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]
Nav: Hello and welcome back to Communicable, the podcast brought to you by CMI Comms, ESCMID's open Access Journal, covering infectious diseases and clinical microbiology. My name is Navaneeth Narayanan. I'm a clinical associate professor and infectious disease pharmacist at Rutgers University in New Jersey in the United States, and an associate editor at CMI Comms.
I'm joined by my co-host Josh Nosanchuk, a fellow editor at CMI, comms and Infectious Clinician and Professor of Infectious and Microbiology and Immunology at Albert Einstein College of Medicine in New York City.
Josh: Thanks, nav. We're all excited to welcome three fantastic guests today to discuss the utility of adjunctive steroid therapy for pneumocystosis in individuals without aids based on two papers they published in 2025.
The first guest is one of our wonderful partners at CMI Communications, Emily McDonald is a rockstar associate editor at the [00:01:00] Journal. Emily is an associate professor of medicine at McGill University Health Center at McGill. She is the Director of Clinical Practice Assessment Unit in General Internal Medicine, and a scientist in the Infectious Disease and Immunity and Global Health Program Center for Outcomes Research and Evaluation.
Emily is also associate chair of Quality and Safety and the director of the Canadian Medication Appropriateness and Deprescribing Network. Emily, welcome back to Communicable.
Emily: Hi everyone. Yeah. Excited to be here today in the guest and host chair.
Josh: Terrific. Our second guest is Elie Azoulay. He's a professor of medicine and director of the Medical Intensive Care Unit at Paris City University and Faculty of Medicine at the St.
Louis Hospital in Paris. He's a leading expert in infections and sepsis in critical care settings. Thank you for joining us on Communicable Elie.
Elie: Hello everyone, and thank you for having me.
Our third guest is Intensivist Virginie Lemiale. [00:02:00] She practices at the Paris City University and Faculty of Medicine at the St.
Josh: Louis Hospital in Paris. She's the lead author on the paper with Elie that brings us all together today. Welcome to Communicable.
Virginie: Hello everybody. I'm pleased to be here.
Nav: It's very nice to have everyone here. just to get us started, as our listeners know, we , start with the get to know you question. So the question for you all, is, where's one place in the world that you've never been but you would like to visit?
And, why is it there? Josh, do you wanna start?
Josh: Sure. So I wanna go a lot of places, but one of the things that I've been speaking about with my wife and some friends is going to Chile. And why Chile? Well, they've got great wine, beautiful country, and I love to ski.
And I've never been skiing in South America. I've been skiing in a lot of parts of the world, but not there. So that's my, next choice. And you, Nav?
Nav: So I think I've inherited this from my wife, but she's a major Lord of the Rings fan. So New Zealand is the place that we want to be so we can, pretend to [00:03:00] emulate that trek to, to return
Josh: the rings.
You gotta go to Hobbiton. It is so cool.
Emily: It's amazing. Yeah, I highly recommend it. well, for myself, I, have never been to Japan. so I would love to go to Japan. If you know me, my life goal is to die with a little bit of mercury poisoning from eating a lot of tuna. and I love sushi, but I've never been to Japan.
So that was an easy question for me. It's number one on my bucket list.
Nav: What about you, Elie?
Elie: Well, I would dream to go to Kenya because this is something I have in mind and just over the weekend I shared, some pictures from friends coming back from there. And this is obviously the next place I would go to visit.
Of course, there are many other places that I love in the world, but Kenya is the next one.
Josh: Sounds great. And you, Virginie
Virginie: Also Japan, but for other reason because I would like to visit the Mt. Fuji because uh, I love mountaineering and then the serenity I [00:04:00] imagine that Japan has. So I would like to go there.
Nav: Alright, thank you all for sharing. and so for this episode, we've asked Emily to join us as she wrote this intriguing paper at CMI Communications that really aimed to contextualize the findings of an important multicenter double-blind, randomized clinical trial, assessing the utility of steroids in the setting of severe pneumocystis pneumonia in patients without HIV Virginie is the first author, and Elie is the senior author on the Lancet Respiratory Medicine paper that presented the data and the results of this trial.
Josh and I thought that it would be really fascinating and, fun, to bring Emily on board and to have her join Virginie Lemiale and Elie to discuss these efforts, together. here we are now and we're gonna learn about, the trial as well as this contextualization paper that Emily put together with her colleagues and to understand what to do about this longstanding challenge.
Josh: But first we want to hear from our critical care guests, Virginie and Elie, about how they got involved in working on infectious diseases challenges in the [00:05:00] ICU Virginie. Could you please share some information about your journey with us?
Virginie: Yeah. I'm, uh, pulmonologist, by training. So I am particularly interested in the pulmonary condition we encounter in the intensive care and at St Louis Hospital.
Also, we are mainly treating immunocompromised patients. So, it was natural for me to try to answer two questions that I have every day about those patients with immunocompromised condition and pneumocystis pneumonia.
Josh: And how about you, Elie?
Elie: Well, you know, we have been raised with, the AIDS epidemics and, we learned from, the Covax paper and, you know, all of these giants of medicine who brought evidence that, pneumocystosis in HIV patients and non HIV patients is actually different disease.
In one case, there is a very, high influx of inflammatory cells in the alveoli, and in the other one, because of the HIV setting, it was a different disease. I was a resident at the time where [00:06:00] we had only the pneumocystis cases in the ICU ventilated and Unfortunately, before 95, most of them were dying, especially if they had a failure of, trimethoprim-sulfamethoxazole. So we learned from the evidence and we, could not figure out, what were the expectations for giving steroids in non AIDS related patients. So we are very much into something that we love and we wanted to have more evidence to guide the practices.
