Methods for predicting response to immunotherapy

Abstract

The present invention provides a method for predicting a subject's response to immunotherapy, comprising the steps of: (a) staining a biological sample disposed on a substrate; (b) imaging the biological sample, whereby one or more images of a high power field (HPF) are generated; (c) detecting a plurality of biomarkers in the biological sample; and (d) analyzing the one or more images, thereby predicting the subject's response to immunotherapy.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 208,829, filed June 9, 2021. The disclosure of this prior application is deemed to be part of the disclosure of this application and is incorporated herein in its entirety. [Technical field]

[0002] The present disclosure relates to the field of biotechnology, and more specifically to tissue-based biomarker assays for immunotherapy.

[0003] [Federally Sponsored Research and Development] This invention was made with Government support under Grant CA142779 awarded by the National Institutes of Health. The Government has certain rights in this invention. [Background technology]

[0004] Patients with multiple solid tumors have shown high rates of tumor regression and improved survival after treatment with immune checkpoint inhibitors. Unfortunately, in most cancer types, fewer than half of patients respond to anti-PD-L1 drugs, so the development of predictive biomarkers to precisely guide each patient's treatment is crucial. PD-L1 immunohistochemistry (IHC) on pretreatment tumor biopsies is a common tissue-based biomarker approach to predict responsiveness to anti-PD-(L)1 drugs and has many companion diagnostic indications; however, its expression as a single marker has limited predictive power. Other approaches include assessment of microsatellite instability, testing the tumor for gene mutation counts, detection of interferon (IFN)-γ gene signatures, and quantification of multiple proteins by multiplex immunofluorescence (mIF) / IHC. A recent meta-analysis showed that mIF / IHC has improved diagnostic performance over other tissue-based approaches when predicting response to anti-PD-(L)1, highlighting the biomarker potential of these new techniques. Summary of the Invention [Problem to be solved by the invention]

[0005] The present disclosure is based on the discovery that by detecting multiple proteins as markers of immune cell type and function, as well as tumor cell markers in fixed tumor samples using immunofluorescence and / or immunohistochemistry, it is possible to predict the response of a subject to the immunotherapy described herein.In particular, this method analyzes the expression levels of some of these markers, as well as the co-expression of markers in individual cells.Without wishing to be bound by any theory, it has been discovered that the analysis of biomarkers in biological samples can be used to predict the response of a subject to immunotherapy.

[0006] Provided herein is a method for predicting a subject's response to immunotherapy, the method comprising: (a) staining a biological sample disposed on a substrate; (b) imaging the biological sample, where one or more images of a high power field (HPF) are generated; (c) detecting a plurality of biomarkers in the biological sample; and (d) analyzing the one or more images, thereby predicting the subject's response to immunotherapy.

[0007] Also provided herein is a method for stratifying a subject and placing the subject into a therapeutic category, the method comprising: (a) staining a biological sample disposed on a substrate; (b) imaging the biological sample, where one or more images of a high power field (HPF) are generated; (c) detecting a plurality of biomarkers in the biological sample; and (d) analyzing the one or more images, thereby stratifying the subject and placing the subject into a therapeutic category.

[0008] In some embodiments, the plurality of biomarkers comprises PD-1, PD-L1, CD8, FoxP3, CD163, a tumor cell marker, or any combination thereof. In some embodiments, the tumor cell marker comprises Sox10, S100, or both.

[0009] In some embodiments, the staining comprises immunofluorescence staining. In some embodiments, the staining comprises immunohistochemical staining. In some embodiments, the biological sample is stained with an antibody. In some embodiments, the antibody is a monoclonal antibody. In some embodiments, the antibody is a polyclonal antibody. In some embodiments, the biological sample is stained with one or more antibodies. In some embodiments, the biological sample is stained with six antibodies. In some embodiments, the biological sample is stained with four antibodies. In some embodiments, the biological sample is stained with a second antibody that detects the antibody. In some embodiments, the second antibody is conjugated to a label. In some embodiments, the label is a detectable label. In some embodiments, the label is a fluorophore. In some embodiments, the imaging step (c) comprises performing immunofluorescence microscopy on the biological sample.

[0010] In some embodiments, the analyzing step (d) comprises: (i) image acquisition and processing; (ii) cell segmentation and phenotyping; and (iii) image normalization. In some embodiments, the image acquisition step comprises compiling one or more images to obtain an image of the entire biological sample within the substrate. In some embodiments, the compiling comprises overlaying and aligning one or more images.

[0011] In some embodiments, the cell segmentation and phenotyping step comprises identifying a cell type in the biological sample. In some embodiments, the phenotyping step comprises detecting at least one expression of a biomarker in the cell type. In some embodiments, the expression of at least one biomarker is designated as low, mid, or high. In some embodiments, the cell type comprises CD163+ macrophages, CD8+ T cells, Treg cells (CD8negFoxP3+), tumor cells, CD8+FoxP3+ cells, or any combination thereof. In some embodiments, a CD8+FoxP3+PD-1low / mid cell type is identified as an indication that the subject will respond to immunotherapy. In some embodiments, a CD163+PD-L1neg cell type is identified as an indication that the subject will not respond to immunotherapy to the same extent as a reference subject identified as not having a CD163+PD-L1neg cell type.

[0012] In some embodiments, the cell segmentation and phenotyping step further comprises determining the density of a cell type in the biological sample. In some embodiments, the density of a cell type in the biological sample is determined by analyzing the distance between a cell and another cell. In some embodiments, the density of a cell type in the biological sample is determined by analyzing the distance between a cell and a tumor-stroma boundary. In some embodiments, a high density of CD8+FoxP3+ cells is identified as an indication that the subject will respond to immunotherapy.

[0013] In some embodiments, the step of image normalization comprises calibrating the fluorescence intensity of at least one of the biomarkers in the one or more images to the tissue microarray.

[0014] In some embodiments, the analyzing step (c) further comprises identifying at least one biomarker in a biological sample from a subject having the disease, and the identification of the at least one biomarker is used to predict the subject's response to immunotherapy and / or to stratify the subject and place the subject in a treatment category. In some embodiments, the disease is cancer. In some embodiments, the cancer is a metastatic solid tumor. In some embodiments, the cancer is melanoma. In some embodiments, the cancer is non-small cell lung cancer. In some embodiments, the cancer is selected from bladder cancer, breast cancer, cervical cancer, colon cancer, endometrial cancer, esophageal cancer, fallopian tube cancer, gallbladder cancer, gastrointestinal cancer, head and neck cancer, hematological cancer, Hodgkin's lymphoma, laryngeal cancer, liver cancer, lung cancer, lymphoma, melanoma, mesothelioma, ovarian cancer, primary peritoneal cancer, salivary gland cancer, sarcoma, gastric cancer, thyroid cancer, pancreatic cancer, renal cell carcinoma, glioblastoma, and prostate cancer.

[0015] In some embodiments, the immunotherapy comprises administration of an immune checkpoint inhibitor. In some embodiments, the treatment categories comprise radiation therapy, chemotherapy, immunotherapy, hormone therapy, antibody therapy, or any combination thereof.

[0016] In some embodiments, the substrate is a slide. In some embodiments, the biological sample comprises a tissue, a tissue section, an organ, an organism, an organoid, or a cell culture sample. In some embodiments, the tissue is a formalin-fixed paraffin-embedded (FFPE) tissue. In some embodiments, the biological sample is fixed before step (a). In some embodiments, the biological sample is fixed with formaldehyde. In some embodiments, the biological sample is fixed with methanol.

[0017] Also provided herein is a method for improving the predictive value of a biomarker, the method comprising: (a) obtaining a plurality of high power field (HPF) images generated from a biological sample; (b) detecting a biomarker in each of the plurality of images; (c) selecting a sub-plurality of images from the plurality of images of step (a); and (d) analyzing the sub-plurality of images, thereby improving the predictive value of the biomarker.

[0018] In some embodiments, the method further comprises generating an area under a receiver operating characteristic (ROC) curve value that is greater than the area under a receiver operating characteristic (ROC) curve value generated when analyzing all of the images.

[0019] In some embodiments, the biomarkers include PD-1, PD-L1, CD8, FoxP3, CD163, tumor cell markers, or any combination thereof. In some embodiments, the tumor cell markers include Sox10, S100, or both.

[0020] In some embodiments, the sub-plurality of images is 30% of the plurality of images of step (a).In some embodiments, obtaining step (a) comprises performing immunofluorescence microscopy on the biological sample.

[0021] In some embodiments, the analyzing step (d) comprises (i) image acquisition and processing, (ii) cell segmentation and phenotyping, and (iii) image normalization. In some embodiments, the image acquisition step comprises compiling multiple images of high power fields (HPF) to obtain an image of the entire biological sample. In some embodiments, the compiling comprises overlaying and aligning the multiple images.

[0022] In some embodiments, the cell segmentation and phenotyping step comprises identifying cell types in the biological sample. In some embodiments, the phenotyping step comprises detecting expression of a biomarker in the cell type. In some embodiments, the expression of the biomarker is designated as low, mid, or high. In some embodiments, the cell type comprises CD163+ macrophages, CD8+ T cells, Treg cells (CD8negFoxP3+), tumor cells, CD8+FoxP3+ cells, or any combination thereof. In some embodiments, a CD8+FoxP3+PD-low / mid cell type is identified as an indication that the subject will respond to immunotherapy. In some embodiments, a CD163+PD-L1neg cell type is identified as an indication that the subject will not respond to immunotherapy to the same extent as a reference subject identified as not having a CD163+PD-L1neg cell type.

[0023] In some embodiments, the cell segmentation and phenotyping step further comprises determining the density of a cell type in the biological sample. In some embodiments, the density of a cell type in the biological sample is determined by analyzing the distance between a cell and another cell. In some embodiments, the density of a cell type in the biological sample is determined by analyzing the distance between a cell and a tumor-stroma boundary. In some embodiments, a high density of CD8+FoxP3+ cells is identified as an indication that the subject will respond to immunotherapy.

[0024] In some embodiments, the step of image normalization comprises calibrating the fluorescence intensities of the biomarkers in the multiple images to the tissue microarray.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this invention belongs.The present invention can be carried out using methods and materials similar or equivalent to those described herein, but the preferred methods and materials are described below.All publications, patent applications, patents, and other documents described herein are incorporated by reference in their entirety.In the event of any discrepancy, the present specification, including definitions, will prevail.In addition, the materials, methods, and examples are illustrative only and are not intended to be limiting.

[0026] The details of one or more embodiments of the invention are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the invention will become apparent from the description and drawings, and from the claims. [Brief description of the drawings]

