cells in the pre-treatment sample) to generate the PRB; (c) Optionally, weighting each gene in the PRB using a logistic regression coefficient obtained by fitting each gene as a predictor to amplification status and / or clinical response status as an outcome.
2. The method according to claim 1, wherein step (a) comprises: (1) generating the matched post-treatment sample by contacting an ex vivo culture of an untreated sample of the cancer with an immunomodulatory compound targeting T cells (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3) and / or an anti-OX40 agonist antibody (such as HFB10)) under conditions and for a period of time sufficient to allow the immunomodulatory effect of the compound on the T cell population in the ex vivo culture to manifest; (2) encapsulating the single T cells separated, purified or enriched from the ex vivo culture into picoliter droplets for single-cell characterization of functional properties, thereby dividing each encapsulated single T cell into a first responding T cell pool and a second non-responding T cell pool based on the presence or absence of the functional properties, respectively; (3) determining the TCR clonotype of each encapsulated single T cell in the first responder T cell library and the second non-responder T cell library, thereby identifying the TCR clonotype of the responder T cell as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample; (4) encapsulating single T cells isolated, purified or enriched from the pre-treatment sample into picoliter-scale droplets to identify the responsive T cells having the same TCR clonotype as the TCR clonotype of the clonally expanded T cells in the matched post-treatment sample; or, (1) obtaining the pre-treatment sample and the matched post-treatment sample of the cancer from a suitable donor, wherein the matched post-treatment sample has been contacted with the compound under conditions and for a period of time sufficient to allow the immunomodulatory effect of the compound on the T cell population in the post-treatment sample to manifest; (2) encapsulating the single T cells separated, purified or enriched from the post-treatment sample into picoliter droplets for single-cell characterization of functional properties, thereby dividing each encapsulated single T cell into a first responding T cell pool and a second non-responding T cell pool based on the presence or absence of the functional properties, respectively; (3) determining the TCR clonotype of each encapsulated single T cell in the first responder T cell library and the second non-responder T cell library, thereby identifying the TCR clonotype of the responder T cell as the TCR clonotype of the clonally expanded T cell in the matched post-treatment sample; (4) encapsulating single T cells isolated, purified or enriched from the pre-treatment sample into picoliter droplets to identify the responsive T cells having the same TCR clonotype as the TCR clonotype of the clonally expanded T cells in the matched post-treatment sample.
3. The method according to claim 2, wherein step (b) comprises: (5) using (whole transcriptome) single-cell RNA sequencing (scRNA-seq) to identify the list of genes that are upregulated in the responsive T cells in the pre-treatment sample, wherein each gene in the PRB has a log2 fold change > 0.2 and is expressed in <40% of T cells with a TCR clonotype that is not clonally expanded in the matched post-treatment sample; Optionally, wherein the scRNA-seq is performed using a gene panel specifically designed for the mechanism of action of the compound or a predetermined gene panel designed to evaluate immune modulation.
4. The method according to any one of claims 1 to 3, wherein (1) the pre-treatment sample is a blood sample; and / or (2) The cancer is renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL) or uterine cancer (UC).
5. The method according to any one of claims 1 to 4, wherein the ex vivo culture is: (1) Freshly isolated from cancer tissue; (2) established from stored (e.g., frozen) cancer tissue; (3) single cell suspension; and / or, (4) Adherent cultures.
6. The method according to any one of claims 2 to 5, wherein: (1) the time period is about 12 hours to 24 hours, 24 hours to 36 hours, 36 hours to 48 hours, up to 3 days, up to 4 days or up to 5 days; and / or (2) The conditions include: one concentration in a series of concentrations of the compound, and / or with or without combination with a second therapeutic agent; optionally, the second therapeutic agent includes cytokines (e.g., IL-2, TNF), TCR stimulation (e.g., through CD3 cross-linking) and / or co-stimulation (e.g., CTLA, B7, CD28, etc.).
7. The method according to any one of claims 1 to 6, wherein the compound is an immunomodulatory antibody.
8. The method according to any one of claims 2 to 7, wherein step (2) is performed by a microfluidic device or a droplet chip device (eg, a device capable of performing phenotypic / genotypic assays).
9. The method according to any one of claims 2 to 8, wherein the single T cells are isolated, purified or enriched by: (1) Isolation, purification or enrichment of specific T cell subsets or populations (such as CD4 + or CD8 + T cells); and / or (2) Extensively isolate, purify or enrich T cells.
10. The method according to any one of claims 2 to 9, wherein the functional characteristics include: IFN-γ secretion, secretion of markers associated with T cell response or activation and / or upregulation of activation markers.
