Subscribe
Share
Share
Embed
Communicable takes on hot topics in infectious diseases and clinical microbiology. Hosted by the editors of CMI Communications, the open-access journal of ESCMID, the European Society of Clinical Microbiology & Infectious Diseases.
[00:00:00]
Erin: Hello, and welcome back to Communicable, the podcast brought to you by CMI Communications, ESCMID's open access journal covering infectious diseases and clinical microbiology.
My name is Erin McCreery, and I'm the senior director of infectious diseases strategy at the University of Pittsburgh Medical Center in Pittsburgh, Pennsylvania, United States. I am joined again today by four other fantastic CMI Communications editors to discuss the clinical trials presented at ESCMID Global 2026.
This is part two of a two-part series, so if you missed the last episode, please be sure to go back and listen, where we covered five trials, and today we're gonna cover five more. We are going to skip bios today. I am joined, by Josh, Marc, Anne-Grete, and Angela from CMI Communications, , and we're gonna go right into our get to know you question.
Have you ever been, or would you ever go skydiving? Marc, what's your gut reaction to that?
[00:01:00] Well, I can't use the exact words, but no. No.
Erin: You can say it, and we'll bleep you out.
Marc: Canyon, entering a cave, underground speleology or something like that, and this. No way.
Erin: Oh, yeah, Marc had some choice words in our outline for this experience. my answer is I have never been, I will probably never go, but I wish I was the kind of person that wanted to go skydiving. Like, it actually seems really cool, and I think if I- if someone literally pushed me out of the plane, I'd enjoy that part.
I just, I know I would never be able to do it, like, to get up there in an open plane and take the step off. I'm too much of a scaredy-cat. Angela?
Angela: Answer is no.
Erin: Josh.
Josh: short and sweet. I think it was Emily who came up with the question,
Erin: our
Josh: girl,
Erin: Emily.
Josh: Yeah. Um, yeah. My answer is, kind of similar to yours.
I can see why it would be fun, but I just- I'm not brave enough to do it. And I know why. It's because I don't have the novelty gene Have you guys heard of this, the dopamine 4 receptor [00:02:00] gene? polymorphisms in this gene have been shown to correlate with, exciting behavior and gambling and high-risk activity. So people that skydive are much more likely to have th-this gene polymorphism than people that don't. And I don't have it- So- ... I'm pretty sure ...
Angela: and they don't self-eliminate. Like, they don't select themselves out of
Josh: the- Yeah, exactly.
You're right ... gene
Angela: pool. You'd
Josh: think it would be an e- You're keeping it going ...
Angela: but it's an evo-
Josh: it's an evolutionary advantage as well, right? Because taking risks- Mm ... sometimes pays off. Mm.
Angela: So nuanced. That's awesome. I love it. Mm-hmm. Anne?
Anne: Well, actually, funnily, I had a gift card to do, skydiving, like, 10 years ago, and I never went.
I don't know what happened. I think because I moved to the Netherlands and something happened in between. And I felt like this was a sign I shouldn't do it. So I never went, but I did want to. But then I was, 10 years younger then, so now I would not do it. No.
Angela: So out of all of us, she's the one who's actually, like, trialed this experiment.
Anne: I've actually planned it, yeah. Nice. [00:03:00] But then it never happened, so. I
also the person who gave you that gift card is listening and like, "What the heck?" You know? Why didn't you- ... why-- never getting a gift again.
Erin: Hopefully. That, that was
Josh: expensive.
Erin: Yeah. Oh. All right. Let's get right into our trials. We got five more trials coming at you today. We're each gonna present one. If one of us has a conflict of interest with the trial we're discussing or choosing to comment on after one of our colleagues presents it, we will declare that before we have our discussion, or we simply won't comment, and all of our conflicts of interest are in the show notes for you.
Marc, take us away.
Marc: Yes. I will start off with the ProCalBan trial, which stands for the use of PCT, procalcitonin, as a point-of-care test to guide de-escalation of empiric antibiotic therapy in adult patients with sepsis. And here it comes. In a tertiary hospital in Bangladesh. It's a randomized, controlled, open label trial and was presented by Forhad Chowdhury.
Now, I must admit that I do have a emotional conflict of interest related to po- [00:04:00] procalcitonin, uh, especially in the Western world, where this has been fiercely debated, especially for guiding treatment in patients with community-acquired pneumonia, and I have always been on the, the contra side. No one has ever-- could ever convince me that this is really a, uh, a beneficial addition to our armamentarium, especially not since we have CRP, which is much more, let's say, much cheaper.
So now- Now we completely change the setting. We're not-- no longer in the Western world. We now go to Bangladesh, where 30 to 60% of the bacterial strains are considered multi-drug resistance according to our standards, where there is massive antibiotic use. And there, in the words of the presenter, there is a lack of quality microbiology labs, so not providing sufficient information to guide clinician and the choices about what to treat with and how long to treat are guided by clinical judgment only.
The utility of [00:05:00] biomarkers, including CRP, to guide antibiotic use is very, very much understudied. So after a, let's say, review study up front for all the possible biomarkers to be used, they made the choice to use procalcitonin as it was well-studied in high-income studies, where it is mostly studied.
That was one of the reasons, and it was available. In high-income countries, it is mostly studied in ICU settings with low AMR prevalence, but not yet has it been studied in low and middle-income countries. And in Bangladesh, critically ill patients are generally treated in general wards, and there's a high prevalence of AMR.
And the presenter actually showed some pictures on the setting where these patients were treated, which was completely different than where we treat patients. And there was also enormous crowding and, very primitive conditions. So that's where the study was executed. And the intervention was that patients that were treated on antibiotics, treated-- [00:06:00] started with antibiotics, had their PCT measured daily.
And if the level fell below point five nanogram per ML, there was a non-binding advice to deescalate antibiotics. That was the intervention. The control was the standard care. And the follow-up of patients was until 28 days after discharge from hospital. Now, they screened more than 1,000 patients and actually recruited 532 in two groups of 266 patients.