[0027] [Figure 1] An exemplary schematic of the AstroPath platform for staining optimization and image processing to generate high-quality datasets. The use of the TSA-based AstroPath workflow for multiplex IF and imaging, and associated data, is detailed with optimization of a 6-plex assay (PD-1, PD-L1, CD163, FoxP3, CD8, Sox10 / S100, DAPI) to characterize PD-1 and PD-L1 expression. Solutions to common limitations and sources of error are outlined. [Diagram 2]Figure 2A shows optimization of staining to achieve high sensitivity and specificity using chromogenic IHC. Trends in staining index (SI) and bleed-through (BT) used to inform TSA fluorophore and marker pairing are shown. Figure 2B shows optimization of staining to achieve high sensitivity and specificity using chromogenic IHC. Sensitivity of IF staining compared to chromogenic IHC is shown, with reduced native signal for PD-1, PD-L1, and FoxP3 using the manufacturer's recommended protocol. Sensitivity was improved by replacing the secondary antibody. Figure 2C shows optimization of staining to achieve high sensitivity and specificity using chromogenic IHC. Dilution of primary antibody to optimize signal-to-noise (S / N) ratio is shown, with 1:100 being the optimal dilution for CD8 IF staining. Figure 2D shows optimization of staining to achieve high sensitivity and specificity using chromogenic IHC. Optimal concentrations for each TSA fluorophore are shown. To ensure the sensitivity of the assay, only dilutions with signals comparable to chromogenic IHC (light grey bars) were considered. To minimize BT between channels, the lowest acceptable TSA concentration was selected for a marker (e.g., CD8 / 540). However, if a fluorophore-marker pair was susceptible to BT, the highest acceptable TSA concentration was selected to raise the true positive threshold (e.g., FoxP3 / 570). Figure 2E illustrates the optimization of staining to achieve high sensitivity and specificity using chromogenic IHC. Detection of each marker in multiplex IF was compared to the respective monoplex IF for final validation, confirming equivalence. [Diagram 3]FIG. 3A illustrates minimizing instrumental errors during field acquisition and stitching whole slides using lessons from astronomy. Images of the whole tissue of interest taken are shown using 20% ​​overlapping HPFs as shown in low and high power images (average of 1300 fields taken per case). FIG. 3B illustrates minimizing instrumental errors during field acquisition and stitching whole slides using lessons from astronomy. Images are shown where each HPF was found to have instrumental imaging errors including lens distortion and field illumination variations. FIG. 3C illustrates minimizing instrumental errors during field acquisition and stitching whole slides using lessons from astronomy. Pixels are shown when overlapping image regions compared to determine field alignment error. To improve alignment, a spring-based model was used to minimize pixel shift. Misalignment errors were reduced from ±3 pixels in the x direction and ±5 pixels in the y direction to less than ±1 pixel in both (ranges reported for 95-5th percentiles). Lighting variability also decreased from 11.2% to 1.2%. [Figure 4]Figure 4A shows the spatially distinct immune cell populations and marker expression in situ. Representative mIF images show that T cells expressing PD-1low are adjacent to cells expressing PD-L1high, with hotspots at the edge of the tumor. Within the tumor parenchyma, PD-1high and PD-1mid cells are observed adjacent to PD-L1low expression, consistent with a more exhausted T cell phenotype. A histogram including all cases in the cohort shows the cell density of CD8+ cells expressing PD-1 as a function of distance to the tumor border. The intensity of PD-1 expression increased as T cells were exposed to tumor antigens. Figure 4B shows the spatially distinct immune cell populations and marker expression in situ. Representative images of metastatic melanoma deposits showing localization of CD8+FoxP3+ cells in areas of dense infiltration of CD8+PD-1neg and CD8+PD-1+ cells adjacent to tumor cells showing adaptive (IFN-γ-driven) PD-L1 expression by the tumor. A histogram including all cases in the cohort shows that CD8+FoxP3+ cells are most likely to localize near CD8+PD-1neg cells. Other cell types in the same relative position to the tumor-stroma border include CD8+PD-1+ cells and PD-L1+ tumor cells. [Diagram 5]Figure 5A shows AUC heatmaps of response to treatment as a function of various immune cell types expressing PD-1 / L1 and PD-1 / L1 expression intensity using two different slide sampling strategies. PD-1 / PD-L1 mIF assay combined with hotspot HPF selection shows that the density of CD8+FoxP3+PD-1low / mid, tumor PD-L1neg and CD163+PD-L1neg cells has the highest value among individual features for predicting response and non-response to anti-PD-1. Approximately 86% of CD8-FoxP3-PD-1pos cells in melanoma are conventional CD4T- cells. Figure 5B shows AUC heatmaps of response to treatment as a function of various immune cell types expressing PD-1 / L1 and PD-1 / L1 expression intensity using two different slide sampling strategies. Similar characterization using representative field sampling is shown, highlighting similar key features associated with responsiveness to treatment. However, the resulting AUC, especially for the CD8+ cell subset, was not as high using this approach. This finding highlights the fact that slide sampling is another component of assay performance; it is also one that can be optimized and standardized. Tumor PD-L1neg, CD163+PD-L1neg are negatively correlated features, while all others are positively correlated features. [Figure 6]Figure 6A shows a multifactorial analysis of a 6-plex mIF assay focusing on the intensity of PD-1 and PD-L1 to predict objective response and long-term survival. Table showing 10 features associated with responsiveness to treatment by univariate analysis in 30% hotspot HPFs. Features are listed in order of decreasing predictive value. Figure 6B shows a multifactorial analysis of a 6-plex mIF assay focusing on the intensity of PD-1 and PD-L1 to predict objective response and long-term survival. Combinatorial ROC curves and corresponding AUC values ​​are shown for the 10 features evaluated in the Discovery cohort and an independent second cohort. Figure 6C shows a multifactorial analysis of a 6-plex mIF assay focusing on the intensity of PD-1 and PD-L1 to predict objective response and long-term survival. TME from a patient is shown, where poor prognosis is characterized by tumor cells lacking PD-L1 expression and a high density of CD163+ cells, with or without the presence of other immune cells (left panel). Patients with an intermediate prognosis have a CD163+PD-L1neg myeloid-rich TME with low levels of immune infiltration (middle panel). Patients with the best prognosis have a highly inflamed TME, with CD8+ and CD8+FoxP3+ T cells expressing various levels of PD-1 and PD-L1 (right panel). PD-L1 expression is also evident on CD163+ cells. Figure 6D shows a multifactorial analysis of the 6-plex mIF assay focusing on PD-1 and PD-L1 intensity to predict objective response and long-term survival. A Kaplan-Meier analysis is shown stratifying patients into poor, intermediate, and good overall survival (OS) and progression-free survival (PFS) in the discovery cohort by distinct TME defined by specific cell types expressing various levels of PD-1 and PD-L1. Similar stratification of patient outcomes was achieved using an independent validation cohort from another institution (OS, p=0.036; PFS, p=0.024; log-rank test). [Figure 7]FIG. 7A is an exemplary schematic diagram of the Tyramide signal amplification (TSA) technique that can be used to amplify signals and visualize multiple markers on a single slide. An exemplary schematic diagram showing that TSA detection allows for a larger amplification of signals (~1000-fold) compared to staining with fluorophore-tagged secondary antibodies. This ability is due to the deposition of multiple TSA fluorophore molecules by an enzyme-catalyzed reaction. FIG. 7B is an exemplary schematic diagram of the Tyramide signal amplification (TSA) technique that can be used to amplify signals and visualize multiple markers on a single slide. The multiplex staining process is shown, which is divided into three stages: slide preparation, sequential staining, and final processing. FIG. 7C is an exemplary schematic diagram of the Tyramide signal amplification (TSA) technique that can be used to amplify signals and visualize multiple markers on a single slide. An exemplary schematic diagram showing that in the sequential staining stage, microwave treatment (MWT) strips the antibodies from the previous staining round while retaining the deposited TSA fluorophore due to its strong binding to the tissue. The staining process can be repeated for multiple markers without cross-reactivity. [Figure 8] Multispectral image acquisition using a Vectra 3.0 system that allows simultaneous visualization of six channels of interest and DAPI. A mercury halogen lamp emits light and is received by five excitation cubes with wavelengths in the visible spectrum. The light is then received by liquid crystal tunable filters that transmit specific wavelengths, each forming an individual monochromatic image plane. The resulting image is unmixed with a library of pure spectra for each fluorophore. The individual images for each fluorophore are then pseudocolored and overlaid to form a composite image. The unmixed images are further processed using inForm™ software. [Figure 9]Figure 9A illustrates the characterization of TSA fluorophores in terms of staining index (SI) and bleed-through (BT). It has been shown that SI is a useful signal / background indicator to quantify the brightness of immunofluorescence reagents. Fluorophores 540 and 620 showed the lowest and highest SI, respectively. TSA fluorophores with lower SI were paired with more abundant markers. For example, CD8 (an abundant and intense antigen) was paired with Opal540 (a fluorophore with lower SI). Figure 9B illustrates the characterization of TSA fluorophores in terms of staining index (SI) and bleed-through (BT). It has been shown that BT can result in the detection of false positive signals in one channel due to spillover from another channel. The BT trends were characterized when each fluorophore was used at a dilution of 1:50. Top left: The log of normalized fluorescence intensity was plotted for each possible pair of TSA fluorophores and TSA fluorophores. A parameterized hyperbolic sine curve was fitted as shown in the graph. Top right: The table shows the trend of BT from each channel to the other, i.e., A*a of the parameterized hyperbolic sine function. BTs between fluorophores ranked from high to low are most pronounced: 540 to 570, 650 to 620, 520 to 570, 540 to 620, 540 to 520. Bottom left: Examples of low and high BTs. Bottom right: Representative BTs from 540 to 570 shown in a micrograph, with membranous signal from CD8 cells in the 570 (FoxP3) channel (actual FoxP3 staining is nuclear). The trend of BTs is further reduced by diluting the TSA fluorophore in subsequent panel optimization steps. [Figure 10]Figure 10A shows that primary antibody optimization is required to maximize the specificity of IF staining using chromogenic IHC as the absolute standard. After selecting the appropriate HRP-conjugated polymer, dilutions of the primary antibody were performed to optimize the signal-to-noise (S / N) ratio, i.e., specificity. Representative image of CD8 monoplex IF staining showing that 1:100 is the dilution with the optimal S / N ratio. Concordance was observed between the three different signal quantification approaches. Figure 10B shows that primary antibody optimization is required to maximize the specificity of IF staining using chromogenic IHC as the absolute standard. After selecting the appropriate HRP-conjugated polymer, dilutions of the primary antibody were performed to optimize the signal-to-noise (S / N) ratio, i.e., specificity. It has been shown that when the appropriate HRP-conjugated polymer is combined with the optimal primary antibody concentration, monoplex IF shows a signal comparable to chromogenic IHC. [Figure 11] FIG. 11A shows a comparison of monoplex and multiplex IF staining. It is a bar graph showing that when optimized as detailed herein, multiplex IF results in a similar percentage of positive cells as monoplex IF. FIG. 11B shows a comparison of monoplex and multiplex IF staining. It is shown that the usable dynamic range of epitopes was reduced by an average of 13% in the multiplex IF format. Each dot represents the difference between the 95th and 5th percentiles of the mean normalized fluorescence intensity of positive cells of a single HPF. [Figure 12]FIG. 12A shows that the optimal overlap of adjacent tiles is x=20% of the tile width and height. It shows that overlapping image tiles (examples shown in red) creates seamless coverage of the entire area, constructed from the central rectangles of each image (blue lines, pink shading indicates one central rectangle). These central rectangles form the statistical sample for analysis and completely cover the tissue. The overlap (dark blue shading) is observed multiple times and used to estimate the intrinsic error. FIG. 12B shows that the optimal overlap of adjacent tiles is x=20% of the tile width and height. It shows that too much overlap is "costly" in terms of data resources and time, while too little overlap does not provide enough information to correct for image defects. The amount of information in estimating the correction (inverse variance) is proportional to the areas T and O, respectively. In the formula, the useful area is T, which represents the tissue area on the slide, and O is the overlap area. 12C shows that the optimal overlap of adjacent tiles is x=20% of the tile width and height. It shows that the optimal solution is x=0.2, or 20%, which corresponds to when the area imaged multiple times (O) is equal to the area of ​​the tissue itself (T). [Figure 13]Figure 13A shows that image processing of individual fields includes flat-field correction of systematic illumination variations. 