Will your cancer treatment be determined by an algorithm?

In the future an algorithm could select the best cancer treatment for your specific cancer by simply analyzing a tumor sample.

Algorithms can find what you're looking for on Google or suggest your next movie on Netflix. Soon they will also find the best type of cancer treatment for an individual patient.

The company Genomic Expression has discovered a method for sequencing samples from cancerous tumors, where they analyze RNA inside the cancer cells and use algorithms to find the best treatment regimen.

"In contrast to DNA, which is the hard disc of the cell, RNA is the software – the program the cell is running" Morten L. Pedersen, CTO Genomic Expression

DNA makes life programable, but what kind of program the cells is running is found in the RNA.

Data from sequencing RNA is run through a newly developed algorithm that can determine the type of cancer it is and find the best treatment by searching large clinical databases.

Cancer manipulates the RNA expression levels; by being able to analyze how RNA changes in the cancer cell, we can treat each individual cancer patient extremely accurately with the right intervention.

The inconvenient truth is that only one out of four cancer treatments prolong life. Genomic Expression aims to change this by assisting doctors as they select the best drug for their patient, and pharmaceutical companies as they select the best patients for their drugs. 

“We are not a new type of medicine – we find the best drug for the individual patient instead of trial and error or standard treatment,” says Morten Middelfart, Chief Information Officer at Genomic Expression.

Right now Genomic Expression's algorithm has been tested on a number of cancer types, and the results are truly amazing, explains Jesper Zeuthen, CMO at Genomic Expression.

“I have worked on curing cancer my whole life by developing new drugs. We will not be able to cure cancer with a magic bullet drug, but we have a chance if we genetically match the tumor up against drugs that actually work on the patient's specific genetic alterations, and focus on immune therapy.” Jesper Zeuthen, Chief Medical Officer, Genomic Expression

Jesper describes a triple negative breast cancer patient who went from having a dire prognosis – most of these patients die within two years of diagnosis – to finding a handful of drugs that were already approved and 30 clinical trials that matched her tumor.

OneRNA™ works by first analyzing a small tumor sample taken at the hospital, and then sending it to Genomic Expression to be sequenced and run through the algorithm, which provides the doctor with an actionable report of possible treatment options.

“In contrast to DNA panels, RNA can identify markers for response to the new immune therapies, furthermore we are not using panels so we can deliver new actionable findings real time to the doctors, while it typically have taken 10-30 years to implement validated markers in the clinic historically” concludes Gitte Pedersen, CEO Genomic Expression.

Precise and less expensive cancer treatment

The algorithm use the identity and quantity of the expressed RNA to calculate a score and list drugs that are approved or in clinical development. The changes in the RNA that are caused by cancer provide this valuable information.

"Doctors have done well so far – it is not criticism of them – but the brain can only handle a certain amount of data, while an algorithm can handle more and thereby figure out which treatment is most promising. It's as if we tried to find specific documents on the Internet: without a search engine, it would take much longer," says Morten Middelfart.

The problem with choosing the right cancer treatment today is that two people can both have lung cancer, but it is not certain that the same treatment works on both patients.

"There are lots of examples of clinical trials in which doctors see this picture that some cancer patients are cured with a treatment while others with the same type of cancer does not respond. I believe that if we create a method that shows the optimal treatment, ultimately the doctors will need to combine the genomic information with the clinical information to select the treatment. For example, some patients may not be physically strong enough for a specific treatment due to the side effects," Middelfart says, stressing that finding the best method can result in a cheaper treatment.

"If we can get the cancer drugs already on the market to work better, then we will be in a situation where treatment per patient are much cheaper," Tanya Kanigan, COO, Genomic Expression says.

Algorithms for the greater good

The big question is whether we can trust the algorithms. Morten Middelfart says that we can. Many systems that we trust are already controlled by algorithms. "We have, for example, developed control systems such as the autopilot for an Airbus aircraft. There must of course be a natural skepticism, but when the system deserves to be trusted, then we need accelerated development. It makes sense to talk about credibility, because most patients trust their doctor, however some patients may feel differently about it if the doctor relies on an algorithm," he points out.

He denies, however, that there is a risk in using algorithms. The risk is not that the algorithm will find the wrong cancer treatment, but rather that the algorithm will get so good that it cannot find a treatment to suit all patients.

"A risk I fully agree with, is that one can imagine that big data will make the algorithms too accurate. You can find details that are so specific that they apply only to one person, while it might be smarter to find people who have something in common, so they can get the same treatment. It will not be possible to develop a billion dollar drug for just that person," he says.

According to Morten Middelfart, there is no doubt that we will use algorithms for optimizing cancer treatment, in part because using such algorithms can save lives and make healthcare more effective.

"I think it is the future and it is right around the corner. The results that we have seen now in some of the cancers that are not driven by simple mutations are amazing. It has always been my passion to work with big data, where algorithms remove significant steps that otherwise would have been hard to solve. Some of our applications would be impossible to resolve without an algorithm," he says.

“The bottom line is that when we use algorithms for optimizing and individualizing cancer treatment, we are making a huge impact in the lives of these patients. Like most people, my family has been touched by cancer. Our dream is to make cancer a treatable disease and not a death sentence.”

This a sentiment that is shared by the founders of Genomic Expression, Gitte and Morten Pedersen, who lost both their parents to cancer as well as Jesper Zeuthen, CMO who arguably spend his whole life on finding a cure.

You can view Morten Middelfart's presentation "Big Data Solutions for Tumor Sequencing” at The Second Symposium on Big Data in Medicine at Oxford University, UK here: