or 3-5% of cancer patients, especially those with widespread metastatic tumors, the site of the cancer’s origin remains elusive. These are known as cancers of unknown primary (CUP).
This mystery presents a significant hurdle, as most cancer treatments are tailored to specific cancer types, which means that CUP patients often miss out on “precision” drugs that could be more effective and have fewer side effects.
The Solution
A collaboration between MIT and the Dana-Farber Cancer Institute has led to a machine-learning model that predicts a tumor’s origin by analyzing the sequence of roughly 400 genes.
This model was able to identify with high confidence the origin of at least 40% of CUP tumors in a dataset of around 900 patients.
Consequently, these findings might pave the way for more effective, targeted treatments for patients previously left in the dark about their cancer’s origin.
How it Works:
The team trained their machine-learning model on genetic data from almost 30,000 patients with 22 known cancer types.
When tested on about 7,000 tumors whose origins were known but had not been part of the training data, the model—dubbed OncoNPC—accurately predicted their origin about 80% of the time. For those predictions made with high confidence, accuracy surged to roughly 95%.
Using this model on a group of CUP tumors, high-confidence predictions about their origin were made for 40% of these tumors.
Clinical Implications:
The model’s predictions aligned well with patients’ survival times and the type of cancer’s typical prognosis. For instance, patients with predicted aggressive cancers had shorter survival times.
The study found that some CUP patients had received targeted treatments based on educated guesses. Those who received treatments aligning with the model’s predictions had better outcomes.
Remarkably, using this model, it was determined that an additional 15% of the patients could have benefited from available targeted treatments.
The Way Forward: While these findings are promising, researchers aim to incorporate more data types into their model, like pathology and radiology images, to refine and expand its predictive capabilities.
This holistic approach could not only determine tumor type but could also suggest optimal treatments, making a significant difference for patients with CUP.
If you care about cancer, please read studies about a new method to treat cancer effectively, and this low-dose, four-drug combo may block cancer spread.
For more information about cancer prevention, please see recent studies about nutrient in fish that can be a poison for cancer, and results showing this daily vitamin is critical to cancer prevention.
The study was published in Nature Medicine. Follow us on Twitter for more articles about this topic.
Copyright © 2023 Knowridge Science Report. All rights reserved.