Nav: Thank you both for sharing, and Virginie I think it resonates with a lot of folks, to ask a research question that comes from their everyday practice. and I think for, clinician investigators that, hits home so much.
So thank you for sharing, especially given the work that you've done. so now we're gonna turn to the papers at the heart of this episode. let's first start with the July, 2025 Lancet Respiratory Medicine paper. It's entitled Adjunctive Corticosteroids in Non-AIDS Patients with Pneumocystis jirovecii pneumonia.
It's a multicenter double-blind randomized control trial. you can find the link in the show [00:07:00] notes for all our listeners. So, Virginie, let's start with you. Let's talk about, what led you to, perform this study. What were the big questions that you were asking, and what was the hypothesis that you had going into this trial .
Virginie: As I said, this study come from the clinical discussion we had regularly in the department when, a patient without, HIV had a pneumocystis pneumonia. And each time we say, I will give him, steroids. people say, no, this is not demonstrated in the literature.
And, first I compared 100 and, uh, 39, non HIV patient, who receive steroids, or not but that's not the answer to the question because it was a retrospective study we don't know when they start steroid. And so we can't answer the question. And in the same way in the literature, no, study can, answer this question.
That's why we want to start, a prospective randomized and, double blind, study [00:08:00] to really answer the question should the patient receive a steroid on no steroid.
I, guess we get to more information a little later on. All right. Now I'm gonna ask a next question. this is to, both of you, you know, how did you bring this large group of collaborators together? 27 hospitals in France, and how did you keep 'em together during the duration of the study, which was from February of 2017 to February, 2024, which obviously includes the COVID-19 pandemic.
Well, I think first we should, mention one thing. I think that Virginie should be commended for having led this, very important trial, looking for guiding the practices with high level of evidence. at the same time, we have a, group of, investigators in France, because this study was performed in France, but also we have another one, all over the world working all together, learning from each other, and, you know, looking for the same, thing, which is guiding the practices.
Elie: And, in France, this group, which is called the Grow, has actually [00:09:00] included the patients, and as you said. Something happened within the midst of the study the COVID pandemic and, the trials survived it only because of, Virginie's willingness to, complete it.
I must say, and, we will discuss this hopefully with Emily later. I think that this is the typical case where systematic reviews and meta-analysis brought so many conflicting data that it was really important to, to have a high level, high quality of evidence, to really guide the practices.
Elie: And this is why, the trial is actually bringing new data to guide the practices, but also is looking at, systematic reviews and meta-analysis showing what was wrong in terms of dosage, in terms of, length, duration of, steroids, but also the underlying groups, especially the level of, of underlying hypoxemia, which will be discussed later.
Josh: Absolutely. We, all agree that this is a, terrific effort to spearhead and lead over time and we're just really impressed with it. [00:10:00] And I think that's one of the reasons why it's gotten a lot of attention, from various groups. Now, Virginie, you shared the questions that launched your study, and could you share about some of the methods you used and how you assessed outcomes?
Virginie: so we want to conduct a, trial with a superiority, study because, it was easier to answer the question and evaluate the short term mortality on day 28. maybe it was too short. But, at the time we designed the study, it seems that it was an evidence for the day 28.
Now, maybe it would be day 90. it was an evidence because we also want to, assess the potential consequence of a corticosteroid like, uh, more infection in ICU, during ICU stay or in insulin or diabetics or more intubation we didn't know. And so, we use, day 28 as primary outcome.
Virginie: And then, we conduct [00:11:00] a double blind study, which was very difficult to perform because we didn't have any, placebo for, steroid, by pills. So, it took a long time to design the study and to perform every, step of the study it was long time for the, other investigator to assess and accept all the, step of the study.
Nav: Virginie, I'm curious, especially as you talk through some of these iterations and things you had to work out in the design. the trial was launched in 2017 to 2024, but when was the actual conception, you know, when you start writing the first draft of the protocol and getting your team together?
'cause I'm sure it took a long time before.
Virginie: Yeah.
Nav: You know, first enrollment,
Virginie: it was in 2015 or 14, we draft the first time, then we discuss with, our research group, we try to find money to perform the study. So it took maybe 1, year and then another one year [00:12:00] after I have the money to start to understand what would be the, placebo and the above line, how we can perform it.
Because, the line today is perform with the pharmacist who, who do the steroid, syringe and so it took a long time to be sure that everybody understand what, we want to do and, how we, we will perform it.
Elie: I think it's important to remind that as opposed to, other countries, we are still very lucky to get, important support for clinical research.
But at the same time, we don't have to apply to an IRB to get the grant. So we have a long time to wait for, authority regulations and the IRBs and all the declarations before starting the trial. So this is why there is an entire year to get the grant and then an additional year to start.
and the other thing is in this situation where we are not, evaluating the benefit from a drug that has, a huge, industrial development. So finding a placebo and getting to the reality of what [00:13:00] it'll be, was an important thing. the last thing to remind that I think is very important is that people have strong beliefs about the use of steroids in, pneumocystis pneumonia, some were against absolutely against.
Elie: And some were very much pro. So, you know, for these hypoxemic patients that we have included in the trial, it was not so easy to convince that, we had a very strong possibility to evaluate, but also that we have an equipoise allowing us to really, perform double-blind randomized control trial.
