Immunogradient Indicators for Antitumor Response Assessment by Automated Tumor-Stroma Interface Zone Detection

Abstract

The distribution of tumor-infiltrating lymphocytes (TILs) within the tumor microenvironment provides strong prognostic value, which is increasingly important with the arrival of new immunotherapy modalities. Both visual and image analysis–based assays are developed to assess the immune contexture of the tumors. We propose an automated method based on grid subsampling of microscopy image analysis data to extract the tumor-stroma interface zone (IZ) of controlled width. The IZ is a ranking of tissue areas by their distance to the tumor edge, which is determined by a set of explicit rules. TIL density profiles across the IZ are used to compute a set of novel immunogradient indicators that reflect TIL gradient towards the tumor. We applied this method on CD8 immunohistochemistry images of surgically excised hormone receptor–positive breast and colorectal cancers to predict overall patient survival. In both cohorts, the immunogradient indicators enabled strong and independent prognostic stratification, outperforming clinical and pathologic variables. Patients with breast cancer with low immunogradient levels had a prominent decrease in survival probability 5 years after surgery. Our study provides proof of concept that data-driven, automated, operator-independent IZ sampling enables spatial immune response measurement in the tumor-host interaction frontline for prediction of disease outcomes. The distribution of tumor-infiltrating lymphocytes (TILs) within the tumor microenvironment provides strong prognostic value, which is increasingly important with the arrival of new immunotherapy modalities. Both visual and image analysis–based assays are developed to assess the immune contexture of the tumors. We propose an automated method based on grid subsampling of microscopy image analysis data to extract the tumor-stroma interface zone (IZ) of controlled width. The IZ is a ranking of tissue areas by their distance to the tumor edge, which is determined by a set of explicit rules. TIL density profiles across the IZ are used to compute a set of novel immunogradient indicators that reflect TIL gradient towards the tumor. We applied this method on CD8 immunohistochemistry images of surgically excised hormone receptor–positive breast and colorectal cancers to predict overall patient survival. In both cohorts, the immunogradient indicators enabled strong and independent prognostic stratification, outperforming clinical and pathologic variables. Patients with breast cancer with low immunogradient levels had a prominent decrease in survival probability 5 years after surgery. Our study provides proof of concept that data-driven, automated, operator-independent IZ sampling enables spatial immune response measurement in the tumor-host interaction frontline for prediction of disease outcomes. One of the most prominent discoveries in modern genetics—analysis of the whole human genome sequence—has made it possible to determine specific genetic mutations associated with cancer. This determination has led to the definition of cancer as a cell disease caused by genetic mutations.1Stratton M.R. Campbell P.J. Futreal P.A. The cancer genome.Nature. 2009; 458: 719-724Crossref PubMed Scopus (2385) Google Scholar,2Wong K.M. Hudson T.J. McPherson J.D. Unraveling the genetics of cancer: genome sequencing and beyond.Annu Rev Genomics Hum Genet. 2011; 12: 407-430Crossref PubMed Scopus (81) Google Scholar Although genetic mechanisms explain many aspects of tumor progression, many hallmarks of cancer, including host immune and inflammatory response, angiogenesis, and metabolic disarrangements, evolve in the context of the tumor microenvironment (TME).3Hanahan D. Weinberg R.A. Hallmarks of cancer: the next generation.Cell. 2011; 144: 646-674Abstract Full Text Full Text PDF PubMed Scopus (42734) Google Scholar In particular, a major driving force of local tumor–host tissue interactions are inflammatory and immune cell infiltrates. Solid tumors are infiltrated by both innate immunity cells (natural killer cells, macrophages, neutrophils, and phagocytes) and adaptive immunity cells (T lymphocytes, B lymphocytes, and dendritic cells).4Whiteside T.L. The tumor microenvironment and its role in promoting tumor growth.Oncogene. 2008; 27: 5904-5912Crossref PubMed Scopus (1479) Google Scholar Recently, discoveries of the mechanisms by which cancer cells inhibit host antitumor immune response and new immunomodulating therapies have shifted focus toward a search for antitumor immune response components and biomarkers.5Taube J.M. Galon J. Sholl L.M. Rodig S.J. Cottrell T.R. Giraldo N.A. Baras A.S. Patel S.S. Anders R.A. Rimm D.L. Cimino-Mathews A. Implications of the tumor immune microenvironment for staging and therapeutics.Mod Pathol. 