11,508 image averages were stacked to define the average illumination variation per image layer across a single HPF. Shown is the uncorrected, smoothed average image of layer 13 (FITC broadband filter, PD-L1). Figure 13B shows that image processing of individual fields includes flat-field correction of systematic illumination variations. The flat-field model was developed and applied to result in a smoothed, corrected average image. Figure 13C shows that image processing of individual fields includes flat-field correction of systematic illumination variations. The relative pixel intensities of uncorrected and corrected images are shown, demonstrating a consistent nine-fold reduction in illumination variation (11.2% to 1.2% for the 5th to 95th percentiles, 3.6% to 0.4% for the mean standard deviation). Here, pixel intensity relative to the average layer intensity across all image layers is shown for one representative specimen. [Figure 14]Figure 14A shows that stitching image tiles generated using the 20% overlap approach into an absolute Cartesian coordinate system creates a whole slide image accurate to a fraction of a pixel without losing information. It has been shown that simply stitching image tiles next to each other can lead to a loss of reliable information for approximately 3-6% of cells. Figure 14B shows that stitching image tiles generated using the 20% overlap approach into an absolute Cartesian coordinate system creates a whole slide image accurate to a fraction of a pixel without losing information. A schematic visualization of how the jumps in the mechanical stage movement and the inaccuracies of the underlying stitching algorithm accumulate in the x and y directions across the slide is shown. The relative displacements in the x and y directions required to seamlessly stitch the image tiles are denoted as dx and dy. It was found that the average accumulated shift across the slide results in an error of 20 μm in both the x and y directions. Figure 14C shows that stitching image tiles generated using the 20% overlap approach into an absolute Cartesian coordinate system creates a whole-slide image accurate to within a fraction of a pixel without loss of information. The contours from the uncorrected stitching are shown overlaid on the image generated using the AstroPath approach. Uncorrected whole-slide stitching results in a misalignment of up to 80 pixels, which corresponds to 40 μm or the diameter of four lymphocytes. Correcting such errors is particularly important when using multiple microscopes, software analysis suites, and / or scans from different systems. Such directional and cumulative misalignments can also lead to inaccuracies in slide registration during Z-stack imaging (i.e., overlaying a second slide image on top of a first image obtained from the same specimen). [Figure 15]FIG. 15A illustrates how the single marker phenotyping approach minimizes errors in the dataset due to over-segmentation of large cells. A representative image showing improved cell segmentation using the single marker approach is shown (red line = cell border, * = over-segmented tumor nuclei). FIG. 15B illustrates how the single marker phenotyping approach minimizes errors in the dataset due to over-segmentation of large cells. A representative image of the mixed phenotyping output after single marker phenotyping (top left) and the corresponding output of the individual single marker phenotyping algorithms before mixing (bottom and right) are shown. FIG. 15C illustrates how the single marker phenotyping approach minimizes errors in the dataset due to over-segmentation of large cells. It is shown that the number of positive cells quantified by the single marker approach reflects an "absolute criterion", while the multi-marker approach overestimates tumor cells and CD163+ cells. The "absolute criterion" is defined as the segmentation / phenotyping performed for each lineage marker in monoplex IF (i.e., individual staining). Figure 15D shows that the single-marker phenotyping approach minimizes the error in the dataset due to over-segmentation of large cells. This systematic error was further characterized by testing the number of cells counted in the CD8 hotspots of 46 specimens with the single-marker approach versus the multi-marker approach. The percentage difference in cell counts shows that the multi-marker approach overcounts tumor cells and CD163 cells by 30% compared to the single-marker approach. [Figure 16]Figure 16A shows representative output from a custom-built algorithm that facilitates visual inspection of segmentation and phenotyping performance. A custom display showing ~1250 cells per view is shown. Each cell in the mIF image is placed with a colored dot indicating lineage. Additionally, a dash is placed over the cell if PD-L1 (green dash) and / or PD-1 (cyan dash) are expressed. Except in rare cases of low tissue availability, 20 views per specimen were visually inspected. Figure 16B shows representative output from a custom-built algorithm that facilitates visual inspection of segmentation and phenotyping performance. Examination of up to 25 randomly selected positive and negative cells / stamps for each marker across all HPFs of a given specimen is shown. The results of the segmentation algorithm are shown in red, with each cell that is positive for a given marker marked with a white "+". Using these stamps, a minimum of 2000 cells exhibiting each marker were visually inspected per specimen. FIG. 16C shows a representative output from a custom-built algorithm that facilitates visual inspection of segmentation and phenotyping performance. An additional representative QA / QC stamp is shown, without the overlaid cell segmentation map. A "+" indicates each cell that was determined to be positive by the algorithm. FIG. 16D shows a representative output from a custom-built algorithm that facilitates visual inspection of segmentation and phenotyping performance. It is shown that the QA / QC stamp viewer can also be used to visually inspect co-expression profiles of interest. A representative image of a CD8+FoxP3+ cell is shown (FoxP3 in red, CD8 in yellow). An average of 200 CD8+FoxP3+ cells were visually inspected per specimen. FIG. 16E shows a representative output from a custom-built algorithm that facilitates visual inspection of segmentation and phenotyping performance. Three examples of CD8+ cells (yellow) that are PD-L1+ (green) are shown. The three images were obtained from three different patient specimens to show generality.The top panel shows only the CD8 channel; the middle panel shows only the PD-L1+ channel; the bottom panel shows both the CD8 and PD-L1 channels together. Specifically, the bottom left panel shows CD8+PD-L1- cells (one yellow asterisk), CD8-PD-L1+ cells (one green asterisk), and CD8+PD-L1+ cells (one yellow and one green asterisk). DAPI is also shown in blue. [Figure 17] FIG. 17A illustrates that accurate comparison of specimens stained at different times requires correction for batch-to-batch variability. Graphs are shown that clearly show batch-to-batch variability in expression intensity for PD-1 and PD-L1. This was corrected by normalization to tissue microarray slides containing tonsil and spleen (n=3 each) run per batch. FIG. 17B illustrates that accurate comparison of specimens stained at different times requires correction for batch-to-batch variability. Bar graphs are shown in which the percent coefficient of variation across 9 batches was 17% for PD-1 and 22% for PD-L1, decreasing by ~50% for both markers once normalized. [Figure 18]FIG. 18A shows the intensity levels of PD-(L)1low, PD-(L)1mid, and PD-(L)1high. Histograms showing PD-1 (left) and PD-L1 (right) intensity cutoffs defined by pooling all PD-1pos or PD-L1pos cells and dividing the population into tertiles. FIG. 18B shows the intensity levels of PD-(L)1low, PD-(L)1mid, and PD-(L)1high. Photomicrographs (bright field top, IF bottom) are shown showing that the location of the PD-1+ population varies by specific region of the tonsil tissue (left). PD-1high cells are mainly located in germinal center T cells in the bright zone, while PD-1low and PD-1mid cells are present in the interfollicular region. Photomicrographs (bright field top, IF bottom) showing the location of the PD-L1+ population also vary by microanatomical location within the tonsil (right). Tonsillar crypts show PD-L1high. PD-L1mid and PD-L1low cells were observed in germinal centers, with scattered PD-L1low perifollicular cells. Anatomical regions with low and moderate expression were preferentially selected for assay optimization of PD-1 and PD-L1 signals in the mIF assay. [Figure 19]Figure 19A shows the density of specific cell populations in responders and non-responders across the TME. The mean tumor area analyzed in 53 patients was 61 mm2 (range 5-308 mm2). It is shown that there was no significant difference in the density of PD-L1 positive cells between anti-PD-1 responders and non-responders when tumor cell expression % was scored using a commercially available chromogenic 22C3 IHC assay and interpreted by a pathologist using light microscopy. Representative photomicrographs of PD-L1 IHC are shown on the right. Figure 19B shows the density of specific cell populations in responders and non-responders across the TME. The mean tumor area analyzed in 53 patients was 61 mm2 (range 5-308 mm2). It is shown that total cell density and tumor cell PD-L1+ cell density across the TME (whole slide analysis using 6-plex mIF assay on the AstroPath platform) were associated with response, while no significant association was observed for CD163+PD-L1+ cell density. Median + / - 95%CI, one-sided Mann-Whitney. Representative micrographs of PD-L1 in all cell types by mIF assay (top row), PD-L1 vs. tumor (middle row), and PD-L1 vs. CD163 (bottom row) are shown in the right column. [Figure 20] Figure 20A shows the proportion of PD-1 expression by cell type in the melanoma TME. 94 melanoma specimens in the archive in TMA format were stained for PD-1, CD8, CD4, CD20, FoxP3, and tumor (Sox10 / S100) using the mIF assay. CD8+ cells contributed the majority of PD-1 to the melanoma TME. Of the non-CD8+ cells that contributed PD-1, 86% were labeled as CD4+ (65% of conventional CD4+ cells and 21% of CD4+FoxP3+ cells). Figure 20B shows a photomicrograph of a representative CD20+PD-1+ cell. [Figure 21]Figure 21A shows that when a similar strategy was applied, analysis of the entire TME (100% sampling) was not as effective in stratifying patients as 30% sampling. With 100% TME sampling, features were identified with an AUC of p-value <0.05 after multiple testing correction (positive features: CD8+PDL1low, CD8+FoxP3+, CD8+FoxP3+PD-1low, CD8+FoxP3+PD-1mid, tumor PD-L1low, and negative features: CD163+PD-L1neg, CD163+ cells) (Table 11). Combinatorial ROC and Kaplan-Meier curves were generated for the discovery cohort using these features. Similar general trends were observed as with 30% slide sampling (Figure 6), but stratification was less effective. This finding highlights the fact that slide sampling is another component of assay development that can be optimized and standardized. Figure 21B shows that when a similar strategy was applied, analysis of the entire TME (100% sampling) was not as effective in stratifying patients as 30% sampling. With 100% TME sampling, features were identified with an AUC of p-value <0.05 after multiple testing correction (positive features: CD8+PDL1low, CD8+FoxP3+, CD8+FoxP3+PD-1low, CD8+FoxP3+PD-1mid, tumor PD-L1low, and negative features: CD163+PD-L1neg, CD163+ cells) (Table 11). These features were used to generate combinatorial ROC and Kaplan-Meier curves for an independent second validation cohort. Similar general trends were observed as with 30% slide sampling (Figure 6), but stratification was less effective. This finding highlights the fact that slide sampling is another element of assay development that can be optimized and standardized. [Figure 22]FIG. 22A shows the association between TME defined by specific cell types and long-term survival by Kaplan-Meier analysis on smaller specimens. The minimum tumor area for inclusion in the study was 5 mm2. Patients with tumor area <20 mm2 on slides are divided into good, intermediate, and poor prognosis using the scoring rules defined in FIG. 12B. The 20 mm2 area was chosen because it was the size of a three core biopsy (each 1 mm×15 mm in size) with ~50% tumor present in each core. FIG. 22B shows the association between TME defined by specific cell types and long-term survival by Kaplan-Meier analysis on smaller specimens. The minimum tumor area for inclusion in the study was 5 mm2. Patients with tumor area >20 mm2 on slides are divided into good, intermediate, and poor prognosis using the scoring rules defined in FIG. 12B. An area of ​​20 mm2 was chosen because this was the size of a three core biopsy (each 1 mm × 15 mm in size) with ~50% tumor present in each core. [Diagram 23]Figure 23A shows the results of reducing the mIF assay from 6-plex to 4-plex for predicting objective response and stratifying overall survival. In the index 6-plex assay (Figure 6), the CD8+ subset was used to predict patients with good vs. intermediate long-term prognosis. Herein, we investigated the possibility of using only total CD8+ cell density for this distinction, reducing the number of markers required from 6 to 4 (CD8, CD163, PD-L1, Sox10 / S100). Using these 4 features, we generated combined ROC and Kaplan-Meier curves for the discovery cohort. We found that combining the highest tertile of total CD8+ density with the densest features negatively correlated with response (CD163+PD-L1neg, tumor PD-L1neg or tumor cells) allowed for stratification of survival. This approach is less effective in predicting outcomes, especially for the prediction of objective response. However, there is an advantage to including additional markers, for example markers that help resolve patients in the intermediate prognosis group or markers that help identify factors within the TME from patients in the intermediate and poor prognosis categories that can help inform new rational treatment strategies. Figure 23B shows the results of reducing the mIF assay from 6-plex to 4-plex for predicting objective response and stratifying overall survival. In the index 6-plex assay (Figure 6), the CD8+ subset was used to predict patients with good vs. intermediate long-term prognosis. Herein, we investigated the possibility of using only total CD8+ cell density for this distinction, reducing the number of markers required from 6 to 4 (CD8, CD163, PD-L1, Sox10 / S100). Using these four features, combined ROC and Kaplan-Meier curves were generated for the validation cohort. We found that combining the highest tertile of total CD8+ density with the most dense features negatively correlated with response (CD163+PD-L1neg, tumor PD-L1neg or tumor cells) allowed for survival stratification. This approach is less useful for predicting outcome, especially with respect to predicting objective response.However, there is an advantage to including additional markers, for example markers that help resolve patients in the intermediate prognosis group or markers that help identify factors within the TME from patients in the intermediate and poor prognosis categories that can help inform new rational treatment strategies. [Figure 24] Figure 1: CD8+FoxP3+PD-1+ cells are strongly associated with objective response in patients with advanced non-small cell lung cancer treated with anti-PD-1 based therapy. Pre-treatment lung cancer specimens from n=20 patients with advanced disease were stained with a 6-plex mIF assay and imaged using mosaic high power fields (HPFs) tiled across the specimen. [Diagram 25] Figure 25A shows the application of the AstroPath imaging approach to pretreatment specimens from patients with non-small cell lung cancer who received anti-PD-1 in neoadjuvant therapy for advanced disease. The median pretreatment biopsy size in this cohort was 3 mm2 (mean 15 mm2). Figure 25B shows the application of the AstroPath imaging approach to pretreatment specimens from patients with non-small cell lung cancer who received anti-PD-1 in neoadjuvant therapy for advanced disease. A second analysis was performed to determine pretreatment features predictive of the degree of pathological response to treatment. Left: Heatmap shows the association of several potential cell populations identified by this method with pathological response values ​​of 10% residual viable tumor (rvt10, an end point in many Phase II / III clinical trials) and 50% residual viable tumor (rvt50). Right: Features identified by this method can be used to predict survival outcomes in these patients (Kaplan-Meier analysis). DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0028] The present disclosure is based on the discovery that the analysis of multiple cell types and their spatial interactions, as well as the expression levels and cellular profiles of biomarkers (e.g., immune regulatory molecules) can be used to predict a subject's response to immunotherapy. In some embodiments, the detection of one or more biomarkers (e.g., PD-1, PD-L1, CD8, FoxP3, CD163, and tumor cell markers) in biological samples using immunofluorescence and / or immunohistochemistry can be used to predict response to checkpoint inhibitor therapy (e.g., checkpoint blockade with anti-PD-1-based therapy) and / or stratify long-term survival after immunotherapy. Immunotherapy (e.g., immune checkpoint inhibitor (ICI) therapy) has revolutionized cancer treatment by improving overall survival (OS), but much effort is being put into developing predictive biomarkers to direct patients who are most likely to benefit to specific treatments while exploring alternative therapies for patients with very low chances of response.

