Pharmaceutical and healthcare organizations have zettabytes of data at their disposal, literally. They have spent the last decade accumulating information from electronic health records, insurance claims, lab results and more. Whatever frustrations one might have with the industry at large, it can’t reasonably be suggested that companies were caught unaware by the data and analytics revolution. If anything, they have veered in the direction of over-collection and analysis.

Thus for a data-centric initiative to capture the industry’s attention and imagination, it has to be spectacular — not just in terms of scientific advancement, but in terms of upending existing schemes and processes. And last April’s label expansion for Pfizer’s breast-cancer drug Ibrance, which was based on real-world data from off-label use of the drug in male breast-cancer patients, checked all the boxes. The data for the approval was culled from three sources: Flatiron Health’s breast cancer database, IQVIA’s insurance database and Pfizer’s global safety database.

“Ibrance represented a really, really important story,” says Craig Lipset, founder of Clinical Innovation Partners. “The FDA updated a label based on real-world evidence — which, from a data science and regulatory science perspective, is amazing. You can’t underplay how big this is.”

At the same time, the Ibrance label expansion troubled many observers. As much as it may have represented a breakthrough for data and regulatory science, it also painted the disharmony between data initiatives and pharma’s patient-centricity in stark terms.

Most notably, none of the men whose data was used for the Ibrance submission had been notified about it, much less educated about their rights or compensated. And that represents a real problem for an industry that continues to struggle with trust and transparency, even as its messaging continues to paint it as a champion of patients’ rights.

“The fundamental tenets of patient-centricity are transparency and trust, and really understanding the notion of consent is at the heart of that,” says Renée Deehan-Kenney, PhD, VP of computational biology at Precision For Medicine’s QuartzBio.

“The FDA updated a label based on real-world evidence. You can’t underplay how big this is.” —  Craig Lipset, Clinical Innovation Partners

She explains that violations of this trust could have a sobering effect on the patients pharma hopes to engage and, ultimately, the volume of data they might agree to share. Along those lines, it’s well within the realm of possibility that once patients have consented to sharing their data, they’d be open to additional requests (for instance, for a biospecimen).

“Most patients I’ve engaged with have been incredibly selfless in terms of their desire to augment research for their indication, even knowing that it might not benefit them – that they might die before they see the benefits,” she explains. “That selflessness is really fragile and shouldn’t be broken through a lack of transparency.”

What Pfizer did vis-à-vis the Ibrance expansion and the data underlying it, it should be stressed, was consistent with current standards as well as with ethical and legal norms. But those standards and norms will continue to shift, especially as patients are further empowered to take control of their data.

pfizer
Source: Getty

That’s why, despite all the enthusiasm about the Ibrance regulatory breakthrough, it comes with a patient-centricity asterisk. “We need to not only make patients aware how their data might be used, but also give them a source of pride that they’re involved in solving problems like this. That’s not a scientific part of the equation, but it’s a part of the equation,” Lipset says. 

Ibrance wasn’t the FDA’s first approval based on real-world evidence and without new clinical trial data. The first time the agency bestowed its blessing in this manner occurred in 2017, when it approved a new heart-valve surgical procedure. That said, the real-world evidence push dates back to the passage of the 21st Century Cures Act in 2016, which set the industry, and especially its data scientists, on a path towards bolder innovation.

“It spurred the FDA to be more aggressive,” notes Aaron Mitchell, a principal at ZS. “Previously, the traditional use of real-world data was for post-market safety surveillance, or maybe in trial design.”

In the wake of the Act’s arrival, the FDA issued a statement in December 2018 that opened the door for Pfizer and its ilk. “Real-world data collected from a variety of sources offer new opportunities to generate evidence and better understand clinical outcomes. These data may be derived from a diverse array of sources, such as electronic health records, medical claims, product and disease registries, laboratory test results and even cutting-edge technology paired with consumer mobile devices. These data are being used to develop information and real-world evidence that can better inform regulatory decisions,” the statement read.

What Pfizer did vis-à-vis the Ibrance expansion and the data underlying it, it should be stressed, was consistent with current standards as well as with ethical and legal norms.

Even prior to the Ibrance approval, the FDA had been at the vanguard of the real-world-data effort. It’s not news that pharma companies tend to move slow, given the regulatory landscape they inhabit. But most observers were surprised — and delighted — by the vigor with which the FDA pushed this agenda.

“The FDA in many cases has been leading the industry,” Mitchell notes. “They’ve been putting guidance documents out and they’ve been out on the speaking circuit and in the media. That points to a greater desire for real-world data to be used in regulatory decision-making.”

But even without FDA enthusiasm, it’s clear that pharma has been building towards the greater use of real-world data for some time. The advantages are obvious: Among others, it allows researchers to look at patient populations that, for whatever reason, were not already studied.

For an example of how this might work in practice, consider the treatment of amyotrophic lateral sclerosis (ALS). As Deehan-Kenney notes, there is “no great mouse model” for the testing of therapies to treat the disease. “Lots of medicines fail during trials because while a company has been able to demonstrate significant efficacy in the lab in mice, it doesn’t work with people.”

But real-world data could potentially upend the treatment calculus. “Now we have access to data from patients with ALS — not just how they do on a particular treatment, but their genomic information. We’re starting to collect deeper information on the biological drivers behind these conditions,” Deehan-Kenney continues. “Not only can we deeply characterize a mechanism of action in a way that we couldn’t even five years ago, but we can look at clinical samples and the deep disease biology. It’s not just about replicating something in a tissue dish.”

Still, don’t expect the FDA to start granting Ibrance-type approvals in droves. As much as it may be in the interest of many different clinical-trial-adjacent constituencies to embrace the data era, potentially we’re looking at billions of data points from billions of patients across billions of indications. Nobody should underestimate the difficulty that comes with pulling together disparate data types from disparate sources and attempting to shape them into a bona-fide clinical asset.

