Disease and treatment

Genetic test may reveal which treatments people need to survive cancer

A newly developed genetic test may become an important tool in the future for assessing whether individuals with cancer have a good chance of surviving with more gentle treatment or whether they need more intensive treatment.

A new genetic test developed by researchers from Denmark and other countries can differentiate between high-risk and low-risk lung tumours and thereby determine people’s risk of dying from cancer.

The test, which is called ORACLE (outcome risk associated clonal lung expression), works by identifying many genetic markers that indicate good opportunities for surviving cancer. The test is intended as a diagnostic tool that will help doctors to distinguish between people who only need standard treatment and the people who require more intensive treatment for their tumours.

“This is the first time that gene expression can be used to accurately divide people with lung cancer into risk groups. I hope that this will help improve long-term survival by identifying the people who are currently not receiving intensive treatment but who may benefit from it,” explains Nicolai Juul Birkbak, Associate Professor, Department of Clinical Medicine, Aarhus University.

The study has been published in Nature Medicine.

Doctors do not know who will thrive after treatment

ORACLE was developed based on genetic studies of people with lung cancer but can theoretically be used with all types of cancer.

“We developed ORACLE based on people with lung cancer, but preliminary studies indicate that the analysis can also be used for other types of cancer. In any case, the approach used in developing ORACLE can be useful,” says Nicolai Juul Birkbak.

Doctors routinely divide people with lung cancer according to the stage of their tumours.

Stage 1 tumours are the smallest and can often be surgically removed without requiring chemotherapy, whereas stage 2 tumours require more intensive treatment, including surgery and possibly chemotherapy and radiotherapy, to eliminate them.

However, doctors face the challenge that one quarter of all stage 1 tumours recur after surgery. Conversely, chemotherapy is used to treat many people with stage 2 lung cancer even though this may not warrant a treatment that has so many side effects.

So far, doctors and researchers have failed to find a reliable way to distinguish between the people who are at high risk of relapse and those who often only require surgery without follow-up chemotherapy.

ORACLE solves this problem.

“The problem is that some people may have a small tumour that turns out to be aggressive. Because the tumour is small, doctors may think that surgery is sufficient. With ORACLE, we can identify people with a small aggressive tumour who are at high risk of relapse and can therefore also benefit from intensive treatment from the outset,” explains Nicolai Juul Birkbak.

Narrowing 20,000 genes down to 23 cancer-associated genes

ORACLE was developed based on comprehensive genetic analysis of 156 tumour samples from 48 lung cancer patients.

In the first part of the study, the researchers examined 20,000 genes and found 1,080 that were very suitable candidates for identifying people at high risk as they were stably expressed and unaffected by variation within the tumours.

Then the researchers used algorithms to find 23 genes most strongly correlated with the probability of surviving lung cancer.

The researchers used these 23 genes to create an ORACLE risk score, which can be used to assess individual risk based on the genetic analysis of cancer biopsies.

“Our goal is to ensure that we can identify the people with an aggressive stage 1 tumour who are initially treated with surgery so that we can provide the appropriate treatment immediately. A second goal is also to identify people with non-aggressive stage 2 tumours so we might avoid treating them with chemotherapy or radiation therapy from the outset,” explains Nicolai Juul Birkbak.

ORACLE identifies people at risk

The researchers tested ORACLE on genetic data from 904 people with lung cancer, and a high ORACLE risk score was associated with an increased risk of dying from the disease.

Then the researchers conducted the same study on 103 people with lung cancer, and a high risk score, independent of other risk factors, was associated with a three-fold higher risk of death within 5 years compared with a low risk score.

A study of 60 people with stage 1 lung cancer successfully used ORACLE to identify the people with the best chance of survival. Existing analytical methods could not identify any difference in risk between the people with cancer.

“We are preparing to validate ORACLE in a group of 500 to 1000 people with cancer. The next step will be to transform ORACLE into a tool that doctors can use clinically and immediately receive information they can incorporate into their planning of how to treat each person with lung cancer. I expect we will have something ready for clinical use within a couple of years,” says Nicolai Juul Birkbak.

A clonal expression biomarker associates with lung cancer mortality” has been published in Nature Medicine. In 2015, the Novo Nordisk Foundation awarded a grant to Jiri Bartek, a co-author, for the project Data-intensive Complex Systems Approach for Cancer Genomics Research: from Theory to Efficient Targeted Therapeutic Intervention.

Nicolai Juul Birkbak
Associate Professor
I have a background in cancer biology, biomarker development, translational cancer research and cancer evolution and heterogeneity based on research undertaken at Technical University of Denmark (PhD and postdoc), Dana-Farber Cancer Institute (postdoc), and University College London & the Francis Crick Institute (senior postdoc). We apply computational approaches to study cancer evolution from a translational perspective. Our mission is to understand cancer evolution at the molecular level, and to build tools and develop methods that use this information to improve patient treatment. An essential question in cancer research today and a focus of our research is understanding the key steps in carcinogenesis: how cells develop from a normal state to malignant cancer through benign, invasive and metastatic disease. Over recent years, exponential drop in Next Generation Sequencing costs coupled with significant investment in cancer research has led to the creation of large cancer cohorts with extensively characterized tumor samples. This effort has improved our understanding of cancer as a molecular disease, but a focus on driver events has so far not led to a breakthrough in patient therapy, and patient survival has not significantly benefited. Our lab utilizes cancer NGS data and computational tools to mine the developmental history on individual cancers, and to determine clonality of events. In this manner, we aim to describe the order of carcinogenic events as probabilities that depend on past driver acquisitions. This will allow us to construct evolutionary trajectories for individual cancer types, potentially informing about likely changes malignant cells may be biased towards when subjected to anti-cancer therapy. This opens the door to therapeutic approaches where treatment may be directed towards likely cancer clones not yet observed in a given sample.