After developing its strategy with a handful of initial patients, researchers at Columbia University Medical Center are planning to raise money to expand a program using whole-genome, deep exome, and RNA sequencing to build models of regulatory networks driving an individual patient's cancer that could inform targeted treatment strategies.
Andrea Califano, professor and chair of CUMC's Department of Systems Biology, who is directing the program to offer patients these "N-of-1" trials, told Clinical Sequencing News that the strategy differs from other projects using sequencing to guide cancer care in that it is able to look beyond known cancer driver mutations.
"If you go around most cancer centers, the way they work to potentially tailor a therapy for a patient is to discover actionable mutations in genes that have been previously reported as having some therapeutic implications. But there are relatively few of these, and though they may appear actionable, they are not always," Califano said. "For instance, HER2 is an actionable mutation, but in a gastric cancer it is not really actionable because [these tumors] do not respond well to trastuzumab."
Because most tumors types can be driven by multiple mutations, Califano said that the most successful treatment approaches are not aimed necessarily at single genes that are mutated, but rather at downstream, or related processes that are key elements of tumor growth and survival but not of normal cell activity.
"Unfortunately, these dependencies, you can't always find them from just the mutations," he said. "So what we do differently [with this program] is we build models that are genome-wide … of regulation in the cancer cells using a variety of algorithms.
These models are supported, he said, by an aggressive sequencing approach using Illumina technology, which is modified depending on the patient and the quality of their samples.
"Typically, we combine sort of a low-coverage, full-genome analysis of both tumor and normal samples with high-coverage exome profiling and high-coverage paired-end RNA sequencing," Califano said. "We try to discover both mutations and other events, including structural changes, copy number changes, and fusion events that may be unique to the patient."
Then the group takes this data and creates regulation models, which, hopefully, reveal potential drug targets within a cancer's genetic mechanisms.
"We take the profiles of the patients, and of course, we first test for actionable mutations. That's a low hanging fruit," Califano said. "But then we take the RNA and actually analyze what are the potential dependencies of the tumors that are induced by the presence of other alterations. We find what we call master regulators, a small number of genes whose activity is absolutely necessary to the cancer cell for survival, and we match them to drugs that can abrogate their activity.
According to Califano, the top 10 potential therapies are then tested in xenograft or mouse avatar populations, or potentially other model systems if a xenograft is not possible.
He said that the process might involve testing experimental compounds in some cases to better understand the pathways involved. But only FDA-approved or late-stage drug candidates would be recommended to a physician.
So far, the center has implemented the process with a small number of motivated independent patients, spurred by an initial, wealthy subject with a neuroendocrine tumor, who contracted the team to do a comprehensive analysis on his cancer.
Califano said the group has worked with three patients in this initial pilot— two of which have gone through the process of being sequenced and had mechanistic regulation models built for them, which have yielded potential therapies to try.
"You don’t want to give a new drug when the patient is responding well to standard treatments," Califano said of these two cases. "So right now we are on a hold and wait situation, but the moment the patient would progress, these approaches would be prioritized."
Moving forward from its three initial cases, Califano said the group is now seeking funding to support using the same approach on 30 or 40 additional patients.
He said the team would like to split that between three or four different cancers — pancreatic, neuroendocrine, and metastatic gastric cancers, and potentially also neuroblastoma — with 10 patients representing each type.
"The first patients were done on an ad hoc basis, to learn what would need to be put in place," Califano said. "Now that we've been doing this for a year and a half, and worked out the details of the pipeline, we are ready to expand."