DarwinHealth
Digging Deeper than Genes

The current Achilles heel of cancer drug discovery is the inability to forge precise and predictive connections between specific patient subgroups and effective drug therapies. In simpler terms, although biopharmaceutical companies have developed very robust pipelines of medicines that are potentially approvable by the FDA, their strategies for identifying patients who might respond optimally to their drugs are far from accurate. For instance, even as immunotherapy is showing great promise, we still do not understand which patients will respond and which patients will fail treatment with immune checkpoint inhibitors such as nivolumab. These barriers to precision-based cancer treatment and drug discovery are made even more complex, due to the heterogeneous nature of cancer.

At the forefront of making precise, data- and experimentally driven connections among cancer drugs and patient cohorts is DarwinHealth, a New York City-based biotechnology company providing its collaborators and partners proven technologies—including those based on novel biomarkers—focused on hard-wiring predictions between investigational and FDA-approved drugs and relevant molecularly-identified patient sub-populations, across the pan-cancer spectrum.

“DarwinHealth’s mission—guided by technologies developed by Professor Andrea Califano and his research team at Columbia University—is to make precise alignments, so that clinical trials can be conducted much more efficiently and at much lower cost than might be possible with conventional technologies,” explains Gideon Bosker, co-founder and CEO of DarwinHealth. Fundamentally, these alignments, relationships, and predictions are based on a mechanistic rather than a statistical association-based stratum. It’s an approach that leverages the ability to “reverse engineer” cell regulatory networks to accurately identify Master Regulator (MR) proteins that, much like a malware virus in a computer, induce and perpetuate aberrant states and destructive behaviors that characterize cancer cells. Critically, Master Regulators are organized into small, ultra-stable modules called Tumor Checkpoints .Much like an air conditioner can keep the temperature of a room constant, despite changes in the weather, Tumor Checkpoints maintain the stability of the cancer cell state which, in too many cases, is resistant to attack by both intrinsic protective responses the body may mount in response as well as to drugs doctors may deploy for treatment.

Not surprisingly, then, DarwinHealth is leveraging Master Regulators—and the Tumor Checkpoints they comprise—as a new class of cancer targets, largely independent of the specific mutations in a particular tumor. In collaboration with Dr. Mariano Alvarez, DarwinHealth’s chief scientific officer and long-time collaborator at Columbia University, Dr. Califano developed many tools for the dissection and interrogation of the cell’s regulatory logic, including the VIPER algorithm, the company’s key technology that allows direct identification of Master Regulator proteins. Today, these tools are deployed at the frontlines of clinical medicine and patient care for cancer therapy, allowing DarwinHealth to operate with authority at the intersection of biotechnology, cancer research, clinical trials, and the biopharmaceutical industry. In its effort to provide solutions to what Bosker and Califano has characterized as “the precision deficit disorder in cancer medicine,” the company allows their partners to better understand and predict the mechanisms-of-action and therapeutic effects of their evolving drugs in very specific patient populations—predictions that focus on targets extending beyond superficial, canonical mutations that have driven precision oncology over the last two decades.

The Foundational Technology

Over the last ten years, the fundamental paradigm for cancer drug discovery has been heavily focused on two therapeutic trajectories—targeting genetic mutations and potentiating the immune system to fight cancer cells. However, recent scientific research published in high-quality journals, including from research organizations that have until now pursued cancer mutations as the ultimate therapeutic target, such as the Broad Institute, for instance, are now showing that the mutational and gene expression profiles of a tumor are the least and most predictive of drug sensitivity, respectively, among all ‘omics layers scientists and cancer specialists use to guide drug discovery and treatment.

By further deconvoluting gene expression profiles via the Master Regulator proteins that regulate them, DarwinHealth has been able to dramatically exceed even the predictive power of gene expression profiles.

“All of our cells have exactly the same genome, but they do completely different things,” explains Dr. Califano, co-founder and chief scientific advisor of DarwinHealth. “Therefore, I have always found it difficult to accept that one may completely characterize or predict the response of a tumor to a drug only based on the presence of specific mutations. Moreover, tumors will adapt to therapy in multiple ways, much as normal cells in our body adapt to changes in temperature or nutrients. As such, we observe many molecularly-distinct tumor cell states emerging from the exact same genome including within the same tumor.” Protein activity inferred from RNA sequencing, on the other hand, proves to be far more predictive of such cell states, including of their drug sensitivity.