Emily: it might be important to really just like drive home the prior evidence, which was either completely derived from HIV patients and non HIV patients or based entirely on observational data up until this pivotal trial and that observational data, when you put it all together and you included all patients.
it looked like it could be harmful to give steroids, and then if you took out the hypoxic patients, it looked like it could be beneficial, and that's why this was so hotly debated in [00:14:00] clinical practice.
Can you share some of the, groups that were very concerned about using the steroids, were there particular types of patients that there was more heightened concern over? As you point out in one of, the critiques of some of the, possible complications of analysis, is that you did have different groups and that was very important.
Josh: But one of the things that we always address is. These are different individuals and steroids do complex things. And which groups were more concerned?
Elie: Yeah, and as we say French, there were groups in the group because one thing that we need to bear in mind, and we were at the beginning in 2016 when we applied for the grant, is that 20 to 25% of the patients have two infections.
They have pneumocystis and pneumocystosis, but they have an additional bacterial, fungal, viral, or any other condition. Sometime a non-infectious conditions that it comes on top of, PJP itself. So I would say that, we were also before the time where we were giving steroids, community [00:15:00] pneumonia. So this is again, another major confounder.
So. To be, very honest. We didn't have to fight so much because people were also very interested to have the answer to this, uh, you know, pendulum question, and , Emily, what you said is fantastic, is that when you look at the systematic review and meta-analysis, there were big differences on where you're giving steroids to hypoxemic or non hypoxemic patients.
And as you know, some hypoxemic patients are not all in the ICU or intermediate care units. Some stay in the wards when they have one or two liters of oxygen per minute. So these patients were also managed by ID specialists, used to give steroids to patients with pneumocystosis. And, the discussion about whether it was different or not compared to HIV patients was not so easy to address.
when you started this to when the trial concluded, it's a 10 year period. it's very sobering and humbling, I think for [00:16:00] folks that aren't really, in the clinical trials world to understand how long these things take.
Elie: Yes.
Nav: And you're in it for the long haul. When you really have a question you want to answer clinically, it takes many years to get it up and going and whatnot. So I genuinely appreciate that. on that note, do you want to take us through some of the main findings about who you actually enrolled and what were the outcomes that you found?
The main results, is for the day 28, uh, mortality. It's negative, but the curves are, quite different after seven days, but there is no significant difference between the two population at day 28.
Virginie: It's not scientific, but, I think the most interesting, important thing is that the day 90, mortality was very different between the two groups. There is not a significant difference, but we improve patient in the first date and then they could continue their, course of their, underlying disease and they could improve, after.
also the most interesting is that, The other secondary, [00:17:00] endpoints are all, positive for, steroid treatment. There is less intubation and there is less intubation in the, five first days, which is very important because in the first five days, this is, the time the, pneumocystis pneumonia, is more and more, severe, usually, during the clinical course of the pneumocystis pneumonia.
Elie: I want to give emphasis to what, Virginie just mentioned. I think that, of course the primary endpoint didn't meet the significance, expectations, but day 90 mortality was different and all the other outcomes were positive. I just want to also, as you say, nav and as you say Josh, about safety.
This trial is actually, very, promising. And, we have a very, good safety data. tolerance of steroids was amazing. And of course, patients were in the ICU and monitored very carefully, but at the same time, The threat that we were potentially expecting from steroids were not met.
And, [00:18:00] patients were doing very well, as mentioned by Virginie. Our challenge as ICU specialists is about the inflammatory rebo that we have by day five with, trimethoprim-sulfamethoxazole, where we have a huge burden of pneumocystis in the alveoli and we know that patient worsen. Maybe it is something that you also encounter in the ID wards, but for us, giving steroids, in this, time window is a very important challenge.
Virginie also believes that maybe giving only seven days of steroids might be sufficient, and she's now working on trying to identify whether, because as you know, patients wear wind from steroids when they were discharged. Whether those patients receiving less steroids than others, maybe were with, the same outcome.
This is something she's working on.
Josh: So that's super interesting. You talked about things that are very important here and, in terms of the safety you brought up earlier that they have other infections and you saw no significant, adverse [00:19:00] outcomes in those individuals with other infections or more infections.
The other thing is the timing of the steroids. And on average, you guys were, it was about, I think, day three that they were started. And so the question is, is that the right time? Is it too late? Is it too early? And the issue there that we saw in the COVID-19 pandemic was the same thing. When do you start steroids and when is the best?
And we learned a lot, but we had so many patients to help us do that. So what do you think is day three the right day? Is it day five? Or should it be as soon as you make the diagnosis?
Virginie: I think the good thing is before day five, because the, the day five is the day where the best patient are, more and more severe.
but day three, I think it's very important to have a diagnosis before we start steroids in that, setting. And so I'm not sure that we could, do less than the three. Maybe it'll be better at day one, but we don't have any [00:20:00] microbiology diagnosis as timing. So, it's difficult to, make a hypothesis, for that.
So, I think as early as possible is, good
Josh: another thing that I, was really struck by in terms of your acknowledging and discussing is the, fungal burden, like Elie just brought up. And, there you did have different CT counts for the PCR based diagnosis and what biomarker to use to figure out which individuals may have more of a risk of this inflammatory process.
The fungal burdens in these individuals is usually, pretty substantial. So, you know, have you thought about that? and I really interested your answer.