[0029] In some embodiments, provided herein is a method of predicting a subject's response to immunotherapy, the method comprising: (a) staining a biological sample disposed on a substrate; (b) imaging the biological sample, where a high power field (HPF) image is generated; (c) detecting one or more biomarkers in the biological sample; and (d) analyzing the HPF image, thereby predicting the subject's response to immunotherapy.

[0030] Various non-limiting aspects of these methods are described herein and may be used in any combination without limitation. Additional aspects of the various components of the methods described herein are known in the art.

[0031] As used herein, the term "administration" typically refers to the administration of a composition to a subject or system to achieve delivery of an agent that is or is contained in the composition. Those skilled in the art will recognize the various routes that may be utilized for administration to a subject, such as a human, in the appropriate circumstances. For example, in some embodiments, administration may be ocular, oral, parenteral, topical, etc. In some particular embodiments, administration is bronchial (e.g., by intrabronchial instillation), buccal, transdermal (which may be or include one or more of, e.g., topical to the dermis, intradermal, interdermal, transdermal, etc.), enteral, intraarterial, intradermal, intragastric, intramedullary, intramuscular, intranasal, intraperitoneal, intrathecal, intravenous, in a particular organ (e.g., intrahepatic), mucosal, nasal, oral, rectal, subcutaneous, sublingual, topical, tracheal (e.g., by intratracheal instillation), vaginal, vitreous, etc. In some embodiments, administration may include only a single dose. In some embodiments, administration may include the application of a certain number of doses. In some embodiments, administration can include administration that is intermittent (e.g., multiple administrations separated by a time period) and / or periodic (e.g., individual administrations separated by a common time period) administration. In some embodiments, administration can include continuous administration (e.g., perfusion) for at least a selected period of time.

[0032] As used herein, the term "antibody" refers to an agent that specifically binds to a particular antigen. In some embodiments, the term encompasses any polypeptide or polypeptide complex that contains sufficient immunoglobulin structural elements to confer specific binding. Exemplary antibody agents include, but are not limited to, monoclonal antibodies, polyclonal antibodies, and fragments thereof. In some embodiments, antibody agents can include one or more sequence elements, such as humanized, primatized, chimeric, etc., as known in the art. In many embodiments, the term "antibody" is used to refer to one or more of the constructs or formats known or developed in the art to utilize the structural and functional characteristics of antibodies for alternative presentation. For example, in some embodiments, the antibodies utilized in accordance with the materials and methods provided herein include, but are not limited to, intact IgA, IgG, IgE or IgM antibodies; bi- or multispecific antibodies (e.g., Zybodies®, etc.); antibody fragments such as Fab fragments, Fab' fragments, F(ab')2 fragments, Fd' fragments, Fd fragments, isolated CDRs or sets thereof; single chain Fvs (scFvs); polypeptide-Fc fusions; single domain antibodies (e.g., shark single domain antibodies or fragments thereof, such as IgNAR); cameloid antibodies; masked antibodies (e.g., Probodies®), Small Modular Formats selected from, but not limited to, ImmunoPharmaceuticals ("SMIPs™"), single chain or tandem bispecific antibodies (TandAbs®); VHHs; Anticalins®; Nanobodies® minibodies; BiTEs®, ankyrin repeat proteins or DARPINs®; Avimers®; DARTs; TCR-like antibodies; Adnectins®; Affilins®; Trans-bodies®; Affibodies®; TrimerX®; microproteins; Fynomers®, Centyrins®; and KALBITOR®.In some embodiments, an antibody is or comprises a polypeptide whose amino acid sequence comprises a structural element recognized by those skilled in the art as an immunoglobulin variable domain. In some embodiments, an antibody is a polypeptide protein having a binding domain that is homologous or largely homologous to an immunoglobulin binding domain. In some embodiments, an antibody is or comprises at least a portion of a chimeric antigen receptor (CAR). In some embodiments, an antibody is or comprises a T cell receptor (TCR).

[0033] As used herein, the term "biological sample" refers to a sample obtained from a subject for analysis using any of a variety of techniques, including but not limited to biopsy, surgery, laser capture microscopy (LCM), and the like, and generally includes cells and / or other biological material from the subject. Biological samples can be obtained from eukaryotic organisms, such as patient-derived organoids (PDOs) or patient-derived xenografts (PDXs). Biological samples can include organoids, which are miniaturized and simplified versions of organs created in vitro in three dimensions that exhibit realistic microanatomy. Subjects from which biological samples can be obtained can be healthy or asymptomatic individuals, individuals with or suspected of having a disease (e.g., cancer) or predisposition to a disease, and / or individuals with or suspected of needing treatment.

[0034] The biological sample may include one or more diseased cells. Diseased cells may have altered metabolic characteristics, gene expression, protein expression, and / or morphological characteristics. Examples of diseases include inflammatory diseases, metabolic diseases, nervous system diseases, cancer, etc. Cancer cells may be derived from solid tumors, hematological malignancies, cell lines, or obtained as circulating tumor cells.

[0035] Biological samples may also include immune cells. Sequence analysis of the immune repertoire of such cells, including genomic, proteomic and cell surface features, can provide a wealth of information to facilitate understanding of the state and function of the immune system. Examples of immune cells in biological samples include, but are not limited to, B cells (e.g., plasma cells), T cells (e.g., cytotoxic T cells, natural killer T cells, regulatory T cells, T helper cells, etc.), natural killer cells, cytokine-induced killer (CIK) cells, myeloid cells such as granulocytes (basophilic granulocytes, eosinophilic granulocytes, neutrophilic granulocytes / neutrophilocytes), monocytes / macrophages, mast cells, platelet cells / megakaryocytes, dendritic cells, etc.

[0036] The biological sample may include any number of macromolecules, such as cellular macromolecules and organelles (e.g., mitochondria and nuclei). The biological sample may be a nucleic acid sample and / or a protein sample. The biological sample may be a carbohydrate sample or a lipid sample. The biological sample may be obtained as a tissue sample, such as a tissue section, a biopsy, a core biopsy, a needle aspirate or a fine needle aspirate. The sample may be a liquid sample, such as a blood sample, a urine sample or a saliva sample. The sample may be a skin sample, a colon sample, a buccal swab, a histological sample, a histopathological sample, a plasma or serum sample, a tumor sample, a lymph node sample, a biological cell, a cultured cell, a clinical sample, such as whole blood or a blood-derived product, a blood cell, or a cultured tissue or cell, including a cell suspension.

[0037] As used herein, the terms "cancer," "malignancy," "neoplasm," "tumor," and "cell type" refer to cells that exhibit relatively abnormal, uncontrolled, and / or autonomous growth, and thus, they exhibit an abnormal growth phenotype characterized by a significant loss of control of cell proliferation. In some embodiments, tumors may be or include cells that are precancerous (e.g., benign), malignant, premetastatic, metastatic, and / or nonmetastatic. The present disclosure specifically identifies certain cancers to which its teachings may be particularly relevant. In some embodiments, the associated cancers may be characterized by solid tumors. In some embodiments, the associated cancers may be characterized by metastatic solid tumors. In some embodiments, the associated cancers may be characterized by hematological tumors. In general, examples of different types of cancer known in the art include, for example, bladder cancer, breast cancer, cervical cancer, colon cancer, endometrial cancer, esophageal cancer, fallopian tube cancer, gallbladder cancer, gastrointestinal cancer, head and neck cancer, blood cancer, Hodgkin's lymphoma, laryngeal cancer, liver cancer, lung cancer, lymphoma, melanoma, mesothelioma, ovarian cancer, primary peritoneal cancer, salivary gland cancer, sarcoma, gastric cancer, thyroid cancer, pancreatic cancer, renal cell carcinoma, glioblastoma, and prostate cancer. In some embodiments, hematopoietic cancers may include leukemia, lymphoma (Hodgkin and non-Hodgkin), myeloma and myeloproliferative disorders; sarcoma, melanoma, adenoma, solid tissue cell types, squamous cell carcinoma of the mouth, throat, larynx and lung, liver cancer, genitourinary cancers such as prostate, cervical, bladder, uterine and endometrial cancer, renal cell carcinoma, bone cancer, pancreatic cancer, skin cancer, cutaneous or intraocular melanoma, cancers of the endocrine system, thyroid cancer, parathyroid cancer, head and neck cancer, breast cancer, gastrointestinal cancer, nervous system cancer, benign lesions such as papilloma, precancerous lesions such as myelodysplastic syndromes, acquired aplastic anemia, Fanconi anemia, paroxysmal nocturnal hemoglobinuria (PNH) and 5q-syndrome.

[0038] As used herein, the terms "therapeutic agent" and "chemotherapeutic agent" can refer to one or more proapoptotic, cytostatic and / or cytotoxic agents, including, for example, specifically agents utilized and / or recommended for the treatment of one or more diseases, disorders or conditions associated with unwanted cell proliferation. In many embodiments, chemotherapeutic agents are useful in the treatment of cancer. In some embodiments, the chemotherapeutic agent may be or include one or more alkylating agents, one or more anthracyclines, one or more cytoskeletal disrupting agents (e.g., microtubule targeting agents such as taxanes, maytansine and analogs thereof), one or more epothilones, one or more histone deacetylase inhibitors (HDACs), one or more topoisomerase inhibitors (e.g., inhibitors of topoisomerase I and / or topoisomerase II), one or more kinase inhibitors, one or more nucleotide analogs or nucleotide precursor analogs, one or more peptide antibiotics, one or more platinum-based agents, one or more retinoids, one or more vinca alkaloids, and / or one or more analogs of the following (i.e., sharing related antiproliferative activity): In some embodiments, the chemotherapeutic agent may be utilized in the context of an antibody-drug conjugate.

[0039] As used herein, the term "stratify" refers to assigning a treatment regimen. In some embodiments, subjects can be stratified and placed into treatment categories, and treatment regimens within the treatment categories are assigned to the subjects. In some embodiments, stratification of subjects can be used in prospective or retrospective clinical studies. In some embodiments, stratification of subjects can be used to assign prognosis or predictions for survival or chemotherapy or radiotherapy sensitivity. In some embodiments, stratification typically assigns subjects to groups based on shared mutation patterns or other observed features or feature sets. In some embodiments, the treatment regimen can be an anti-cancer treatment. In some embodiments, the treatment regimen can be within a treatment category, and the treatment category includes an anti-cancer treatment. In some embodiments, the treatment category can include radiation therapy, chemotherapy, immunotherapy, hormone therapy, antibody therapy, or any combination thereof.

[0040] As used herein, the term "subject" refers to an organism, typically a mammal (e.g., a human in some embodiments, including prenatal human forms). In some embodiments, the subject is afflicted with the relevant disease, disorder, or condition. In some embodiments, the subject is susceptible to the disease, disorder, or condition. In some embodiments, the subject exhibits one or more symptoms or characteristics of the disease, disorder, or condition. In some embodiments, the subject does not exhibit symptoms or characteristics of the disease, disorder, or condition. In some embodiments, the subject is a person who has one or more characteristics characteristic of a susceptibility to a disease, disorder, or condition, or a risk of a disease, disorder, or disorder. In some embodiments, the subject is a patient. In some embodiments, the subject is an individual for whom diagnosis and / or treatment is and / or has been performed.

