Lately the internet has been abuzz with talk of artificial intelligence (AI). While AI is certainly not new, the recent development of new applications for the technology – most notably ChatGPT – has made headlines, sparked debate, and paved the way for rapid innovation across a variety of industries and uses. Healthcare, and specifically the cancer arena, are no exception.
Over the past year, AI has served as the catalyst for the creation of new vaccines and therapeutics, transformed diagnostics, and aided in prioritizing and triaging cases among numerous other advancements.
Though the competition is stiff for the title of “revolutionary healthcare application of AI,” there is one area in which machine learning is uniquely positioned to address informational gaps and expand treatment options – cancer of unknown primary (CUP).
Researchers at the Massachusetts Institute of Technology (MIT) and Dana-Farber Cancer Institute have put an AI-powered model to work on solving some of cancer’s biggest riddles, and the results show significant promise.
What is AI?
First things first – before delving into the role of AI in enigmatic cancers, it is important to first understand what exactly AI is.
John McCarthy, renowned computer scientist and one of the founders of the discipline of artificial intelligence, described AI as, “the science and engineering of making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable.”
IBM offers an even simpler definition – “a field, which combines computer science and robust datasets, to enable problem-solving” (remember, IBM created the supercomputer that beat Jeopardy! champions…so they know a thing or two about AI).
What is cancer of unknown primary?
Cancer of unknown primary (also referred to as “cancer of unknown origin” or “enigmatic cancer”) is cancer that has spread from another part of the body, but the point of origin/where the cancer began is unknown.
Cancer of unknown primary makes up 3-5% of all US cancer diagnoses.
The challenge of CUPs
Naturally, CUPs present a number of challenges in the treatment of cancer. Perhaps the most significant obstacle CUPs create in the care process comes in the form of therapeutics, as lacking a tissue of origin limits treatment options and access to clinical trials.
Patients who are precluded from key, lifesaving treatments and clinical trials as a result of a CUP diagnosis often experience shorter life expectancies.
Enter OncoNPC
In an effort to provide better targeted therapies and improve overall health outcomes for CUP patients, researchers at MIT and the Dana-Farber Cancer Institute developed a machine-learning classifier known as “OncoNPC.”
The model analyzes a sequence of 400 genes to predict where a tumor originated in the body.
The AI model was trained on targeted next-generation sequencing (NGS) data from 36,445 tumors across 22 cancer types from three different institutions.
OncoNPC – an acronym for “Oncology NGS-based primary cancer-type classifier” – was applied to 971 CUP tumors collected at the Dana-Farber Cancer Institute and predicted primary cancer types with high confidence in 41.2% of the tumors.
Among their findings, which appeared in the August 7 edition of Nature Medicine, researchers also noted that the tool identified CUP subgroups with, “significantly higher polygenic germline risk for the predicted cancer types and with significantly different survival outcomes.”
Perhaps the most compelling measure of the model’s success is its ability to better inform treatment options. Researchers found that OncoNPC, “enabled a 2.2-fold increase in patients with CUP who could have received genomically guided therapies.”
The takeaway
By all indications, OncoNPC is a powerful tool; providing doctors with the ability to recommend more focused therapies and patients with more information and decision-making power in their treatment.
Researchers concluded that the model, “provides evidence of distinct CUP subgroups and offers the potential for clinical decision support for managing patients with CUP.”
Amid all of the AI noise, OncoNPC stands out as an innovative approach to care – an application of cutting-edge technology with the ability to aide both patient and provider.
