Moving targets

Amazon advertises itself as ‘delivering smiles’. It’s a reference to the logo on the company’s packages. Once upon a time – before next day delivery, perhaps – there were people who smiled when they received them, too. It’s unclear if they still exist. Here and now, the newly arrived package that Gary Whale, vice-president of operations for EUSA Pharma, holds up to his webcam makes him look a little dour by comparison. To him, working in supply is “a hiding to nothing”.

Amazon and other logistics giants like it are so good at delivering so many consumer goods so quickly that in many cities a same day arrival is more likely to prompt a shrug than anything else. Responsible for a clinical supply operation that is increasingly expected to match Amazon for convenience, though, Whale has more reasons than most not to reciprocate the company’s grin – as much as he, like almost everyone else who has spent much of 2020 locked indoors, might rely on the packages that bear it.

“Shipping this around the world is easy because it doesn’t go off,” he says, opening his latest cardboard box to reveal a towel. “You can have it sitting in a warehouse until it’s bought – and there are no controls, no regulations around the product.” The materials and networks Whale and his peers deal with are something different altogether – but there’s a gap between knowing that to be the case and acting like it.

That said, as reliant as they might be on rapid delivery, most consumers still make sure they have a towel in the house before they get in the shower. Clinical trials are sometimes lacking in that kind of common sense. Whale recently had to avert a crisis at a site that was so confident in a delivery estimate that it sedated a neuroblastoma patient before their assigned treatment had arrived. They might have gotten away with it, too – if the material they were waiting for hadn’t missed its flight. That meant a panicked flurry of phone calls and £20,000 in last-minute tickets, Covid-19 tests, and payment for a courier to personally carry the material on the next available plane.

“[The site] said, ‘You said it would be here on Wednesday,’” recalls Whale. “It was due on Wednesday, but whether it would get there on Wednesday is another thing. Wait until you’ve got [the] drug before you bring a patient in.”

That unintended just-in-time (JIT) issue arose from a scheduled delivery for a traditional trial. As trials become more complex and reactive, it’s likely that such problems are going to become more common. Covid-19 increased the world’s dependence on online shopping, and it pushed adaptive trials into the centre of clinical development. For them, the supply chain has to go one further than supplying the towel. It needs to be as responsive as the valve controlling the water.

Just-in-time demand

Of course, Amazon’s supply chain already is. With permanent warehouses the world over, continually optimised routes, long-lasting, lightly regulated products and the best data around, more than a few pharmaceutical operations teams would contend theirs could be too. As it is, adaptive trial design means they have to deal with unstable, unpredictable demand on top of unstable, temperature-sensitive products, and some of the world’s strictest regulation. Not that anyone else in the industry realises what that means.

“The general premise,” explains Whale, “is that an adaptive trial requires just-in-time shipping. So, you don’t know where you’re shipping to or when you’re shipping until the shipment needs to leave.”

“Clinical operations just see it as the same thing [as more traditional trials],” says David Northrup, head of supply chain at Theravance Biopharma. “It’s no challenge; the same but better. But we as an operational line look at it and go, ‘Holy shit, this is different’.”

Northrup, who co-hosted a presentation on adaptive trial supply at the 2019 Global Clinical Supplies Group conference, was at Pfizer when talk of adaptive designs first set the industry “buzzing” in the late 2000s. Supply difficulties limited the initial impact, and much of that excitement faded, but in the decade since, he explains, simulation and modelling tools have advanced to the point that adaptive protocols are feasible on a wider scale. “A decade ago, it was all spreadsheets and it took forever to figure out scenarios,” he says. “Now you can click a button to run 1,000 scenarios in a split-second. You can really start to frame-up solutions and answer some of the questions about how much you need and where you need to put it.”

Simulations improve, but they don’t make anything happen. Modelling how different adaptive trials really are – not just from traditional RCTs, but from each other – isn’t even the first step, as operations teams need to gather the data and win the flexibility to use them effectively. That means consulting with trial designers and clinical teams on different recruitment strategies, site locations, randomisation schemes and dosing plans before protocols are developed. “I would love to start with the simulation,” says Northrup. “I think data speaks beautifully. But the reality is that you may not have that and you’ve got to charm them. The challenge is to get them to understand that the supply perspective is much more complicated. We’re still asking them for the same information and they don’t see how the sausage is made – but it’s got to be made very differently in order to get them what they need.”

Anh Ninh, a mathematician specialising in clinical trials operations management at the College of William and Mary in Virginia, US, is continually surprised that more people in the industry don’t share his interest.

