Improving breast cancer treatment prognoses with liquid biopsy

Cancer cell DNA can be isolated from a breast cancer patient’s blood sample, and the degree of fragmentation of the DNA can be assessed to determine whether the patient has a poor or good prognosis for treatment.

Breast cancer is the most common cancer in women, with over two million women being diagnosed with it in 2020. Fortunately, the prognosis for breast cancer has improved, because it can be detected in the early stages. In addition, treatment methods have advanced. One of these is liquid biopsy, which is becoming an increasingly important diagnostic technique for cancers.

Professor Arto Mannermaa’s research group, specialising in personalised medicine and biobanking, has been studying liquid biopsy since 2015. Liquid biopsy is based on the fact that cells in the body release DNA into the bloodstream and bodily fluids, a form known as cell-free DNA (cfDNA). In other words, DNA from cancer cells is released into the patients’ bloodstream, containing mutations specific to cancer. The cfDNA is sequenced, revealing the genetic alterations present in the tumour.

“We have investigated the concentration, fragmentation level and mutations of cell-free DNA that are associated with the prognosis of breast cancer patients. Similar connections can be found in several other forms of cancer,” says researcher Jouni Kujala. Kujala works in Mannermaa’s research group at the University of Eastern Finland.

“This is a research topic that involves many computational aspects, especially processing sequencing data,” Kujala says. In the future, Kujala plans to focus on cell-free microRNA.

“It is an entirely different type of nucleic acid that can be isolated from the blood samples of cancer patients. Cell-free microRNA regulates gene function, but its predictive value is not yet fully understood.”

Cell-free DNA

Mannermaa’s research group has studied the fragmentation of cell-free DNA. Based on this research, the prognosis of breast cancer patients can now be assessed. The result is important, because the method helps identify breast cancer patients with a poor prognosis earlier and more accurately than before. Early detection is one of the key ways to reduce breast cancer mortality.

The researchers have analysed the causal association of cell-free DNA integrity with breast cancer treatment prognoses.

“When cancer cells release cell-free DNA into the bloodstream, it gradually begins to fragment into smaller pieces until it breaks down completely.”

Integrity reflects the degree of DNA fragmentation – that is, how much DNA has fragmented in the blood.

“The more intact – less fragmented – cell-free DNA is, the more it is associated with a poor prognosis for breast cancer.”

The results of the study by Mannermaa’s research group were made possible by extensive patient data, gathered by the Kuopio Breast Cancer Project (KBCP).

“The KBCP includes over 500 breast cancer patients, and comprehensive data has been collected from them. We know their lifestyles and the cancer treatments they have received. We have follow-up data for up to 25 years, which is exceptionally long even on an international scale.”

In this study, the sample included breast cancer patients who had not yet started any form of cancer treatment.

“The sample consisted of early-stage breast cancer patients who were initially considered to have a good prognosis.”

There was a clear rationale for selecting such a sample, as breast cancer recurs in up to a third of patients and is the most common cancer-related cause of death in women. The goal of Mannermaa’s group is that in the future, patients with aggressive breast cancer could be identified even earlier through integrity measurement, and directed to enhanced treatment and monitoring if necessary.

Resources for teaching AI

The measurement of DNA integrity is a simple one. The isolated sample is placed in a measuring device that determines the relative proportion of DNA fragments in the sample. Then it is possible to calculate the degree of integrity of the cell-free DNA in the sample.

According to Kujala, even such a simple value describing the quality of a DNA sample can be useful in predicting cancer outcomes. In the future, this method could be put to further use in training artificial intelligence.

“When measuring the concentration and integrity of cell-free DNA, these are purely quality indicators. They are not currently used in assessing the patient’s prognosis. The actual diagnostic aspect has largely focused on mutations and other features of DNA. Machine learning could enable more effective use of this data – it is collected from all samples that are examined, but it is hardly utilised at all.”

Mannermaa’s team is developing algorithms that learn from genomic data and clinical information to identify and predict risk factors for breast cancer. Genomic and clinical data are combined to form an AI model that not only helps to determine the risk of illness, but also in drawing up individual treatment plans.

The amount of data in Mannermaa’s team’s study is so huge that it requires the supercomputing capacity of the Finnish ELIXIR node of the CSC – IT Center for Science.

“We have access to CSC’s resources specifically for machine learning purposes. So far, we have developed cancer risk analytics, but the same models are utilised in the further work with these liquid biopsy results. This data has not been further processed yet,” says Mannermaa.

Ari Turunen

23.1.2024

Read article in PDF