[0041] As used herein, the term "treatment outcome" refers to an evaluation performed to evaluate the results or outcomes of management and treatments used in combating a disease to determine the efficacy, effectiveness, safety and feasibility of the treatment given to the subject. In some embodiments, determining the treatment outcome can include whether the subject responds to a particular treatment administered to the subject. In some embodiments, determining the treatment outcome can be used to stratify patients with a disease into groups with different treatment outcomes (e.g., overall survival rate, disease control rate). In some embodiments, determining the treatment outcome can include analyzing overall survival rate, disease control rate, changes in psychological state or changes in physical state (e.g., tissue damage, pain level). In some embodiments, subjects exhibiting a given cell type (e.g., CD8+ T cells, CD8+ FoxP3+ cells) are predicted to have an improved outcome compared to reference subjects identified as not having that cell type (Figure 1).

[0042] A platform for predicting response to immunotherapy In some embodiments, provided herein is a method of predicting a subject's response to immunotherapy, the method comprising: (a) staining a biological sample disposed on a substrate; (b) imaging the biological sample, where a high power field (HPF) image is generated; (c) detecting one or more biomarkers in the biological sample; and (d) analyzing the HPF image, thereby predicting the subject's response to immunotherapy.

[0043] immunotherapy As used herein, "immunotherapy" refers to treating disease (e.g., cancer) by activating or suppressing the immune system. For example, cancer immunotherapy uses the immune system and its components to initiate antitumor responses through immune activation. In some embodiments, immunotherapy may include immune checkpoint inhibitors, oncolytic virus therapy, cell-based therapy, CAR-T cell therapy, or cancer vaccines. In some embodiments, immunotherapy may include immune checkpoint blockade, in which an immune checkpoint inhibitor is administered. In some embodiments, immunotherapy includes administration of an immune checkpoint inhibitor. In some embodiments, the immune checkpoint inhibitor is a PD-1 inhibitor. Examples of PD-1 inhibitors include, but are not limited to, pembrolizumab, nivolumab, semipilimab, JTX-4014, spartalizumab, camrelizumab, sintilimab, tislelizumab, toripalimab, and dostallimab. In some embodiments, the immune checkpoint inhibitor is a PD-L1 inhibitor. Examples of PD-L1 inhibitors include, but are not limited to, atezolizumab, avelumab, durvalumab, KN035, CK-301, AUNP12, CA-170, and BMS-986189. In some embodiments, the immune checkpoint inhibitor is a CTLA-4 inhibitor (e.g., ipilimumab, tremelimumab). In some embodiments, the immune checkpoint inhibitor is a CTLA-4 inhibitor used in combination with a PD-1 inhibitor or a PD-L1 inhibitor. In some embodiments, the immune checkpoint inhibitor can be any checkpoint inhibitor, for example, as described in Mazzarella et al., Eur J Cancer (2019) 117:14-31, which is incorporated herein by reference.

[0044] Biomarkers In some embodiments, detection of one or more biomarkers in a biological sample can be used to predict a subject's response to immunotherapy. In some embodiments, detection of one or more biomarkers in a biological sample can be used to monitor a subject's response to immunotherapy. As used herein, the term "biomarker" refers to a measurable indicator of the severity or presence of a disease (e.g., cancer) condition. In some embodiments, a biomarker can be used to aid in the diagnosis of a condition (e.g., identifying early cancer). In some embodiments, a biomarker can be used to determine the overall survival of a subject without treatment or therapy. In some embodiments, a biomarker can predict a subject's response to a treatment (e.g., immunotherapy). In some embodiments, one or more biomarkers can be detected in a biological sample. In some embodiments, the one or more biomarkers can include PD-1, PD-L1, CD8, FoxP3, CD163, a tumor cell marker, or any combination thereof. In some embodiments, the biomarker can include a tumor cell marker. In some embodiments, the tumor cell markers may include AFP, BRAF V600E, S100, Sox10, cytokeratin, Melan-A, HMB45, vimentin, desmin, myogenin, smooth muscle actin, GFAP, synaptophysin, chromogranin, CD45 / LCA, or any combination thereof. In some embodiments, the tumor cell markers may be a combination of Sox10 and S100.

[0045] Multiplex staining In some embodiments, the method described herein includes staining the biological sample disposed on the substrate. To facilitate visualization, the biological sample can be stained using a wide variety of stains and staining techniques. In some embodiments, the biological sample can be stained using any number of biological stains, including, but not limited to, acridine orange, bismarck brown, carmine, coomassie blue, cresyl violet, DAPI, eosin, ethidium bromide, acid fuchsin, hematoxylin, Hoechst stain, iodine, methyl green, methylene blue, neutral red, Nile blue, Nile red, osmium tetroxide, propidium iodide, rhodamine, or safranin.

[0046] The biological sample may be stained using known staining techniques, including Kangrunwald stain, Giemsa stain, Hematoxylin and Eosin (H&E) stain, Jenner stain, Reischmann stain, Masson's trichrome stain, Papanicolaou stain, Romanowsky stain, Silver stain, Sudan stain, Wright stain, and / or Periodic Acid Schiff (PAS) staining techniques. In some embodiments, the biological sample may be stained using Tyramide Signal Amplification (TSA) techniques. In some embodiments, the biological sample may be stained using a pan-membrane stain. In some embodiments, the biological sample may be stained using a cell membrane stain.

[0047] In some embodiments, the staining comprises immunofluorescence (IF) staining. In some embodiments, the staining comprises immunohistochemistry (IHC) staining. In some embodiments, the biological sample may be stained using detectable labels (e.g., radioisotopes, fluorophores, chemiluminescent compounds, bioluminescent compounds, and dyes). In some embodiments, the biological sample is stained using only one type of stain or one technique. In some embodiments, the staining comprises a biological staining technique, such as H&E staining. In some embodiments, the staining comprises using a fluorophore antibody. In some embodiments, the biological sample is stained using two or more different types of stains or two or more different staining techniques. For example, a biological sample can be prepared by staining and imaging using one technique (e.g., H&E staining and bright field imaging) and then staining and imaging using another technique (e.g., IHC / IF staining and fluorescent microscopy) on the same biological sample.

[0048] Multiplex staining methods are described, for example, in Bolognesi et al., J. Histochem. Cytochem. 2017; 65(8):431-444, Lin et al., Nat Commun. 2015; 6:8390, Pirici et al., J. Histochem. Cytochem. 2009; 57:567-75, and Glass et al., J. Histochem. Cytochem. 2009; 57:899-905, the entire contents of each of which are incorporated herein by reference.

[0049] In some embodiments, the biological sample is stained with an antibody. In some embodiments, the antibody is a monoclonal antibody. In some embodiments, the antibody is a polyclonal antibody. In some embodiments, the biological sample is stained with one or more antibodies (e.g., 1 antibody, 2 antibodies, 3 antibodies, 4 antibodies, 5 antibodies, 6 antibodies, 7 antibodies, 8 antibodies, 9 antibodies, 10 antibodies). In some embodiments, the biological sample is stained with 6 antibodies. In some embodiments, the biological sample is stained with 4 antibodies. In some embodiments, the biological sample is stained with a second antibody that detects the antibody. In some embodiments, the second antibody is conjugated to a label.

[0050] In some embodiments, the label is a detectable label. In some embodiments, the label is a fluorophore. In some embodiments, the detectable label can be directly detectable by itself (e.g., radioisotope label or fluorophore), or in the case of enzyme label, can be indirectly detectable, for example, by catalyzing chemical change of chemical substrate compound or composition, which chemical substrate compound or composition can be directly detectable. In some embodiments, detectable labels include, but are not limited to, radioisotopes, fluorophores, chemiluminescent compounds, bioluminescent compounds, and dyes.

[0051] In some embodiments, the substrate is a slide. In some embodiments, the biological sample comprises a tissue, a tissue section, an organ, an organism, an organoid, or a cell culture sample. In some embodiments, the tissue is a formalin-fixed paraffin-embedded (FFPE) tissue. In some embodiments, the biological sample is fixed before the staining step. In some embodiments, the biological sample may be fixed using formalin fixation and paraffin embedding (FFPE). In some embodiments, the biological sample may be fixed with any of a variety of other fixatives to preserve the biological structure of the sample before analysis. For example, the sample may be fixed by immersion in ethanol, methanol, acetone, formaldehyde (e.g., 2% formaldehyde), paraformaldehyde-Triton, glutaraldehyde, or a combination thereof.

[0052] In some embodiments, a suitable fixation method is selected and / or optimized based on a desired workflow. For example, formaldehyde fixation may be selected as compatible with a workflow using an IHC / IF protocol for protein visualization. As another example, methanol fixation may be selected for a workflow that emphasizes the quality of an RNA / DNA library. Acetone fixation may be selected in some applications to permeabilize tissues. In some embodiments, the biological sample is fixed with formaldehyde. In some embodiments, the biological sample is fixed with methanol.

[0053] Image analysis and processing In some embodiments, the method described herein further comprises imaging the biological sample disposed on the substrate, and an image of a high power field (HPF) is generated. As used herein, "high power field (HPF)" refers to a slide area of ​​view under high magnification of a microscope. In some embodiments, the imaging step can generate one or more HPFs. In some embodiments, the imaging step can generate up to about 5000 (e.g., about 4500, about 4000, about 3500, about 3000, about 2500, about 2000, about 1500, about 1400, about 1300, about 1200, about 1100, about 1000, about 900, about 800, about 700, about 600, about 500, about 400, about 300, about 200, about 100, about 50, about 40, about 30, about 20, about 10, about 5, about 4, about 3, or about 2) HPFs. In some embodiments, the imaging step comprises performing immunofluorescence microscopy on the biological sample.

[0054] In some embodiments, provided herein is a method for improving the predictive value of a biomarker, the method comprising: (a) acquiring a plurality of images of a high power field (HPF) generated from a biological sample; (b) detecting a biomarker in each of the plurality of images; (c) selecting a sub-plurality of images from the plurality of images of step (a); (d) analyzing the sub-plurality of images; and (e) generating an area under the curve value that is greater than the area under the curve value generated when analyzing all images, thereby improving the predictive value of the biomarker. In some embodiments, the sub-plurality of images can be about 30% (e.g., about 5%, about 10%, about 20%, about 40%, or about 50%) of the plurality of images. In some embodiments, the sub-plurality of images can be up to 100% (e.g., up to 5%, up to 10%, up to 20%, up to 30%, up to 40%, up to 50%, up to 60%, up to 70%, up to 80%, or up to 90%) of the plurality of images.

[0055] In some embodiments, the analysis step of the method described herein can further include (i) image acquisition and processing; (ii) cell segmentation and phenotyping; and (iii) image normalization. Methods for analyzing and processing images of biological samples are described, for example, in PCT Application WO 2020 / 061327 and US Patent Application No. 17 / 278112, the entire contents of each of which are incorporated herein by reference.

[0056] In some embodiments, the method may include acquiring, by the apparatus, a plurality of field images of the specimen. In some embodiments, the plurality of field images may be captured by a microscope. In some embodiments, the method may include processing, by the apparatus, the plurality of field images to derive a plurality of processed field images. In some embodiments, the processing may include applying spatial distortion correction and illumination-based correction to the plurality of field images to address imperfections in one or more of the plurality of field images. In some embodiments, the method may include identifying, by the apparatus, a key region in each of the plurality of processed field images that contains data useful for characterizing cells or subcellular features, identifying, by the apparatus, overlapping regions in the plurality of processed field images, and deriving, by the apparatus, information regarding the spatial mapping of one or more cells of the specimen. In some embodiments, the derivation of information may be based on performing, by the apparatus, image segmentation based on data contained in the key region in each of the plurality of processed field images, and obtaining, by the apparatus, flux measurements based on other data contained in the overlapping regions. In some embodiments, the method may include causing the device to perform operations related to identifying, based on the information, characteristics associated with normal tissue, factors used to diagnose or prognose a disease, or select a treatment.

[0057] In some embodiments, the device may include one or more memories and one or more processors communicatively coupled to the one or more memories and configured to acquire a plurality of field images of the tissue sample. In some embodiments, the plurality of field images may be captured by a microscope. In some embodiments, the one or more processors may be configured to apply spatial distortion correction and illumination-based correction to the plurality of field images to derive a plurality of processed field images, identify key regions in each of the plurality of processed field images that contain data useful for cell characterization, identify regions in the plurality of processed field images that overlap with each other, and derive information regarding the spatial mapping of one or more cells of the tissue sample. In some embodiments, the one or more processors may be configured to perform segmentation at a subcellular level, a cellular level, or a tissue level based on data contained in the key regions of each of the plurality of processed field images, obtain flux measurements based on other data contained in the regions that overlap with each other, and load the information into a data structure to enable statistical analysis of the spatial mapping to identify predictors of immunotherapy.

[0058] In some embodiments, a non-transitory computer-readable medium may store instructions. In some embodiments, the instructions, when executed by one or more processors, may include one or more instructions that cause the one or more processors to acquire a plurality of field images of a tissue sample, apply spatial distortion correction and / or illumination-based correction to the plurality of field images to derive a plurality of processed field images, identify key regions in each of the plurality of processed field images that contain data useful for cell characterization, identify overlapping regions in the plurality of processed field images, and derive spatial resolution information for one or more cells or subcellular components of the tissue sample. In some embodiments, the one or more instructions that cause the one or more processors to derive spatial resolution information cause the one or more processors to perform image segmentation based on data contained in the key regions in each of the plurality of processed field images, and obtain flux measurements based on other data contained in the overlapping regions. In some embodiments, the one or more instructions, when executed by one or more processors, may cause the one or more processors to add spatial resolution information to a data structure to enable statistical analysis useful for identifying predictors, prognostic factors, or diagnostic factors for one or more diseases or associated therapies.

[0059] In some embodiments, the image acquisition step includes compiling a plurality of HPF images to acquire an image of the entire biological sample within the substrate. In some embodiments, the image acquisition step includes compiling a plurality of HPF images to acquire an image of a portion of the biological sample. In some embodiments, the plurality of HPF images are from the same tumor. In some embodiments, the plurality of HPF images are from different tumors. In some embodiments, the plurality of HPF images are generated from the same microscope. In some embodiments, the plurality of HPF images are generated from different microscopes. In some embodiments, the plurality of HPF images are generated from image data from scanning from chromogenic IHC slides. In some embodiments, the plurality of HPF images are generated from imaging data from tissue-based mass spectrometry. In some embodiments, the plurality of HPF images are generated from image data collecting spatially resolved single cells for genomic and transcriptomic analysis. In some embodiments, the plurality of HPF images are sorted / ranked by features in the images. In some embodiments, the features can be expression of a biomarker. In some embodiments, the features can be expression of a CD8 marker. In some embodiments, the features can be expression of CD163 cells, FoxP3 cells, CD163 PD-Ll neg cells, tumor cells, tumor PD-Ll+ mid cells, FoxP3CD8PD-l+ low cells, FoxP3PD-l low +PD-Ll+ cells, FoxP3CD8PD-LI+ mid Cells, other cells PD-l low +, PDLI+ cells, FoxP3CD8+PD-l+ mid cells, CD163PD-LI+ cells, or any combination thereof.

[0060] In some embodiments, compiling includes aligning the multiple HPF images in an overlapping manner. In some embodiments, each HPF image of the multiple HPF images can overlap with an adjacent image by about 20% (e.g., about 5%, about 6%, about 7%, about 8%, about 9%, about 10%, about 11%, about 12%, about 13%, about 14%, about 15%, about 16%, about 17%, about 18%, about 19%, about 21%, about 22%, about 23%, about 24%, about 25%, about 26%, about 27%, about 28%, about 29%, or about 30%). In some embodiments, each HPF image of the multiple HPF images can overlap with an adjacent image by up to 100% (e.g., up to 10%, up to 20%, up to 30%, up to 40%, up to 50%, up to 60%, up to 70%, up to 80%, or up to 90%).

[0061] In an embodiment, the cell segmentation and phenotyping step includes identifying cell types in a biological sample. In some embodiments, cell segmentation can be performed by defining the membranes of larger cells apart from highlighting smaller lymphocytes. In some embodiments, the phenotyping step includes detecting at least one expression of a biomarker in the cell type. In some embodiments, the expression of at least one biomarker is designated as low, medium, or high. In some embodiments, the phenotyping can include detecting the expression of a single biomarker, where the cells are designated a low, medium, or high status for the single biomarker. In some embodiments, the phenotyping can include detecting the expression of multiple biomarkers, where individual phenotypes from a single biomarker are combined to determine the phenotype of a cell with multiple biomarkers. In some embodiments, the phenotyping can include detecting the expression of PD-1 expression. In some embodiments, the phenotyping can include detecting PD-L1 expression (FIG. 4A). In some embodiments, the cell type comprises CD163+ macrophages, CD8+ T cells, Treg cells (CD8negFoxP3+), tumor cells, CD8+FoxP3+ cells, or any combination thereof. In some embodiments, the cell type is determined to be negative for a biomarker, where the biomarker is not expressed in the cell. In some embodiments, the cell type is determined to be negative for a biomarker, where the biomarker is not expressed in the cell. low / mid In some embodiments, the CD163+PD-L1 cell type is identified as an indication that the subject will respond to immunotherapy. neg The cell type is CD163+PD-L1 neg These subjects are identified as an indication that they will not respond to immunotherapy to the same extent as reference subjects identified as not having this cell type.

[0062] In some embodiments, the cell segmentation and phenotyping step further comprises determining the density of a cell type in the biological sample. In some embodiments, the density of a cell type can be used as an indicator of a subject's response to immunotherapy. In some embodiments, the density of total PD-L1+ cells and tumor PD-L1+ cells can be identified as an indicator of response to immunotherapy. In some embodiments, the density of CD163+PD-L1+ cells does not correlate with response to immunotherapy. In some embodiments, a high density of CD8+FoxP3+ cells is identified as an indicator that a subject will respond to immunotherapy.

[0063] In some embodiments, the image normalization step comprises calibrating the fluorescence intensity of at least one of the biomarkers in the plurality of HPF images to the tissue microarray. In some embodiments, the image normalization comprises calibrating the fluorescence intensity of PD-1 intensity. In some embodiments, the image normalization comprises calibrating the fluorescence intensity of PD-L1 intensity.

[0064] In some embodiments, the analyzing step further comprises identifying at least one biomarker in a biological sample from a subject having the disease, where the identification of the at least one biomarker is used to predict the subject's response to immunotherapy.

[0065] The present disclosure is further described in the following examples, which do not limit the scope of the disclosure as described in the claims. EXAMPLES

[0066] Example 1 - Case Selection Staining optimization of the mIF assay was performed on archival formalin-fixed paraffin-embedded (FFPE) sections of tonsils and melanomas. After optimization of the index mIF assays (PD-1, PD-L1, CD8, FoxP3, CD163, S100 / Sox10), a retrospective analysis was performed on a discovery cohort of pretreatment FFPE tumor specimens from 53 patients with metastatic melanoma who had received anti-PD-1-based therapy. Thirty-four patients received anti-PD-1 monotherapy (nivolumab or pembrolizumab) and 19 patients received anti-PD-1 / CTLA-4 dual inhibitor therapy (nivolumab and ipilimumab). Patients were classified as responders (complete or partial response) and non-responders based on RECIST 1.1 criteria. Five-year overall survival and progression-free survival information were also determined. Clinicopathological characteristics of the cohort, including age, sex, and stage of disease, were additionally collected (Table 1). One representative FFPE block was selected for mIF staining. A PD-L1 IHC companion diagnostic assay (22C3) was also performed on these specimens. An independent validation cohort of pretreatment FFPE tumor specimens from 45 patients with metastatic melanoma was also investigated (Table 2). The optimized 6-plex mIF assay was applied to these specimens and correlated with objective response and long-term survival. Cases from both the discovery and validation cohorts were reviewed by a board-certified dermatopathologist to confirm the diagnosis of melanoma. Cases with tumors <5 mm on the slide, extensive necrotic or folded tissue, or pure detumescent histological subtypes were excluded from the analysis.

[0067] A separate tissue microarray (TMA) was used to characterize lymphocyte subsets expressing PD-1 in the melanoma TME by a second mIF assay (PD-1, CD8, CD4, CD20, FoxP3, Sox10 / S100). The TMA contained tissue from 94 patients with metastatic melanoma. One representative formalin-fixed paraffin-embedded (FFPE) block from each tumor specimen was selected for inclusion in the tissue microarray. Six 1.2 mm cores representing both the center and periphery of the tumor were taken from each block and tiled in the format of a tissue microarray. The resulting TMAs were reviewed, and cores with tissue folds, excessive necrosis, and / or <10% of the surface area occupied by tumor cells were excluded from the analysis.

[0068] [Table 1]

[0069] [Table 2]

[0070] Example 2 - Reagents and multispectral microscope Fluorophore Reagents and Multiplex Staining FFPE slides were stained using the tyramide signal amplification (TSA) technique to achieve superior amplification and high plexing compared to standard IF detection (Figures 7A-7C). 7A-7C. Compared to detection of primary antibodies using directly labeled secondary antibodies, the TSA technique utilizes secondary antibody-mediated detection of HRP polymers. A single HRP polymer secondary can catalyze the activation of several fluorophore-labeled tyramides (TSA fluorophores). After activation, the TSA fluorophores covalently bind to surrounding tyrosine residues and remain deposited on the tissue during a heat treatment step that strips the primary and secondary antibodies. Using sequential staining and stripping, a single FFPE tissue section was labeled with six markers and DAPI.

[0071] Slide scanning and multispectral decomposition Images were scanned using a Vectra 3.0 Automated Quantitative Pathology Imaging System (Akoya Biosciences) and processed using the digital image analysis software inForm (Ver 2.3, Akoya Biosciences). A schematic diagram of the multispectral imaging microscope system is shown in Figure 8. The system captures 20x multispectral images consisting of a multilayer image "cube" of 35 image planes. These planes correspond to wavelengths selected by liquid crystal tunable filters and are acquired across the visible light spectrum. Images of multiplex-stained samples are then decomposed using an inverse least-squares fitting approach that minimizes the squared difference between the characteristic emission spectrum of each fluorophore and the measured value. The decomposition separates the autofluorescence and overlapping emission signals of each fluorophore, removes the autofluorescence background, and creates eight signal-specific "component" planes for each fluorophore plus DAPI and autofluorescence.

[0072] To decompose the multispectral image cube, a decomposition library is created using spectra representative of the TSA fluorophores, known characteristic emission spectra of DAPI, and background autofluorescence. To obtain pure spectra for the library, 4 µm thick FFPE tonsil sections were stained with anti-CD20 (dilution 1:400, clone L26 Leica Microsystems) by monoplex IF (see Monoplex IF section) with each fluorophore. TSA concentrations were adjusted to give pixel normalized fluorescence intensity (NFI) counts of 10-15 for each TSA fluorophore (520 1:150, 540 1:500, 570 1:200, 620 1:150, 650 1:200, 690 1:50). DAPI was not added at the end of the protocol. One tonsil section was stained with DAPI only to extract the DAPI spectrum, while the autofluorescence spectrum was extracted from an unstained slide of the tissue of interest. Slides were imaged and spectra were extracted in inForm using automated tools for library creation, as well as DAB and hematoxylin spectral libraries for spectral decomposition of chromogenic stains.

[0073] Example 3 - Optimization of dyeing Characterization of TSA fluorophores - staining index (SI), bleed-through (BT) and marker pairing To investigate the staining index of fluorophores, consecutive slides of five archival tonsil specimens were stained by monoplex IF with anti-CD8 (dilution 1:100, clone 4B11) and each TSA fluorophore (dilution 1:50). Single-cell data were exported from inForm. SI was calculated as the difference in mean fluorescence intensity between the positive and negative cell populations divided by two standard deviations of the negative cell population.

[0074] The same tonsil specimens were used to characterize bleed-through or spillover of the emission spectra of fluorophores, a frequent limitation of multiparametric fluorometry. Dot plots of the logarithm of normalized fluorescence intensity counts versus the normalized fluorescence intensity counts were generated for all channels. A linear relationship at low intensities and an exponential relationship at high intensities were consistently observed (Figure 9A-9B). To account for this duality, hyperbolic sine curves were parameterized and fitted to each of the paired data sets using a nonlinear least-squares model. To improve the accuracy of the fitting, outliers in the noise population were removed. The data were then inverted and centered on the median of the original noise. The trend in BT was calculated as the linear term*nonlinear term of the fitted curve.

