TTHealthWatch is a weekly podcast from Texas Tech. In it, Elizabeth Tracey, director of electronic media for Johns Hopkins Medicine in Baltimore, and Rick Lange, MD, president of the Texas Tech University Health Sciences Center in El Paso, look at the top medical stories of the week.
This week’s topics include predicting preeclampsia, improving ablation for atrial fibrillation, wastewater surveillance in aircraft for infectious disease, and platform trials for amyotrophic lateral sclerosis (ALS) treatments.
Program notes:
1:39 All agents negative
2:40 In trials like ALS why not use historical controls
3:40 In favor of such controls
4:10 Predicting preeclampsia
5:09 Represented 85% of the area under the curve
6:09 Representative population at large
6:40 Using AI to treat atrial fibrillation
7:40 Measure electrical activity inside the heart
8:40 Effective in those who’d had longest
9:06 Pandemic monitoring by looking at aircraft lavatories
10:09 Networks of 10-20 sites can provide information
11:10 Plug in additional information to predict
12:38 End
Transcript:
Elizabeth: Can we predict preeclampsia from a prenatal sample?
Rick: Using artificial intelligence to treat atrial fibrillation.
Elizabeth: A new method for trying to determine whether a pandemic is about to start.
Rick: And trials of agents to treat ALS.
Elizabeth: That’s what we’re talking about this week on TTHealthWatch, your weekly look at the medical headlines from Texas Tech University Health Sciences Center in El Paso. I’m Elizabeth Tracey, a Baltimore-based medical journalist.
Rick: And I’m Rick Lange, president of Texas Tech University Health Sciences Center in El Paso, where I’m also dean of the Paul L Foster School of Medicine.
Elizabeth: Rick, you’ve got two of them from JAMA. These are related to clinical trials relative to ALS and they introduce us to something different, so-called platform testing.
Rick: And 2 others, by the way, in JAMA Network publication. So, this is four trials, all undertaken within what’s called the HEALEY ALS Platform.
Let me first say these studies looked at four different agents to treat ALS, a fatal neurodegenerative disorder. The median survival from when the symptoms occur to death is usually 2 to 3 years and, unfortunately, the only treatment we have is supportive.
So, four different agents were looked at: an injectable form that targeted central nervous system inflammation, another oral medication that attacked some of the maladaptive microglial activation. There’s another agent that was supposed to improve cellular energetics and another agent that somehow is supposed to stabilize the mitochondria. Unfortunately, all four of these agents were not successful in delaying progression of ALS or in improving survival.
But as you mentioned, what’s interesting is this is the first use of platform trials in the US. Here’s how that works. If across all of those studies you have the same inclusion and exclusion criteria, you might construct a trial that looks like this. One of the agents you recruit 100 people to get that agent and you have 25 that receive placebo. And the next agent you have 100 that receive that agent and 25 that get the placebo, and you do that for all 4 agents. So at the end, each agent had 100 people that received it and you could combine all the placebo groups together to get one placebo group. It makes the trial go faster. It’s less costly. It’s more efficient.
Elizabeth: To me, it seems like it also reduces the number of people who experience the disappointment for people, particularly with life-threatening illnesses like ALS, being randomized to a control group versus getting the intervention agent. Because those things, of course, offer people hope and control groups often don’t offer that kind of hope. It’s interesting here and it begs the question for me, especially in trials like ALS, why can’t we use historical controls?
Rick: The disease progression is highly variable. So to make sure that the agents and the patients you’re giving the agents to match the general population, you have to compare them at the same time.
One of the criticisms of this study is maybe we need to extend it longer. I mean, these studies were just done for 24 weeks, and maybe if we extended the trial for longer, we might see some benefit. But the point is you have to have concurrent treatment and placebo to make sure that the patients are matched adequately.
Elizabeth: I’m going to push back on you on that because I believe that, sure, on an individual basis ALS proceeds at a highly variable rate — unquestionably, that’s true — and we have witnessed, measured, observed so many of those folks over the years that — especially in this condition that has a really short period of time between diagnosis and death. I would be in favor of using historical controls in this group. I’m just going to put that out there. You can disagree with me, but that’s what I would do.
Rick: Does it give you a definitive answer? Because sometimes, especially in a short period of time, the benefit you’re looking at may not be huge. The other thing you have to remember, Elizabeth, is you’re assuming that these agents could be beneficial. They could, in fact, be harmful and they also have side effects. So that’s why I think it’s best if you can do a trial where you can actually have a comparative group concurrently. But I understand your points.
Elizabeth: We’ll agree to disagree.
Let’s turn to Nature Medicine and let’s talk about preeclampsia. At least in the United States, African-American women experience this at a much higher rate than Caucasian women or other groups and it can have some pretty dire outcomes, including maternal and fetal death.
These folks were taking a look at risk prediction, who’s going to develop preeclampsia from prenatal cell-free DNA screening; that’s already taking place. They draw blood and they take a look at this prenatal cell-free DNA. The maternal and the fetal contributions to those can be discerned. They took a look at 1,800+ of these routinely collected prenatal DNA samples at a median of 12.1 weeks of pregnancy. They developed a framework. What’s in there? Can we develop a prediction model for preeclampsia risk?
Their model had an 81% sensitivity, 80% specificity, and represented 85% of the area under the curve. It’s not great, but it’s pretty good. If they were able to employ this in order to identify women who were at a higher risk for preeclampsia, then clearly that’s going to enable some interventions to take place.
Rick: Yeah, and Elizabeth you’re right. The value of this particular study is this is DNA we’re already collecting, making sure that the fetus looks fine — that is, in terms of the chromosomal makeup. They took a bunch of women early in pregnancy and then followed them, a number of which had preeclampsia, and they went back and looked at their particular DNA. Can we identify some differences?
