Cancers are generally caused by changes in one’s genetic code, which scientists call mutations. Some cancers with common mutations have very effective treatments. Since the mid-20th century, the standard treatment option for most cancers has been chemotherapy. However, oncologists are now moving away from this one-size-fits-all approach to cancer treatment.
Researchers have recently recognized that every person’s tumor has unique mutations, even among the same type of cancer. Therefore, they require a customized treatment regimen to optimize effectiveness. Scientists have been particularly slow in developing drugs that treat rare cancers, where less is known about the genetic framework supporting tumor growth. Now, researchers aim to understand the genetics of each patient’s tumor so they can design personalized treatments directly targeting it.
Genetic mutations code for specific proteins that tumors use to grow, known as oncoproteins. If tumor growth is like the speed of a car, then an activated oncoprotein is like a gas pedal stuck to the floor. It could be stuck due to a manufacturing blueprint error, similar to the mutation. Previous methods of inhibiting tumor growth only studied the blueprint, not the real pedals. This means they’ve missed many broken accelerators, or in this case, the oncoproteins.
Researchers have successfully linked oncoproteins to specific gene mutations. However, not all mutations create dangerous oncoproteins. To create personalized treatments targeting specific tumors, researchers need to understand which mutations are of concern for each tumor. Two main problems arise when researchers measure oncoprotein abundance and activity. The first is knowing whether hundreds of individual oncoproteins are activated. The other is knowing which of those activated oncoproteins cause dangerous tumor growth.
One way researchers find which oncoproteins lead to tumor growth is by examining all the other genes each oncoprotein controls, known as its regulon. All the messenger RNA strands in an organism are known as the transcriptome. By analyzing the genes in the entire transcriptome of a tumor, researchers can examine each oncoprotein’s regulon and predict the activity of oncoproteins in a single tumor sample.
Researchers from Columbia University used this method to develop a computational technique that utilizes gene expression information, known as Virtual Inference of Protein activity Enriched by Regulon, or VIPER. They used VIPER to determine which oncoproteins cause tumor growth from a single sample. This technique is useful in the field of personalized medicine, as it can aid in the development of treatments targeting active, dangerous oncoproteins.
The researchers needed to test whether VIPER could accurately predict which drugs would work for specific groups of cells grown in plastic dishes under certain conditions, known as a cell line. They used VIPER to predict whether cells would live or die with a given treatment. Then they compared the VIPER predictions with predictions based only on which mutations were present.
They focused on a well-known gene associated with lung cancer, known as EGFR. The scientists used VIPER and mutation-state predictions to examine 79 lung cancer cell lines. In comparison, they also predicted the efficacy of drug treatment using the traditional method, by sequencing the DNA in the sample and assessing the mutations present. Then they administered a drug to inhibit EGFR to every cell line and observed which cells survived and which died. Finally, they compared the survival of each cell line to the predictions made by VIPER and the predictions based on the mutations present.
The researchers found that the cell lines VIPER predicted would have high EGFR activity responded to the drugs and lived, even without gene mutations. This method improved on the mutation-based predictions, which didn’t detect that those cell lines would be treatable. In addition, cells where VIPER predicted low EGFR activity didn’t respond to the drugs, even when the gene was mutated. This suggested that VIPER predicted drug response better than mutation data alone.
The team concluded that VIPER could help move oncology a step closer to personalized cancer treatment. Future applications could enable doctors to biopsy a patient’s tumor for any cancer as long as the transcriptome is sequencable. From this, oncologists could get the gene expression data, run it through VIPER, and have a clear idea of which oncoproteins are supporting the tumor, which could help them choose effective drugs for treatment. Since the study’s publication, researchers have continued to improve VIPER’s computational methods, and it has been used on small scales at facilities in New York City.
