Welcome listeners to this presentation for the 51st Congress on Critical Care in 2022 organized by the Society for Critical Care and Medicine. My name is Heijder de Vries. I am a research fellow at Amsterdam UMC in the Netherlands. And I'm also a PhD candidate with a focus on diaphragm function, lung protective ventilation, and weaning from mechanical ventilation. And it is my great pleasure to talk to you today about the trial that will be published in Critical Care and Medicine at the same time as this presentation is published. And the trial is titled Lung and Diaphragm Protective Ventilation by titrating inspiratory support to diaphragm effort. A randomized controlled trial. I have no conflicts of interest to report related to this work. So let's get right into it. The diaphragm, my favorite muscle in the human body, is the most important muscle for inspiration. As you all know, it's a dome shaped muscle that separates your thorax from your abdomen. And to measure the pressure generated by the diaphragm, the reference method is the transdiaphragmatic pressure, which can be obtained by subtracting the gastric pressure from the esophageal pressure, essentially calculating the pressure difference caused by this muscle between the two compartments. In healthy subjects, the transdiaphragmatic is between three and eight centimeters of water per breath during rested breathing to slight exercise. But this muscle has a great reserve of power and can generate up to 100, maybe even 150 centimeters of waters in some subjects. Sadly, the diaphragm can become weakened. And this happens quite commonly in ICU patients. Observational trials have found that one half to three quarters of ICU patients meet the criteria for severe diaphragm weakness at some point during their admission. And importantly, this weakness is also related to worse ICU outcomes. Here on the right, we see a bar graph showing subjects with diaphragm weakness in black and without weakness in white. And if diaphragm weakness was present, up to 50 percent of these subjects experienced difficulty weaning from the ventilator. If there was no weakness, this number was only about 10 percent. So a logical train of thought would be to assume that preventing diaphragm weakness might also improve ICU outcomes. But of course, this requires validation in studies. Now, what do we know about the pathophysiology of diaphragm weakness in the ICU? I'd like to start by discussing one of the landmark papers that has shown that disuse of the diaphragm during mechanical ventilation can lead to atrophy. This beautiful study was published in 2008, and they compared the thickness of diaphragm fibers in brain dead organ donors who had been ventilated on controlled ventilation for one to four days. And they compared these with subjects that had not been ventilated for a prolonged period of time. Usually subjects that immediately died in car accidents and the likes. And what you can see here quite beautifully is these are cross sections of diaphragm fibers in the critically ill patients. And the area, the cross sectional area of the muscle fibers is much lower in these ICU patients compared with these control patients. And you can see that clearly in this graph over here, where the darker color is the ICU patients, over half of their thickness is gone compared to control patients that have not been ventilated. Another study that demonstrated that this leads to weakness was published in 2011. This is a bar graph showing the strength of the diaphragm measured with magnetic stimulation on the y-axis. And on the x-axis, we say days of ICU admissions. Now the white bar is subjects that have been ventilated for only half an hour during, for instance, surgery, usually not critically ill patients. And the black bars are critically ill patients. And already at the first day of admittance, critically ill patients have significantly lower total diaphragm strength compared to the subjects undergoing surgery. And then during the course of the first week of admittance, they lose about 30 to 40 percent of their strength. And this dotted line shows the target point after which we say that a patient has diaphragm weakness. And what we can clearly see is that on average, most patients in this cohort actually develop diaphragm weakness by the end of the first week of admittance. Our own group has confirmed these preliminary findings in actual ICU patients. So here we see another biopsy of a control subject's diaphragm with large fibers. And here on the right, we see the diaphragm of a critically ill patient. Again, fibers are much smaller, but there's also a certain degree of chaos in this fiber. They're not nicely arranged. There's a lot of infiltration of inflammatory cells. Those blue little cells are inflammatory cells. And there's a total disarray of the muscle. And this does not only fit the idea of atrophy, but also it raises the possibility of load-induced injury, where the diaphragm has to work very hard for a prolonged period of time in a muscle-toxic environment during ICU