Marc: It was a population with a median age between around 42 to 45 years, so a young population with mostly respiratory tract infections, about 50%. 25% had urinary tract infection, and enteric fever accounted for 12%. 18% of these patients, despite the suboptimal microbiology, had bacteremia, mostly E. coli or Klebsiella.
The primary outcome was the length of treatment, and the length of treatment had a median duration of 4.7 [00:07:00] days in the intervention group compared to nine days In the standard of care group. The secondary outcome was the days on therapy. So, if you have two antibiotics on one day, you have two days on therapy, the DOT, and the median values of those were 7.2 in the intervention arm and 13.2 days in the standard of care.
Marc: So there clearly was a reduction in the number of antibiotics per day and in the number of days, antibiotics were given. The median length of stay of patients in the hospital, though, was comparable, although statistically they said it was a significant difference, but it, it was 4.03 versus 4.73, so a half a day difference between the groups.
And the mortality, as observed in the follow-up period, was 8.6% versus 7.9%, so not a difference. So the conclusion was that in this apparently ongoing project, the procalcitonin was a highly promising tool in this low and middle income [00:08:00] setting to achieve a clinically relevant reduction in antibiotic use and that it also was safe for relevant patient outcomes, and that's what I like very much about these types of studies.
We can easily say, " We can reduce antibiotic exposure, but at the same time we must demonstrate that it is still safe for the patients." They had not a formal non-inferiority co-endpoint, but here at least it was very obvious from the data that there was not a big difference in outcome. And that was a question I really would like to have addressed to the presenter, but this was the last presentation in the session, and it looked like the chair had to catch a plane, so he did not allow any questions and immediately closed the session.
So I really would have liked to ask how this, let's say, non-inferiority for safety will be demonstrated, as I assume that its trial is still ongoing. And the other question is, how would this change their practice now? Because in the abstract, but it was not presented, they also [00:09:00] showed us the price and the costs of both strategies per patient day.
And there the PCT testing test, strategy, including the antibiotic use, was more expensive, 20% more expensive than in the comparator. And with no difference in length of stay, which is the major cost driver for cost effectiveness, let's say for cost analysis in hospitals, this would imply a 20% higher costs for a strategy.
So it would've been very interesting to hear from the presenter how this is, taken into consideration in his country. But actually, I think it was a very well done study, nicely presented, and, giving it quite a different, view on the possibilities of biomarker-guided therapy in other settings than we are used to working.
Josh: Yeah. I guess one thing I noticed on this is that the duration of therapy in the control arm was a median of nine days, meaning, you know, 50% of people got more than nine days of therapy. and these are for patients with pneumonia, [00:10:00] UTI, and bacteremia for the most part. And we know that seven days of therapy from the balance trial is sufficient for those with bacteremia and three to five days for those with pneumonia.
So their baseline antibiotic use is already probably too long. and therefore, yes, they've shown a shorter duration with procalcitonin but I wonder if there would be the same impact if they changed their practice, you know, to accord- Yeah ... with, the latest evidence.
Marc: I could also not, address the question why they decided to use procalcitonin and not CRP because I think CRP would be the, the cheaper option.
also that was not clear to me.
Erin: Yeah. I love that with these podcast episodes we have now made emotional conflict of interest a thing 'cause I'm gonna use that all the time. I also have an emotional conflict of interest with this. I am team-- I joke but we've coined this my friends at the Mayo Clinic and I are team anti-calcitonin because the Mayo Clinic did this beautiful huge real world database showing that procalcitonin guided therapy patients actually in the wild usually have worse outcomes.
Um, [00:11:00] and here there was actually a higher mortality signal in the procal treated arm despite less antibiotics. and I think when you're trending procals it may actually have this like antithetical effect. and we've removed procalcitonin at my institution because of what Josh said. So I think in the early 2000s when antimicrobial stewardship wasn't a thing and we didn't have a lot of infectious diseases trials to guide duration of therapy, I think it probably was super helpful and it was what we see in this trial.
We saw durations going from, you know, 14 to 12 days or 10 to eight days but now no one should get more than seven days for most things. and if you need more than seven days you have a legitimate complex infection that it doesn't matter what your procal is. So I think now we have better data and you don't need a test to tell you what duration of therapy you should give.
All right. I'll go next. I am going to review the beneficial trial which is a multicenter randomized clinical trial to study the impact of bedside model informed precision dosing of vancomycin in severely ill neonates and children. so this was presented by Peter De [00:12:00] Cock from Belgium and let me just say, I mean, they did an RCT in neonates.
I can't even imagine how challenging that is to consent parents into a study like this. So unreal effort that was not talked about at all during the session but when I saw the title I, I don't know how many RCTs exist in neonates at all and so I think this is just Unbelievable effort to start off.
But why did they do this? So in 2020, the vancomycin guidelines were updated, and that shifted from trough-based monitoring for vancomycin to AUC-based monitoring in both adults and children. And by and large, this was a safety move because if you have a, a trough greater than 15, you will have an AUC greater than 400 just from the way math maths. But if you do AUC-guided dosing, you are much more likely to be under that toxicodynamic ceiling of AKI with vancomycin. So AUC-guided dosing, really a safety thing. So what is model-informed precision dosing for our listeners who are not PK nerds [00:13:00] like Anne Grete and myself?
So this is essentially you take a population pharmacokinetics models, you take data from all the humans in a particular kind of space, You take that model as kind of what we call a prior, and then when you apply it to the patient in front of you, you can also take into account patient-specific factors such as their serum creatinine and their weight, and you can use this to calculate or estimate your empiric, like your dose zero initial starting dose.
So we call that the a priori dose optimization. And then you can do therapeutic drug monitoring to get patient-specific concentrations, layer that on top of the model as well, and that's a posterior dose optimization. So the proposed benefits to model-informed precision dosing is that you are getting the more correct dose earlier than you would with a nomogram or some other kind of, you know, package insert or labeled dosing approach, because you don't have to wait to steady state to dose optimize.