[0075] Chromogenic staining Four-micron-thick sections were stained individually for CD8, CD163, PD-1, PD-L1, FoxP3, Sox10, S100, and Sox10 / S100 cocktail. Briefly, slides were deparaffinized, rehydrated, and subjected to heat-induced antigen retrieval (HIER) in target antigen retrieval buffer, pH 6 (S1699, Dako) at 120°C for 10 min (Decloaking chamber, Biocare Medical). Blocking for endogenous peroxidase (3% H2O2, H325-500, Fisher Scientific) and protein (ACE Block, BUF029, Bio-Rad) was performed. For protocols using biotinylated secondary antibodies, endogenous biotin was also blocked (Avidin / Biotin Blocking Kit, SP-2001, Vector Labs). Primary antibodies were incubated for 22 h at 4°C, followed by secondary antibodies for 30 min at room temperature (RT), as described in Table 3. Protocols using biotinylated secondary antibodies used the tyramide signal amplification (TSA) system as previously described. Antigen-antibody binding was visualized using 3,3′-diaminobenzidine (D4293, Sigma). Slides were counterstained with hematoxylin and coverslipped (VectaMount, H-5000, VectorLabs).

[0076] [Table 3]

[0077] Monoplex IF Monoplex IF staining was performed sequentially on three tonsil and melanoma (for Sox10 and S100) slides, titrating each primary antibody (Table 4). Briefly, slides were deparaffinized and subjected to microwave HIER (Haier1000W) in pH 9 and then pH 6 buffers (AR900 and AR600, respectively, Akoya Biosciences) at 100% power for 45 s and 20% power for 15 min. After endogenous peroxidase removal (3% H2O2, H325-500, Fisher) and protein blocking (Antibody Diluent Background Reducing, S3022, Dako), primary antibodies were incubated at RT with serial dilutions starting at twice the optimal concentration used for chromogenic staining. All secondary antibodies were incubated for 10 min at RT. Then, TSA fluorophores paired with the given markers (Opal 7 color kit, NEL811001KT, Akoya Biosciences) were applied for 10 min. A final microwave step was performed at pH 6, and slides were stained with DAPI (Opal 7 color kit, NEL811001KT, Akoya Biosciences) and coverslipped (ProLong Diamond Antifade Mountant, P36970, Life Technologies). For comparison of the primary titrations, 10 corresponding high power fields (HPFs) were selected for each dilution to assess the signal-to-noise ratio (SNR) for both pixel- and cell-based approaches (Figure 10A). We also selected 10 corresponding HPFs for comparison with chromogenic IHC. HPFs were specifically selected to capture the wide dynamic range of PD-1 and PD-L1 expression (see Figures 18A-18B).

[0078] [Table 4]

[0079] After the optimal primary antibody concentration was identified, TSA titration was performed on five melanoma tumor sections for all markers (Table 5). HIER steps were performed both before and after staining according to how the slides would be processed in the final multiplex assay. For each IF condition and associated chromogenic IHC, 10 corresponding HPFs were selected for analysis. The optimal TSA concentration was selected for each marker, taking into account signal comparability compared to chromogenic IHC and bleed-through between fluorescent channels.

[0080] [Table 5]

[0081] Multiplex IF Single sections from five FFPE melanoma specimens were stained for all six markers in the multiplex panel (Table 6). In addition, three 4 μm thick tissue sections were stained for individual markers before and after the slides used for the 6-plex panel. Ten HPFs were compared between multiplex IF and the corresponding monoplex IF.

[0082] [Table 6]

[0083] Approaches to signal quantification The signal was quantified by a number of different approaches, including cell-based, pixel-based approaches, with or without machine learning. The manufacturer recommends a cell-based approach combined with machine learning, which labels individual cell types and assigns them to Cartesian coordinates, facilitating the analysis of cell density, fluorescence intensity of markers in different cellular compartments, marker co-expression and distance metrics between cells. Cell-based quantification was performed using the Cell Segmentation Module of inForm software, which identifies and maps individual cells, followed by machine learning-based phenotyping, i.e., cell type assignment.

[0084] We also used a cell-based approach to quantify the signal without machine learning. We used the Cell Segmentation module to output the mean fluorescence intensity of each fluorophore in the compartment of interest for each cell. We then binned the data into 10% relative intensity intervals, extracted the top 10% median as signal and the bottom 10% as noise, and performed a quantile-based cell analysis.

[0085] The pixel-based approach does not rely on cell identification, i.e. cell segmentation, but simply measures the number of pixels that are positive for a marker in a given area. When comparing IF and IHC staining, this approach was used because the same cell segmentation algorithm cannot be applied to both techniques. Pixel-wise data were extracted and analyzed with the R package mIFTO (compiled and developed for AstroPath, available at https: / / github.com / AstropathJHU / mIFTO). Positive pixels (signal) and negative pixels (noise) were assigned using thresholds determined using the Colocalization Module of inForm. Tumor cell expression was investigated using a machine learning algorithm that classifies pixels into tissue categories. This was necessary for accurate tumor quantification due to the variability in tumor cell size and the use of a dual marker (Sox10 / S100) cocktail, which does not allow for thresholding of single marker intensity.

[0086] To compare monoplex IF staining with chromogenic staining, a pixel-based approach was used. For Sox10 / S100 staining, a machine learning algorithm was also used. For all other markers, no machine learning was used for this particular comparison. The number of positive pixels with chromogenic staining was considered as the baseline, and the percent deviation of positive pixels with IF staining was calculated.

[0087] Positive signals from monoplex and multiplex IF staining were compared using pixel- and cell-based approaches. Potential changes in marker intensity between multiplex and monoplex IF were assessed by comparing the usable dynamic range of each epitope, defined as the difference between the 95th and 5th percentile mean cellular fluorescence intensity per HPF.

[0088] statistical analysis For comparison of staining between corresponding fields from consecutive slides, paired t tests were performed and data are reported as mean ± SEM.

[0089] Example 4 - Image acquisition, phenotyping, and inter-batch normalization Image acquisition The entire slide was acquired by tiling the HPFs with a 20% overlap as in Figure 3A and Figures 12A-12C. The midpoint of the overlap was used to determine the boundaries of the corrected HPF (Figure 3B). Flat-field corrections for each of the 35 layers were derived from an average of 11,000 HPFs and smoothed with a Gaussian to reduce the influence of outliers (Figure 3B and Figures 13A-13C). A mathematical correction was also applied for each HPF for the "pincushion effect" due to lens distortion (Figure 3B). The fields were then stitched together using a spring-based model that removes "jitter" due to microscope stage movement (Figure 3C and Figures 14A-14C).

[0090] tissue annotation Tumor and stromal boundaries were manually annotated using HALO (Indica Labs, NM) image analysis software. Necrotic areas, tissue folds, and other artifacts were excluded from the analysis.

[0091] Single marker phenotyping and associated quality assurance / quality control (QA / QC) inForm software typically assigns phenotypes to individual cell lineages simultaneously, such as CD8 vs. CD163 (i.e., "multi-marker" phenotyping). "Single-marker" phenotyping was also performed, whereby cells were assigned a positive or negative status for each marker individually. The six individual data sets were then merged into a single Cartesian coordinate system using cell centroids.

[0092] The quality of the final phenotyping was verified by a board-certified pathologist who visually inspected an average of 25,000 phenotyped cells per specimen using a custom viewer (Figure 16A). Specifically, the 20 densest CD8 HPFs containing at least 60 tumor cells, 50% tissue coverage, and a total of 400 cells were selected for each specimen for visual QA / QC inspection of the performance of the phenotyping algorithm. A second custom viewer facilitated the inspection of up to 25 randomly selected positive and negative cells for each marker from the same HPF (Figure 16B-16D). For each specimen, a minimum of 2000 cells displaying each marker were visually inspected using this second viewer. Custom QA / QC code for both viewers can be found at https: / / github.com / AstropathJHU / MaSS.

[0093] Normalization for batch-to-batch variation Tissue microarrays (TMAs) containing three normal spleens and three tonsils were run with each multiplex staining batch, and PD-1 and PD-L1 staining intensities from control tissues were used for batch-to-batch normalization.

[0094] Computer Hardware and Software Configuration Images were acquired using a local desktop computer associated with the Vectra that had been upgraded to include two 2TB M.2 NVMe SSDs allocated as a single drive for maximum storage and transfer efficiency. Multispectral image tiles were transferred from the local computer to a cluster of four servers dedicated to processing the Vectra data. Two of the servers were configured for compute performance with nine 2TB nVME SSDs, 128GB RAM, and 24 physical cores. The remaining two servers were configured for storage and included six 6x6TB HDDs configured as a RAID5 array. This resulted in a net usable HDD capacity of 313.3TB. At peak times, the study consumed 32.27TB of this storage capacity.

[0095] One compute server was specifically dedicated to image enhancement and segmentation, running multiple virtual machines, each with its own inForm instance. The interactive portion of inForm was overridden using automation tools so it could be run as a batch process. The other compute machine was dedicated to storing the database. One of the storage machines contained compressed backups of the raw data. To improve accessibility, each image was individually compressed using 7-Zip software settings for optimal speed and compressed size of the image file. The final storage server stored the data as it was being processed.

[0096] Intermediate data products are reproducible and can be discarded during or after processing; the minimum storage requirement for this project is approximately 15 TB without compression. Although this configuration uses a lot of parallelism to speed up image processing and analysis by 12-15 times, it is important to note that the general workflow described herein can be performed on a single computer equipped with one inForm license.

[0097] Example 5 - Assessment of cell type density according to distance to tumor-stroma boundary The density of specific cell types expressing PD-1 or PD-L1 was determined versus distance from the tumor-stroma border. PD-1 levels (negative, low, intermediate, high) were determined by dividing the PD-1 positive signal into tertiles. To allow comparisons between cell types with different abundance levels, a probabilistic density was calculated by dividing the cell density within each distance bin by the total surface density of that cell category.

[0098] Example 6 - Density assessment of specific cell populations and association with response to anti-PD-1 The density of specific cell types, including assessment of PD-1 and PD-L1 expression levels (negative, low, medium, high), was determined for each specimen and examined for association with response to treatment. Assessment of low, medium, and high PD-1 and PD-L1 expression levels was determined by grouping all positive cells for either marker across all cases and dividing the dynamic range of each positive signal into tertiles (Figures 18A-18B). The density of cells showing different PD-1 / PD-L1 expression levels for each cell type was then compared between responders and non-responders using a one-sided Wilcoxon rank sum test. Rank sum values ​​were converted to AUC values.

[0099] To determine the impact of sampling HPFs on the resulting AUC, increasing proportions of the tumor microenvironment were assessed iteratively. Field sampling was performed in one of two ways: 1) CD8+ cell density was determined for each HPF, and then fields were ranked in order of decreasing CD8+ cell density for inclusion in the "hotspot" analysis; 2) Fields were randomly ranked and selected with increasing proportions (Figures 5A-5B). To avoid bias, 100 random rankings were generated and the average AUC reported at each proportion step. These were randomly selected for the "representative" analysis. Reported p-values ​​were corrected for multiple comparisons using the Benjamini-Hochberg correction.

[0100] Each feature that showed an association with response in the univariate analysis of the 30% hotspot HPF sampling and the whole TME (100% sampling) (adjusted p-value < 0.05) was combined in a multivariate model. Specifically, a binary logistic regression model was applied to evaluate the combined ROC curve and the corresponding AUC was calculated to assess the prognostic accuracy of the combination of the top 10 features in the discovery cohort for predicting objective response. These same 10 features were then validated in an independent validation cohort. These features were also used to develop a combined model for predicting long-term survival by Kaplan Meier analysis. In this combined model, patients whose specimens had a high density (top 20%) of any one of the features negatively correlated with outcome were first grouped together, regardless of other expression factors. The remaining patients were then divided among those with a high density (top 15%) of any one of the features positively correlated with outcome.

[0101] Example 7 - mIF assay of PD-1 expression by lymphocyte subsets A 6-plex mIF assay for PD-1, CD8, CD4, CD20, FoxP3 and tumor (Sox10 / S100) was developed and validated on an automated platform (Leica Bond Rx). The staining order and staining conditions are shown in Table 7. It was used to assess the proportion of PD-1 expression contributed by individual lymphocyte subsets to the melanoma TME.