Elie: we are thinking about that because we, well, patients were here treated by trimethoprim-sulfamethoxazole as pointed out by Virginie. I think it's very important to consider, the therapeutic effects of steroids also based on the diagnostic strategy as we know that currently our laboratories are no more, performing [00:21:00] staining or immunofluorescence.
everything is based on PCR and PCR is certainly at the same time a very good way to have, the same diagnosis for everyone. But on the other hand, we also need to balance that with, some very difficult cases in new forms of disease, rising in patients treated by ibrutinib and other new drugs.
and I say immunotherapy being one of them bringing new forms of disease that some consider to be atypical, but you know, everyone here has forgotten what is a typical pneumocystis system with what you say, Josh. I think that, in a close future where, echinocandins will be part of the routine, part of, PJP treatments, I think we also need to consider that.
It should be tailored not only based on the level of beta-D-glucan, which as you know, is, one of the major, uh, biomarker to consider giving echinocandins. But at the same time, we also need to know whether we are treating, clinical forms where cysts [00:22:00] are the most important bodies of pneumocystis acting and giving the disease as opposed to patients with having the trophic forms also responsible for the disease.
And we know that all the treatments have not the same effect. So I think that introducing Beta-D-glucan to oppose the. Different clinical forms will also be a very interesting form, especially in the time where, and I consider in a very close future where, combination therapy for pneumocystis will be given in these severe forms.
And I will be very interested to know what is the ideal day to start steroids when we will introduce also echinocandins for these patients and we will change the scope of the disease and the fungal burden as we just discussed.
Nav: So it sounds like Paris City University is gonna be the pneumocystis center for global trials in terms of biomarkers, in terms of combination therapy. So I'm very excited to see all the clinical trials that you will have [00:23:00] very soon. With that,
Elie: we will do that all together. We need to think global.
Nav: No, I love it.
this back and forth does just show how many more questions spawns out from a clinical trial, right? It doesn't end there. It spawns out many more questions to help optimize, to understand what you need to do at the bedside , for these patients.
Nav: So, I really love this discussion.
Elie: And nav, if I may, there is one very specific thing to remind is that, for some patients having, long-term steroids and developing pneumocystis in this setting, we know the CD-4 recovery when we taper the dose of steroids is one of the major determinants for the inflammatory, alveoli, development.
So by saying that I think that in this very specific group of patients, steroids might be different, not only in terms of, the duration of therapy, but also the importance of the onset of steroids when we start steroid therapy in these patients with, a, CD-4 recovery. So we should increase [00:24:00] also monitoring of, the underlying immunity and especially the CD four cells in these patients.
Josh: Yeah, it would be great to get a lot more immunological data. Cytokine data IL-17 and other things. This wraps up to a tough question for Virginie. If you could go back in time and change things in your trial design, would you, and if so, what would you do?
Virginie: So maybe at first I will, change the primary endpoint, as I said before, because I think the day 90 mortality or later is one of the most important in those patient because, patient may survive from ICU, but they survive with good, performance status, it's better for their, other treatment because they need other treatment after, ICU.
So maybe 28 days, it may be a little bit, too short the. Question I could not answer, but I don't know how to do it in the future is that the length of, steroid [00:25:00] treatment in those patients because, in the protocol, all the patients should be treated with 21 days, of, steroid treatment.
Virginie: But we saw some patients who, improve very quickly and then they left, uh, hospital and this was it. They don't have 28 days, of, steroid. And maybe, it could be enough, for them to be treated for less than, 15 days. so maybe I will perform some exams before they left hospital if they're, treated from the, pneumocystis pneumonia, they will stop, steroid treatment, with the good, accuracy.
I think it's difficult to perform another study because as you say, it took, 10 years to perform it, 10 years more. It'll be a little bit difficult, but, uh.
Emily: maybe I can jump in here. Elie, Virginie, in Canada, myself and my colleague were just funded by the government to, run a PCP platform, international trial, to answer three questions, low dose septra versus regular [00:26:00] dose septra.
The, how long study where we look to see if the person is better on their steroids by day seven, and then randomize them to stopping or continuing to the full duration. and adjunctive echinocandin. So that'll be our first, three questions, and we're gonna start to collect some randomized data on PCP alternatives.
So for people who can't take septra, we'd be randomizing them, head to head to some of the alternative treatments. so hopefully we can work together to answer this question. I think that would be really cool. It's one of my grand reveals on this podcast today. the other one is, I don't know if you know, I was one of the reviewers for your trial.
I can't remember if we talked about this before. I wrote to the Lancet Respiratory Medicine and I said they should send it back up to Lancet. So I was very convinced by your trials and actually advocated that it should go to Lancet proper. because I, think that the day 90 mortality data, is extremely convincing.
Most important. and you'll see [00:27:00] in the CMI communications paper that I, co-authored, there's been a lot of advancements in ICU care. And so mortality in the ICU may be delayed now, because we've improved care in the ICU because we've improved care in general of the patients who are getting PCP, how we ventilate people has changed entirely over the last 20 to 25 years.
so people may be surviving longer in the ICU. And so in order to see the impact of that treatment, we may need to follow them longer. And, we're gonna come to the CMI communications paper, but one of the debates that we had was, did your trial need a bigger sample size or did it need a longer follow up period?
And my argument was that it, just needed a longer follow up period and that you had the sample size.
Elie: Well, that's a very interesting point, Emily, because you are raising so many, hot topics in your project. so of course, I think we were powered enough, but we should have a longer follow up that's obvious.
in your, trial, which echinocandin are you going to use rezafungin or another one?
