Artificial Intelligence-based Radiomics in the Era of Immuno-oncology

Abstract

The recent, rapid advances in immuno-oncology have revolutionized cancer treatment and spurred further research into tumor biology. Yet, cancer patients respond variably to immunotherapy despite mounting evidence to support its efficacy. Current methods for predicting immunotherapy response are unreliable, as these tests cannot fully account for tumor heterogeneity and microenvironment. An improved method for predicting response to immunotherapy is needed. Recent studies have proposed radiomics—the process of converting medical images into quantitative data (features) that can be processed using machine learning algorithms to identify complex patterns and trends—for predicting response to immunotherapy. Because patients undergo numerous imaging procedures throughout the course of the disease, there exists a wealth of radiological imaging data available for training radiomics models. And because radiomic features reflect cancer biology, such as tumor heterogeneity and microenvironment, these models have enormous potential to predict immunotherapy response more accurately than current methods. Models trained on preexisting biomarkers and/or clinical outcomes have demonstrated potential to improve patient stratification and treatment outcomes. In this review, we discuss current applications of radiomics in oncology, followed by a discussion on recent studies that use radiomics to predict immunotherapy response and toxicity.

Tumor response assessment on imaging following immunotherapy

“…The use of machine learning based blood biomarkers, such as neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio, have already been shown promising results in for patient selection and predicting the treatment outcome in non-small cell lung cancer patients treated with nivolumab ( 104 ). There is, moreover, an increasing interest in radiomics models trained the clinical outcomes ( 105 ). Radiomics uses large numbers of features extracted with data characterization algorithms from medical imaging to define tumor patterns and features that are not visible to the human eye.…”

“…The field of radiomics may also offer additional information for the prediction of prognosis in patients treated with immunotherapy based on established molecular biomarkers ( 73 , 105 ). A recent multi-center retrospective study of large populations of non-small cell lung cancer patients demonstrated the utility of a deep learning score based on radiomics features extracted from 18 F-FDG-PET/CT to predict the PD-L1 expression status on immunohistochemistry ( 73 ).…”

Unlocking the Potential of the Human Microbiome for Identifying Disease Diagnostic Biomarkers

“…The identification of “ideal biomarkers” is considered a daunting task for many diseases, including some cancer types. Most of the current sampling techniques for cancer tissues cannot identify individuals who will lack response to therapy, and they fall short in classifying cancer types correctly, owing to the inter- and intra-tumor heterogeneity of tumors [ 80 ]. A biomarker should be easily measurable, non-invasive, and cost-effective.…”

“…Recently, some clinical trials have used diverse approaches to define characteristics of the patients who develop primary or acquired resistance to immunotherapy (e.g., NCT04243720) [ 168 ]. Such trials are aiming to develop an integrated model to predict drug resistance relying on multimodal data including radiomics, genomics, transcriptomics, epigenetics, immunophenotypic data, and fecal microbiome data [ 80 ]. There is promise that artificial intelligence models combining microbiome-based biomarkers with other omics data (e.g., radiomics) will be able to provide a more comprehensive view of the tumor microenvironment, aiding in better cancer diagnosis and allowing clinicians to non-invasively track changes in cancer phenotypes [ 80 ].…”

Prognostic Value and Quantitative CT Analysis in RANKL Expression of Spinal GCTB in the Denosumab Era: A Machine Learning Approach

“…An accurate assessment of preoperative RANKL levels may also help clinicians and patients make treatment decisions, including the determination of whether drug therapy, extended resection, postoperative radiotherapy, intensive follow-up, and reasonable clinical expectations. Although tissue biopsy before surgery can be used for disease detection, aside from invasive risk, a small amount of biopsy tissue cannot be used enough to evaluate the entire tumor heterogeneity, and making assessments based solely on small tissue biopsies can lead to bias in diagnosis and evaluation [ 11 ].…”