The great progress of immunotherapy has changed the current mode of cancer treatment. However, since only a few patients respond to immune checkpoint blocking and other immunotherapy strategies, more new technologies are needed to decipher the complex interaction between tumor cells and tumor immune microenvironment (TME) components.
Tumor immunogenomics refers to the comprehensive study of TME by using the multi-group data reflecting tumor immune status such as immunogenomics, immunoproteomics, and immunobioinformatics. It depends on the rapid development of next-generation sequencing technology. High-throughput genome and transcriptome data can be used to calculate the abundance of immune cells and predict tumor antigens. However, since batch sequencing represents the average characteristics of heterogeneous cell populations, it is impossible to distinguish different cell subtypes. The technology based on single cell can better analyze TME through accurate study of immune cell subpopulation and spatial structure. In addition, the deep learning model based on radiology and digital pathology contributes to the research of tumor immunity to a large extent, and these artificial intelligence technologies perform well in predicting the response of immunotherapy. The progress and breakthrough of these new technologies have far-reaching significance for cancer treatment.
Tumor immune microenvironment
In the past few years, the research progress of tumor immunity has fundamentally changed our understanding of cancer. The definition of tumor has also evolved from simple tumor cell aggregation to a complex organ-like structure, consisting of tumor cells, immune cells, fibroblasts, vascular endothelial cells and other surrounding stromal cells. The structure composed of various cells and components near tumor, such as immune infiltrating cells, blood vessels, extracellular matrix, etc., also known as tumor immune microenvironment, has become one of the most popular research topics in oncology. TME has been proved to play a decisive role in carcinogenesis, tumor progression, metastasis and recurrence.
TME contains an extremely diverse subset of immune cells, including T lymphocytes, B lymphocytes, natural killer (NK) cells, macrophages, dendritic cells (DC), granulocytes and myelogenous suppressor cells (MDSCs). In general, T cells, B cells, NK cells and macrophages help to inhibit tumor growth, while MDSC and regulatory T cells (Treg) tend to inhibit anti-tumor immunity. However, existing studies have confirmed that given the complex interaction with tumor cells, the specific role of immune cells may change dynamically, or even become completely opposite.
In short, all kinds of immune cell types, even different functional states of specific immune cell types, may have diametrically opposite effects against tumor immunity. Therefore, we need to use the most advanced bioinformatics technology to systematically describe the immunological characteristics of tumors to the greatest extent and provide more information to enhance our understanding of tumor immunity.
Immunogenomics in the NGS era
In the past two decades, NGS, including whole-genome sequencing (WGS), whole-exome sequencing (WES) and RNA sequencing (RNA seq), has been successfully developed and applied to obtain human whole-genome information. NGS produces high-throughput genomic and transcriptional data, which lays the foundation for the study of multi-step immune response.
Quantify immune cells in TME
TME is composed of a variety of immune cells. For the quantitative analysis of tumor immune cell components, traditional methods, such as flow cytometry and immunohistochemistry (IHC), are not suitable for large-scale analysis due to their high cost and low tissue availability. With the rapid development of NGS, we can estimate the abundance of dozens of immune cell types through NGS data, which are also proved to be reliable. The sources of these analyses are mainly DNA and RNA sequencing, especially the latter. As for RNA sequence data, the principle of calculation methods is mainly divided into gene set enrichment analysis (GSEA) and deconvolution.
Generally, representative algorithms based on GSEA include ESTIMATE, xCell and MCP counting. A common feature of GSEA-based methods is the need to establish a specific gene set for each interested immune cell subgroup. The deconvolution of cell components is the reverse process of cell subtype convolution in body tissues based on gene expression characteristics. Tools based on deconvolution include decoraseq, PERT, CIBERSORT, TIMER, EPIC, quanTIseq and deconf
Identification of tumor antigen
Somatic DNA mutations, including single nucleotide mutation (SNV) and insertion and deletion (INDEL), are the main sources of abnormal antigens. At present, the Genome Analysis Kit (GATK) is an industry standard for identifying SNV and INDEL by analyzing WES, WGS and RNA sequence data. Its scope is also expanding to cover copy number variations (CNVs) and structural variations (SVs).
In addition, the abnormal peptide needs to bind with HLA to assist the recognition of T cell receptor (TCR), thus triggering the immune response. Predicting HLA typing is crucial for recognizing tumor antigens. HLA miner and Seq2HLA are two early tools for HLA typing from NGS data. PHLAT, HLAReporter, SNP2HLA, HLA-HD, optype and HLA-VBSeq perform well in four, six and eight resolution of different cancers. Among these tools, Polysolver is currently one of the recognized standard tools that use low coverage WES data.