Emily: Yeah. So we're [00:28:00] gonna do it pragmatically because there's different echinocandi ns that are funded based on your country or reimbursed. And so we are actually thinking to just permit whatever echinocandin you use in your center.
Elie: Okay. And are you going to stratify based on the beta-d-glucan concentrations?
Oh yeah. Great question. So, very sadly in my center, when you send a beta-D-glucan, you get the result in three weeks. in some places the turnaround time for beta-D-glucan is very long. So I think that what we could do is after the fact, take a look at the beta D glucan result, but I know for ourselves, we wouldn't be able to stratify on it because the result takes so long to come back.
Emily: It is still an expensive test and not available everywhere.
Elie: Yeah, it can be done posthoc here. I agree with you. But at the same time, there are point of care, uh, beta-glucan that are also being developed by many, different, countries. So, to address, what we should have done.
I think that Josh, one of the point that has a lot of importance here is how we teach pneumocystis to the young generations, because there is no exclusion that in [00:29:00] the future we will have more cases. I remind you that all the patients that Virginie included are failure of prophylaxis. And I think that this is something we need to remind to everyone.
many of the cases that we have included in this trial are cases were either, we missed, indications for prophylaxis or patient didn't take their prophylaxis. because I'm sure that most of us didn't really see, many cases of pneumocystis in patients really undergoing and taking their prophylaxis.
And the other thing is we had something to do. I remind you that we are after all pulmonary and critical care physicians is to also gain in understanding by looking at the cells in the alveoli. I would be interested in this trial. Some patients has RDS forms with a lot of neutrophil accumulated in the alveoli, and some other patients had more, maybe a, kind of a hyper sensitivity to the p nemesis body.
And these were certainly patients who were having a high benefit from steroids. [00:30:00] And this is something that we looked at, but not in every patients, uh, because, not everyone had a bl even if, it was about 80%, but also, because not everyone had, alveolar cells identified in the bl. So this is something for the future of course.
Nav: Thanks for sharing. I think this is a good time, Emily, maybe to talk about your analysis in your paper and what you've done with your colleagues.
And so what was it about, Virginie and, and Elie's team's trial that led you to do this analysis in the first place?
Emily: So, Viriginie and Elie's trial was, a frequentist trial and sometimes in a frequentist trial, when we do the analysis and we look at the primary outcome, we tend to be very concrete in our interpretation, such that, if the P value is not, less than 0.05, that's a negative study full stop and we lose all of the nuance.
And I was very concerned, that that could happen with this trial. First, not only was this the first RCT to be [00:31:00] done in steroids and pneumocystis, it was this one of the first RCTs to be done in Pneumocystis in 20 years. Full stop. and so, we've talked a number of times about how challenging it was to set up and run this study, how long it took, how much effort it took, the patients who contributed their data.
And I thought, it's a shame that some people are gonna read the bottom line of the conclusion of the abstract and stop there. whereas I believe that from the day 90 data that steroids in non HIV pneumocystis, save lives, that they decrease mortality. And so, my colleague and I, we were discussing this and we thought, there's a new kid on the block, Bayesian analysis where you can actually take prior data and prior information that you know about the intervention.
and you can apply it to the current results and you can help to contextualize, the result and you get a bit more nuanced answer and maybe an answer that is sometimes easier and more helpful for [00:32:00] clinicians, to interpret in our day-to-day practice. So if you think about how we practice medicine, we're actually using a Bayesian approach all the time.
Emily: So when you see a patient in front of you, for example, and you're trying to decide what their diagnosis is and you're trying to decide between three different diagnoses at the beginning, maybe those three diagnoses are equally probable because you don't have any information about that patient.
But then you find out, for example, that there's a consolidation on their chest x-ray, and now suddenly pneumonia is a lot more likely than pulmonary embolism or pneumothorax. And so we've used that information that we gained from the chest x-ray to change our probability of what the diagnosis is. So you can kind of apply that to statistics now.
And so we talk about, just as like a little primer, the prior probability, which is the probability of, in this case, corticosteroids decreasing mortality before any data is collected. Okay. And then we talk about the posterior probability. So when you do a Bayesian analysis, you can update that [00:33:00] probability and include data that was collected, before or during the trial and help change your posterior probability.
And so because steroids had been tested in other populations, in a sense we had information about whether or not steroids can impact that outcome of mortality in patients with pneumonia. So for example, we knew now from COVID that steroids, could be helpful in COVID pneumonia. We knew from community acquired pneumonia that steroids could be helpful.
We knew from patients with HIV that steroids could be helpful. And so what we decided to do is we decided to take all the RCTs, that have been conducted in patients with HIV and we decided that it was, and this is what happens in your analysis when you're doing a Bayesian analysis, we decided that Those patients were probably in some way similar to non HIV patients. And so we redid the analysis and we used that information, to contextualize the results, to give a probability that steroids would decrease [00:34:00] mortality in, patients without HIV with pneumocystis. So you'll see that's the analysis that we did, but we had to make some assumptions.
maybe I'll pause there and see if you have any questions. ' it's a new approach that not everybody is familiar with, so we kind of have to wrap our heads around this and there's a lot to discuss.
Elie: Emily, just uh, first comment from myself on which I, fully agree, is that, the Bayesian approach is really mimicking the clinical approach much more than any, very formal randomized control trial with, exceptional, inclusion, exclusion criteria.
So what you said is also very important to translate the results into the bedside. And I think it's, a very important point.