admissions, containing steroids, toxins, sepsis, that the muscle might become inflamed and also lose function, not only because of disuse, but because of very high use. But this was only a hypothesis when we saw infiltration of inflammatory cells in the diaphragm, which is not a pattern that fits atrophy. Indeed, we found that there were significantly more macrophages and neutrophils present in the diaphragm biopsies of critically ill patients compared to control subjects. Another important observational trial in this regard used ultrasound to study the thickness of the diaphragm in ICU patients over the first week of admittance. And they studied about 100 subjects. And on the y-axis here, we see the change in thickness compared to baseline. So everyone starts at 0%. And then what we clearly see in this image is that about 40% of the subjects in this cohort developed a decrease in diaphragm thickness, most prominently in the first four days. And then it seems to kind of plateau. 40% remained almost equal in diaphragm thickness. And about 10% to maybe 20% showed an increase in thickness. And indeed, the group that increased in thickness was the group with high effort. The group that remained equal had physiological diaphragm effort, meaning measured with ultrasound, it had diaphragm effort that can also be observed in healthy subjects during rested breathing or during slight exercise. And the group that tended to decrease in thickness had the lowest amount of effort, meaning effort below the values that we could observe in healthy subjects. And what they further showed in another paper is that this change in thickness is related to the chance to be liberated from mechanical ventilation. Here on the y-axis, we see the daily hazard of being liberated from mechanical ventilation. And indeed, this chance to be liberated is highest when there's no change in diaphragm thickness from baseline. It gets substantially lower when there's a decrease in thickness, but it also gets lower when there's an increase in diaphragm thickness. So an increase in thickness in this regard seems to not be just a physiological adaptation of having to work harder during mechanical ventilation, but it seems to be correlated to worse outcome. Based on all the studies I've mentioned before and others in animals, a new approach has recently been advocated called a lung and diaphragm protective approach to mechanical ventilation. And this is basically an extension on lung protective ventilation. It focuses on obtaining low lung stress, so using low tidal volumes, using low driving pressures, while also trying to keep diaphragm activity in the physiological range, the range observed in healthy subjects during slight exercise. And what we have done is we have conducted a trial in which we had the hypothesis that we can use the inspiratory support setting on the ventilator to titrate diaphragm effort, with the intention of obtaining physiological efforts, which we defined as a PDI between 3 to 12 centimeters of water, in critically ill patients on partially supported mechanical ventilation. And the design was a randomized controlled physiological trial. We included all patients with acute respiratory failure on supported mechanical ventilation, meaning either pressure support or NAVA, and we included 40 of them. We randomized them equally between the intervention group in which we tried to titrate support and the standard of care group in which the attending physicians were given total control over the ventilation settings without any feedback from us. And then in the end, after 24 hours, we compared both groups. The primary outcome was a percentage of breaths where the PDI was in the physiological range that we set, and secondary outcomes were measures of lung protective ventilation, including tidal volume, driving pressures, transpulmonary pressures, and biomarkers for lung injury. To do this, every patient received a specialized nasogastric catheter. This catheter has two balloons attached to it. One will end up in the stomach and one will end up in the lower third of the esophagus. And these balloons allowed us to measure gastric pressure and esophageal pressure. And then taking gastric pressure minus esophageal pressure is the method to calculate the transdiaphragmatic pressure. And they're also a feeding catheter, so we didn't have to insert extra catheters into our patients. Next, these were connected to a personal computer, which allowed us to store and collect all the breaths that a patient took in the 24-hour period. Here we see flow, airway pressure, esophageal pressure in green. And this red line over here is the transdiaphragmatic pressure. The blue line is the transpulmonary pressure. And we would then measure the average transdiaphragmatic pressure in the first two minutes. And the results we would enter into a very simple titration protocol. And depending on the height of the PDI, we would take either decrease support by two centimeters water or increase it. And we also took blood samples at the beginning, after 12 hours, and