SHANGHAI – Denovo Biopharma LLC, a cross-border biotech registered in San Diego and Hangzhou, has dosed the first patient in a phase III trial of DB-102 (enzastaurin) to treat diffuse large B-cell lymphoma (DLBCL). DB-102 is a serine/threonine kinase inhibitor of the PKC beta and AKT pathways.

Unusual for a global phase III trial, the first patient was dosed in China, not the U.S. That marks a milestone for the industry, demonstrating that the CFDA's recent regulatory reforms are having the intended effect to streamline and hasten the review of much-needed drugs.

But the real story is that Denovo is going for something more improbable: to prove that the third time is the charm.

Enzastaurin has failed already twice before, not hitting efficacy targets in DLBCL or in glioblastoma in phase III trials run by Eli Lilly and Co., of Indianapolis. But it is precisely those failures that give Denovo reason for optimism. (See BioWorld Today, May 13, 2013, and Sept. 25, 2014.)

Those two phase III trials and numerous other studies generated a vast amount of patient data, which Denovo has mined using its biomarker platform and algorithms to discover de novo – new – genomic biomarkers.

"We stand on the shoulders of giants. Lilly spent more than 15 years on this drug and ran 60-plus clinical trials. We are sitting on rich information that can guide us to make the best decisions. These data and drug would have been shelved otherwise," Wen Luo, Denovo CEO, told BioWorld Asia. "We are doing new drug discovery. But while most have to do it with the lights off, we have the lights on with the biomarker illuminating the way."

Focusing on a large subset of patients that had improved survival on enzastaurin, the company drilled down to discover a novel biomarker and patented it as DGM1 (Denovo Genomic Marker 1). That led to developing and patenting a diagnostic test, giving the firm protection now that enzastaurin is no longer patent-protected.

"Our approach changes the patent game because the biomarker is completely novel, and our trial will take less than five years," said Wen. Because DGM1 is, in fact, a combination of biomarkers that does not occur in nature, Denovo was able to get it patented.

Distressed asset

While taking a biomarker approach makes sense in theory, the practice of discovering a biomarker is remarkably difficult.

For one reason, to successfully execute a biomarker-driven approach requires a hard-to-find set of skills that marries drug discovery and genomics. When Wen left his job at Ligand Pharmaceuticals Inc. to set up Denovo, he had already begun to develop a biomarker platform. It was a natural fit for his experience. Two decades ago, Wen started his career doing kinase drug discovery at Sugen (acquired by Pharmacia, which was then by Pfizer Inc.) followed by years at Incyte Genomics, the largest genomics company at that time and predecessor to Incyte Corp.

Even so, when Denovo formed six years ago, few thought a startup Chinese company would be successful at acquiring an asset from a big pharma.

"There are several things that we do that are so new. The concept is very straightforward but to do it is complicated and uncertain," Wen said. "There is no guarantee that a biomarker will be found."

Wen said it took two years to convince Lilly to hand over the global rights for DLBCL (and for glioblastoma as well) for what amounted to a rock bottom price since enzastaurin could be considered a distressed asset. In exchange, Denovo could offer the promise of a payout down the road in milestone and royalty payments. Yet, even with so little to lose, it took two years to convince Lilly to do the deal.

It took another six months to complete the technology transfer and about a year to run the tests that would ultimately uncover the biomarker.

Using the DNA from blood or plasma samples, Denovo's biomarker team scans the genome for 5 million biomarkers using machine learning and data mining. In oncology, looking for a mutation in a tumor biopsy is the more common route – but Denovo also scans the germline DNA, to try to identify the genetic difference between the responding patients and nonresponding patients.

"Most people focus on the drug target, based on the success of Iressa to hit EGFR," explained Wen. "This is a narrower approach. We scan the whole genome and perform big data analysis.

"We are elevating the game to a new level. It requires looking at the phenotype, all the clinical data and investigating the study from different angles."

The process requires advanced DNA chip technology and gene sequencing equipment. Being located in San Diego, Denovo has worked with Illumina Inc. to use its most advanced DNA chips as well as other genomics tools at companies headquartered nearby. Wen said the company has been able to shave months off the biomarker discovery process and, in some cases, it can happen within 90 days.

Once the DGM1 biomarker was identified, Denovo received a pleasant surprise: the biomarker population that responds to the drug is the majority of DLBCL patients.

"We did not anticipate the biomarker population to be so high in DLBCL; it is actually over 80 percent," said Wen. "Potentially, we have a blockbuster drug on our hands."

The biomarker-negative population, while only less than 20 percent of the patients, had results that were worse than the control arm of the study. That brought down the overall efficacy of the study that led to the phase III failure.

Déjà vu: Phase III all over again

Good fortune struck in a second way as well.

Denovo is able to replicate Lilly's phase III study with only a few differences: It will be aided by the diagnostic test to identify patients and will use China as a key site for the global trial. The company can follow the same protocol because, in the intervening years, there has been little change to the standard of care (SOC) for DLBCL. It remains R-CHOP – the monoclonal antibody rituximab with chemo drugs cyclophosphamide, doxorubicin, vincristine and prednisone.

"Our two oncology indications for DLBCL and glioblastoma have not been affected by the I-O and PD-1 storm of the last few years," said Wen. "We have the highest chance of success by following what Lilly has done and luckily the SOC has not changed in 20 years."

Before starting the clinical trial, Denovo has been running virtual trials to show proof of concept, and Wen said that work "gives us tremendous confidence and helps us to use the same trial design as Lilly."

The phase III trial, referred to as ENGINE, is a global randomized, double-blind study evaluating the efficacy of DB-102 in combination with R-CHOP vs. R-CHOP alone in DLBCL patients with or without the biomarker DGM1.

The study will enroll 235 patients. The primary outcome measure is overall survival in patients with DGM1.

The number of patients is relatively small for a phase III study, but the biomarker approach for selecting responsive patients is expected to generate statistically significant results.

There are more than 25,000 people in the U.S. with DLBCL. Wen said that for about 60 percent of patients, the current R-CHOP treatment is effective but for the remaining 40 percent high-risk patients, there are no viable options.

In three years, Denovo has reason to believe it will see a significant percentage hit overall survival to make enzastaurin a new gold standard for treatment – not as a standalone treatment but in combination with R-CHOP.

"We have faced a lot of skepticism. The first challenge was could we get drugs from big pharma. The second has been whether we could actually find a biomarker," Wen said. "Now, we are facing the question of whether we have the capacity to run a phase III trial. As a new company, we progress by taking it step-by-step, and delivering."