Nav: when I was looking, over Elie and, Virginie's trial, the initial sort of assumptions going in, in terms of the baseline mortality rate, Versus, the effect size, right? So the effect size assumption was about 20%. What you found was about 10%.
And so from a frequentist view, they teach you in school, like, don't look at the P value. If it's close to 0.05, it becomes binary, which I [00:35:00] think really muddies the water in terms of interpretation and, it's not really helpful, to actually make some decisions. And so Emily, I guess, how is it that your meta-analysis helped bridge that interpretation of what the presumed effect size was of 20%, which was larger, which would be easier to sort of find, even with the day 28.
mortality as an outcome, whether that was too short of a followup or not, versus what was actually found and how, that p value was borderline and now this meta-analysis broadens that ability to understand what that difference is between the corticosteroid group and the control group.
Emily: I mean, one way to run the analysis like a frequentist approach is to ignore all the prior data, right? And just to assume, that there's the null hypothesis that the two corticosteroids and no corticosteroids are equal. but another way to look at the data is to not ignore the prior information that we have and actually use it, to help come up with, a new probability.
And so, what we did was we [00:36:00] took a pretty, what's called a neutral prior. So we didn't heavily influence the result of our analysis with the prior data. But then we also even took, a pessimistic approach. So we said, well, what if corticosteroids are harmful? So we are able to model using different assumptions what the new probability is.
and that's the probability of benefit. And so even when we took a moderately pessimistic prior and we said, well, maybe corticosteroids could actually be harmful, there was still a high probability that, corticosteroids would be beneficial. And so what it does is it allows you to kind of test different hypotheses.
And so, how we talked about in the clinical environment that some clinicians thought steroids were bad, and some clinicians thought steroids were good. Well, you can actually, make everybody happy because, taking this approach allows you to model those different clinical circumstances.
And then you can see how much the prior data informs your new probability.
Nav: Elie, Virginie, what were your first thoughts when you, came [00:37:00] across Emily's paper, her analysis and what she tried to do from that perspective?
Virginie: so when I saw this paper, I was very interesting because, yes, I thought that the trial was, written with negative study, but I agree. the most important is the other, secondary outcome were very consistent with the, steroid use. And so, I was very happy that, you can model the, results and, I was sure that it was working and then you demonstrate by this, study.
Emily: And I guess like when we do a journal club, let's say we take a frequentist trial and we do a journal club. we're doing this intuitively right? We're looking to see pathophysiologically, does the outcome make sense? based on what we know of other studies, does it align with other studies?
And, you know, all this analysis does is it actually allows you to really quantify that approach that we're all sort of doing intuitively when we take the time, to do a full in depth [00:38:00] analysis of the results of a study.
Elie: Well, I, think that there is a number of points to raise and I'm very happy about the discussion.
I'm really enjoying after all this podcast, I'm very, very happy. first when I read the paper, the first thing is I said, well, there is Emily McDonald and I'm so happy. Why? Because the person who is doing that, is an expert. And she knows what is a patient, and she's dealing with the problem every day.
She's not about just to produce another paper on steroids in pneumocystis and we had so many, so this one is really helping. So I, might say that the two papers are very interconnected, linked to each other, and I think that Emily extended the impact of the trial to reach the bedside. So this is one thing.
The other thing is, and you raised something that I love about knowledge transfer. I've been editor in chief of intensive care medicine for many years and I'm now editor-in-chief, critical care, sepsis and severe infection. And you are all editors in, the ESCMID Journal. So we are [00:39:00] saying the same thing.
I think it's. Common responsibility to bring the knowledge at the bedside. And the classic paper, the classic designs with introduction methods result discussion are boring for many readers. So we are now committed to bring the data to the bedside. And the visual abstract is a big advance, but as you just mentioned, people were looking at the last line in the abstract, now they're looking at the last picture on the left bottom part of the, of the visual abstract.
We need to do something else. And there are things that people expect. So I would say let's do together a cartoon on pneumocystis where we would bring in a different way, this data to the bedside. This is what I'm doing today, summarizing the literature on pneumocystis, not only steroid, but also other treatments with the echinocandins, with the biomarkers and the way we can stratify.
And also with the pathophysiology, with everything that leads to hypoxemia. [00:40:00] And, I would be so happy we can do that together. We need to think global. I think we were working on our own until today and now we are together.
Emily: I'm so happy to hear that. one thing that came to mind, as you were talking was that the reason we have arrived at a point where we're, reading the last line of the abstract and that we don't wanna put.
Anything other than the primary outcome in the conclusion is a reaction, right? there's a bias to publish positive studies. And so as researchers, sometimes what's happening is you're running many secondary analyses and then you're quote unquote cherry picking something positive, even though you've done multiple comparisons.
And as a reaction, we said, okay, no, we can't do that. That's wrong. it may be a bit of an overreaction. We also have to use our common sense when there is a secondary outcome that clearly aligns pathophysiologically and with prior data, yeah, we have to use our common sense and highlight that.
Nav: I, do, think it's overcorrection is the right word. you have to use some, thoughtfulness, right? Thinking deeply about what the [00:41:00] results are, how you interpret them.
Because at the end of the day, the point, Virginie, that you made, , the initiation was to answer a question on how to better take care of patients. And so the results of it, all of the results together, not just one very specific line from the results, should then therefore inform how to take care of patients at the end of the day.
Emily, I, love your approach and I love the way you've explained it. we have a phenomenal microbiologist at, my institution, Tom Kern, he's our state lab director and our microbiology director, when he teaches some of these things too, he just, he says, we're natural bayesians.