You don't need [00:14:00] fixed time points for sampling, so that's practically very valuable, especially in pediatric patients where drawing blood is so precious and you have to kind of negotiate for what labs you're drawing with other clinical service lines because you can only take so much blood. So with model-informed precision dosing, - you don't have to have an exact trough, and so you can time your therapeutic drug monitoring with other biochemical labs that are being ordered per routine care.
And so I think that's a huge advantage in pediatrics. And so when the 2020 guideline, VANCO guidelines came out, they came out in April of 2020, so honestly, like some people missed them. We were all a little busy at that point in time . And it was terrible timing because we went from one level to two concentrations in a COVID world where we were trying to minimize blood draws and minimize access into rooms.
So it was like really bad timing to be like, "Oh, hey, we're gonna switch to two concentrations in everybody." But there was a lot of dialogue about the switch, but the bottom line was actually that AUC guided dosing did not have evidence of superiority to trough-based dosing. [00:15:00] And people don't like to change, right?
And so this group astutely said, "Let's do a randomized clinical trial to show if standard dosing and conventional therapeutic drug monitoring is different than model-informed precision dosing They used InsideRx as their platform and, they randomized patients to these dosing strategies, which is so cool.
So the standard arm, they got the starting dose per the institutional guideline. They dose adjusted on the institutional nomogram, and they did, you know, patient-specific trough values for the most part, what we were all doing. For model-informed precision dosing, like I said, they used a Bayesian software and a pharmacokinetics model.
They used the Collin model for neonates and pediatrics and then, like I said, patient factors go into that model. They did do TDM faster in these patients because there was no need to reach steady state for sampling, and they specifically asked the physicians who had patients randomized to this arm to order the therapeutic drug monitoring with other labs and not have a separate blood draw.
And their target was an AUC of 400 to 600. They [00:16:00] did this at seven hospitals, 14 wards, randomized one to one. There was partial blinding, so the statistical analysis people were blinded, but the clinicians obviously were not because that would be impossible. They had 150 patients in each arm, 300 kids. This is wild.
60% of them were neonates. 60%. I, I can't get over that. Like how... and kudos to these parents. You know, if you happen to have been a parent who consented your neonate into the study, I cannot even imagine how hard that is when you have a baby in the NICU and then you're like, "Sure. Let's-" Learn from them.
I mean, that is just amazing. So they included neonates and children up to 18 years old, either in ICUs or on oncology units, and they had to have planned to start continuous or intermittent vancomycin with a suspected or confirmed gram-positive infection. If you were on ECMO, had severe AKI or known CKD, you were out.
Not gonna lie, I don't love that. I think it would be much more applicable if they included the AKI and the CKD because practically there's a lot of pushback on investing in these kinds of softwares [00:17:00] because they're unreliable with fluctuating renal function as is anything, But that's most of our hospitalized patients.
And so to kinda just like take out the bucket of AKI, it's like that's, that's sick patients. So I would've liked to see how these softwares play in patients with AKI compared to our standard care. But alas, the primary endpoint was the proportion of patients receiving the target AUC between 24 and 48 hours after the start of vancomycin, and that was achieved in 53.9% of patients in nomogram dosing versus 71.8% of patients in the model-informed AUC dosing, which was an absolute difference of 18.9%, which was significant.
Their secondary outcomes were patients with new or worsening AKI, mortality and reaching targets after 48 and then after 72 hours. So kind of these buckets of timestamps day by day. There was a 22% difference in favor of model-informed precision dosing in achieving your target at 48 to 72 hours and at greater than 72 hours.
So it was like pretty consistent that you were more likely to achieve your [00:18:00] target. And then here's an important one. AKI was 16.9% in the standard arm versus 12.4% in the model-informed arm, which was not statistically significant, but I would think we would all think a 5% difference is clinically significant.
Indeed, that's sometimes the safety threshold used as endpoints in other trials for power calculations and whatnot. So I find 5% to be actually quite significant. A tertiary endpoint was the number of blood samples and the cumulative vancomycin dose, and I really love that they looked at this because we see that anecdotally, but it's true, and the author saw this too, that using model-informed precision dosing led to quite large decreases in cumulative vancomycin dose.
So you can use much less drug when you're doing AUC guided dosing, which somewhat cost justifies the software quite frankly, 'cause just the raw savings in milligrams of drug often kind of lend to that. So their conclusions were AUC dosing is feasible. It led to sustained improvements in [00:19:00] PK/PD target attainment.
It had a very low risk of harm, no extra patient burden, potentially less acute kidney injury, and definitively you use less drug. Thoughts?
Anne: Yeah, I can also say Peter is amazing. I know he does a lot of studies in children, and it's really nice to see. Also nice to see the MIPD, is useful here because indeed we debate on that a lot. and there is actually even in adults we do not have that many studies. So I really like this study as well, and, using less drug and having less, toxicity, I mean, that's already great.
having a huge impact of MIPD, I think is really hard. I think it's really hard to see a bigger clinical benefit here.
Yeah. from an adult medicine point of view, I thought it would be kinda obvious that using, these Bayesian modeling approaches would get you to a target, level faster than using [00:20:00] the clunky, conventional methods.
Josh: But I guess having demonstrated in a randomized trial is valuable, particularly in a special population like neonates. and I agree, when one switches over from using trough-based dosing to using AUC-based dosing using a Bayesian model, which we've done in my hospital, it's quite noticeable how much lower the doses are.
It's a bit hard to kind of be comfortable with it when you first start doing it.
Anne: I think it also makes your practice easier because you can take any sample. Yeah. You do not need to wait steady state. You do not need to, take a particular trough or you can really- Yeah
like, if you have a good model- In- ... you can actually ease the practice for nurses ...
