By John Russell
April 2, 2009 | Merrimack Pharmaceuticals is frequently seen as a poster child for the systems biology approach to drug discovery. Founded in 2000, it has made few missteps. Last spring, Cambridge, Mass-based Merrimack had the good fortune to raise $65 million in a series F financing round before the economy soured. In July, the company’s first compound developed using its systems biology approach, MM121 (oncology), entered Phase One.
Merrimack hit a minor speed bump near the end of 2008 when its most advanced compound, MM093 (rheumatoid arthritis), did not meet its Phase Two primary end-point and work on the program was suspended. But MM093 is an in-licensed compound, and not the product of its distinctive “network biology” approach which pairs experiments with heavy use of in silico modeling to derive detailed mechanistic understanding of pathways, targets, and compounds. Currently the company’s focus is on biologics.
Clearer evidence for or against Merrimack’s brand of systems biology is expected late this year when data from MM121’s Phase One is released. The broad idea, of course, is that these systems approaches will reduce pipeline attrition and perhaps be more naturally suited for developing targeted therapies and companion diagnostics.
Last month, Merrimack elevated Ulrik Nielsen, a company founder, to SVP and Chief Science Officer, and Predictive Biomedicine editor John Russell took the opportunity to talk with Nielsen and Kathleen Petrozzelli, senior manager, corporate communications, about Merrimack’s progress and course adjustments. One recent change: Merrimack is seeking a partner to help it bring MM121 to market, a departure from its past go-it-alone strategy.
JR: Congratulations on your promotion to CSO. How will your role change?
Nielsen: It hasn’t really sunk in yet. I’m spending more and more of my time exploring how what we call network biology can be applied to areas of medicine beyond oncology where we’ve spent most of our efforts. We’re looking at the field of regenerative medicine and also exploring areas of safety as it relates to targeted therapies and predicting safety issues.
JR: You mean biomarkers around safety issues?
Nielsen: Right but really driven through a systems understanding based on specific molecular profiling in non-target tissues for predicting potential toxicity.
JR: Could you provide a quick review of MM121 and how its development reflects Merrimack’s network biology approach?
Nielsen: So MM121 is sort of the most advanced product to come out of that approach and the insight there was that ErbB3 was not really appreciated when we started as a therapeutic target. It highlighted that and gave us confidence to move forward on two programs around inhibiting ErbB3. There’s now a second program we made public this fall which has to do with a second ErbB3 therapeutics. It’s designed specifically for her2/erbB2 over-expressing tumors which is a different sort of network setup than what MM121 targets. We had the insight a long time ago. It has very, very potent anti-tumor activity in ErbB2 over-expressing tumors. On that drug we actually took network biology another layer down and kinetically optimized the molecule because it’s a bi-specific antibody so it binds to two different targets on the cell surface.
We are able to use models to suggest the optimal affinities of the two different ends because now you have two different antibodies you bind to. We had some non-intuitive insights there from modeling that you could lower the affinity of the second binding arm tremendously before you lost activity—something that would be incredibly hard to show experimentally. We were able to show we could lower the binding for the second arm, maintain the activity, and potentially limit inhibition of ErbB3 in non-tumor tissue at the same time. We are still in the process of filing an IND on this program and hope to be in the clinic in the middle of this year.
JR: How did Merrimack’s in silico expertise make a difference in that program?
Nielsen: I’m not sure we would have developed this kind of molecule without the modeling insight. It was very early on. We were modeling these types of molecules in the ErbB3 pathway doing every possible combination and it would have been very time-consuming to create all those combinations; we would have done nothing else in the last three years except different combinations of antibodies in that pathway. [Instead] we were able to simulate it in a matter of weeks and go ahead with confidence we were making the right molecule.
JR: Let’s shift back to MM121, how is it doing?
Petrozzelli: It’s in phase one [begun last July] now and we’ll probably be able to release the data on the phase one in the fourth quarter. We’re also planning to initiate a phase one combination study with MM121 in the middle of this year (June). So the news here is the product itself is moving through phase one pretty well at this point and we should be entering phase two by the end of the year and we’ll also have a combination study going.
JR: As you know many observers are watching to see how well Merrimack’s “network biology” approach fares. What’s Merrimack’s perspective?