That's what humans are the way they think. And so I think you putting it in that way is really helpful.
Emily: one other kind of reassuring thing that we found in our analysis was, and I love that we did this, but we didn't just calculate the odds ratio of benefit, but we translated it into, a risk difference.
and so we saw that giving corticosteroids, looked like there was about an 11% improvement. And then when we compared [00:42:00] our risk difference to, the risk difference found in Virginie's study that was 10.9%. And so what's really nice and harmonizing is that these two risk differences were almost, superimposed.
It's almost the same number that we found with our analysis as the primary study, which helps you have maybe a bit more confidence in the findings. So there's other. Things that we can do with this type of analysis. That's really interesting. You can also do an analysis where you predict and you model.
If you were to do another study, what's the likelihood that it would, shift or move the dial in the opposite direction. so that's something else that if you're interested in, you can read about it in this study, but it looked like doing another trial in this sphere, probably based on modeling, wouldn't change, and that you would still see a mortality benefit.
So it also helps you to sort of say, okay, maybe we can move on from this question and start to answer other questions like, what type of steroid or how long do we give them for?
Elie: And if I may, Emily, I think that you really should be commended for [00:43:00] leading this because, uh, well, we still are in a time where there are more than half a million cases every year with 40%, mortality rates.
and, I think that, bringing treatments like steroid that are available almost everywhere, including in low and medium income countries that are not spared by the disease. I think that this is something very important and we, should consider that this is also part of our, common responsibility and the way you brought the data are not into raising controversy inside.
Something that is quite clear is really. Giving a voice to a trial and putting it in a setting, in a context where people are more able to understand and make their own opinion of that. And to my view, it's much more than a 10% difference. It's also bringing and reinforcing evidence, which is what we need to really be confident with what we do and the decisions that we are making.
Josh: I want to go back to the safety issue that, you brought up before, Elie, that this appeared to be safe, so you were doing no [00:44:00] harm, but there was a very strong and clear signal of benefit,
Elie: including in populations with a 20 to 25% co-infections or co associations with other infections, including other fungal infections, severe viral infections, et cetera.
Nav: And Elie on that point, and Virginie, if, you could, I, don't think we talked about this and maybe it's not as clear to, listeners that haven't read the study at all, but who was the population? in the title it's non AIDS patients. So these are patients without HIV.
Who were these patients in terms of the at-risk condition that predisposed 'em to pneumocystis in the first place.
Virginie: those patients without, HIV had high risk of, pneumocystis pneumonia because half of them took steroids for more than three months. mostly the, solid organ transplant patient.
And with so many, solid organ, organ transplant, patient with, pneumocystis pneumonia, but also other, patient with, immune disease and then took steroid and [00:45:00] not all those patient are, prescribed, prophylaxis. And so we saw many of those patients. And I think also the higher improvement is in those patients with hematological disease.
And then we took them small amount of steroid and then they could, improve very, quickly. And so that's all. I think these are the most frequent, patient in our studies. There is also some solid cancer patient, but I think it's more difficult to analyze because those patients who took steroid, before they have pneumocystis are also the most severe patient for cancers as you see, I think.
So it's more difficult to analyze it.
Elie: And nav your question is a very important one because, non HIV patients is a very heterogeneous group of patients and that there are, recent studies in autoimmune and autoinflammatory disease, that pointed out that, the risk benefit ratio for prophylaxis, Was actually not in favor of routine prophylaxis in this setting. So we are back to the point [00:46:00] that we discussed with Josh. I think that, addressing a risk also needs to include immunity, assessment, and also a, case by case discussion. And I think that this is a very important point.
All the patient that have been included in this trial were a failure of prophylaxis for many different reasons. And we need to update also these indications for prophylaxis based on such data. there are very nice data in lupus, for example, showing that, the risk benefit ratio was not in favor of routine, prophylaxis in patients with lupus.
But at the same time, there are lupus, patients, requiring prophylaxis because they have either. Other, risk factors, like, acute kidney injury or chronic renal insufficiency, but also there are other risks that should be monitored for example, monitoring CD four cells, et cetera, et cetera.
Josh: Getting back to the basic fungal biology, the. Pneumocystis dies very quickly with this therapy. And, the fungal burdens, as we talked about, were very high. And it gets [00:47:00] back to what biomarkers can we have in future trials. And we typically do not repeat, pneumocystis PCRs over several days.
And yeah, the beta D glucan, but that lags, what do we need to develop for the next round of trials that really shows us information that may help us guide which patient should be treated with, what therapy? And Emily, it sounds like you are well on the way to answering more of those questions, which is super exciting.
Emily: Yeah, hopefully we'll be working on this together. I'll be reaching out to both of you after the podcast. the fact that you can do a non-invasive nasal swab now and pick up the PCP PCR is something really interesting. Right?
Before, to detect PCR, we had to do much more invasive tests. Of course, the nasal swab, you know, there can be issues with sensitivity and specificity, but it also, because it's so easy to get, could be incorporated into trials, to look, to see, to predict who's getting better, [00:48:00] to see how quickly the fungal load is decreasing to decide duration of steroids, to see who needs secondary prophylaxis and for how long.
I think that that test, is something really neat that we should be incorporating into future studies.
Elie: Yeah, we are still doing a lot of, induced sputa in these patients. we just need to change the threshold with which we decide whether we are, making differences between colonization and infection on whether we use a nasal swab or nasopharyngeal aspirate or, an induced sputa.