Josh: in some ways that's kinda harder though, because for decades we've been, like, hammering into people, "You've gotta take the trough level. You've gotta take the level just before the dose at, you know, this time in the morning, et cetera."
now they're confused. They're saying, "Wait. What? So I can take it any time? I don't understand." So there's a bit- No, fair enough ... of change management to, to be done.
Erin: I [00:21:00] will say though that- And not with
Anne: other drugs. With other drugs please keep doing it.
Erin: Well, but that's true with... Okay, so like isavuconazole has a half-life of, like, 400 years, right?
And so once isavuconazole's at steady state you can check its concentration at any time. It doesn't matter, right? Peak, trough, they're all the same. I find that when I tell clinicians when they call, they're like, "It wasn't done at exactly at trough. Is it okay?" And I'm like, "Yeah, it's fine.
No worries." they don't like that answer. They're uncomfy when you're like- Mm-hmm ... "It's fine," you know? Yeah. So that's a valid point. The other thing I'll say, Josh, to your point, is actually one of the original observational studies in vanco AUC dosing was done out of Detroit- And Natalie Finch, a pharmacist, was the author.
And So the DMC Medical Center had been doing AUC dosing for quite some time long before the guidance. And one of the pearls of that paper, it was in AAC, I think 2014, was they, you know, they looked at like time to AKI and whatever. But in like one of the supplementary tables they showed the cumulative vancomycin dose and it was substantial, like two grams a day less.
And that has always been something I've, thought was so neat. And so it's cool to see that pan out in a, in an RCT as [00:22:00] well. but staying on the gram positive train, Anne, do you wanna start the next study?
Anne: Yeah. I'm very lucky to present SNAP to Josh, Josh in this call.
So this is a really cool, a nice trial that was, presented at the late-breaking clinical trials in sepsis management. by Asha Bowen from Wembley, Australia. the study, is called Adjunctive, Clindamycin for Treatment of Staphylococcus Aureus Bacteremia, a randomized control trial within the S.
Aureus Network, Adaptive Platform, the SNAP platform I don't think needs a long introduction, but basically there are the domains that are the different trials, run at the same time and silos look at different antibiotic susceptibilities. this clindamycin trial, was in the adjunctive treatment domain.
It was basically clindamycin versus no clindamycin. the thinking behind the trial was basically, due to the virulence factors that, contribute [00:23:00] to Staphylococcus aureus pathogenicity. So, for example, the hemolysins, nucleases, proteases. limiting expression and exotoxin release could reduce, Staphylococcus aureus virulence and improve outcomes.
So because clindamycin, is an inhibitor of, ribosomal protein synthesis, . Using, this drug leads to reduction in ex-exotoxin production. so they also did a pilot RCT in 34, patients, and there was no difference in the primary, outcomes, the seros free survival at day 14.
However, the secondary outcome, which was the mortality at day 90, was 0% in clindamycin, so 0 out of 17, and 24% in no clindamycin, so 4 out of 17 So the inclusion criteria for this, , was a Staphylococcus aureus in blood cultures admitted to a participating hospital.
Exclusion was, more than seventy-two hours from index blood culture, polymicrobial [00:24:00] bacteremia, known positive, blood culture, seventy-two hours to hundred and eighty days previous, end of life care or clinicians, deems not appropriate. So there was also clindamycin specific criteria, which was no domain, specific inclusion criteria.
However, the exclusion criteria was obviously the allergy to lincosamides, receiving clindamycin or linezolid that cannot be ceased or substituted, necrotizing fasciitis, current CDAD, known CDAD in previous three months, current severe diarrhea, or less than four hours since platform enrollment.
So the intervention was then clindamycin or no clindamycin, so that was five days of clindamycin. the intravenous dose was six hundred milligrams per dose, three times a day, oral was, four hundred and fifty milligrams per dose, three times a day.
The primary outcome was all-cause mortality at ninety days, and there was a pre-specified subgroups, uh, for ICU [00:25:00] cohort, and outcome was, by clindamycin susceptibility. So, the secondary, outcome was all-cause mortality at fourteen, twenty-eight, and forty-two days, microbiological treatment failure, all-cause diarrhea, Clostridium difficile diarrhea, uh, CRP change and some other, outcomes there.
Anne: And there was an interim analysis after every five hundred participants, uh, in platform. So again, very impressive, of course, this platform trial, the numbers of the patients are, huge. So this, study that they presented was done from February twenty twenty-two to July twenty twenty-five.
This was seven thousand four hundred and thirty-two randomizations more than 14,000 patients screened and, uh, four thousand one hundred and eight participants were then in the adjunctive clindamycin trial. between, the participants, there was obviously then randomization between the groups and the clindamycin and no clindamycin group were comparable within the [00:26:00] nine countries.
comorbidities were similar between the groups. primary, outcome of all-cause mortality at ninety days, for clindamycin, the intention-to-treat, population was 16.9%, and no clindamycin was 15.2%. So the adjusted, odds ratio was one point fifteen, with a confidence interval of zero point ninety-five to one point thirty-nine.
So there was a low likelihood of superiority. The posterior probability was, zero point zero seven, futility zero point zero zero four. So in the ICU subgroup, clindamycin, had 30.4% mortality, and no clindamycin had 24.7%. And the adjusted odds ratio was there one point forty-seven, and likelihood of superiority also only four percent.
secondary outcomes, the mortality was higher also, uh, [00:27:00] at twenty-eight, forty-two, and ninety days in the clindamycin group. another very important outcome was looking at the Clostridium difficile diarrhea. clindamycin does not drive, this, Clostridium difficile diarrhea more than other antibiotics, with both arms at one point nine percent.
The all-cause, diarrhea was, higher in the clindamycin group, and, SIRS and CRP changes were also no different between the groups. at this point, there was no susceptibility data actually, analyzed or still to be performed. however, the presenter did mention that it is unlikely to affect the results well, the overall conclusion was that for Staphylococcus aureus bloodstream infection, there is no role for adjunctive clindamycin.
likely it can be harmful for patients, currently, its unknown impact on gut microbiome because of obviously the increased diarrhea in the clindamycin group. so [00:28:00] adjunctive antitoxin antibiotics are unlikely to have benefit.