Nielsen: Well, we’re convinced internally at this point that this is the way forward for the industry, and certainly for us. If you want to reduce attrition you’ve got to bring these simulation tools into early stages of thinking about what’s the best therapeutic approach to a pathway, to get the systems understanding, and then create the best therapeutic. In terms of proving [the concept] to the outside world, our ErbB3 program has raised the awareness. There’s been incredible interest. I think the real proof will be real clinical data as we get it progressing through clinical trials and as MM111, as the next program is called, this bi-specific program hits the clinic later this year. I think it will be the first bi-specific [antibody] in the clinic.
I recently went back and looked at the original pitch we made to seed investors and the title of the presentation was ‘Drug Discovery in Complex Protein Pathways’. I looked at the goals we set out for ourselves in terms of moving this forward and you know we’ve pretty much done what we set out to do. I think that path that we’ve taken to get there is probably a little different than we thought originally.
JR: What was the biggest change?
Nielsen: We embraced modeling and simulation to a greater extent than we initially envisioned. Initially the focus was on the need to generate the data to allow us to understand the system. But once we had that data we started seeing there was a very nice interplay with the modeling and simulation that benefited us tremendously and set much of the direction for what we’re doing now.
JR: Given the success of your modeling efforts, has the modeling group grown?
Nielsen: So we don’t actually have a modeling group. You may be surprised. We saw early on that this type of approach to very complex problems was best solved in multidisciplinary teams so every team is set up without a department structure. We’re bringing modelers and engineers and biologists to do drug discovery together in pretty independent teams. We’ll have one team around one pathway, or one therapeutic area.
JR: Do you think this works for Merrimack primarily because of the company’s modest size versus a larger company?
Nielsen: I think it’s the best way period. Without the close interaction on a daily basis between modelers, experimentalists, and the people who design the drug, we could not accomplish what we’re doing today. In fact, what we’re seeing is that there is much more of focus on the actual application of systems and network biology to drug discovery this way as opposed [focusing on] let’s get better at modeling and let’s get better at doing the biology.
Generally the insight we have from trying to structure a whole company around the systems approach is that the traditional departmental way of organizing things will work against putting what’s really engineering at the center of the effort—not engineers, but the way engineering is done in other industries. That really requires the multidisciplinary approach. In some ways threatens the way medicine is traditionally organized. Or drug discovery is traditionally organized.
JR: How is the difficult macro economic situation affecting Merrimack?
Petrozzelli: We closed our series F financing ($65 million) at the end of May (‘08) so we got in before the economy really started to go down. That put us in a really good position. What we’re doing going forward—since we realize the opportunities to finance in traditional ways of equity financing or even potentially going to the public market don’t really exist right now—is we are looking to partner our lead program MM121 and are in active discussions on that right now. We hope to be announcing something on that later in the year.
JR: That’s a change, isn’t it?
Petrozzelli: I think every company wants to own their products totally, right, so from the outside this [could be seen] as a slight change. But this is a great path for us to go down. As we move MM121 through the development, what we can get from a partner we couldn’t get from an equity financing. So there are definitely strategic things that are pushing us in the partner direction outside of just the economy. I think for our company to sustain itself we have to start building some of this infrastructure around moving a product from phase two into phase three and working with a partner is probably the best way to do that.
JR: So what else is new?
Nielsen: We haven’t talked about diagnostics. For all of our programs that grew out of the systems approach, we also have diagnostics efforts that are also rooted in the insight that got from our systems analysis of those pathways. The first one is with the MM121 program. The plan there is to start using that in patients on an experimental basis and we have similar programs for other therapies that we are developing.
Petrozzelli: I think the main take-away is that long-term, every therapeutic that comes out of the company will have a paired companion diagnostic. The first signs of that will be later in 2009 when we can start using what we’re calling a phase one “tail of our phase one” program to better understand the diagnostic we currently have in place and we could potentially put it into place in our phase two program.
Nielsen: There was one more thing. We’ve now recognized that cancer drugs are rarely going to be monotherapy, even if they are systems-designed like MM121. As such we’ve used modeling for some years now to help predict the best combinations of targeted therapies to bring into the clinic. And we will be starting the first trial based on that type of insight, in the MM121 program, with another target inhibitor. We haven’t disclosed which target inhibitor that is yet. It was actually something we knew that this combination would be great before we even created MM121.
We think there will be some opportunity for MM121 as a monotherapy, especially in combination with a diagnostics to find those patients mostly driven by this ErbB3 induced signaling, but we think there are many more opportunities for using this therapy in combination with other drugs and sort of preempting the development of resistance to therapy.
JR: Thank you both for your time.
This article first appeared in Bio-IT World’s Predictive Biomedicine newsletter. Click here for a free subscription.