TRANSCRIPT
Deborah Borfitz:
Hello and welcome to the Scope of Things podcast, a no-nonsense look at the promise and problems of clinical research based on a sweep of the latest news and emerging trends in the field and what I think is worthy of your 30 or so minutes of time. I'm Deborah Borfitz, senior science writer for Clinical Research News, which means I spend a lot of time with my ear to the ground on your behalf and a lot of hours every week speaking to top experts from around the world. Please consider making this your trusted go-to channel for staying current on things that matter, whether they give us hope or cause for pause. In another 10 minutes or so, you'll be hearing from a few big pharma executives speaking at the recently concluded Scope Europe event about regulatory requirements for AI literacy training.
Deborah Borfitz:
But first, the latest news, including efforts to reduce excessive data collection studies, the cardiovascular benefits of popular weight loss drugs, whole genome sequencing of breast cancer, upcoming basket trials for pan-cancer immunotherapy, a promising youth molecule, a virus cocktail to combat superbugs, an AI system for annotating medical images, how Fitbits are aiding precision health, latest findings from the Human Epilepsy Project, and how imposter study participants can threaten research integrity.
Deborah Borfitz:
The Tufts Center for the Study of Drug Development has teamed up with Trancelarate on the biggest collaborative project yet, uncovering opportunities for optimizing protocol design by drawing attention to core standard and required procedures in clinical trials that were not essential for assessing the key objectives and endpoints crucial for a study's success. The non-essential procedures were study-specific and commonly related to collecting data on patients more often than necessary. More fit-for-purpose data collection, which is explicitly supported by global regulatory guidelines, would take some of the onus off study sites of participants, even if sponsors make only a tiny reduction in the overall volume of information they've been amassing in their trials. You can look for the full story on the findings in planned initiatives of Transcelerate to raise awareness of how study data overload occurs and identify opportunities for improvement in clinical research news.
Deborah Borfitz:
One of the largest and longest running clinical trials to date, examining the cardiovascular benefits of the weight loss drugs Ozempic and Magovi, find that they lower the risk of major heart problems in people who are overweight or obese and already have heart disease, even if they don't have diabetes, and that this protection happens regardless of how much weight a person loses or their baseline body weight. Only one-third of the drug's protective effect on the heart after two years could be attributed to a shrinking waistline. The research team previously found the medication led to a 20% reduction in heart attacks, strokes, and other major cardiac events in the same patient population.
Deborah Borfitz:
Researchers in the UK report that annually, over 150,000 women with breast cancer could benefit from whole genome sequencing to either guide immediate treatment or help match those patients to clinical trials. Although a human genome can now be sequenced for as little as $100, the technology is not yet widely used, in part because the information can be too overwhelming to make sense of. Using the data assets of the National Genomic Research Library, the research team found that 27% of breast cancer cases had genetic features that could help guide personalized treatment. Among them homology directed repair deficiency, unique mutations that could be targeted with specific drugs, signs of resistance to hormone therapy, and mutational patterns suggesting cancer weaknesses that treatments could exploit. Importantly, the analysis also provided insights into prognosis. One big win here is that it could massively open the potential for recruitment to multiple clinical trials in parallel.
Deborah Borfitz:
Researchers at the University of California Irvine are prepping for first-in-human basket trials, focusing on a pair of glycans dubbed glitter 1 and glitter 2 that are highly abundant in solid tumors, but previously considered untargetable. The means are Velcro-like sugar-binding proteins that overcome the tumor microenvironment's immunosuppressive properties and trigger T cell-mediated cancer killing. Both dose escalation studies will initially enroll a small number of patients with a solid cancer that is refractory or metastatic, people with the highest density of the target, as well as those in the greatest need, and expand to more patients once the safest, highest dose is identified to look for efficacy. In preclinical studies, the biologically engineered immunotherapies killed all types of solid cancers, including breasts, colon, lung, ovaries, pancreas, and prostate without harming normal tissue. The consensus opinion of more than 25 researchers from around the world is that a tiny molecule known as NAD could be key to her healthier aging. The vital youth molecule helps the body produce energy, repair damaged DNA, and keep cells functioning properly, but levels of it naturally decline with age, a fall that has been linked to memory loss, muscle weakness, and many age-related diseases. Early results of clinical trials endeavoring to increase NAD plus levels, often by giving patients a special vitamin-like compound, are encouraging but need to be validated in larger and longer studies targeting different conditions. NAD Plus supplements are quite popular on the commercial market globally, but now we have a much-needed scientific roadmap to help guide ongoing research and future clinical use.