In addition to recognizing abnormal peptides and HLA typing, antigen MHC binding affinity is the next focus of tumor antigen prediction. Many peptide-MHC-I (pMHC-I) affinity prediction tools are based on artificial neural network (ANN) training methods and site-specific scoring matrix (PSSM), such as the widely used tools NetMHC and NetMHCpan. However, due to the diversity of the length of MHCII binding peptide and the “openness” of the binding region, it is more challenging to predict the affinity of pMHC II. The number of available methods for predicting the affinity of pMHC II is far less than that of pMHC-I.
Immunohistochemistry in the single-cell era
Although the research on tumor immunity using NGS technology has greatly promoted the development of oncology, batch sequencing may lead to signal dilution below the detection limit and mask the response of individual cells. This may mask many important biological phenomena. Until recently, technological breakthroughs in single-cell-related methods have completely changed our understanding of tumor immunity, and the research level has been transferred from the regional level to the single-cell level.
Multicolor flow cytometry
The ability of multi-parameter analysis to distinguish different subsets of immune cells functionally and physically has led to the development of flow cytometry into a routine 8-parameter flow cytometry. In addition, with the progress of technology, instrument design that can measure more parameters has been realized, such as 30 parameters and 50 parameters flow cytometry. However, due to the lower accuracy of the more measurable parameters, or the limited accuracy of the more measurable parameters, especially due to the overlap between the emission spectra of fluorescent dyes, these shortcomings to some extent limit the application and further development of multicolor flow cytometry.
Mass flow cytometry
Mass spectrometry is the latest innovation in this field, also known as time of flight (CyTOF) flow cytometry, which combines flow cytometry with mass spectrometer. Compared with traditional flow cytometry, the mass spectrometer uses metal isotopes instead of fluorophores to label antibodies, and then uses a time-of-flight detector to quantify the signal. The detector can detect at least 40 parameters and avoid spectral overlap. CyTOF has been proved to be an accurate high-dimensional analysis method of tumor tissue for exploratory immunoassay and biomarker discovery.
Although in theory, mass flow cytometry allows us to detect up to 100 parameters per cell, the processing speed and flux are limited by ion flight. After atomization and ionization, the cells were completely destroyed in the pretreatment process, resulting in the infeasibility of subsequent cell classification application. In addition, CyTOF may not be suitable for measuring some low expression molecular characteristics due to its low sensitivity.
Spectral flow cytometry
Spectral flow cytometry is another latest technological progress to promote the efficacy of traditional flow cytometry. Unlike mass spectrometer, spectral flow cytometer still uses fluorescent dye to label antibody, but it replaces traditional optics and detectors with dispersion optics and new detectors for measuring total emission spectrum. Based on the same principle, traditional flow cytometry and spectral flow cytometry maintain quite good compatibility, especially in the availability of commercial antibodies, but can better eliminate confusion factors, such as spectral overlap, to improve efficiency. With the development of compensation technology, spectral flow cytometry may replace multicolor flow cytometry.
Single cell RNA sequencing
The technology based on flow cytometry combines a specific tag with the corresponding cell subgroup and identifies the tag, indicating that the target must be determined before sample collection, and the initial target limit limits the information obtained from these technologies, and only through these technologies can “known unknown” be found.
The emergence of single-cell sequencing technology has pushed the single-cell field to a new height. It is no longer limited by the predetermined goals such as flow cytometry, and can use the standard NGS protocol to sequence a single cell to obtain unbiased multi-group analysis that can be used to identify “unknown”.
At present, the application of scRNA-seq is more mature than other methods. The field of tumor immunotherapy has provided us with many valuable discoveries and enlightenment. However, the technical noise generated by the amplification of trace substances remains the most significant challenge. How to separate a single cell and maintain its biological activity, how to solve the huge technical noise caused by amplification and improve sensitivity, how to obtain the highest number of measurable genes at the lowest price, and how to analyze data more effectively, all of these greatly raise the threshold of single cell sequencing and limit its wide application.
Immunology and artificial intelligence
The technical progress of artificial intelligence in tumor immune research mainly involves the following aspects: (1) reducing the workload of artificial recognition of immune infiltration on pathological sections; (2) An alternative technology is provided to identify immune cell subpopulations and spatial structures that are difficult to recognize by the naked eye; (3) To provide a non-invasive method to predict the TME characteristics and response to immunotherapy of specific patients.