2018; 31: 214-234Crossref PubMed Scopus (206) Google Scholar,6Parra E.R. Francisco-Cruz A. Wistuba II, State-of-the-art of profiling immune contexture in the era of multiplexed staining and digital analysis to study paraffin tumor tissues.Cancers (Basel). 2019; 11: E247Crossref PubMed Scopus (73) Google Scholar Tumor-infiltrating lymphocytes (TILs) and their distributions within TME compartments have been reported as potential prognostic and predictive biomarkers in various types of cancer. Studies in clinical and experimental settings have revealed the prognostic role of CD3+, CD4+, CD8+, and FOXP3+ TILs in many types of solid human tumors, such as melanoma, colorectal, and J.M. Galon J. Sholl L.M. Rodig S.J. Cottrell T.R. Giraldo N.A. Baras A.S. Patel S.S. Anders R.A. Rimm D.L. Cimino-Mathews A. Implications of the tumor immune microenvironment for staging and therapeutics.Mod Pathol. 2018; 31: 214-234Crossref PubMed Scopus (206) Google L.M. and PubMed Scopus Google Galon J. The immune contexture in human on clinical Rev 12: PubMed Scopus Google in tissue a inflammatory Google D. A. D. A. in cancer 2019; PubMed Scopus Google Scholar TILs are by of and B cells with interactions and their in the TME both and spatial aspects to their as major components of antitumor cell as immunotherapy for human PubMed Scopus Google Scholar study of colorectal cancer by Galon J. A. the a PubMed Scopus Google Scholar based on digital image analysis of immunohistochemistry revealed that the of and TILs in the tumor and the with the of the This led to a which to the staging J. A. the of the in the of Pathol. PubMed Scopus Google Scholar have prognostic and predictive of TIL in and human breast cancer of tumor-infiltrating density a method based on and in breast 2018; PubMed Scopus Google J. J. J. A. A. A. P.A. lymphocytes and in of breast cancer: a analysis of with 2018; Full Text Full Text PDF PubMed Scopus Google Scholar cells are an important in the concept of of a antitumor response to in various The cancer as a for immunotherapy in 2019; Full Text Full Text PDF PubMed Scopus Google A. J. for the towards immunotherapy of 12: Full Text Full Text PDF PubMed Scopus Google A. the PubMed Scopus Google Scholar The of has a search for novel image to image biomarkers for colorectal, and of the in cell 2019; PubMed Scopus Google J. A. for prognostic in and 2018; PubMed Scopus Google automated analysis of tumor and their spatial prognostic in colorectal 2019; PubMed Scopus Google Scholar the of method that to a of tumor immune contexture is the of the immune cell and their to the tumor Galon J. The immune contexture in human on clinical Rev 12: PubMed Scopus Google P.A. tumor-infiltrating lymphocytes in solid a for and for a method the biomarkers the host immune response, TILs in breast and in tumor and areas for Pathol. PubMed Scopus Google P.A. tumor-infiltrating lymphocytes in solid a for and for a method the biomarkers TILs in melanoma, cell and and cell of the and and Pathol. PubMed Scopus Google Scholar The of immune cells in the TME of the tumor and a for a and of the Galon J. The immune contexture in human on clinical Rev 12: PubMed Scopus Google Scholar of visual for the are the of the of the by J. J.M. The of cancer: and a analysis of PubMed Scopus Google Scholar in prognostic indicators in and of the and is a of the tumor and host tissue is possible of the with an have a in which it is to host tissue tumor cell J. J.M. The of cancer: and a analysis of PubMed Scopus Google J.M. new prognostic of Full Text PDF PubMed Scopus Google Scholar used the definition by A. A. A. M.R. Galon J. The tumor microenvironment and are of to PubMed Scopus Google a the the host tissue this definition an explicit of the it an to This a of it to and in tumors with and and it the of analysis are The to a is to the which the in the the within the tumor as of the the and clinical of TILs and assays Recently, J. A. for prognostic in and 2018; PubMed Scopus Google Scholar applied a tissue to search for biomarkers in their study of cancer In particular, their used to tumor and areas and the tumor as a to both tumor and A. of immune contexture 2018; PubMed Scopus Google Scholar a method to immune contexture it a of tumor In this a novel set of immunogradient indicators based on a new method of automated of the tumor-stroma interface zone The method the tumor a set of explicit based on the IZ is and by the distance the to TIL density profiles across the The TIL density toward the independent prognostic of overall survival in with hormone receptor–positive and This study in and The of with hormone receptor–positive breast as A. A. is an independent prognostic of overall survival in with hormone breast PubMed Scopus Google A. A. D. J. profiles of breast analysis of digital image analysis Pathol. PubMed Scopus Google Scholar the to years in to the of and the of The with to of and by in and the of The and to years The and of the and patient are in and cancer cancer years in a new The patient by and in with the by the for the patient and for patient to in the tissue used for the tumor tissue that the of tumor tissue The tissue and on by automated CD8 used by of an The with staining in human The a of the whole images tissue to and of and and and of and in and the in used to and extract of image analysis by a whole images of the to a specific tissue for the of tumor of and analysis images are in and B The automated the of the tissue by and grid is in a to the as in A. A. is an independent prognostic of overall survival in with hormone breast PubMed Scopus Google A. A. D. J. profiles of breast analysis of digital image analysis Pathol. PubMed Scopus Google Scholar a set of data and cells and tissue is to the method on the grid of of many by the are of tumor and the by and as and grid the of data for tumor as is with the in the grid is the of for is the of tumor is by the of the gradient the in tumor used for interface is as The of the gradient is to the as to gradient for image Scopus Google J. in a Scopus Google Scholar used a to for to the gradient to the of In is in the and it of tumor. the of the the tumor of are of a tumor-stroma and a by it with the in and the in and are as The is that the of across in tumor as caused by in and that that is The are and and by and the of tumor tumor-stroma and on the grid as as the and the tumor is used as a probability for a of the The set of for is as with in tumor are by for which this is the the of and determine the is of the the of the by are by the of of and and are as as tumor in and the tumor and areas are of the is of the areas of the of a distance distance to the the a for distance the tumor and aspects of the are The has the the is the distance and in it is the The tissue and are in The of with tumor and for of width. IZ of is to toward it is possible to of by In this of and of are to a IZ with tumor and aspects as and their distance to the an of cell within and across the IZ by the data for The a interface that cell and across the of cell density profiles are for tumors in set of both the and the IZ and immunogradient indicators in various tumor and density profiles are in of and data cell and B and and in IZ of are of in are and of CD8 density in CD8 in tumor indicators of interface in a new of interface the of to tumor indicators indicators of cell density within and across the IZ for to predict of the cell density indicators to the most of as in the for for and of for of cell The a the which have to the for it to in it as a of which of the interface a has a as This in the cell density in the of the and is as the of the in and an decrease in cell density to the data and indicators are in The for the image analysis of the whole images and the of the and IZ for of the indicators of and IZ are in for and IZ of and of CD8 indicators cell interface zone (IZ) and tumor settings and for the breast cancer and and for the colorectal cancer and are in the and tumor breast of for CD8 density by in of for CD8 by in of colorectal cancer. in a new The interface zone (IZ) and tumor settings and for the breast cancer and and for the colorectal cancer and are in the and tumor breast of for CD8 density by in of for CD8 by in of colorectal cancer. and with tumor areas in the whole image to extract IZ The distributions of cell and revealed used for the of the is used in the and distribution with based on analysis of by for and a used to of analysis the method of based on the of the of of the The J. a and PubMed Scopus Google Scholar used to determine a for to and of The distributions for the the The used to assess the of the analysis of the data a by a to the independent prognostic of TIL density indicators in the context of variables. both cohorts, the of the of CD8 density and CD8 gradient a by the immunogradient CD8 with the IZ cell response and in with the and in with for the of Patients with and in with for the of Patients with by by IZ cell response and IZ cell response and and breast of for of for density of CD8 colorectal interface zone of tumor in a new breast of for of for density of CD8 colorectal interface zone of tumor The and of data in and used to for with data and and immunogradient are for both the and cohorts, including pathologic and survival The of and used to the prognostic of the cell in the and tumor of the IZ and the density profiles across The density indicators and IZ of and as and with of both and The most prognostic in with for and in with for the The of the indicators are in The distribution of cells within the IZ in both the and most in the and within the and and and in the tumor of the IZ in both patient cohorts, a of IZ cell density decrease toward the tumor independent of IZ cell density indicators for both the and the and of the in the data set in and and the the by strong of the that the of density within IZ aspects and as the IZ density