NEW YORK – Precision oncology company DarwinHealth has published a drug discovery and clinical trial matching protocol, which it has already begun fielding in several partner-led cancer trials.
The method, called the Patient-to-Model-to-Patient (PMP) protocol and published in Current Protocols, consists of a computational and experimental workflow for identifying patients most likely to respond to treatments in clinical trials based on their individual master regulator activity profiles.
The protocol broadly consists of systematically identifying patient master regulator proteins and tumor checkpoints through tumor gene expression analysis and identifying the drugs/compounds whose mechanism of action is predicted to have the greatest effect on those checkpoints. This creates what DarwinHealth calls "mechanism-based biomarkers" that can be used for drug development and clinical trial matching.
"The PMP protocol … expands the concept of oncogene addiction to the transcriptional regulators of the cancer cell state, which are universally present," said DarwinHealth CSO Mariano Alvarez, lead author on the study.
The protocol begins with identifying the master regulators that comprise an individual's tumor checkpoint, based on the gene expression profiles taken from tumor samples. It then identifies in vitro and in vivo models that functionally recapitulate that checkpoint, enabling researchers to identify tissue context-specific drug mechanisms of action and carry out biomarker-based preclinical validations of drug efficacy.
DarwinHealth and clinical trial collaborators are using the PMP protocol to match mechanism-based biomarkers to those patients most likely to respond.
"One of the most important challenges in cancer drug discovery is aligning the mechanism of action and the particular cohorts in which the drug is likely to work with the cohorts in the real world in clinical trials that potentially are responsive to those drugs," said Gideon Bosker, cofounder and CEO of DarwinHealth, adding that the likelihood of gaining US Food and Drug Administration approval for drugs can often depend on biomarker-directed therapy.
James Chen, senior VP of cancer informatics at Tempus Labs, said that while the PMP method "has the potential to truly customize therapy for cancer patients," certain key limitations may need to be addressed for the protocol to realize its full potential.
The protocol, for example, necessitates large databases with enough genotype-phenotype correlation data to adequately anticipate patient heterogeneity, he commented.
Alvarez explained, however, that regulatory network architecture is largely tissue- and lineage-independent.
Tempus competes in the clinical trial matching space, having developed a predictive machine learning model to improve enrollment in biomarker-driven trials, in collaboration with Janssen Research & Development.
Chen also commented that by focusing on bulk RNA transcriptome analysis, the protocol doesn't account for other parts of the tumor ecosystem, such as DNA alterations, DNA methylation, and aspects of the tumor microenvironment, all of which can impact an individual's response to therapies.
"Particularly in this era of immunotherapy, the host microenvironment cannot be overlooked," Chen said.
Alvarez argued, however, that DarwinHealth believes that transcriptome analysis coupled with an accurate regulatory network model and the VIPER (Virtual Inference of Protein activity by Enriched Regulons) algorithm to be sufficient.
"At the end of the day," he said, "the functional effect of changes in chromatin remodeling and genetic alterations will be reflected in the transcriptome."
VIPER and the Master Regulator Inference algorithm (MARINa) together form the basis of the PMP protocol. The latter algorithm analyzes transcriptional regulation models to identify master regulators, and the former evaluates differential gene expression signatures to estimate how "active" a master regulator is, based on the differential expression of the genes that it regulates.
DarwinHealth also employs these tools in drug repurposing efforts, such as finding new cancer treatments from compounds already approved for use in other therapeutic areas.
Alvarez noted that the protocol's main limitation for the time being lies in deconvoluting intratumor heterogeneity, for which the company is working to incorporate single-cell RNA sequencing.
"Coincidentally," Alvarez said, "the use of scRNA-seq data with the PMP protocol opens the door to its application to the tumor microenvironment."
DarwinHealth sees its strategy of identifying therapeutic targets based on master regulator-driven mechanisms of action as being what sets it apart from others in the field.
"The ability to dissect the transcriptional identity state of a cancer cell, with that transcriptional identity state being characterized by a very specific master regulator activity signature, is unique to DarwinHealth," Bosker said.
Andrea Califano, cofounder and chief scientific adviser of DarwinHealth, added that the strategy of targeting whole transcriptional programs adds "complementarity to oncogene therapy by targeting an entire repertoire of tumor vulnerabilities, as represented by master regulator proteins that are responsible for maintaining the aberrant state and regulatory programs of the tumor cell."
DarwinHealth is currently engaged in several clinical studies, in which partner companies have been using the PMP protocol.
"There are about six or seven partners utilizing different aspects of this technology," Bosker said.
Karyopharm Therapeutics, for example, employed it in several trials of its cancer drug Xpovio (selinexor) and Curis has been using it to test its B-cell lymphoma candidate therapy fimepinostat.
DarwinHealth does not license its technology to partners but rather analyzes partner data using its computational platforms, typically under a royalty and milestone sharing agreement.
"We are a co-discovery company," Bosker said.
The IP underlying the PMP protocol is licensed exclusively to DarwinHealth by Columbia University, where the algorithms and databases used by DarwinHealth were developed in the lab of Andrea Califano, who also directs the JP Sulzberger Columbia Genome Center.
"We have undertaken a very aggressive patent prosecution program related to different elements of the [PMP] pipeline," Bosker said, adding that the company has been allowed claims for the majority of its technology in the US, Europe, and China.
The development of the PMP protocol also represents another step in DarwinHealth's eventual transition toward becoming a composition-of-matter biopharma company. The company plans to evaluate some 500 compounds in various cancer subtypes and to prioritize them based on how they affect tumor checkpoints and suppress tumor growth in animal models. To this end, the company plans to expand its R&D efforts.
"We are taking the PMP model and we are leveraging it for our own internal drug discovery and drug screening," Bosker said.
Despite the planned R&D ramp-up, Bosker said that DarwinHealth has no fundraising plans, as it is currently self-sustaining and profitable.