A game-changing new test: Artificial intelligence predicts cancer recurrence years in advance.

 

Artificial intelligence
cancer

A game-changing new test: Artificial intelligence predicts cancer recurrence years in advance.

Hormone receptor-positive, HER2-negative breast cancer is the most common type of breast cancer, but it is also one of the most complex in terms of its long-term course. The risk of cancer recurrence does not end with the completion of treatment; it can persist for years. Data shows that more than half of distant recurrences occur five years or more later, even in patients who have not had lymph node involvement.
A recent study published in the journal "npj Breast Cancer" indicates that this pattern is responsible for about 60% of breast cancer deaths, reflecting the need for more accurate tools to estimate long-term risks.

What are the limitations of the current test?

Physicians currently rely on the Oncotype DX test (the 21-gene relapse score) to determine the likelihood of disease recurrence and to assess the need for adjuvant chemotherapy. However, this test is most effective at predicting relapse within the first five years, leaving a significant gap in assessing late-onset risk.
This gap prompted researchers to develop a new AI-based model called “IICM+”, which aims to improve the ability to predict distant relapse, especially after the first five years of diagnosis.

Remarkable results in long-term forecasting

The model's development was based on data from 2,808 patients in the TAILORx trial, before being tested on an independent group of 1,621 patients. To assess performance, the researchers used the C-index, which measures prediction accuracy, where a value of 1 represents perfect accuracy, while a value of 0.5 reflects performance close to random guessing.
The results showed a clear advantage for the new model, scoring 0.735 in predicting distant relapse, compared to 0.578 for the Oncotype DX test. According to experts, this difference is clinically significant because it indicates a better ability to differentiate between patients based on their risk level.

Why does prediction after 5 years matter?

The significance of this advance lies in the nature of the disease itself, as cancer cells can remain dormant for years before reactivating in other organs. Therefore, the ability to identify patients most at risk of relapse after 5, or even 10 or 15 years, allows physicians to revise treatment plans and long-term follow-up.
This type of prediction may also help in tailoring treatment more precisely, whether by extending the duration of hormone therapy or considering additional interventions for higher-risk groups.

Potential impact on treatment decisions

One of the most prominent potential applications of the new model is its ability to reclassify some patients who are considered low-risk according to Oncotype DX, especially within the category whose results range between 0 and 25, which is the category from which chemotherapy is often excluded.
IICM+ may reveal that some of these patients carry a higher risk than appears, which may open the door to more precise and personalized treatment discussions, without negating the role of clinical assessment or joint decision between the doctor and the patient.

A long road before the application

Despite these promising results, experts emphasize that the model still needs further validation before being adopted in everyday practice. Testing its performance in larger and more diverse groups is essential to ensure it offers genuine added value compared to existing tools.
It must also be proven that its use actually improves patient outcomes, in addition to ensuring its ease of application and reasonable cost within different health systems.
In this context, experts remind us that Oncotype DX itself required years of studies before it became an approved standard, and this is the path that the new model is expected to take before it finds its place in clinics.


Post a Comment

0 Comments