Facts and Hopes on the Use of Artificial Intelligence for Predictive Immunotherapy Biomarkers in Cancer

Abstract

Immunotherapy by immune checkpoint inhibitors (ICI) has become a standard treatment strategy for many types of solid tumors. However, the majority of cancer patients will not respond and predicting response to this therapy is still a challenge. Artificial intelligence (AI) methods can extract meaningful information from complex data, such as image data. In clinical routine, radiology or histopathology images are ubiquitously available. AI has been used to predict the response to immunotherapy from radiology or histopathology images, either directly or indirectly via surrogate markers. While none of these methods are currently used in clinical routine, academic and commercial developments are pointing towards potential clinical adoption in the near future. Here, we summarize the state of the art in AI-based image biomarkers for immunotherapy response based on radiology and histopathology images. We point out limitations, caveats, and pitfalls, including biases, generalizability, and explainability which are relevant for researchers and healthcare providers alike, and outline key clinical use cases of this new class of predictive biomarkers.

Self-supervised attention-based deep learning for pan-cancer mutation prediction from histopathology

“…Initial studies demonstrated this predictability in lung cancer 3 , breast cancer 4 , and colorectal cancer 5 . Subsequently, several "pan-cancer" studies showed that DL-based prediction of biomarkers is feasible across the whole spectrum of human cancer [6][7][8][9][10] . However, these studies were overwhelmingly performed in a single large cohort without externally validating the results on a large scale.…”

Predicting melanoma survival and metastasis with interpretable histopathological features and machine learning models

“…The current clinical gold standard is PD-L1 staining of tumor tissue, but it is poorly predictive of patients who will respond to immunotherapy, nor those who will have adverse events due to the immune checkpoint inhibition (39). AI approaches to predict these by analyzing histopathology and radiological images have been published, but most employ DL learning approaches and interpretability/explainability is still a key issue (40). As in our discussion of melanoma prognosis, we believe that deep-learning and more interpretable approaches are both needed for effective clinical translation.…”

An overview and a roadmap for artificial intelligence in hematology and oncology

“…Traditionally, in the 2010s, "Radiomics" machine-learning methods have used sets of expert-defined visual "features", coupled with simple ML models (Aerts et al 2014). More recently, end-to-end Deep Learning has been increasingly applied to such tasks (Ghaffari Laleh et al 2022).For example, AI has been used to prognosticate the course of colorectal cancer from digitized histopathology image data (Skrede et al 2020), or to predict the response to immunotherapy from radiological imaging data (Trebeschi et al 2019;Wu et al 2019;Ligero et al 2021). Also, AI has been used to predict the presence of genetic alterations from image data (Shmatko et al 2022;Kockwelp, et al 2022;Kather et al 2020), and is being discussed as a potential way to prescreen patients for targeted molecular testing (Shmatko et al 2022).…”