“I talk to people in design groups and they have no idea how important the supply process is,” he says. “That’s the biggest problem: the lack of communication. I think the supply way of thinking has to be shared between everyone involved in the trial process so that they know that even a small change can really affect how the supplies should be done. People propose all sorts of fancy ways to adapt their trials, so that they can make them quicker. But they don’t really think that it’s going to affect the supply cost.”

As Northrup sees it, operations teams need to translate their concerns into terms that interest their clinical counterparts. “You can say, ‘look, guys, if you don’t tell me what we’re gonna do, it’s going to cost us ten times more than what you need’, which they don’t care about because it’s money,” he says, before reframing the point. “‘But you can say: ‘you’re not going to be able to achieve what you want to achieve in a timely fashion’ – that’s the key thing.”

From strategies to tactics

Sometimes, however, there isn’t any flexibility. Whale points to a 16-country, 120-site, 126-patient Crohn’s disease study EUSA ran in 2019. “The biggest difficulty is not getting drugs to sites – it’s getting patients,” he says. “You open a site wherever you’re going to get a patient, not necessarily because it’s easy to ship to.” For that trial, EUSA established one of its five depots in Ukraine specifically to deal with the difficulties of getting material into the country. With interactive response systems calling for a prescription to be released as soon as a patient has been enrolled, there’s no leeway for customs issues.

That’s one of a range of what Northrup calls “tactical approaches” supply teams can use in tandem with their simulations. Another is what he calls “small-scale manufacturing; frequent delivery” – which helps to prevent excessive levels of overage, though it increases the risk of stock-outs. Although EUSA does manufacture, as Whale puts it, “not just-in-time, but close to when we need it”, for the most part, the company tries to maintain a six to 12 month stock buffer. That’s helpful if a study that started with a 1:1:1 treatment allocation is changed to 0:2:1 after the first interim analysis. Still, as valuable as such an adaptation makes excess stock for the second treatment arm, anything left over from the first instantly becomes waste. Batch sizes and trigger levels need to be finessed in order to split the difference.

“From an operational perspective, it’s a lot of work,” admits Northrup. “But if you know the quantities from your simulation tools, you can do it. And then you’ve got your IRT in place, which means you can use JIT packaging or secondary packaging and labelling to get it to the right depot or to the right site at the right time as well.” This helps to delay committing material to dosing regimens or protocols until as late as possible, ensuring the supply chain remains flexible enough to keep up with adaptations and avoid write-offs. For the same reason, Northrup recommends manufacturing smaller doses, which can be combined to make larger doses for different treatment arms or when escalations occur.

That said, requiring patients to take three injections instead of one is likely to negatively impact compliance. Equally, the choice between using unlabelled ‘brite’ stock and partially labelled materials also needs to take into account whether or not they require specialist packaging facilities at the depot-level, which might actually reduce flexibility.

The greater good

For all the complexities, Northrup stresses that clinical supply professionals don’t need to fear the steady uptick in the use of adaptive trials. Given the benefits they can bring in accelerating development timelines, waste is not as big an issue as it might be for more traditional RCTs. “You can always overproduce or overwhatever in order to make this work,” he says. “When you hear ‘adaptive trial’, don’t be afraid. It’s a problem that can be solved. It’s not a knife edge.” Indeed, the success of adaptive Covid-19 trials has shown that, even without extended start-up phases, money and will can make complex, constantly changing trial designs work, and work well. EUSA is currently launching one in the US for an ARDS drug.

Then again, as Ninh points out, with all the attention on stopping the pandemic, the supply costs that might be incurred have been the least of anyone’s worries. That will change as the world emerges from the worst of it. Ninh suggests that the long-standing clinical logic for managing supply well – that if it’s done poorly it will slow a drug’s progress to the market – could be complemented with another better suited to optimising adaptive trials. “You should try to do everything as costefficiently as possible,” he explains. “And even if time is the main concern, then you still need to try to save money. Why just throw it away? It’s not a trade-off between the running time or the recruitment time and the supply cost, but it’s more like you’re trying to save as much money as you can. You can use the money you save on one trial when you need it for another.”

That’s something for clinical supply professionals to tell their colleagues. Northrup also thinks they need to reframe their own motivations. “The fact that the costs are going to be astronomical is hard for operations folks to swallow,” he admits, “but you have to see the greater good – the motivation has to be for the patient.” Let them think it’s easy.