to as density by strong of for both and of the and strong of the and the density across the of reflect toward the tumor of the as the IZ density IZ density gradient toward to as gradient both important of the immune response, the of and as IZ cell response to their prognostic probability based on are in and prognostic by are in for both used the set of IZ cell density indicators with the and the IZ cell density indicators with the IZ The in both patient the indicators a with for the in the a with prognostic of the in the In the by the for the density and of the density in the tumor of the IZ as as the IZ cell response in the context of by the in the the for the density to predict in the IZ cell response in the context of by the patient and the in both and revealed strong independent prognostic of the IZ density indicators the density gradient toward the tumor in with density of in the IZ as the IZ cell response Our study that automated of the tumor-stroma IZ in digital microscopy images of tumor with of immune cell density profiles provides novel indicators to assess antitumor immune response in the frontline of tumor-host Although the IZ in method is to the concept of used in and in P.A. tumor-infiltrating lymphocytes in solid a for and for a method the biomarkers the host immune response, TILs in breast and in tumor and areas for Pathol. PubMed Scopus Google P.A. tumor-infiltrating lymphocytes in solid a for and for a method the biomarkers TILs in melanoma, cell and and cell of the and and Pathol. PubMed Scopus Google J. A. A. A. and of immune cells within human colorectal tumors predict clinical PubMed Scopus Google A. A. A. A. J. D. of and immune in and cell PubMed Scopus Google density is associated with cell and in human 2018; PubMed Scopus Google immune PubMed Scopus Google Scholar the IZ is data and independent of visual the to tumor is by spatial within the which enables of immune cell density indicators across the IZ to reflect the gradient of the and the grid subsampling of the of the indicators within of the The set of immunogradient indicators provides strong and independent prognostic in the and patient the method is and for various assays of inflammatory and immune response in tumor and Our method subsampling with explicit to extract and the tumor-stroma digital images of microscopy in the probability of is In for and the to a a the of immune cell distribution in the tumor microenvironment spatial analysis for digital image PubMed Scopus Google Scholar In the on a of J. A. the a PubMed Scopus Google J. A. the of the in the of Pathol. PubMed Scopus Google J. A. A. A. and of immune cells within human colorectal tumors predict clinical PubMed Scopus Google immune PubMed Scopus Google Scholar and most immune profiling based on of the D. D. J. A. J. D. of immune cells in human solid 2018; PubMed Scopus Google J. A. J. of tumor-infiltrating lymphocytes in human cancers by image 2018; PubMed Scopus Google and automated image analysis for of tumor lymphocytes in Pathol. 12: PubMed Scopus Google Scholar is and of and are in tumors with including tumor and in colorectal 2008; Google of tumor in colorectal and to clinical Pathol. 2018; PubMed Scopus Google Scholar In is data and independent of visual of a tumor and to tumor cell method tissue areas to the tumor-host interface and its spatial to the tumor host of the it is on a specific tumor and the IZ in tumors with in with and the the method provides the important of of the IZ and spatial ranking the and tumor aspects within the This method enables of IZ in immune cell density indicators that reflect of the infiltrates. The in IZ ranking and of tissue aspects are of the of is The biomarkers as a of on the tumor and immune cells as by P.A. tumor-infiltrating lymphocytes in solid a for and for a method the biomarkers the host immune response, TILs in breast and in tumor and areas for Pathol. PubMed Scopus Google P.A. tumor-infiltrating lymphocytes in solid a for and for a method the biomarkers TILs in melanoma, cell and and cell of the and and Pathol. PubMed Scopus Google J. A. A. A. and of immune cells within human colorectal tumors predict clinical PubMed Scopus Google Scholar reported prognostic of TIL density for of the A. A. A. A. J. D. of and immune in and cell PubMed Scopus Google Scholar used an that within the tumor to the tumor in and cell density is associated with cell and in human 2018; PubMed Scopus Google Scholar the as a tumor of the tumor and the in immune PubMed Scopus Google Scholar the as a with a of the tissue and the tumor in study a survival for with and TIL within the to distance a tumor A. N.A. A. J. D. analysis spatial cell in the of colorectal cancer associated with PubMed Scopus Google Scholar that TIL in with the tumor cells and antitumor J. A. A. A. and of immune cells within human colorectal tumors predict clinical PubMed Scopus Google Scholar TIL in prognostic with