after 24 hours of study inclusion. Now, I'd like to show the titration protocol in a little bit more detail. It was actually very simple. So it always started with calculating the transdiaphragmatic pressure. And if pressure was between 3 and 12, there would be no action. And the protocol would be repeated one hour later. If PDI was too low, meaning there's too little activity of the patient's diaphragm, they might be at risk for atrophy. We would decrease the amount of support given by the ventilator, hoping that this would mean that the patient would use their diaphragm more. Then we would check whether we had not decreased pressure support by too much. So if there was a very high rise in breathing frequency or if tidal volumes became too low, lower than 4 milliliters per kilogram projected body weight, or if there were signs of dyspnea, we would halt the titration and come back in an hour. However, if they tolerated the decrease in pressure support, we would calculate PDI again. And we would keep repeating this until PDI would be between 3 and 12. On the other hand, if PDI was too high, we tended to increase support, hoping that the patient would do less, the ventilator would do more. We would check whether tidal volumes became too large, meaning higher than 10 ml per kilogram, or if plateau pressure became higher than 30 centimeters of water. If this was not the case, we would calculate PDI again and keep on repeating this loop until PDI was between 3 to 12 or until we met one of the stopping criteria. We assessed a total of 451 patients, and of this group, about a third of the patients met all the criteria for inclusion. Most patients that did not meet the criteria for inclusion were either never ventilated in a partially supported mode because they got transferred or because they died, or they were assisted in a supportive mode, but the physician expected that they could be extubated within 24 hours. So then the study would not have much point. So about 1 in 3 patients met all the criteria. We could only approach about 60 for consent because if we already had a subject included, we could not measure another one because we only had one measurement set up. 40 of these 60 accepted and were randomized. 20 were randomized to control, 20 to the intervention, but we sadly lost one patient in the intervention group because he self-extubated and did fine, and we did not replace the subject. In total, 68% of our inclusions were male, and they were on average 65 years old. About two-thirds were medical admissions and one-third were surgical admissions, and a single patient had just a neurological problem. The PF ratio was on average slightly below 200 millimeters of mercury, and over 90% of all subjects met all the criteria for ARDS according to the Berlin definition. Expected hospital mortality was 40 to 45% based on the Apache and subscores, and subjects were ventilated on average three days on controlled ventilation before inclusion and a median of four days of partially supported mechanical ventilation before inclusion. Now here we see a smoothed histogram of the PDI at baseline. So the x-axis is BDI, and the dotted lines represent the target range that we set for the study between 3 to 12. And what we can see over here is that only about 7% of all breaths were below the threshold at the baseline, and about 37% were above the baseline, and some breaths were very high above the baseline. So we had subjects generating 20 to 30 centimeters of water with their diaphragm, which is very high effort in healthy subjects even. So these patients were working very, very hard. And then, of course, in half of these patients, we would titrate support, and the other half we would not intervene. And here are the primary outcomes. So the percentage of breaths with a PDI between 3 to 12 centimeters of water, two box plots, red is the control group, bluish-green is the intervention group. And it is immediately apparent that the intervention group has much higher percentage of breaths in the target range than the control group, so 81 versus 35%. Every dot is an individual subject. And what you can also see is that most subjects in the intervention group have between 75% and 100% of breaths within range, but two subjects over here had less than 50% of breaths within range. And those are the subjects in which the titration was not effective and in which we met the stopping criteria that I've shown before. Here on the left, we see the control group. And here we see a very widespread. Some subjects were very, very fine and have more than 75% of their breaths in range, while most are below 50%. Here we see a dot graph showing the fraction of breaths within range on the y-axis over the hours of the study on the x-axis. Both groups start about at 50% at an equal point. And the intervention group, again in green, tends to very quickly increase the percentage of breaths that they have within range because of the titration protocol and then fluctuate around a plateau at about 80% of breaths in range. While the control group