Independent life sciences consultant Kerry Coffee helps companies evaluate investments, craft growth strategies, and restructure their organizations, with a particular emphasis on bridging the gap between clinical science and commercialization.
In our latest Expert Q&A, she shares best practices for building efficient pharmacovigilance organizations, leveraging artificial intelligence, and accelerating clinical trials.
Many pharmaceutical companies are trying to evolve the operating models of their pharmacovigilance functions. Let’s talk about some of the dynamics at play.
With the rise social media and the internet, there are a growing number of channels through which potential adverse events can be reported. So the number of reports that safety organizations must evaluate has grown exponentially. At the same time, there is a continuous pressure to manage costs. Companies are primarily using two levers to address these pressures. The first is moving operations to lower cost locations globally and making sure that knowledge is shared and understood. The second is introducing AI and other workflow technologies that can handle the growing volume of work.
Another consideration for larger companies is the frequency of M&A activity. As companies acquire and divest different parts of their portfolio, they often have to grow or shrink their pharmacovigilance organizations, which requires flexibility in the operating model.
What are some best practices for building efficient global pharmacovigilance functions?
Pharmacovigilance organizations have three primary focuses: regulatory compliance, quality, and cost. We often think that outsourcing is a great way to create efficiency, however for large, global organizations, the companies that perform most efficiently will keep the majority of their operations in-house. They may build those operations in lower-cost geographies, but keeping the expertise and processes in-house gives them more control over work quality and cost, leading to more efficient operations.
Also, with the volume of work and variability in the workload, the resource model must be flexible. There are different ways of accomplishing that. One model is to create regional centers that concentrate expertise and absorb overflow.
How does that fit into broader clinical operating structures?
A lot of it depends on the size of the company, their portfolio, and what they’re reasonably able to do in-house. What is their geographic presence? Companies today are marketing their products all over the world, and local knowledge is important. If you already have a base of operations in China, you can evaluate whether it’s cost-effective to use that to establish a greater regional presence—whether it’s to conduct clinical trials or do pharmacovigilance work.
It also gets back to your earlier point of making sure that information flows efficiently throughout the organization.
In pharmacovigilance, as a product comes to market, you go through a process of evaluating and understanding signals. If you launch a year later in another country, you want to make sure that knowledge is available and the information that you’re reporting to regulatory bodies is consistent.
That’s something that hasn’t always been done well in the past, especially in organizations where you’ve got strong command and control at the country level as opposed to headquarters.
What best practices are there with respect to governance?
Governance has to be global. Pharmacovigilance collects information on chemical and biological entities. If you’re trying to understand a product and evaluate the signals, you want to have as much information as possible. By its nature, the safety function is always going to be global because biology crosses borders. Most organizations, if they don’t already have a global governance structure, are moving towards one.
You brought up artificial intelligence. How have you seen it deployed in pharmacovigilance and/or broader R&D?
In drug discovery, there’s a lot of potential. AI gives you a more robust and faster way to screen potential candidates. It also enables you to do systems modeling, so you can understand not just how a drug acts locally or at the level of, say, the renal system, but within the whole body.
True AI has yet to be deployed at scale in pharmacovigilance, because the technology is not quite there. Many pharma companies have AI programs and partnerships, and things are moving very quickly. Safety is a great area for AI because of the large amount of data and common variables that you use to evaluate risk. In the future, a case will come in and be run through an algorithm to look for similarities with past cases and assess risk. If enough things match, it will tell you, “We’ve seen this before, and the characteristics tell us it’s low risk,” and no further review will be necessary. If we can do that with even half of the cases, it’s huge from a resource standpoint—and I think the hope is that we’ll eventually be able to do it with the great majority of cases. But it’s at least a couple years away.
Are the companies you’ve worked with preparing for that eventuality as they reevaluate their pharmacovigilance operating models?
They are certainly planning for it from an investment standpoint, and they are starting to plan for it from a resource forecasting standpoint. But they can’t deploy it until the technology has been validated.
You’ve been involved with several projects that were aimed at accelerating clinical development. Can you talk about some of the best practices you’ve learned? Where do companies stumble?
Some things have been around for a while, but it takes companies a while to put them into play. An example is adaptive trial design, which helps accelerate development by enabling you to make real time modifications to the protocol. You don’t have to wait for the full results of a phase II study before adjusting dosage, and often you can move into the next phase of development without interruption.
Many of the newer drugs that are coming to market are in rare diseases or very specific types of cancers, including gene-based therapies where you’ve got much smaller patient populations. You need to be smart about how you’re recruiting patients to avoid delays. This is an area where outsourcing can be a very effective strategy because you’re accessing a larger network of investigators which can accelerate development timelines. But it’s a challenge.
Organizations like TransCelerate emphasize the creation of standards throughout the industry so that things like investigator training and site monitoring are consistent and there’s not as much friction around trial execution. All of these things help to get therapies to market faster.
What about some newer innovations?
One thing that interests me are the new types of data that are being leveraged in clinical development and regulatory approval. Registries, for example, offer a way of collecting meaningful data on marketed products, often in disease states for which a product isn’t indicated. There are companies that develop and house disease registries. Corrona is probably the best—they run registries for several immune disorders. And it’s terrific, because their data is often better than anything you’re going to get in an EMR or through a clinical trial. It’s really targeted to the disease state.
Recently, a large pharmaceutical company was submitting for drug approval for an autoimmune disorder and was told by regulators that they needed more safety data. They were able to use registry data instead of going back and running more people through the trial, which might have taken another year.
The more that companies can utilize alternative data sources like registries, the more comfortable regulators will become with the data. I think there’s often a perception that regulators require you to jump through hoops. But the reality is that they are looking for better ways of doing things, too. They’re looking for best practices, and they’ll often take industry’s lead if industry has found a best practice that allows products to get to market sooner.
About the AuthorMore Content by Leah Hoffmann