They also reference another trial looking at this. I'm gonna throw this to Josh to comment, on this, study.
Josh: Yeah. . Thanks, Anne. it was really exciting to get these results when we found that the statistical trigger of futility had been hit and, eventually for us to see all the results and to present them. so I obviously have a conflict of interest being, one of the global co-leads of the trial, along with Steve Tong.
But also worth mentioning, there are literally hundreds of investigators around the world that have worked on this trial, including, people that are on this podcast, Mark and Erin and, and Emily, who would have been here. So, I guess a couple of takeaways from this. One is, uh, it's very clear that clindamycin was of no benefit in the population overall and was of no benefit in the ICU subgroup.
And, in fact, yeah, the likelihood of superiority in the ICU subgroup was four percent, which really is [00:29:00] mathematically exactly the same as the likelihood of it being inferior was ninety-six percent. The likelihood of, mortality being higher in the clindamycin, in that ICU subgroup was ninety-six percent based on the Bayesian posterior probability model.
So that's pretty worrying. and we don't know why that is the case, if it's even true, but we think it may be something to do with disturbing the gut microbiome because there's more and more emerging data that, exposing people to unnecessary, anaerobe-disturbing antibiotics is bad for outcomes animal models and human models and human observational studies of sepsis.
And that's why the latest Surviving Sepsis guidelines has recommended avoiding, meropenem and pip-tazo if, you don't need them. And then the other thing we were kind of, I guess, pleasantly surprised about is that the rate of C. difficile diarrhea was, like, exactly the same in the two groups, and it was quite low, [00:30:00] 1.9%.
and When we were setting up the trials, a lot of, colleagues, at some sites and even whole regions said, " There's no way I'd randomize patients to get clindamycin 'cause of C. diff." as far as I know, this is the largest ever RCT of clindamycin for any indication, and There's very clear data here that if used in these doses for just five days, it doesn't increase your risk of C. difficile more than other antibiotics. so yeah, that's all I'll say here. there's more, analyses ongoing, and we're, preparing the paper for publication.
Do you think, Josh, that there could be, like, more subgroups here?
Anne: Because obviously the population is really heterogeneous, big population. would there be benefit in looking at some other subgroups, do you think?
Josh: It's a pragmatic trial, so we've purposely not collected a whole lot of detailed data, and we're also very careful to limit our subgroup analyses to a priori ones, so to avoid kind of false discovery.
the key subgroup that [00:31:00] we hypothesized would benefit the most was the subgroup that are critically ill, the ICU subgroup. other subgroups such as necrotizing pneumonia, for example, we don't really have sufficient data collection to look at that.
Anne: thanks, Josh, for the commentary,
Marc: so this notion that maybe the disturbance of this anaerobic flora has an effect on patient outcome that you will only see with so large numbers actually is a concept that was proposed in the 1970s by a Dutch medical microbiologist, Dirk van der Waaij, and this actually was the basis of what is now called selective digestive decontamination.
It started with the notion that if you eradicate the anaerobic flora from gnotobiotic mice, mice without other flora, that it will kill them. So it will make them more susceptible to invasive infections with other bacteria if the anaerobic flora is disturbed.
Marc: It's the concept of colonization resistance. [00:32:00] So maybe, maybe, maybe that's the reason.
Anne: Thanks, Marc.
Angela: I will actually go even further, Marc. Marc wrote a very nice pro/con piece with Josh, right? Jo- Was it Josh and Marc? You guys wrote a pro/con piece on S
Josh: Yeah, I think so. All
Angela: right. Well, yeah. It's good. Plug it.
Okay. So we've been talking about gram positive so far, the vancomycin and the, staph aureus, and now we're gonna go to gram nothing.
Josh: The mycoplasma. Ooh. Okay. That was good.
Angela: So- That
Josh: was good. Good enough ... Angela, do you wanna tell us about this mycoplasma pneumonia trial?
Angela: I do. I really do. So this next trial is entitled Adjunctive Betamethasone Treatment of Hypoxemic Adults Hospitalized With Mycoplasma Pneumonia, Community-Acquired Pneumonia: An Open-Label, Multicenter, Randomised Controlled Trial. It was presented by Dr. Carl Hagman of Gothenburg, Sweden in the ESMT Global's new late-breaking research from The [00:33:00] Lancet session.
So this trial has been published already. You can find the entire report in Lancet Regional Health. So as we know, there are now several trials with conflicting results on whether steroids as adjunctive therapy improve clinical outcomes in people with pneumonia or not. and in fact, we are going to be doing an episode on this.
That'll be coming up not too far from now. the theory behind this, of course, is that our own inflammatory reaction in the lungs is what needs subduing. And since the pathophysiology of severe community-acquired pneumonia with mycoplasma pneumonia involves an exaggerated host response, these investigators thought they would have the right population to test their hypothesis that corticosteroids would be beneficial.
So in this trial that took place in eight Swedish hospitals, 70 hypoxemic adult patients were randomized one-to-one to either standard of care, which was antibiotic therapy only, [00:34:00] or standard of care plus five days of adjunctive betamethasone, dosed at three milligrams on days one and two and two milligrams on days three to five.
And the antibiotic in both arms was standardized to doxycycline once patients were enrolled. So the primary outcome was a continuous one, which I love. Time from randomization to regression of hypoxemia, which was defined as an o-oxygen saturation of at least ninety-three percent and a respiratory rate that was below twenty without needing any supplemental oxygen.
So to get into this study, patients had to be eighteen years or older, non-pregnant, and they had to have at least an X-ray or CT scan confirmed pneumonia with an airway sample showing mycoplasma pneumonia by PCR. And patients had to be hypoxemic to start. And again, that, cutoff was at ninety-three percent. Main exclusion criteria were an alternative airway pathogen, asthma and COPD as was [00:35:00] compromised immunity, which even included diabetes.
