Skip to main content
Premium Trial:

Request an Annual Quote

Under the Hood in Pharma

Premium

Trevor Mundel heads up the exploratory development group, which bridges research and development at Novartis. GT’s Jeanene Swanson caught up with Mundel to talk about the learn-and-confirm strategy, pharmacogenomics in physicians’ offices, and more. What follows are excerpts of their conversation, edited for space.

Genome Technology: Why is Novartis moving away from the traditional three-phase model and toward the learn-and-confirm paradigm?

Trevor Mundel: Toxicology, pharmaceutical sciences, pharmacokinetics — those are all traditional departments. The biomarker and the translational medicine would be the newer types of departments. But it’s more how those departments actually interact, and how they are deployed which is really new.

It is clear that we have an early stage where we don’t know very much — and we really have to explore what is going on — followed by a phase where some of those explorations actually lead to definitive hypotheses, which we can guard and confirm. As a critique of the way some of us operated previously, I would say that there was a minimization of the real exploration, so people would very prematurely jump into the phase of development confirmation. Hence there were a number of these late phase 3 failures, because you just don’t know enough about these therapies that you’re developing.

GT: What are some of the pharmacogenomics strategies that Novartis is developing?

Mundel: One is around, on the early stage, trying to be as comprehensive as we can. So we do look at mutations [or] polymorphisms in our targets. You have to do that fairly systematically because obviously it is something that has to be set up ahead of the time, so that when you get your responses in at the end of the day, you can do a correlation between your response and the mutation or the polymorphism status of the target. You have to have an entire system set up whereby systematically across your studies you are capturing that data, which is not in and of itself an easy thing to do.

Then after we get to a result we like or an interesting result, we are able to pull the data out of the database and look at the correlation between that pharmacogenomics data that we have and the responses that we got, good and bad. So we have a pretty systematic and all-encompassing strategy around that aspect of things.

GT: We’ve all heard the term “blockbuster drug.” But what’s a niche-buster?

Mundel: There are a number of ways to understand that. To some extent, it’s finding a lot of small, targeted indications, none of which is anywhere near a blockbuster, but because of the multiplicity of them in sum total, you now have a highly successful product.

The way we’re thinking of things is, we go back to the pathways that we’re working on. So we’re working on not just any pathways — there are obviously hundreds — [but] we’ve been trying to pick pathways which are really fundamental to driving important processes, whether that be proliferation or inflammation pathways downstream from IL-1, this MAPK pathway, [and more]. It’s almost as if we’re looking for blockbuster mechanisms in pathways. Once you have those pathways, then the issue is, how are you going to show that your therapies actually work, and that’s where we look at these niche indications. So we look at the pathway, and we say, what would be the most succinct way of showing that our therapy actually is working effectively against this pathway, or even enhancing this pathway. And then we often come up with some rare diseases. So that’s where these niche diseases come in, often in terms of proving that our mechanisms, which have got broad application across many diseases, actually work.

GT: What kind of changes has the shift from the blockbuster to the niche-buster model brought to Novartis?

Mundel: Actually, it’s had a host of changes. One is around this notion that we are obliged to these patients to actually provide therapies to them, if they participate in our studies and it’s uniquely effective for them. We can’t just walk away without that extra effort.

A lot of companies are spending all their time thinking about how do you segment markets or not segment markets. What we’re thinking about is, scientifically, how do you segment patients? We’re thinking about the science of where do you produce the best therapy for these target indications, whereas in the previous mode and still what most people are doing in the blockbuster mode, they’re doing much more marketing at an early stage.

As the issues of blockbusters have ratcheted up, everybody is out there looking for alternative strategies. I haven’t really seen a lot of effective approaches to actually making progress with these targeted therapies. You could take some of the traditional groups, you have to add some new capabilities which were never there before, but there’s a certain degree of integration, and how you actually use those groups and deploy them, which is really very difficult to set up. We’ve had some success in that, but I would say that we still have a lot to learn. We have a lot of areas that we are still working on to bring all of these new technologies to bear.

GT: Will pharmacogenomics affect how physicians prescribe drugs?

Mundel: It’s clear we’re going to be converging, and I think the long-term future is around the every patient has a unique therapeutic approach. They may not have a unique therapy, but it could be that they have a unique combination of therapies, or they have a unique dose, and that the physician would have the diagnostic tests performed on that patient which will optimize the couple of available therapies for their condition based on a number of factors.

Physicians obviously are much more confident in prescribing the drug if they know that the patient actually has the target of the drug. There’s always been this empirical aspect to therapeutics, where physicians just basically use trial and error: did you get the dose right, did you try it long enough, when do you stop it, when do you add in combinations? But we now have some tools available which give you some hint at [whether] this patient should be a responder and how long do you wait. Well, if you have the diagnostic test, you know that they should [respond], and you can just accelerate your escalation of dose rates, for instance.

GT: Where do we go from here?

Mundel: [MicroRNAs are] very promising for us. In terms of technology, we found that just the naïve application of gene expression profiling is often non-informative, because gene expression often doesn’t correlate with protein levels and the effects tend to be mediated via proteins and not the genes. Those effects in gene expression can be quite subtle, but they can have dramatic effects on protein levels. So because of that, broad-scale gene expression profiling has been not as productive as people would have liked.

Now you get into miRNA where there’s not a protein correlate that you really have to worry about, so the speculation is, if you can really understand miRNA, we may get some very effective gene-level diagnostics which have a much better correlation with disease in response than just the full frontal type of gene expression profiling work. We hope that we might have much more concise descriptions … of miRNA than we have with gene expression when we’re sort of really trying to ferret out minute signals in a tremendous sea of noise.

The Scan

Booster Push

New data shows a decline in SARS-CoV-2 vaccine efficacy over time, which the New York Times says Pfizer is using to argue its case for a booster, even as the lower efficacy remains high.

With Help from Mr. Fluffington, PurrhD

Cats could make good study animals for genetic research, the University of Missouri's Leslie Lyons tells the Atlantic.

Man Charged With Threatening to Harm Fauci, Collins

The Hill reports that Thomas Patrick Connally, Jr., was charged with making threats against federal officials.

Nature Papers Present Approach to Find Natural Products, Method to ID Cancer Driver Mutations, More

In Nature this week: combination of cryogenic electron microscopy with genome mining helps uncover natural products, driver mutations in cancer, and more.