There was a definite fingerprint. Then they took that information. They applied it to women early on to see how predictive it was. In individuals that have preeclampsia, it’s an issue with the placenta and the blood vessels, so this fingerprint can be very helpful.
Elizabeth: It also nails down this notion that is an endothelial dysfunction on the maternal side that’s probably responsible for the development of preeclampsia.
Rick: And by the way, this is in a population that represents what we see in it, is they didn’t eliminate women that had preexisting high blood pressure or diabetes or other maternal conditions. So, the fact that they included these women indicates that this test could really have benefits.
Elizabeth: They do note, of course, that aspirin prophylaxis is something that they say, “Well, all right, if we identify this endothelial dysfunction maybe that’s what we end up putting women on.”
Rick: You could just give aspirin to everybody, but why give it to everybody if you try to target therapy?
Elizabeth: Let’s turn to your next one and that’s in Nature Medicine also.
Rick: Using artificial intelligence to treat atrial fibrillation. That’s an irregular heart rhythm that affects about 33 million people worldwide.
You can treat that with some medications in some individuals, but most of them need an ablation. You actually deliver burns in the area around where these electrical activations occur to try to isolate them. Radiofrequency ablation is used worldwide and is one of the more effective therapies. However, in as many as 40% to 50% of people, the atrial fibrillation may recur. It’s more likely to recur in individuals that have atrial fibrillation for a long period of time or those that have particularly enlarged heart chambers.
When you deliver these burns, we use what are called intracardiac maps. We actually put a catheter in the heart to try to determine where these signals may be coming from and we have specific anatomic locations that we deliver the burns. We’ve tried to improve the effectiveness by using different anatomic locations, but what this particular study did is said, “Hey, if we can measure the electric activity inside the heart, all of this information we can feed into artificial intelligence and maybe it can tell us a more precise location of where we need to deliver these burns.”
To test this out, they took about 360 people and in half of them they delivered the routine radiofrequency ablation, and the other half they used this tailored approach. In those that received the usual activity, it was about 70%. Those that received the targeted, using AI to determine where it should be, it was 88% successful at 12 months.
Elizabeth: I guess I’m wondering, in extension of this technique, if it would actually shorten the procedure time.
Rick: It doubled the procedure time to about 178 minutes. You spend a lot more time afterwards mapping and applying additional burns. However, what happens is you’re less likely to need a second or third procedure. It was particularly more effective in those that had had atrial fibrillation the longest because those are the hardest to cure.
Elizabeth: Clearly, there’s going to be a need for longer-term follow-up of these folks to see whether or not it recurs. Then, of course, that hard outcome that we’re really interested in, does it reduce strokes in those who are at risk? Which clearly those with Afib are.
Rick: You’re right. What you hope to do is get someone out of atrial fibrillation, put them into a regular rhythm, to get them off of blood-thinning medications.
Elizabeth: I think we’ll see more of this.
Finally, staying in Nature Medicine, we’re taking a look at pandemic monitoring with global aircraft-based wastewater surveillance networks. Kind of a fancy way to say, “Gosh, we’re going to go in there and check in the lavatories, and see whether or not we’ve got all these emerging viruses in there.” And can that be an early sentinel for “There’s something that’s coming, folks, that we need to be paying attention”? That’s my colloquial impression. What do you think?
Rick: Elizabeth, that was pretty good. I mean, this is bathroom talk we’re talking about.
Elizabeth: That’s exactly right. That’s what we’re talking about.
So, taking a look at aircraft wastewater surveillance, it’s amazing how really hideously complex this computational model is. They try to take into account all of these different factors. Respiratory diseases of varying transmission potential. Where are you in the infection process? Have you just been exposed? Are you fecally shedding the virus or not? Are you convalescent? All kinds of other factors, like whether or not the lavatory was adequately cleaned in between flights.
So what they found is that networks of 10 to 20 strategically placed wastewater sentinel sites worldwide can provide additional timely situational awareness and function as an early warning system that it’s possible that a pandemic strain of something is starting to make its way around the world.
They find that increasing the number of sentinel sites beyond this critical threshold really does not significantly improve the capacity of this particular model. They conclude that it could notably shorten detection time as these things start to come across the transom.
Rick: They picked 20 sentinel airports that they selected for high international passenger volume and also for geographical diversity. That model can plug in what the agent is, what the expected doubling time is, how many people were on the flight, where they flew to, where it was first detected, and they can use this to predict where the transmission is going to be globally. You can plug in all this different information and it provides additional predictive value. Pretty powerful model.
Elizabeth: They do note that their time to first detection varies widely from 14.2 days in Geneva to 66.5 in Kalemie. In Europe, it’s average 15 to 25 days before their first detection. That’s a pretty variable number. I’d like to see that honed in. And maybe if this gets rolled out and real data starts to get plugged in, we’re going to see what it really is. Because, gosh, an awful lot of stuff can take place in 66 days, a lot more infection.
Rick: You’re right. You do want to shorten the pathogen detection time and part of that detection relies upon how quickly it’s done in whatever the country could be. Do they recognize it? Do they have national ways of both reporting it and providing that information? So, there is a lot of things that this particular model can take into account.
Elizabeth: And they validated it using the SARS-CoV-2 alpha variant.
Rick: It’s clear that this model — this is a framework for modeling wastewater surveillance at airports — will help us to support public health making decisions through planning and through surveillance both.
Elizabeth: On that note then, that’s a look at this week’s medical headlines from Texas Tech. I’m Elizabeth Tracey.
Rick: And I’m Rick Lange. Y’all listen up and make healthy choices.
Please enable JavaScript to view the