11. The method according to any one of claims 3 to 10, wherein the gene set comprises: (a) proliferation-related genes (such as Ki67); stem-like feature-related genes (such as TCF7); T cell activation-related genes (such as CD25, granzyme B, perforin, CD28, TNF, IL2, IFNG, 4-1BB, CD38, CD69, OX-40, GITR, IL4, IL6); T cell exhaustion-related genes (CD39, CTLA-4, EOMES, PD-1, TIGIT, 2B4, LAG-3, T-bet, TIM3, TOX, CD160, IL-10); migration-related genes (such as CD31, CD103, CXCR5, CXCR4, CCR7, CCR3, CCR4, CCR8, CCR5, CXCR3); and combinations thereof; optionally, the gene set further includes CD4, CD8, FOXP3, CD62L, CD44, CD127, CD27, IL33R and / or PTPRC; or, (b) CST7 (optionally with a weight of 10.8), GZMA (optionally with a weight of 10.7), GZMB (optionally with a weight of 8.2), GZMK (optionally with a weight of 11.3), HAVCR2 (optionally with a weight of 10.7), IFNG (optionally with a weight of 10.0), LAG3 (optionally with a weight of 11.2), NKG7 (optionally with a weight of 9.9), PRF1 (optionally with a weight of 9.9) and TMEM155 (optionally with a weight of 10.8).
12. The method according to any one of claims 2 to 11, wherein in step (2), the functional characteristic (1) for physically separating the responder T cells and the non-responder T cells based on the presence or absence of the functional characteristic using a microfluidic device or a droplet chip device; or, (2) is an activation marker or a response marker (either or both of which can be determined by, for example, single-cell RNA sequencing), and wherein the responder T cells and the non-responder T cells are not physically separated, but are actually distinguished based on the expression or lack of expression of the activation marker or response marker.
13. The method according to any one of claims 2 to 12, further comprising characterizing the pre-treatment sample and / or the post-treatment sample to characterize, for example, the cell type composition of the sample (e.g., via flow cytometry) and target expression on T cells.
14. A method of selecting a patient for treatment of cancer (e.g., renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL), or uterine cancer (UC)) using an immunomodulatory compound targeting T cells (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3), and / or an anti-OX40 agonist antibody (such as HFB10)), the method comprising: Obtaining a clinical response score for the patient by evaluating the expression status and / or expression level of a gene in the predictive response biomarker (PRB) for the cancer obtained by the method of any one of claims 1 to 13 to determine whether the clinical response score of the patient exceeds a predetermined threshold clinical response score; wherein patients whose clinical response scores exceed said threshold score are identified as being beneficial to said treatment and are selected for said treatment, and / or, Patients in which the clinical response score is below the second threshold score are identified as not benefiting from the treatment and are not selected for the treatment.
15. A method for treating a patient's cancer (e.g., renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL), or uterine cancer (UC)) using an immunomodulatory compound targeting T cells (e.g., an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3), and / or an anti-OX40 agonist antibody (such as HFB10)), the method comprising: (i) obtaining a clinical response score for the patient by evaluating the expression status and / or expression level of a gene in the predictive response biomarker (PRB) for the cancer obtained by the method of any one of claims 1 to 13 to determine whether the clinical response score of the patient exceeds a predetermined threshold clinical response score; Patients whose clinical response scores exceed the threshold score are identified as being worthy of treatment and are selected for treatment; and, (ii) administering a therapeutically effective amount of said compound to said patient identified in (i) as being beneficial to said treatment and selected for said treatment.
16. The method of claim 14 or 15, wherein the cancer is renal cell carcinoma (RCC), non-small cell lung cancer (NSCLC), melanoma (MEL), or uterine cancer (UC).
17. The method according to any one of claims 14 to 16, wherein the immunomodulatory compound targeting T cells comprises an anti-PD-1 antagonist antibody, an anti-PD-L1 antagonist antibody, an anti-TNFR2 agonist antibody (such as HFB3) and / or an anti-OX40 agonist antibody (such as HFB10).
18. The method of claim 17, wherein the T-cell-targeted immunomodulatory compound is an anti-PD-1 antagonist antibody or an anti-PD-L1 antagonist antibody, and wherein the PRB comprises CST7, GZMA, GZMB, GZMK, HAVCR2, IFNG, LAG3, NKG7, PRF1, and TMEM155.
19. The method of any one of claims 14 to 18, wherein the anti-PD-1 antagonist antibody comprises pembrolizumab, nivolumab, cemiplizumab, dotalimumab, spartalizumab, carrelizumab, sintilimab, tislelizumab, toripalimab, INCMGA00012, JTX-4014, AMP-224 or AMP-514; and / or the anti-PD-L1 antagonist antibody comprises atezolizumab, avelumab, durvalumab or KN035.
Citation Information
Patent Citations
IFN-gamma gene signature biomarkers of tumor response to PD-1 antagonists
US20160304969A1
Debondable adhesives and uses thereof
US20210037916A1
Anti-ox40 antibody and use thereof
WO2021129872A1
Predictive response biomarker discovery process
WO2021257890A1
