Artificial intelligence (AI) can help determine which patients with certain head and neck cancers could benefit from scaling back the intensity of treatments such as chemotherapy and radiation therapy, according to new research. A team led by researchers from Case Western Reserve University has developed OP-TIL, a biomarker that characterizes the spatial interplay between tumor-infiltrating lymphocytes (TILS) and surrounding cells in histology images. With this study, the researchers sought to test whether OP-TIL can segregate Stage I HPV-associated oropharyngeal squamous cell carcinoma (OPSCC) patients into low-risk and high-risk groups and aid in patient selection for de-escalation clinical trials (J Natl Cancer Inst 2021; https://doi.org/10.1093/jnci/djab215).
"Prognosis plays an important role in patient management, so there is a lot of interest in the development and discovery of prognostic markers that can help in clinical decision making," noted James Lewis, Jr., MD, Professor of Pathology, Microbiology, and Immunology at Vanderbilt University Medical Center, and a co-author of the study.
As Lewis told Oncology Times, previous studies have shown a connection between a high density of tumor-infiltrating lymphocytes and better patient prognosis in oropharyngeal cancer. However, he noted, "we hypothesized that not only density, but also the spatial organization of TILS within the tumorous tissue, could provide clues of patient prognosis."
With this in mind, the team used a combination of computer vision and AI techniques to extract quantitative measures from the spatial organization of TILS and measured their association with patient outcome. The researchers explored the association between OP-TIL and patient outcome on whole slide hematoxylin and eosin images from 439 Stage I HPV-associated OPSCC patients across six institutional cohorts. One (n=94) was used to identify the most prognostic features and train a Cox regression model to predict risk of recurrence and death. Survival analysis was used to validate the algorithm as a biomarker of recurrence or death in the remaining five cohorts (n=345), the authors wrote in the research article.
OP-TIL separated Stage I HPV-associated OPSCC patients with 30 or less pack-year smoking history into low-risk and high-risk groups, even after adjusting for age, smoking status, and other factors. Overall, the research team concluded that OP-TIL can identify Stage I HPV-associated OPSCC patients that are likely to be poor candidates for treatment de-escalation. The authors found that, following validation on previously completed multi-institutional clinical trials, OP-TIL "has the potential to be a biomarker, beyond clinical stage and HPV status, that can be used clinically to optimize patient selection for de-escalation."
Patients with p16 positive (HPV-associated) OPSCC and individuals with certain other cancer types "seem to have worse outcomes when there are significant numbers of multinucleated tumor cells," noted study co-author German Corredor, PhD, a research associate at Case Western Reserve University's Case School of Engineering.
"Further, it is well-known that tumor-infiltrating lymphocytes are predictive of tumor behavior in many different tumor types and organs," Corredor added. "This has been shown by visual analysis in p16-positive oropharyngeal SCC patients, but we wanted to do so via computer-aided quantitative image analysis."
Previous studies have demonstrated the prognostic potential of TILs are not being used in clinical routine practice. "A main reason for that is that quantification of TILs is very time-consuming, highly subjective, and prone to error. The use of AI helped us to detect and quantify TILs automatically," Corredor said. "Additionally, AI helped us to identify what specific patterns of spatial organization of TILs were more correlated with patient prognosis. This analysis could hardly be done by a human, since it would require a very long time and because the AI is able to capture patterns that are not evident for the human eye."
Lead study author Anant Madabhushi, PhD, Professor and Director of the Center for Computational Imaging & Personalized Diagnostics at Case Western Reserve University, added that digital pathology "will be coming along as standard in clinical practice." As such, "pathologists will already have digitally scanned hematoxylin and eosin and immunostained slides that can easily be subjected to image analysis algorithms like ours," he said. "We imagine that oncologists can use this test to stratify patients into different treatment groups that are more tailored to their particular cancers."
Noting that this group of researchers is currently analyzing slides from completed, randomized, controlled clinical trials as a validation of TILS + multinucleation index (MuNI) in p16-positive oropharyngeal SCC patients, Madabhushi concluded that "these techniques have a lot of potential for evidence-based medicine.
"Clinicians may be able to make decisions based on their experience and objective metrics," he noted. "More specifically, they could find the patients with lower risk of death or disease recurrence who may benefit from therapy de-intensification, as well as patients with higher risk who may benefit from more strict treatment regimens."
Mark McGraw is a contributing writer.