The thing is with, our laboratories no more doing, immunofluorescence and staining. I think that the benefit from BL is very conceptual on maybe understanding what are the cells involved in the alveoli to better maybe expect what we can gain from steroids. but of course, there is a, a very clear benefit from non-invasive tests, and Virginie is about to start a trial comparing.
Elie: Invasive, which is semi invasive actually with bronchoscopy and BL and non-invasive diagnostic tests. Maybe Virginie, you want to say a [00:49:00] word on that?
Virginie: it's not only for, pneumocystis. pneumonia, it's for, Pulmonary disease or acute respiratory failure in immunocompromised patient. And we will randomize patient, who will care, will be care with, bronchoalveolar lavage, and, the patient will, have also non-invasive, test.
the test would be, decide, according to the underlying disease, according to the CT scan, pattern, as you said before, Emily, because the probability of, Maybe bacterial infectious is the same before and after the CT scan. And so we'll randomize, more than, 150 patient per group, to compare patient with or without, bronchoalveolar lavage.
Virginie: And also we are, doing a study with a laboratory in, our hospital and we compare, the amount of, the, PCR in nasal swabs, by Washing. And then induced sputum or BL it depends on the patient. And then we will compare the, [00:50:00] amount of, PCR for pneumocystis, in all those patients. But it's ongoing study. So,
Elie: and Emily, we have a paper and the review that should come up. I hope very soon that really works on the pretest probability for every disease, not only for pneumocystis, but as you mentioned, and Josh also said a word about that, we are very quickly moving from one diagnosis to one to two diagnosis.
We are trying to understand, and even before sampling, just at the bedside, what is the pretest probability? Because I think that it makes a very big difference on when we are starting appropriate treatment. And I'm sure that this is also a major confounder when we speak about, uh, day 90 or day 30 outcomes.
And in the trial that, Virginie is now, day 90 mortality is the primary endpoint.
That's fantastic.
Josh: Thank you so much for joining us to share insights from your clinically impactful papers. We want to take a moment to ask if you have any last messages to give our listeners and [00:51:00] we'll start with you, Virginie.
Virginie: you know, when I start the study, one of the investigators said to me, you are brave because you'll never finish this study.
But, now I know that study will going to change your practice. So I will say to the younger that if you have a clinical question in your practice and you don't hesitate to perform a study, even if it's took a long time, because it's very, uh, interesting to have the answer.
Josh: I would say brave and fabulous.
Elie.
Elie: Well, I'm very happy, to have been part of this, conversation and these discussions. I was saying that we should work together and I think that this is the essence of what we do. I was also about to mention to Emily that we already work together because together we are a chain of people working together to change the practices at the bedside.
And I love this, inter specialty, collaborations where ID people, ICU people, pulmonary people, and other specialties work [00:52:00] together with the same aim. And the thing is today, I think that, pneumocystis brings passion. A lot of passion people working on it are really passionate and together.
I think that we share the same passion, so we should move on.
Josh: I think it also underscores how important science is in terms of driving change and patient care, and how we need to continue to support the use of rational scientific data. Emily, thoughts?
Emily: Yes. I love what Virginie said about, you know, inspiring the younger generation of, clinician trialists to, you know, go out there and try and answer the question.
It's hard, but you can do it. don't let people tell you no. and I love what, Elie said about how, you know, we're, we're working together already. Maybe I'll just end on something like a, lighter note. You know, one thing we didn't, debate was the acronym for pneumocystis pneumonia. So, are you guys PCP people, which I'm told we can still use because of the C and the 'cystis', or have [00:53:00] you converted to PJP?
Elie: Well, I like pneumocystosis because this is what it is.
Virginie: I usually, use PCP
Emily: Josh.
Josh: I think there's an age, there's an age thing here for those of us that, that dealt with this, daily, if not multiple times daily in the, late eighties and early nineties. we still are scarred by the, old name of the rat carinii so many people still do use PCP, but I use Pneumocystis jirovecii though .
Nav: I'm PCP, but I try to say pneumocystis more in my courses when I'm teaching my students, just so they're not confused and I have to spend an extra five minutes that I don't have to explain that. So, you know, I tried to, find a middle ground, Emily, as you said.
Emily: Wow. These were very diplomatic answers. I am PCP all the way. Oh, thanks everyone. Maybe
Elie: this is a French corner.
Josh: Well, PCP confuses some people because of the drug that, is associated with the same acronym. So.
Emily: Good point.
Nav: I really want to thank you all again and [00:54:00] especially thank you to our guests, Emily McDonald from McGill University Health Center in Montreal, Canada.
Elie Azoulay and Virginie Lemiale from the Paris City University and St. Louis Hospital in Paris, France. thank you for listening to communicable the CMI Comms Podcast. This episode was hosted by. Josh Nosanchuk and me Navaneeth Narayanan, both from the USA and Associate editors at CMI comms ESCMID's Open Access Journal.
It was edited and produced by Dr. Katie Hostettler-Oi and peer reviewed by Dr. Arjana Zerja from the Mother Theresa University Hospital Center, Tirana, Albania. Theme music was composed and conducted by Joseph McDade. This episode will be citable with the written summary referenced by A DOI in the next eight weeks, and any literature we've discussed today can be found in the show notes.
You can subscribe to Communicable wherever you get your podcast or find it on ESCMID's website for the CMI Comms Journal. Thank you for listening and helping CMI Comms and ESCMID move the [00:55:00] conversation in ID and clinical microbiology further along.