Deborah Borfitz:
Researchers in Australia have set the stage for clinical trials using phage products with the development of a five-phage cocktail called Entelio 2, now available for compassionate use at Alfred Hospital in Melbourne. The virus cocktail targets a group of bacteria responsible for deadly antimicrobial resistant superbugs that have emerged in hospitals around the world. In infected mice, the therapeutic grade phage product succeeded in reducing bacterial loads by over 99%. The model could be replicated at institutions anywhere facing similar threats. An AI system has been developed that quickly and accurately annotates regions of interest in medical images, which is often one of the first and most critical steps in running a new study involving biomedical images. The segmentation process is typically done manually and can be extremely time consuming. With the new system, users need only click, scribble, and draw boxes on new biomedical imaging data sets, and the AI model predicts the segmentation. And as they mark additional images, the number of interactions needed for the model to perform decreases and eventually drops to zero with no user input required. Users need not have machine learning expertise or extensive computational resources, and the system flexibly works with any number of images, making it applicable to a range of applications. Researchers now plan to test the tool in real-world situations with clinical collaborators to further improve its utility.
Deborah Borfitz:
A first of its kind longitudinal population study handed out Fitbits and tablets to participants to promote equity and precision health. The reality is that leveraging smartwatches and other data logging wearables that people already have only serves to overrepresent wealthy, urban, white, and fit individuals who tend to buy these products. Unlike typical convenience samples that underrepresent minorities, older adults, and lower income groups, the so-called American Life in Real Time study achieved broad demographic representation across race, education, and income levels. The authors showed that a COVID-19 detection model trained on data from this study performed consistently across demographic subgroups, while an identically trained 2019 detection model using wearable device data collected from participants in the NIH's All of Us Research Program, who already owned a wearable device, showed 22 to 40% worse performance in older women and non-white populations.
Deborah Borfitz:
The International Human Epilepsy Project has published more good news for patients. Resistance to treatment may wane over time for individuals with the most common form of the disorder who respond poorly to available therapies. Their latest study, which looked at 150 treatment-resistant men and women for up to three years, revealed that almost 70% of participants saw reductions in their average monthly seizure frequency, challenging the notion that once patients have failed a certain number of anti-seizure drugs, their chances of finding relief are small. In fact, the medications appeared to prompt longer periods between seizures, in a handful of cases, a year or longer. In an earlier study, the Human Epilepsy Project found that it can take doctors at least a year to identify the right drug and dose for people with a treatment-sensitive form of the disease.
Deborah Borfitz:
And finally, imposter study participants are threatening the integrity of health research and, by extension, the policies and clinical decisions built on it, according to an editorial published in the BMJ, pointing out that impostor participants were found in the datasets of 18 out of 23 studies examined, with rates ranging from 3% to 94%. The impostors here are people who provide deceptive or inaccurate data to take part in studies, as well as computer bots, mimicking human behavior and responses. The issue has grown in tandem with the growing popularity of online recruitment and can impact all types of studies. Recommended remedial actions include the use of identity verification procedures or human verification tools and transparent reporting of the safeguards used along with an acknowledgement of their limitations. As a reminder, links to the studies, stories, and news releases referenced in this month's news segment can be found in the show notes.
Announcement:
Are you enjoying the conversation? We'd love to hear from you. Please subscribe to the podcast and give us a rating. It helps other people find and join the conversation. If you've got speaker or topic ideas, we'd love to hear those too. You can send them in a podcast review.
Deborah Borfitz:
I am now going to transport you to the stage of the recently concluded Scope Europe event in Barcelona, where executive editor Allison Proffitt got to speak with a panel from Pharma. As part of the conversation, she asked them about the EU AIX requirement for AI literacy trading. Here's what they had to say, and we're going to be hearing answers from Jim Weatherall, AstraZeneca, Janie Hansen, Daiichi Senkyo, Francis Kendall, Biogen, and Jonathan Crowther, Merck K GaA, Darmstadt, Germany.
Allison Proffitt:
One of the things that came up, uh, Jim, in your talk yesterday, that I had heard it elsewhere is the um EU AI Act requires us to train in AI literacy. Um what does that look like? How do we do that for everybody?