Put simply, how a cell will ultimately respond to a drug will depend far less on its “master” genome than on how the information in its genome is interpreted by its protein-based cellular logic. Indeed, proteins are the “foot soldiers” on the front lines of key biological functions and, therefore, govern virtually every aspect of cellular function, including those that go awry in the process of cancer transformation.

Identical genomes can produce completely different protein landscapes, with equally different properties, such as drug sensitivity.

The challenge of deploying this methodology for cancer research, however, has been that protein activity is very difficult to measure. And this is where DarwinHealth’s proprietary technology has stepped in to provide an entirely new, validated approach to taxonomizing tumors based on their regulatory networks; and, more specifically, linked to tumor checkpoints.

Our platform turns cancer from a very complicated disease where every patient would need a different drug to essentially 112 molecular targets that you would want to hit

DarwinHealth has leveraged Califano’s findings into a transformational technology that accurately measures protein activity, even based on single cell analysis, through gene expression profiles. He illuminates the company’s platform technology with a simple analogy. “Planets outside our solar system are very difficult to detect because they do not emit their own light. Rather, planets are identified by studying the fluctuation in the position of the stars around which they orbit, which are predictive of both their mass and distance from the star. Essentially, then, you can infer the presence of something by looking at something else. That is precisely what we do to quantify protein activity. We study the gene expression or RNA sequencing of the targets regulated by a protein in a specific tumor to accurately predict its activity.” A recent study that validates this approach was published in the New England Journal of Medicine. DarwinHealth identified a novel, protein-based biomarker predictive of response to selinex or in patients with advanced multiple myeloma and dozens of similar studies have confirmed the accuracy and reproducibility of VIPER predictions.

Re-Taxonomizing the Cancer Spectrum

A recently published article in the journal Cell—with lead author Evan Paull, and Drs. Alvarez, and Califano as senior authors—highlights the merits of this technology. Titled “A modular master regulator landscape controls cancer transcriptional identity,” the article demonstrates that master regulator proteins are the “glue” or “biological malware” through which highly distinct mutational patterns induce aberrant activation pushing the cell toward a cancer state.

With the VIPER algorithm, which is licensed for commercial applications by Columbia University exclusively to DarwinHealth, the study group discovered 112 different tumor types characterized by discrete sets master regulators.

To put things in perspective, triple-negative breast cancer patients with very different mutational profiles show aberrant activation of virtually the same set of Master Regulators. The article postulates that, based on these findings, patients with different mutational landscapes would respond to the same drugs targeting their Master Regulators, rather than having to develop different drugs for every mutational pattern. “Our platform turns cancer from a very complicated disease where every patient would need a different drug to a very finite number of more universal molecular targets that you would want to develop drugs for,” adds Califano. Moreover, these targets have emerged as a consequence of mutational patterns, not to be misconstrued as mutations themselves.

More importantly, through this platform, VIPER analysis identifies the unique targets and drugs that are precisely aligned with each other. Bosker relates the representation of genes to the 88 keys of a piano and the ten fingers of the person playing the piano to the master regulators. “When you have ten fingers functioning as master regulators and 88 keys as the genes, that piano can play anything from Richard Wagner’s Das Rheingold to John Coltrane’s ‘A Love Supreme.’ In other words, you have the same genes, but you have given your fingers a different set of master regulators, and now they play a different tune. Our technology offers a data-driven roadmap for taxonomizing all cancers into 112 checkpoints and the associated master regulators that play a sad tune in the cancer cell,” says Bosker.

DarwinHealth’s diagnostic products—in particular, Darwin OncoTarget and Darwin OncoTreat—are tests designed to provide therapeutic guidance by analyzing the gene expression profiles (RNA sequencing) of bulk tumor samples or through single cell analysis. Darwin OncoTarget leverages the VIPER algorithm to identify druggable Master Regulators that are aberrantly over-activated. In complementary fashion, Darwin OncoTreat identifies drugs that are able to target the entire Tumor Checkpoint—i.e., the auto-regulated module of Master Regulator proteins responsible for integrating the genetic mutations and for driving the cancer cell state. Currently, these are the only two RNA sequencing-based molecular diagnostics available that can align cancer drugs with aberrations in protein activity level. The technology is being used for numerous clinical trials.

While the aforementioned tests are designed to align drugs with patients and their molecular fingerprints, the Darwin OncoMarker Program develops biomarkers predictive of therapeutic response based on protein activity.

