Deciphering Cancer Is Messy and Complex. We’re Here for It.

How systems biology is embracing the complexity of cancer to predict and improve treatment.

Andrea Califano
June 14, 2022

"Disruptors" is a series sharing new and innovative ideas and viewpoints from Columbia Cancer researchers and clinicians that challenge conventional thinking about cancer care, research, and beyond.

Precision medicine has been a buzzword across the medical field for over a decade. But what does it really mean for cancer care and how is it influencing new therapies for patients? Initially, precision cancer medicine focused on targeting specific mutated genes. We thought that understanding the genetic mutations of a tumor would help us develop targeted drugs that would solve the problem, one broken gene at a time.

What we found instead is that when you build an inventory of all the broken parts in cancer, the number of mutational patterns that could give rise to cancer is larger than the number of atoms in the universe. Each mutation may even determine a different response to treatment, varying from individual to individual, and so targeting genetic mutations appears more and more as a staggeringly difficult task. Aside from the challenge created by such a vast number of possibilities, we’ve also discovered that genetics doesn’t tell the full story of a person’s cancer because cancer cells with the same exact mutations can also have different drug sensitivity.

Humans have about 20,000 genes working together in ways that are different from cell to cell and from individual to individual. The enormous amount of data that we have been able to collect on cancer has helped us build computational models that, rather than trying to explain things one gene at a time, explain how all these genes work together in a system.

Move Over DNA: RNA’s Role in Cancer

Approaching cancer as an ex-physicist, I wanted to open the cancer “box”, look inside, and understand precisely how it works; not just one gene at a time, but based on all the gene products working together. With that, my passion has been to create the “assembly manual” of the cancer cell—a map of the complex network of molecular interactions that determine its behavior and response to treatment. This way, much like looking at the manual of a complex piece of machinery, when something is broken, we know exactly where to find the root cause and how to fix it.

The basis for this manual starts at the interface between two key molecules in the cell, DNA and RNA.

DNA, which I call the “book of what could be”, contains information for all the possible things that a cell could ever be or do. In contrast, RNA represents “the book of what is” because it provides faithful copies only of the genes that a specific cell needs at a given point in time. RNA is directly translated into proteins, which are the molecules that actually “do things” in the cell, by carrying out critical functions. For instance, a liver cell and a brain cell in the same individual will have the same DNA but different RNA and protein expression patterns that allow them to perform different functions.

The big paradox in cancer is that not only the same DNA mutations can produce very different RNA landscapes, but, equally important, the opposite is also true. That is, different DNA mutations can produce virtually indistinguishable RNA landscapes with identical responses to certain drugs. And the latter may hold the key to successful cancer treatment.

Yes! The problem is just as complex as it sounds.

Take for example, the BRAF gene. While this is the most frequently mutated gene in melanoma we know now that this genetic mutation can drive several other cancers, including a small subset of colon cancer, for instance. Drugs that effectively target BRAF mutant melanoma, albeit for a short amount of time, have virtually no effect in colon cancer. Same genetic mutation, yet different drug response.

Because RNA shows us specifically what is going on in a cancer cell, at a specific point in time, we have developed algorithms that accurately predict the proteins that drive the cancer cell, based only on RNA measurements. While this is far more complex than looking for DNA mutations, it also promises to be much more effective when it comes to dismantling cancer, because the protein activity state of a cancer cell provides the most informative data in terms of predicting whether a drug will kill it or not.

As Cancer Evolves So Does Our Approach to Problem Solving

In my lab, we assemble computational networks of molecular interactions between proteins and genes then analyze them to identify and target a handful of “master regulator” proteins, essentially the “pillars” that determine the cancer cell behavior and represent its most critical vulnerabilities.

We have shown that these master regulator proteins work together to power the cancer cell, akin to a building standing up on a small number of load-bearing pillars. You target one or more of these pillars and the entire building collapses. We have developed methodologies to identify precisely which proteins in each cell are the load-bearing pillars of the cancer cell state and which drugs can best target their activities.

Precision cancer medicine has so far not fully delivered on its promise. Patients with advanced metastatic breast cancer have no more specific options than they did 10 years ago. In the next wave of precision cancer medicine, treating cancer will depend on our ability to be even more predictive with our approaches, staying one step ahead of cancer’s terrifying ability to mutate or adapt to its environment, including adapting to therapies.

Beating Cancer at Its Game to Mutate and Thwart Treatment

The only way to stay one step ahead is to surrender to the complexity of cancer and understand that each cancer cell is different, even the same cancer in the same patient. Our algorithms identify groups or types of cancer cells that will respond to certain treatments so that we will know which cells are killed and which are spared when we treat with a specific drug. In that way we can identify the combination and sequence of drugs that will kill each of those specific population of cells within an individual patient’s tumor, removing the guess work to identify drugs or drug combinations that are truly personalized to the patient. This moves us past treating cancers one genetic mutation at a time but rather as a complex networks of broken genes and proteins that determine their drug response.

Cancer is complicated and cannot be simplified beyond a certain level in our research, just as we cannot simplify cancer for our patients. We need to embrace its complexity, matching its intricacy and sophistication with approaches that are equally complex and sophisticated.

Andrea Califano is the founding chair of the Department of Systems Biology, the Clyde and Helen Wu Professor of Chemical Biology (in Systems Biology), and professor of medicine, biomedical informatics and of biochemistry, and molecular biophysics at Columbia University Vagelos College of Physicians and Surgeons. He directs the Columbia JP Sulzberger Genome Center and co-leads the Precision Oncology and Systems Biology program at the Herbert Irving Comprehensive Cancer Center.