These patients were followed for fifty-six days, and there were several secondary outcomes, including a patient-reported symptom questionnaire, which is called the CAP score. So overall, two hundred and twelve patients were assessed for eligibility, of which seventy were randomized.
Thirty-six were assigned to betamethasone and thirty-four to standard of care.
So they nicely report intention to treat results. This population was fairly young, median age of thirty-six in the betamethasone group and thirty-eight in the control group. men than women, fifty-three percent men in the betamethasone group, sixty-two percent men in the control group. They were fairly healthy.
Sixty-five percent of patients had no comorbidities. Median symptom duration was eleven days, quite long, and patients were randomized at a median of two days after admission, so they were getting to them pretty early. So what were the big results? Time to regression of hypoxemia [00:36:00] was shorter in the betamethasone group, with the estimated median time being two point three days in the betamethasone group and three point six days in the control group. This leads to a hazard ratio of one point eight, with the ninety-five percent confidence intervals between one point one and three point zero. or by P value, it was a P value of point o two So small groups.
The length of stay was also significantly shorter in the betamethasone group. Hazard ratio of two point one with ninety-five percent confidence intervals being one point two and three point four. But the CAP score didn't differ significantly. Adverse events were very few. There were essentially no safety signals.
So all in all, this is basically more evidence for the pro-arm in the steroids or not debate, but it's a nicely pathogen-specific, argument here. I think overall, it's quite a nice, elegant trial. it does, of course, come with some limitations. It's very small. the authors do acknowledge [00:37:00] them well in their full publication, which we have access to.
The biggest limitation, of course, is the open label design coupled with a primary outcome that does look objective at first appearance but could easily be a little influenced by someone with a bias. As the authors themselves note, they say, you know, you never know, an unblinded nurse could have initiated earlier oxygen weaning and monitored oxygen saturation and respiratory rate more frequently.
So there are some things to keep in mind as you look at their results. But overall, I would say it's a very nice addition to the repertoire and certainly more fuel for the debate. Marc, tell us, what have you got?
Yeah. So I'm, puzzled by the, extreme effect on the length of stay with an hazard ratio of two point one.
Marc: what were the actual numbers of these days? Because effect on the length of stay is even longer. And let's say the, the hazard ratio is even higher than for the, the shortness of breath or whatever.
Angela: So estimated median duration of [00:38:00] hospitalization was two point nine days from randomization for the betamethasone group and three point nine for controls. a full day. Yeah, so a full day, which, we can safely say has to matter to patients.
I mean, the worst thing is- Yeah ... being stuck in the hospital, right?
Yeah.
Angela: I think overall-- I would say this is an open label trial with, not really an objective primary outcome, I really people could have been behaving differently And that could have led to people going home earlier.
Marc: Well, s-steroids make you feel better.
Angela: They do. They do. Yeah. the counterargument would be, yeah, that's the point. Look- Yeah ... we get to go home- That's why- ... whether we're biased or not.
Marc: That's why we give them.
Josh: I guess one thing I'd add is this is a very niche group of patients, right?
Like, most people who get mycoplasma pneumonia do not get hospitalized or get hypoxemic. they have a headache and a dry cough, and they're treated as an outpatient. So, obviously these results do not apply to those patients and should not be applied to those patients. [00:39:00] Secondly, most patients with mycoplasma pneumonia are never diagnosed with it, um, because they don't have a PCR done, or they may get diagnosed by serology later on.
and thirdly, excluding people with diabetes seemed a bit weird. Like, I, I almost never have any patients in hospital that don't have diabetes. Although I guess these are- Well, these were
Angela: younger ...
Josh: these are young people. Yeah. Yeah. Mycoplasma tends to affect a different population. I think it's nice.
It, as you say, adds to the weight of evidence, supporting steroids. There's, some caveats here, but I mean, I think if I had a young non-diabetic person with PCR-confirmed mycoplasma who's hypoxic, I'd probably give them steroids. But, you know, I can't remember the last time I saw that exact-
Angela: type of patient. No, fair enough. Fair enough. that's always how representative is this trial, let alone this patient? But on the other hand, I like that we're moving to pathogen-specific questions and investigations, and I think we do need to be more individualized in general [00:40:00] in our thinking and our approaches to patients and to their therapies, you know?
I do think it's quite elegant in that they say, "Look, we're going to find the population where we could show this difference," you know, on a low budget. It's a very- Mm ... small trial. And I think that is a, a very smart approach, right? We often do these big trials where we dilute out any signal because we're including everybody.
Josh: Yeah. Yeah. And to be fair to them, a lot of multiplex PCR panels are available now that do include mycoplasma. More and more. So yeah.
Angela: I know that there are periods and places where there are epidemics of mycoplasma, right?
It does tend to be very sporadic, I know there was a quite a big outbreak in Israel, right? 10 years ago. Yeah. And they, they really were able to put out quite a lot of output.
Josh: Yeah. That's a good point. It tends to come in, epidemiological clusters.
so we've gone through some gram-positive trials, then we've gone no cell wall, and so let's end up on some gram-negative focused, work now.
Josh: So, I'm gonna talk about the [00:41:00] EMBRACE trial which, uh, I noticed when I was reviewing this that Mark Bonton is the senior author on the abstract, and, turns out that he was chair of the study steering committee. And also Angela was involved in, the earlier phase trials developing this vaccine and in this trial. So to both of them, kudos for like a huge trial which, had a definitive result, which we'll talk about in a sec. But also, no thanks to you because it was a very hard trial to do as a site investigator.
That's my, not conflict of interest, but involvement was I was in-- a site investigator in this trial, and it's probably the most difficult, hard work trial I've ever done. we can talk about that at another time. But, anyway, so the trial looks at people, with recurrent UTIs or at least with one, at least one previous urinary tract infection.