“It’s not for the faint of heart, right?” Deehan-Kenney quips. W2O arcus partner Mary Pao Seideman frames the issue just as succinctly: “Being expert at this doesn’t exist yet.”

The trust and transparency issues are likely easier to solve, but it could require pharma to throw its weight around a little bit. “Companies in this space are the buyers, the way that Walmart is a buyer. That means they can dictate a little bit about what they want,” Lipset says.

This could expedite the transfer of power to patients, he adds. “Let’s say that you’re a company and you have the opportunity to buy two equivalent data sets. You could preferentially buy the one that includes transparency and engagement with patients. Signaling to the market that way would certainly send a message.”

The worry on the transparency/trust fronts is that the industry faces pressure to act sooner rather than later. What usually happens when companies kick privacy and consent problems down the road, in the hope that somebody else will deal with them? The government steps in.

“If healthcare and life sciences companies don’t start to think about those issues and how to make patients comfortable, it’ll be left to regulators,” Mitchell warns. “Nobody wants that. You can end up with regulation that will be quite diverse from country to country and even state to state.”

Of course, pharma companies need to get their own houses in order first. Seideman cautions that it takes a team to manage the technical demands of any such data initiative. “You need people who can bridge the clinical side with the data,” she notes. Left unsaid? That such individuals are in extremely short supply — not just in pharma and the clinical-trial universe, but everywhere else.

“It takes a team. It takes deep thinkers, people with experience with data and people with some understanding of the drug development cycle.”Mary Pao Seideman, W2O arcus partner

“Think about it: You have EHRs, claims data, billing activities, registries, patient-gathered data and more. How does somebody look at all of that and pull out something that makes sense?” Seideman continues. “It takes a team. It takes deep thinkers, people with experience with data and people with some understanding of the drug development cycle.”

At the same time, pharma companies can avail themselves of a burgeoning community of scientist/entrepreneurs investing considerable time and money in “solving” the data problem. There’s Aetion, currently working alongside the FDA on a project designed to determine whether real-world evidence can replicate the result of randomized clinical trials. One company, Ciitizen, creates a unified medical record out of any/all data individuals upload into its system; another, Hugo, facilitates the collection of real-world data for use in studies and elsewhere. Anyone who has navigated a medical institution’s electronic health record — or, rather, attempted to navigate one — can tell you about the myriad frustrations that comes with the exercise. But let’s say you take the information contained therein, add data from your Apple Watch or your 23andMe results, and hand it over en masse to an organization motivated, financially or otherwise, to present it in comprehensible and shareable form?

“What you want to remember is that we have a lot of really good people who got into the industry for the right reasons,” Mitchell says. “It’s easy for us to tell the typical capitalist story, but the reality is there are many people who want to solve this specific problem. Ultimately, that’s going to win out, that altruistic mindset.”

The industry itself has stepped up to an extent, with Janssen assuming a leadership role. The company’s Global Trial Community program shares participants’ data with them during trials; it reportedly hopes to someday connect trial participants with one another. As part of that larger effort, Janssen spearheaded the creation of the Patient Data Access Initiative (PDAI), an industry collective that has set as its goal finding ways to truly patient-centrify the collection and use of data.

Lipset believes both components represent ambitious moves in the right direction. “It creates some cool infrastructure for PDAI members to leverage knowledge and things that are already out there,” he says. “Some companies know how to navigate the regulatory, legal and compliance issues around real-world data. The idea is that the companies that don’t won’t have to build tools and processes de novo.”

This feels like an underemphasized discussion point, given the volume of monetary and temporal resources it takes to assemble an A-list data operation. “It’s a big budget item for a small biotech to build out a team capable of handling this. In big companies, teams like this are slow to come together. There’s no ‘right’ formula,” Seideman says.

Speaking of budgets and entrepreneurs, the conversation around the sharing and use of real-world data was further muddied last year when the prospect of financially compensating individuals for their information was raised anew. The notion itself isn’t irrational: There’s no arguing the fact that data has value, nor that data aggregators charge life-science companies a whole lot of money for that data. Why shouldn’t the consumers whose data is being trafficked be invited into the monetization loop?

Woman examining laboratory samples
Source: Getty

Lipset feels that this debate misses the point. “You bring up compensation and everybody automatically jumps to ‘we can’t pay people!’” he explains. “How about just making them aware of what’s going on with their data? This can be done much more incrementally than jumping straight into financial compensation.”

Deehan-Kenney agrees, advocating for better education specifically around the issue of patient consent. “In my utopia, there’s a better way to do this than by rushing through 50 pages of a document that’s smothered in legalese,” she says. “When patients are considering being a part of something like this, they need to really understand what the differences are between their information being blinded and not being blinded. You can’t eliminate all the risk, but you can at least give them an idea of the risk that’s involved.”

So while there are limitations as to what real-world data can do in the clinical-trial context, and a host of other related obstacles, most everyone believes those challenges can be surmounted. Whatever issues data wonks may have with pharma’s aversion to risk, they believe that the industry is set to build on the triumph of real-world data that Ibrance’s approval represented.

“Randomized control trials are the backbone of drug development and will always be the backbone of drug development,” says Mitchell. “But as we continue to learn about how drugs operate in the world, about new patients and new indications for therapies and new dosing — all of that is certainly within the scope of what real-world data can inform. The bottom line is that it will be better for clinical decision-making.”

And it will likely be better for pharma’s beloved patient-centricity as well. “For patients, the ability to directly broker and own their own data is coming,” Deehan-Kenney says. “Empowering them could be incredibly beneficial. We’re eroding the scientific and medical patriarchy that has existed for some time.”