Likewise, the company’s Compound-to-Clinic (C2C) Drug Discovery Program, also based on VIPER’s ability to characterize a drug’s mechanism-of-action at the protein level, serves as a platform that analyses up to thousands of drugs in a company’s pipeline and identifies their mechanistic impact on protein activity to optimize alignment between early and late-stage compounds and specific patient cohorts. DarwinHealth’s Novel Cancer Target Initiative (NCTI) is a platform that serves as a foundational technology and database to identify novel molecular targets, against which new drugs can be designed for testing in clinical trials.

Although the challenges are formidable, DarwinHealth has partnered with a number of the world’s most innovative biopharmaceutical companies to embark on a lofty mission to advance cancer therapy and press into service a novel strategy for drug discovery: to use industrial-strength computational horsepower, combined with comprehensive profiling of protein activity, to construct an actionable model of cancer cell behavior so that the foundational drivers—that is, the tumor checkpoints—of malignant misbehavior can be pinpointed and decommisioned with precision.

With the VIPER algorithm, which is licensed for commercial applications by Columbia University exclusively to DarwinHealth, the study group discovered 112 different tumor types characterized by discrete sets master regulators.

To put things in perspective, triple-negative breast cancer patients with very different mutational profiles show aberrant activation of virtually the same set of Master Regulators. The article postulates that, based on these findings, patients with different mutational landscapes would respond to the same drugs targeting their Master Regulators, rather than having to develop different drugs for every mutational pattern. “Our platform turns cancer from a very complicated disease where every patient would need a different drug to a very finite number of more universal molecular targets that you would want to develop drugs for,” adds Califano. Moreover, these targets have emerged as a consequence of mutational patterns, not to be misconstrued as mutations themselves.

More importantly, through this platform, VIPER analysis identifies the unique targets and drugs that are precisely aligned with each other. Bosker relates the representation of genes to the 88 keys of a piano and the ten fingers of the person playing the piano to the master regulators. “When you have ten fingers functioning as master regulators and 88 keys as the genes, that piano can play anything from Richard Wagner’s Das Rheingold to John Coltrane’s ‘A Love Supreme.’ In other words, you have the same genes, but you have given your fingers a different set of master regulators, and now they play a different tune. Our technology offers a data-driven roadmap for taxonomizing all cancers into 112 checkpoints and the associated master regulators that play a sad tune in the cancer cell,” says Bosker.

DarwinHealth’s diagnostic products—in particular, Darwin OncoTarget and Darwin OncoTreat—are tests designed to provide therapeutic guidance by analyzing the gene expression profiles (RNA sequencing) of bulk tumor samples or through single cell analysis. Darwin OncoTarget leverages the VIPER algorithm to identify druggable Master Regulators that are aberrantly over-activated. In complementary fashion, Darwin OncoTreat identifies drugs that are able to target the entire Tumor Checkpoint—i.e., the auto-regulated module of Master Regulator proteins responsible for integrating the genetic mutations and for driving the cancer cell state. Currently, these are the only two RNA sequencing-based molecular diagnostics available that can align cancer drugs with aberrations in protein activity level. The technology is being used for numerous clinical trials.

While the aforementioned tests are designed to align drugs with patients and their molecular fingerprints, the Darwin OncoMarker Program develops biomarkers predictive of therapeutic response based on protein activity. Likewise, the company’s Compound-to-Clinic (C2C) Drug Discovery Program, also based on VIPER’s ability to characterize a drug’s mechanism-of-action at the protein level, serves as a platform that analyses up to thousands of drugs in a company’s pipeline and identifies their mechanistic impact on protein activity to optimize alignment between early and late-stage compounds and specific patient cohorts. DarwinHealth’s Novel Cancer Target Initiative (NCTI) is a platform that serves as a foundational technology and database to identify novel molecular targets, against which new drugs can be designed for testing in clinical trials.
Although the challenges are formidable, DarwinHealth has partnered with a number of the world’s most innovative biopharmaceutical companies to embark on a lofty mission to advance cancer therapy and press into service a novel strategy for drug discovery: to use industrial-strength computational horsepower, combined with comprehensive profiling of protein activity, to construct an actionable model of cancer cell behavior so that the foundational drivers—that is, the tumor checkpoints—of malignant misbehavior can be pinpointed and decommisioned with precision.