And this is a common and difficult problem, and our current tools for prevention are pretty limited and suboptimal. we sometimes end up putting people on long-term suppressive [00:42:00] antibiotics as much as we don't like to do that. methenamine hippurate has some good evidence but a modest benefit.
So in that, context, people have been trying to develop, other strategies such as vaccines. most UTIs are caused by E. coli. and so Johnson & Johnson, correct me if I say anything wrong, Angela. Johnson & Johnson developed, this vaccine that's directed against the nine most common serotypes of E.
coli. we'll just refer to it as their nine-valent, E. coli vaccine. so this was a huge trial. It was a registrational trial, i.e., with the aim of registering this vaccine. so it was very rigorous. they enrolled people aged over 60 years who'd had had at least one UTI in the past two years.
And then I think the protocol was changed partway through to enrich the trial for people with at least one additional risk factor for recurrent UTIs. people were excluded if they had immunosuppression or were [00:43:00] on long-term antibiotic suppression, which, from my point of view, a lot of the patients I see that I kind of was my motivation for getting involved in this trial couldn't go in the trial because they were on long-term suppressive antibiotics, or they were immunosuppressed.
so the trial was terminated early at an interim futility analysis. I'm not sure on what the total planned sample size, but, at this interim analysis, there were seventeen thousand six hundred and thirty-seven people. So a huge number of people, were already enrolled and randomized with follow-up when this futility, decision was made.
and these seventeen thousand odd people were enrolled at over three hundred and thirty-two sites in twenty countries. I can't imagine how much this whole exercise cost, but it must have been exceedingly expensive. People were randomized one to one to either receive this nonvalent E. coli vaccine, a single intramuscular dose, or a placebo single intramuscular dose.
and the mean follow-up [00:44:00] was eighteen months. the original planned follow-up was four years, but at the time of this, futility stopping, it was eighteen months on average. the primary endpoint was invasive extraintestinal disease caused by vaccine serotypes of E. coli. So that was basically bacteremia.
It could be other sterile site isoles, but it was pretty much bacteremia caused by vaccine serotypes of E. coli. And disappointingly, out of over eight thousand patients, there were only eleven patients who met the primary endpoint in the vaccine group and thirteen in the placebo group. So really, there was no difference there, and that gave a calculated vaccine efficacy of fourteen point six percent when you take into account the denominator of number of patient days of follow-up.
and there were similar results for really all the other analyses, including, numerous secondary endpoints, and this includes, UTIs caused by E. coli, not just bacteremia, UTIs caused by [00:45:00] any organism, and then also looking at subgroups of those who were at higher risk. so in all these secondary analyses, there was really no benefit of the vaccine.
so just for example, UTI caused by vaccine serotypes, there were four hundred and fifty-four of these events in the placebo group and four hundred and fifteen in the vaccine group. slightly numerically less, but not significantly different at all, and that gave a vaccine efficacy of eight percent, against UTI by vaccine serotypes.
So the vaccine did induce appropriate antibody responses in a subset where that was measured, and it had an acceptable safety profile, uh, and the expected safety profile. So a small proportion of patients got a sore arm, for example, but, there was nothing unexpected there. interestingly, some of the patients I was looking after who are on the trial had told me, "Oh, I'm sure I got the vaccine because, you know, [00:46:00] I was getting recurrent UTIs, and then they stopped."
And other patients told me, "I'm sure I got the placebo because I'm getting just as many UTIs as ever." But now it's all unblinded, and I've been able to tell the patients what they got, and none of them got what they thought they got which is just a good advertisement for why we need blinding in trials.
overall, this is a, huge disappointment. I do wonder if it was the right outcome. They set a very high bar for themselves, the outcome being invasive, you know, basically bacteremia with vaccine serotypes of E. coli, and you could have used something more common like, recurrent cystitis, for example.
However, this also showed no difference, so it wouldn't have in the end made a difference. I mean, Marc and Angela, you know this well. Like, why do you think this failed?
Well, I can say a few things. First of all, let me, describe how this all went because this is a, a story ongoing for more than 10 years. Also, my involvement in, advising them how to set up the [00:47:00] study and. the vaccine was initially developed by Janssen Vaccines. That was then taken over by Johnson & Johnson.
Marc: And I think six months before the futility analysis actually, the whole vaccine program was taken over by Sanofi. So actually think of what they bought. there was a long discussion with FDA on the primary outcome of the study. How should the primary outcome be defined?
And With these vaccine producers, they want to target an endpoint that allows them to get into a national immunization program. That's their goal. So an endpoint of, a less recurrent infections in patients with recurrent infection would limit the population eligible for their vaccine enormously, and that then would be considered probably internally as not worth the effort to go all the way to get a, a vaccine with a label for a very small patient population.
So the [00:48:00] enormous costs that come with such a study, I think, forces them To go for an endpoint that would allow the vaccine to be included in an immunization program if demonstrated to be cost-effective for the standards for including such vaccines. And I think that's one of the problems that we see with these vaccines for, the elderly, also with the respiratory tract vaccines.
The goal is to get it into an in- national immunization program, and maybe that bar is too high and that we are missing vaccines that can be beneficial for some indications, but probably the indications and the populations go with that indication are too small to make it worthwhile for the producers to make such vaccines.
Josh: Mm. Now,
Marc: what now is the reason that it failed? Actually, I don't know, because there is, of course, no correlate of protection for the antibody activity. There was a, let's say, significant increase in antibodies that could be [00:49:00] measured between the vaccine group and the placebo group. But apparently, that difference was not sufficient to cause a difference in this primary outcome and in all the other outcomes.
So I'm afraid we will never know, unless Angela is now gonna telling us why.
Angela: first of all, I would, beg to differ. It was not Janssen who initially developed the vaccine. No, no, they bought the- Yes ... Swiss
Marc: company.
Angela: Yes, we are very proud here in Switzerland. It was a, Swiss startup that actually, created the vaccine, and that's how I got involved.