Tumor antigen prediction based on deep learning method
The first step in deciphering tumor antigens is to predict abnormal peptides. In addition to various algorithms for SNV recognition, the recently designed CN learning tool has also been designed to detect CNV, showing good performance. Regarding HLA typing, Bulik et al. generated a large comprehensive data set, including HLA types and HLA peptides of various types of cancer tissues, and published data that can be used to train the complete mass spectrum in-depth learning model EDGE, which has been verified in patients with non-small cell lung cancer (NSCLC). In addition, two promising computational deep learning methods, MARIA and MixMHC2pred, have been developed recently, greatly improving the prediction accuracy of MHC-II.
Application of Radiology in Tumor Immunology
With the development of artificial intelligence in medical imaging, imaging is not only a picture, but also a large-scale digital data. The process of analyzing imaging data using AI technology is radiomics. Radiohistochemical techniques applied to tumor immunity are mainly used to identify biomarkers reflecting immune infiltration and predict the treatment response of patients treated with ICB.
Computational pathology in tumor immunity
AI in pathology, or so-called digital pathology, provides new insights into the interaction between immune cells and tumor cells and the relationship between key behaviors of cancer biology through computational analysis.
Similar to radiology, digital pathology combined with deep learning excavates invisible information from images, enabling us to understand TME at the cellular or molecular level. Digital pathology may be a promising method to study the structure of TME and the relationship between cancer biology and treatment.
Application of immunohistochemistry in tumor immunotherapy
Identify biomarkers of ICB for patient stratification
As a target of ICB, the expression level of PD-L1 detected by IHC is the first predictive biomarker found. However, some clinical trials show that ICB has only slight effect on some patients with high expression of PD-L1, and ICB will also respond to patients with low expression of PD-L1. Therefore, other biomarkers are urgently needed to fill this gap.
In 2014, researchers first linked the tumor mutation load (TMB) with the clinical survival rate of patients treated with CTLA-4 inhibitors through WES. Subsequently, other retrospective studies also demonstrated that high TMB was associated with lasting clinical benefits. As for the method used to evaluate TMB, due to the high cost and complexity of WES, FDA approved two alternative NGS platforms, namely FoundationOne CDx (F1CDx) and MSKCC Operable Cancer Target Integrated Mutation Profile (MSK-IMPACT), which were verified by multiple cancer prospective studies.
On the other hand, immune cell infiltration, especially TIL, plays a key role in immune response. In order to find more ideal biomarkers for treatment and prognosis, single cell sequencing was used to identify more immune cell subsets. It has been found that TCF7+memory-like T cells are associated with the clinical improvement of melanoma patients after anti-PD1 treatment, while stem cell-like TCF1+PD1+T cells have been proved to be helpful for tumor control in ICB treatment. Through single cell sequencing technology, more T cell subsets and functional status related to treatment and prognosis were determined.
Prediction of new antigens in ACT treatment
Adoptive cell therapy (ACT) is the retransfusion of transgenic or expanded autologous or allogeneic T cells into patients to enhance anti-tumor immunity. Immunohistochemistry is mainly used to identify ideal tumor antigens in ACT treatment.
At present, new antigen specific TCR-T cells have not yet entered clinical application. However, it is gratifying to see that some case reports show the effectiveness of T cells from colorectal cancer, breast cancer and bile duct cancer patients predicted by immunohistochemistry in recognizing new tumor antigens. Tran et al. conducted WGS on samples from patients with metastatic cholangiocarcinoma and identified 26 somatic mutations. Transcription and transfection of the tandem microgene composed of mutant genes into the autologous APC, then co-culture the new antigen presenting APC with the TIL from patients, and finally identify the antigen-specific CD4+Vb22+T cell clone to induce the regression of epithelial cancer.
The traditional selection of new antigens based on co-culture of autologous APC and T cells is limited due to its low throughput, high cost and time-consuming characteristics. In order to eliminate these obstacles, more high-throughput immunogenic new antigen detection technologies have been developed. Li et al. established a platform based on trogocytosis. In this platform, when TCR is combined with pMHC, the surface marker protein is transferred from APC to T cells. Therefore, the ideal new antigen can be identified by analyzing the labeled protein positive cells. In the future, these emerging immune genomics technologies will achieve high-throughput new antigen selection.
Selecting new antigens for individualized tumor vaccines
Immunohistochemical methods have been widely used in vaccine development in clinical research. In general, the new antigens used to generate personalized vaccines are analyzed by the WES and RNA sequences of tumor and normal tissues, and by algorithms (such as NetMHCpa