James Weatherall:
Great question. Um I'm assuming that was directed at me.
Allison Proffitt:
Everybody's gonna take us away.
James Weatherall:
Well, it would be interesting to hear everybody's view on it, actually, because a lot of the uh stipulations in the act, and that's not the only one, so there's other things we're responding to at the moment, whether it's the you know the USDOJ ruling on, you know, data sovereignty, there's a whole bunch of things that are kind of hitting us. A lot of them are quite open to interpretation. So, what does it mean to have trained every single one of your employees in AI literacy? I think it's largely the what you want to be wary of is that if an external auditor came through your doors tomorrow, that you can demonstrate some kind of record that yes, we rolled out some central training, it kind of generally speaking explains what AI is and what it means to be compliant with AI. I don't think it has to be a ton of detail at the moment, but I think you have to demonstrate that you're doing something to create awareness in your workforce. So that that's that's the simple answer. As I said, the sort of the the the kind of threat is, oh my goodness, we've all got to do something, but the kind of relief is it it can be lots of it can be whatever's fit for your organization.
Allison Proffitt:
So it can be a little fairy dust. Everybody went to an AI Zoom call.
Janie Hansen:
So to add to that, if um I think as long as you're using responsible AI, um you'll be okay, right? So what is responsible AI? Um it's explainable, um, how, why, um with traceability, and then you know, always using who human in the loop and having the human make the decision. So, you know, have some guiding principles around the tools that you roll out.
Francis Kendall:
I'd I think there's poor elements that everybody in this room should really understand is understand what AI is. Use AI, so whether it's just Chat GPT or something like that, uh, and have that ability to understand whether you can evaluate AI. So know what what your limit is and then when you should be going to your data scientists, your IT people, and then to the you know, responsible or ethical use of AI. If those are the four elements I think really important.
Jonathan Crowther:
Maybe to throw like a contentious point in. Have we even have we even mastered data literacy? I mean, there was a whole thing about data literacy three years ago, and now it's just moved to AI literacy. And the reason why I say that is Prasanna and I had a workshop on Monday, and we asked at the beginning to plot people's competence versus their confidence. I was surprised. Everyone was very competent and very confident, and then we started asking questions. And we started asking, oh, put your hand up if this, and less and less hands started going up. So my concern is that we're just accelerating past that point of training, and I think there needs to be a much bigger awareness on not just the AI literacy but the data literacy because that's what's behind it.
Jonathan Crowther:
So, as you say, I agreed, let people use it, and I'm glad to see many companies roll out their versions of AI, but nobody is rolling out prompt training. So, for instance, your core people in pharma and in other corporates, how many of those crazy trainings do you have when you're on board? I joined what four months ago. I think I still have about 120 trainings to do. But one of them, or sorry, let me res say, none of them are on prompt. None of them are on our data background or even how to access the tool. So I have to figure that out on my own.
Allison Proffitt:
Do you have a recommendation if your company doesn't have prompt training? Is there somewhere you can go for that?
Jonathan Crowther:
Yeah, so there's a great company called Data Camp. I don't know if you've seen it, yeah. It's fantastic, right? So we we did it on with our team in Pfizer before I moved. We rolled this out to all of our performance analysts to say we just bought it. It's incredibly cheap for an organization for enterprise license, and you can do so much, and you can build your own um training platform, your own training um learning curves or whatever. And um, that's very effective. Sorry.
James Weatherall:
Yeah, we built something called the Data Science Academy that everybody has access to, and DataCamp is actually behind a lot of our training courses. So I would also recommend.
Deborah Borfitz:
Thank you, Allison, and most especially our esteemed group of panelists for sharing their thoughts regarding AI literacy training. And as always, a big thank you to everyone out there for listening in. If you're not subscribed to this podcast yet, please consider going to Apple Podcasts and doing so right now so you don't miss your monthly dose of news and perspectives you'll be hard pressed to find anywhere else. And if you're up for it, I'd be so very grateful if you'd leave a rating and review while you're there. One final note if you liked today's show, and I know you did, it's only a glimpse of news and insights we have on tapped for the next Scope event taking place on this continent. More specifically, sunny Orlando, Florida. Please plan to join me there from February 2nd through the 5th, 2026, and be sure to use discount code SOT10 for a 10% discount off any current rate. For more information, visit scopesummit.com. Bye for now.