Scientists led by a team at the London Cancer Institute (ICR) have used artificial intelligence (AI) and machine learning (ML) to identify five new subtypes of breast cancer that can help clinicians provide the most effective treatments, including immunotherapy. For individual patients, and potentially guide the development of new anticancer drugs. The tool can find patterns in the genetic, molecular and cellular composition of primary luminal type a breast tumors, and analyze them together with the survival data of patients. The team has previously used the same method to identify subtypes of colorectal cancer (CRC).

“For many years, doctors have used the current classification of breast cancer as a treatment guideline, but this approach is very crude, and it seems that patients with the same type of disease usually respond very differently to drugs,” commented Dr. Maggie Cheng, the team leader. A researcher from the genomic analysis clinical trial group of ICR in London is a co-author of the researcher’s book npj breast cancer. “Our research uses AI algorithms to find patterns in breast cancer that have not been found in human analysis so far, and to find other types of diseases that respond in a very specific way.”

The ICR researchers, led by Dr. anguraj sadanandam, head of the systems and precision cancer medicine team, and colleagues at Royal Marsden Hospital, reported their findings in a paper entitled “heterogeneous cell genetic characteristics reveal heterogeneity and different treatment responses in luminal type a breast cancer.”

Most breast cancers occur in the inner lining of the breast, with estrogen and / or progesterone receptors positive and HER2 negative. These tumors, known as luminal type a breast cancer, tend to have the best cure rate, but patients respond differently to standard treatments (such as tamoxifen) and newer immunotherapies, which may be due to recurrence. “Even this relatively well-defined subtype of breast cancer has heterogeneity in hormone receptor expression, therapeutic response, and genetic variability, which needs further understanding,” the authors write

Intelligent recognition technology may contribute to personalized medical diagnosis

Even among different type a tumors, the factors affecting tumor heterogeneity are complex, including genetic changes, tumor microenvironment and the interaction between different cell types. Moreover, although immune related genes are often expressed in different subtypes of breast cancer, including luminal A-type tumors, “…” Unlike colorectal cancer and pancreatic cancer, it has been reported that there is no exclusive immuno enriched breast cancer subtype

In order to study the differences between different patients with luminal type a breast tumors, the research team turned to the ml / AI tool previously used, and divided CRC into five different subtypes, namely inflammatory, intestinal epithelial, globular, stem like, and transfer amplification (TA). “The researchers said We tried to use our CRC xenogeneic cell characteristics as a substitute to re characterize breast cancer subtypes, especially luminal type a breast cancer, and to understand their phenotypes based on their differentiation, stem cells, fibroblasts and immune characteristics

The researchers were also surprised to find that the stem like subtype of luminal type a breast tumor showed a good recurrence free survival (RFs), “indicating that the presence of stem cells and fibroblasts (rich in stem like subtypes) does not mean that the results also show that TA tumor is characterized by the change of chromosome 8, and the survival rate of patients with this kind of tumor after tamoxifen treatment is worse than other groups. These patients also tended to relapse earlier – 42 months on average, compared with 83 months in patients with high levels of stem cell tumor types. Patients with late recurrence or delayed treatment may benefit from the results of new studies.

The researchers also studied tumors in a group of triple negative breast cancer (TNBC) patients. This type of breast cancer does not respond to standard hormone therapy. Analysis of samples from this group of patients suggests that their tumors may also respond to immunotherapy. “Although no immunotherapy has yet been approved, clinical testing of immune checkpoint inhibitors has been carried out in breast cancer patients, but we have linked some of the underlying breast cancer with CRC inflammatory subtypes, suggesting a way to identify patients who may respond to immunotherapy.”

The authors said that their results provide new insights for luminal-a subtypes, which can help personalized diagnosis and treatment of patients with different types of breast cancer. “Our new research shows that AI can recognize patterns of breast cancer that are beyond the ability of the human eye and provide a new treatment for those who have stopped responding to standard hormonal therapy,” sadanandame said “We’re at the cusp of the healthcare revolution because we’ve really grasped the possibilities that AI and machine learning can open up AI has the ability to be used more widely, and we think we will be able to apply this technology to all cancers, and even open up new treatment possibilities for cancers that currently have no successful choice. “

In addition, the discovery may help to guide the discovery of new drugs even for patients who may tend to relapse after many years. “The AI used in our study can also be used to find new drugs for those most at risk of late recurrence (more than five years), which is common in estrogen related breast cancer and can cause great anxiety in patients,” Cheng concluded. ICR took the lead in using AI to understand the complexity and evolution of cancer, and raised the final 15 million pounds in a 75 million pound investment in the new cancer drug discovery centre, which will be described as “the world’s first” anti evolutionary “treatment program.”


Leave a Reply

Your email address will not be published. Required fields are marked *