We trialed it. it's a conjugated vaccine, so quite immunogenic because it's conjugated. so we did a that phase one early study. In healthy women who had histories of recurrent UTI. so we were able to look at-- Actually, it was really cool. It's probably the only time in my life where I'll do a phase I vaccine trial where you can look at, safety and immunogenicity as you must, but also clinical outcomes.
So Janssen came along and [00:50:00] bought that vaccine because that first trial was showing some signal of, clinical efficacy. It was very slight, and the study was not powered by any means for clinical outcomes, but the vaccine group was doing better with their recurrent UTI. so that's what started everything, I believe.
and yeah, to, answer why it failed, as Marc says, the, the primary outcome, that is a really tough one. You know, going for invasive E. coli disease, it's rare overall on a population level. We all know it's very rare. And like you were saying, Josh, the person who's gonna get E. coli bacteremia, that person is not likely to be immunocompetent or young, right?
We know these patients. These are the urosepsis patients in our hospitals who are immunosenescent or medically immunosuppressed. So it's a catch-22. I mean, you're giving a vaccine to a population that is already disadvantaged in its response to that vaccine, [00:51:00] right?
and then what I've heard from investigators on the ground, I don't know how you feel about this, Josh, but Merel Lambrechts in the Netherlands as well, she told me she was, an investigator that recruited quite a lot of patients, and she's really convinced that they were just missing a lot of events because the patients weren't able to report every single, cystitis or whatever.
they're sure that they missed a ton. bacteremias, maybe not. You know, again, the primary outcome was such a hard one. but the UTIs, she's convinced that the study didn't really equip them to find every UTI that was happening on vacation on a Friday.
Angela: As we know, UTIs always happen on the weekend, you know, when people are away. so she's pretty convinced that they just missed a whole lot of events. I don't know if you, if you had that experience.
Josh: probably not. I mean, we, biased it towards patients we knew we'd be able to follow up and weren't going to be traveling and that kind of stuff.
And they were on the phone to my [00:52:00] research nurses all the time, every time they got a stomach ache, you know, and were sending urine. It was just a huge amount of work. So I don't feel like we, in the trial in general or at some sites at least missed many UTIs. we may have missed getting urine collected in time to know if it was a vaccine serotype, but-
Angela: Yeah.
Yeah. Marc, correct-- I don't remember exactly what the cutoff was, but I think they were going with the old FDA guidance because, like Marc says, this dates back like 10 years. So I think they needed very high, colony counts in the urine to be able to call it an actual, you know, positive culture. E.
coli, as we know in real life, you know, E. coli can cause symptoms, a real UTI, with 10 to the 2, you know, with very few colony counts, which many labs can't even measure. So that may have also- I don't- ... played a role.
Marc: I don't know if I had, but I think we can assume that this is loss of endpoints was, let's say, [00:53:00] was non-differential.
So there was no reason to believe that the intervention itself resulted in extra loss or vice versa. So that would not change, to me, would not reduce the validity, scientific validity of the comparison between the two intervention. It just takes much longer to come to an answer if you miss a lot of endpoints.
And sure, , there would be a loss of endpoints in any such trial. Missing endpoints.
Angela: Yeah. No, I, agree with you that things should have been missed equally, right? 'Cause these patients were randomized. Yeah. So events should have been missed everywhere.
Marc: And it was very double-blind, as we heard from Jos.
Speaker 6: Yeah.
Angela: Yeah. Yeah, but still, I think, the trial was stopped for futility because they just didn't have enough events, right? On either side, on, you know-- if you just had a richer-
Marc: No, it's not, not stopped because of too-- because the analysis was based as planned.
It only took much longer to get to that number in which the first interim analysis was done. So I think the total, sample size [00:54:00] was twice the number- Yeah ... of these endpoints.
Angela: That
Marc: was what was- But
Angela: that's just it, because there were so few events, right? They couldn't- Yeah, yeah ... come to this first analysis.
They had to get to- No, no, no. This was the- ... 17,000 patients to be able
Marc: to- Oh, yeah, yeah ...
Angela: do the, the
Marc: interim. I think the enrollment was still going, so I think the plan was to go- Yeah ... to over 20,000 to get- Yeah ... in total, I think to about between 50 and 60 primary endpoints. So when they came to About 30 that they have now, that was the planned first interim analysis. Yeah. And there were stopping rules with that interim analysis, so no one had expected that the estimate vaccine effectiveness would be, let's say, 14% at the first interim analysis, which hit the trigger to stop
Josh: the trials and tribulations of clinical trials, hey. Like we never know when we're starting a clinical trial what's going to happen, and that's one thing that's fun and exciting about doing them. Yeah. Sometimes, you know, disappointing and sometimes exciting. But, so that was the last trial we were going to discuss.
[00:55:00] one little side comment I think I might make is that you might have noticed these late breakers we've been discussing, a lot of them were presented in journal branded sessions. So JAMA had a session, The Lancet had a session, CMI and CMI Comms had a session, but notably absent was New England Journal of Medicine.
So he-here's a challenge that I'm throwing down. New England Journal, come along to ESMID like all the other major journals do, and, sponsor a session and come and, commission some excellent studies to, be published in your journal. In return, we promise the New England then to send our best papers there.
Yeah.
Anne: thank you so much for the conversation, and thank you for listening to Communicable, the CMI Comms podcast. This was part two of a two-part episode about top trials presented at ESMID Global 2026.
This, episode was edited by Katie Hostettler-Oi. theme music was and conducted by Joseph McDade. This episode will be citable with a written summary referenced by a [00:56:00] DOI in the next eight weeks. And any published literature we've discussed today can be found in the show notes.
You can subscribe to Communicable on Spotify, Apple, wherever you get your podcasts, or you can find it on ESCMID's website for the CMI Comms journal. Thanks for listening and helping CMI Comms and ESCMID move the conversation in ID and clinical microbiology further along.