[0102] [Table 7]

[0103] Example 8 - Multiplex IF staining of slides In the staining process, potential sources of error arise when signals are not detected completely or false positive signals are detected in a given channel due to spillover, or "bleed-through," from another channel. Therefore, the design and optimization of a 6-plex panel includes 1) determining the staining index (SI) of each fluorophore and TSA fluorophore-marker pairing based on bleed-through calculations, 2) selecting secondary / amplification reagents, as well as 3) choosing the primary antibodies and 4) the concentrations of fluorophores to maximize sensitivity and specificity. In the final step, all optimized monoplex protocols are combined with a multiplex assay format such that comparable staining is achieved for each marker between 6-plex mIF, monoplex IF and single-stain chromogenic IHC (Figure 2A-2E and Table 8).

[0104] [Table 8]

[0105] First, the tendency of each marker to bleed-through was determined (Figure 9A-9B). Then, markers were paired with TSA fluorophores using SI and bleed-through information (Figure 2A). For example, fluorophores with high SI were paired with markers with low expression intensity, e.g., TSA fluorophore 520 and PD-L1. Fluorophore pairs "at risk" of bleed-through were assigned to markers found in different cellular compartments, allowing the removal of possible bleed-through during image analysis, e.g., CD8, a membrane stain, was paired with TSA fluorophore 540, and FoxP3, a nuclear stain, was paired with TSA fluorophore 570.

[0106] An important next step was the evaluation of secondary antibodies / amplification reagents. For example, when a “less potent” secondary antibody / HRP polymer system was used, only 50% of PD-1 expressing cells were identified compared to chromogenic IHC (Figure 2B). PD-L1 and FoxP3 also showed low expression levels, but all other markers showed comparable staining between monoplex IF and chromogenic IHC. To address the relative reduction in detection of PD-1, PD-L1, and FoxP3, we modified various components of the assay, including primary and secondary antibody reagents, incubation times, and alternative amplification methods. New secondary antibodies (PowerVision Poly-HRP, 1:1 dilution, Leica Biosystems) improved assay sensitivity for these markers (Figure 2B) and were therefore adopted for PD-1, PD-L1, and FoxP3 in the panel. Importantly, we found it important to select secondary antibodies for each marker prior to optimization of primary antibodies or TSA dilutions. Next, the primary antibody concentration is determined (Figure 2C, Figures 10A-10B), and then the TSA concentration for each fluorophore is selected (Figure 2D). The latter two steps serve to optimize the signal / noise ratio and prevent signal bleed-through or blocking, respectively.

[0107] The final step of assay validation is to combine all optimized monoplex protocols with a multiplex assay format. Following the approach described herein, comparable staining for each marker was achieved between 6-plex mIF, monoplex IF and single-stain chromogenic IHC (Figure 2E and Figure 11A). Of note, although the total cell numbers in the multiplex format matched those in the monoplex format, the dynamic range of the immunofluorescence signal (representing the spread of intensity between the 95th and 5th percentile cells expressing a given marker) was lower in the multiplex format compared to the monoplex format (Figure 11B).

[0108] [Table 9-1] [Table 9-2] [Table 9-3]

[0109] [Table 10]

[0110] [Table 11]

[0111] Example 9 - CD8+FoxP3+PD-1+ cells are strongly associated with objective responses in patients with advanced non-small cell lung cancer receiving anti-PD-1 based therapy Pretreatment specimens from n=20 lung cancer patients with advanced disease were stained with a 6-plex mIF assay and the entire specimen was imaged using a tiled mosaic of high power fields (HPFs). In this example, the 20 HPFs with the highest density of CD8 cells were selected for further analysis. The density of PD-1 and PD-L1 expressing cell populations within the HPF tiles was assessed for their predictive value for objective response (as determined by the area under the curve (AUC) of the receiver operating characteristic curve). Of the populations studied, the population that correlated most closely with a positive response to treatment were CD8+FoxP3+ cells expressing PD-1. When this population was divided into different tertiles of PD-1 expression levels (low, mid, high), PD-1 low and PD-1 mid expressing cells were most closely associated with response (Figure 24). The image correction facilitated by the approach described herein enabled such a robust and reproducible assessment of marker expression intensity in situ. This finding is noteworthy as it extends the findings observed in melanoma to non-small cell lung cancer. Similar findings were also observed in advanced Merkel cell carcinoma, supporting the relevance of this cell population and imaging and analytical approach as a pan-tumor biomarker.

[0112] Example 10 – AstroPath Imaging applied to pre-treatment specimens from non-small cell lung cancer patients treated with anti-PD-1 The median pretreatment biopsy size in this cohort was 3 mm 2 (Average 15mm 2). Evaluation of HPF sampling showed that the highest AUC was obtained when 100% of pre-treatment HPFs were sampled. In this analysis, the density of all CD8+FoxP3+ cells had the highest predictive value for a positive response for the individual features identified in the mIF assay (Figure 25A). A second analysis was performed to determine pre-treatment features predictive of the degree of pathological response to treatment (Figure 25B). Heatmaps showed associations between several potential cell populations identified by this method and pathological response values ​​of residual viable tumor 10% (rvt10, the end point of many phase II / III clinical trials) and residual viable tumor 50% (rvt50) (Figure 25B; left). Features identified using this approach were used to predict survival outcomes in these patients (Kaplan-Meier analysis) (Figure 25B; right).

Claims

1. A method for predicting the response of a subject to immunotherapy, or stratifying a subject and assigning the subject to a treatment category, comprising: (a) staining a biological sample disposed on a substrate; (b) imaging the biological sample, wherein one or more images of a high-power field (HPF) are generated; (c) detecting a plurality of biomarkers in the biological sample; (d) analyzing the one or more images to thereby predict the response of the subject to immunotherapy, or analyzing the one or more images to thereby stratify the subject and assign the subject to a treatment category. A method comprising the above steps.

2. (i) The plurality of biomarkers includes PD-1, PD-L1, CD8, FoxP3, CD163, a tumor cell marker, or any combination thereof, and optionally, the tumor cell marker includes Sox10, S100, or both. (ii) The staining includes immunofluorescence staining. (iii) The staining includes immunohistochemical staining. (iv) The biological sample is stained with an antibody, and optionally, the antibody is a monoclonal antibody or a polyclonal antibody. (v) The biological sample is stained with one or more antibodies, and optionally, the biological sample is stained with 6 antibodies or 4 antibodies. (vi) Any combination of (i) to (v). The method according to claim 1.

3. (i) The biological sample is stained with a second antibody that detects the antibody, and optionally, the second antibody is conjugated to a label, and optionally, the label is a detectable label, and optionally, the detectable label is a fluorophore. (ii) The imaging step (c) includes performing immunofluorescence microscopy on the biological sample. (iii) the analysis step (d) is: (I) acquisition and processing of an image, optionally, the acquisition of the image includes compiling one or more images to obtain an image of the entire biological sample in the substrate, and further optionally, the compiling includes overlapping and aligning the one or more images, acquisition and processing of an image; (II) cell segmentation and phenotyping; (III) image normalization and includes, or (iv) any combination of (i) to (iii), The method according to claim 1 or 2.

4. The step of cell segmentation and phenotyping includes identifying cell types in the biological sample, or The step of image normalization includes calibrating the fluorescence intensity of at least one biomarker in the one or more images against a tissue microarray. The method according to claim 3.

5. (i) The phenotyping step includes detecting the expression of at least one biomarker in the cell type, and optionally, the expression of the at least one biomarker is designated as low, medium, or high. (ii) The cell types include CD163+ macrophages, CD8+ T cells, Treg cells (CD8negFoxP3+), tumor cells, CD8+FoxP3+ cells, or any combination thereof. (iii) The cell type of CD8+FoxP3+PD-1low / mid is identified as an indicator that the subject responds to the immunotherapy. (iv) The cell type of CD163+PD-L1neg is identified as an indicator that the subject does not respond to the immunotherapy to the same extent as a reference subject identified as not having the cell type of CD163+PD-L1neg, or (v) the step of segmenting and phenotyping the cells further includes determining the density of cell types in the biological sample, The method according to claim 4.

6. The density of cell types in the biological sample is (i) determined by analyzing the distance between a cell and another cell or the tumor-stroma boundary, or (ii) high-density CD8+FoxP3+ cells are identified as an indicator that the subject responds to the immunotherapy, The method according to claim 5.

7. The analysis step (c) further includes identifying at least one biomarker in a biological sample from a subject having a disease, and the identification of the at least one biomarker is used to predict the response of the subject to immunotherapy and / or to stratify the subject and place the subject in a treatment category. Optionally, the immunotherapy includes administration of an immune checkpoint inhibitor, and / or the treatment category includes radiotherapy, chemotherapy, immunotherapy, hormone therapy, antibody therapy, or any combination thereof. The method according to claim 1 or 2.

8. The disease is cancer, optionally the cancer is selected from metastatic solid tumors, melanoma, or non-small cell lung cancer, or the cancer is selected from bladder cancer, breast cancer, cervical cancer, colon cancer, endometrial cancer, esophageal cancer, fallopian tube cancer, gallbladder cancer, gastrointestinal cancer, head and neck cancer, blood cancer, Hodgkin lymphoma, laryngeal cancer, liver cancer, lung cancer, lymphoma, melanoma, mesothelioma, ovarian cancer, primary peritoneal cancer, salivary gland cancer, sarcoma, stomach cancer, thyroid cancer, pancreatic cancer, renal cell cancer, glioblastoma, and prostate cancer, The method according to claim 7.

9. (i) the substrate is a slide, (ii) the biological sample includes a tissue, tissue section, organ, organism, organoid, or cell culture sample, and optionally, the tissue is a formalin-fixed paraffin-embedded (FFPE) tissue; (iii) the biological sample is fixed prior to step (a), and optionally, the biological sample is fixed with formaldehyde or methanol, or (iv) any combination of (i) to (iii). The method according to claim 1 or 2.

10. A method for improving the predictive value of a biomarker, comprising: (a) obtaining a plurality of high-power field (HPF) images generated from a biological sample; (b) detecting a biomarker in each of the plurality of images; (c) selecting a sub-plurality of images from the plurality of images of step (a); (d) analyzing the sub-plurality of images to thereby improve the predictive value of the biomarker.

11. (i) further comprising generating an area under the ROC (receiver operating characteristic) curve value that is greater than the area under the ROC curve value generated when analyzing all images, and / or the biomarker includes PD-1, PD-L1, CD8, FoxP3, CD163, a tumor cell marker, or any combination thereof, and optionally, the tumor cell marker includes Sox10, S100, or both, and / or (ii) the sub-plurality of images is 30% of the plurality of images of step (a), and optionally, the obtaining step (a) includes performing immunofluorescence microscopy on the biological sample. The method according to claim 10.

12. The analyzing step (d) is: (i) image acquisition and processing; (ii) cell segmentation and phenotyping; The method according to claim 10 or 11, including (iii) normalization of the image.

13. (i) The image acquisition step includes compiling a plurality of images of a high-power field (HPF) to obtain an image of the entire biological sample, and optionally, the compilation includes overlapping and aligning the plurality of images. Optionally, the cell segmentation and phenotyping steps include identifying cell types in the biological sample, or (ii) The image normalization step includes calibrating the fluorescence intensity of biomarkers in the plurality of images against a tissue microarray. The method according to claim 12.

14. (i) The phenotyping step includes detecting the expression of biomarkers in the cell type, and optionally, the expression of the biomarker is designated as low, medium, or high. (ii) The cell types include CD163+ macrophages, CD8+ T cells, Treg cells (CD8negFoxP3+), tumor cells, CD8+FoxP3+ cells, or combinations thereof. (iii) The cell type of CD8+FoxP3+PD-1low / mid is identified as an indicator that the subject responds to the immunotherapy. (iv) The cell type of CD163+PD-L1neg is identified as an indicator that the subject does not respond to the immunotherapy to the same extent as a reference subject identified as not having the cell type of CD163+PD-L1neg, or (v) The cell segmentation and phenotyping steps further include determining the density of cell types in the biological sample. The method according to claim 13.

15. (i) The density of cell types in the biological sample is determined by analyzing the distance between cells and other cells. (ii) The density of cell types in the biological sample is determined by analyzing the distance between cells and the tumor-stroma boundary, or (iii) high-density CD8+FoxP3+ cells are identified as an indicator of the subject's response to the immunotherapy, The method according to claim 14.

Images