How they use artificial intelligence against pancreatic cancer

Experts are working on technologies that can help an early diagnosis of the disease, which in Argentina is detected in 5,000 patients per year

Pancreatic cancer has a high mortality rate – which is around 95% of patients. This is because, in general, when it is detected, it is usually very advanced and even in a state of metastasis. Then, the challenge is to speed up its diagnosis and it is at this point where artificial intelligence and machine learning play a key role.

Although in recent years the knowledge of this silent tumor, with “camouflaged symptoms” in its early stages has increased, one of the main objectives continues to be to identify populations at high risk of developing pancreatic cancer to incorporate them into programs of screening and early detection, thereby increasing patient survival.

To advance in this direction, epidemiologists, geneticists, clinicians, surgeons, pathologists, molecular biologists, mathematicians and experts in artificial intelligence and machine learning from scientific centers and hospitals in several countries joined forces. Led by Núria Malats, from the Spanish National Center for Oncological Research (CNIO), they have been scrutinizing this tumor for years, the third with the highest mortality in Spain, managing to generate a large amount of data, according to what was reported by the EFE agency.

Thus, to known risk factors, such as obesity, smoking, alcohol consumption, diabetes or chronic pancreatitis, genetic biomarkers – rare and more common variants -, immune or microbiome, are being added to the map of this cancer.

In addition, Malats’ team published in the journal Gut that type 3c diabetes, which accounts for between 5 and 10% of all diabetes cases in Western countries, is an early manifestation of pancreatic cancer, which is celebrated today its world day.

All this data has begun to be tested thanks to innovative strategies of statistical and bioinformatic analysis with data and biological samples from 2,500 patients with pancreatic cancer and 1,500 controls within the European PanGenEU project. The goal now is to go further and translate this knowledge into algorithms that help in early diagnosis.

Artificial intelligence and algorithms can help early detection of pancreatic cancer

What’s Next in Pancreatic Cancer Detection and Treatment

The challenge is to combine clinical, genomic and microenvironmental factors -the genome of microorganisms present in the human body- because there is no single cause powerful enough to identify those individuals highly susceptible to suffering from this tumor -as is the case with smoking and lung cancer.

“My dream is that this artificial intelligence and machine learning tool that we are working on can be accessible from anyone’s mobile devices, with the aim of making a first assessment,” says Malats.

In turn, the expert adds that with data, for example, from a person with diabetes, obesity and a smoker, a first algorithm allows estimating the baseline risk; if it is high, the person would go to their family doctor and he would assess whether genetic biomarker analysis is necessary – almost 100 genetic variants linked to the risk of suffering from this type of tumor have been identified.

But also there will be a second algorithm, which will integrate risk factors and genetic and immune biomarkers designed for clinical practice: “Bringing these algorithms to primary care would be very good to speed up diagnosis”, sums up the scientist.

The algorithms are advanced, especially the first. To develop the second, explains Malats, the biomarkers must be integrated: “As we are already quite clear about what these are, I estimate that in about four years we could have the two algorithms ready for validation.” And the validation phase can last as many years. For this, says Malats, studies are needed that include large cohorts of people with some risk factor, but also smaller ones.

What we have to do is to increase the proportion of patients who could benefit from a new generation treatment and for this they must be diagnosed in much earlier stages, and this is when they have not yet presented symptoms. ”

Therefore, “what must be achieved is to define this high-risk population to incorporate it into screening programs,” summarizes Malats, who recalls that none of this would be possible without the altruism of the patients and without all that basic information, correctly collected and ordered daily by hospital staff and other centers.

What are the figures for pancreatic cancer in Argentina

In Argentina they are detected near 5,000 cases per year (4878 in 2018, according to statistics from the National Cancer Institute). And, although it is the fourth leading cause of death from malignant neoplasia after lung, colorectal and breast cancer, almost all cases are fatal (4,683, according to the INC), so prevention and early screening in people of risk takes on a notorious relevance.

In Argentina around 5,000 cases of pancreatic cancer are detected per year

In Argentina around 5,000 cases of pancreatic cancer are detected per year

Every November 15, World Pancreatic Cancer Day is commemorated, a date used to spread the reality of this disease. As in several types of cancer, if it is detected in early stages, there is a possibility of cure, although very little is disseminated about its risk factors, which means that the investigation does not have a wide adherence.

World statistics are conclusive: detected in early stages, the possibility of tumor removal and cure is around 35%. However, only 10% of patients are diagnosed at this early stage. Once the cancer has spread, survival drops to 12%, while in more advanced stages (when the tumor has metastasized to other organs), only 3% have any chance.