based on tumor Our in analysis settings to the of tissue sampling and IZ sampling width. We various analysis settings in study with The prognostic with IZ in and IZ in which by the to IZ in hormone receptor–positive with TILs to data for the immunogradient applied to tumor cell and J. A. for prognostic in and 2018; PubMed Scopus Google Scholar of immune contexture to predict disease in with cancer. used to and the tumor the the tumor as a that to the and of the tumor of as as TME the set of density and distance indicators for CD3+, CD8+, and cells within prognostic indicators the of biomarkers within the the of their indicators in cancer a density across the A. A. of cells tumor cell in breast A. 2019; PubMed Scopus Google Scholar spatial aspects of cell tumor cell in with Although of the their study an automated tumor cell based on This method to across the to potential of TIL by and that TILs the tumor cell In both the tumor-stroma within the TME the which is to definition of the the IZ is based on subsampling of the tissue and is on the tumor and in tumor it is to a of a in a image of an We the prognostic of the immune cell density profiles in the IZ in the and cohorts, which led to a set of immunogradient indicators that enabled strong and independent prognostic in both patient We various analysis settings to the most prognostic for of the and that a IZ prognostic for IZ data in both the and cohorts, it that the indicators a of cells toward the tumor within the IZ independent of In particular, the which the gradient of cells toward the tumor within the associated with in both patient the which an decrease in cell density in the tumor of the IZ in the of the associated with the is independent of the of the IZ it the this in of low tumor in which tumor ranking The indicators that reflect the spatial of the IZ cells toward the tumor revealed prognostic to the density of IZ cells TIL and their in the tumor and areas Both IZ density and IZ density gradient associated in with in both the and and and the IZ cell response by the of the an independent prognostic of in both patient This that density and gradient aspects of the immune cell across the IZ prognostic of the of antitumor immune the IZ of the with prognostic with probability of and which is to the with by based on and CD8 as reported by of the for the of cancer: a prognostic and 2018; Full Text Full Text PDF PubMed Scopus Google with the and in with for the of Patients with and in with for the of Patients with indicators and density gradient IZ CD8 by breast of for of for density of CD8 colorectal interface zone of tumor in a new breast of for of for density of CD8 colorectal interface zone of tumor a prognostic in hormone receptor–positive which is a disease an role for immune a prominent in the prognostic of the with of the after 5 years in both the this to and probability after years in the with and low IZ cell response this is revealed with the indicators and in the by the cell density on tumor of the IZ This the potential of the immunogradient a and for a to predict of with this it that antitumor immune response which determine the of the are in the TME and by the immunogradient of this to the that the immune response are in the of and are to cancer cells this the of with hormone receptor–positive with low immunogradient immunotherapy In this proof of concept and a of a new method that provides prognostic in independent patient are to the analysis settings and to clinical of the immunogradient indicators for various cancer types and prognostic In to in with applied to the patient In study the potential of automated, and sampling of tumor-stroma in microscopy images to extract indicators of antitumor immune The IZ immunogradient enables strong and independent prognostic of with and with prominent of probability in with hormone 5 years after surgery. and the interface zone and R.A. the interface zone and in with and of the and and the in and of the data and and of the for a patient with breast cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with breast cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with breast cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with breast cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with breast cancer interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with colorectal cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with colorectal cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with colorectal cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with colorectal cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with colorectal cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image for a patient with colorectal cancer The interface zone of on the whole tumor tumor and CD8 density within the the to the in and of and by a within the of the for the of with are by the The of the for and for image

Topics & Concepts