tended to decrease slightly during the study and fluctuate at about 40% of breaths within the preset, predefined diaphragm protective range. So how many support adjustments did we have to use to attain these results? Well, here again, box plots on the left, the control, on the right, the intervention. And what is striking to me is that in the control group, most subjects tend to get between zero and two adjustments to their ventilatory settings in a 24-hour period. While in the intervention group, only a single subject did not require any changes to their ventilator settings according to our protocol. But most of the subjects required around 10 adjustments and some even required 16, showing that subjects tend to behave very differently and really shows how dynamic diaphragm effort can be in a 24-hour period necessitating monitoring. How much was support changed? Again, two box plots, but now on the y-axis, we show the total change in support compared to baseline, meaning that if we tended to decrease support, subjects would end up negatively. And if support was mostly increased, they would end up with a positive number. In the control group, the mean is perfectly zero with just as many adjustments upwards and downwards. While in the intervention group, again in green, it's clear that most subjects tended to get an increase in support, while two subjects tended to get a sharp decrease in support. Probably these are the subjects with low effort and these are the subjects with high effort. How did all these adjustments, these increases mostly affect lung protective ventilation? We always assume that increasing support increases tidal volume, but this graph over here shows that on average, the titration protocol did not affect tidal volume. On the y-axis, we have tidal volumes in mLs per kilogram. And on the x-axis, we have the time of the study and both groups start exactly at the same point and tend to remain perfectly equal on average over time, meaning that most likely increasing support led to a reduction of patient effort, but the net effect on the minute ventilation was close to zero. The same can be seen for the transpulmonary pressure, the gold standard to assess tidal lung stress. Equal in both groups, slightly higher perhaps at baseline in the intervention group, but this did not reach statistical significance and the group tended to remain equal over time. Same deal with the driving pressure here, slightly lower in the intervention group, not significant and no change over time at all, showing that titrating support is really effective at changing your patient's respiratory effort without having too much of an impact on average on parameters of lung protective mechanical ventilation. Additionally, none of the 16 biomarkers that we tested differed significantly between the groups in the 24-hour period, which is, of course, a very reassuring fact. So to conclude, titration of pressure support led to more breaths with a PDI between 3 to 12 centimeters of water, 81% in the intervention group versus 35% in the control group. And it did not affect parameters of lung protective ventilation, including tidal volume, transpulmonary driving pressures, and protein biomarkers for lung injury. Titration was effective in 15 out of 18 subjects in the intervention group. One subject did not require titration and two subjects did not react to titration. So titration is effective in the vast majority of subjects. However, caution is still warranted because there was no effect at all of titration in two subjects. And I've shown over here on the right the PDI, the amount of support, and the tidal volume of one of these subjects. And what can be seen here is that at baseline, trans-cyclic pressure was slightly elevated in this subject. So we tended to increase support, which led to a slight reduction in effort, but also an increase in tidal volume from about 500 to 750 ml, which is a very substantial increase. And then later on, PDI tended to increase again. We tended to increase support again. It had some effect on effort, but we could not get the average transpulmonary and trans-cyclic pressure into the ranges that we deemed safe. So it did not have a positive effect in this patient, which really shows that you really have to monitor the effect of your intervention in each patient individually. Interestingly enough, both patients had an elevated pH even at the onset of titration, suggesting that their elevated drive was not related to ventilation, not related to CO2, but it might have originated from agitation or pain. And in such circumstances, it cannot be expected that increasing the support of the ventilator reduces their respiratory drive. Strength of our study includes the fact that we used the reference standard to assess diaphragm effort. We measured continuously for 24 hours and analyzed each breath in this period, and we only used support titration to get diaphragm effort in the range that we wanted. We did not use sedatives or muscle relaxants. And I think that's an upside because both of these categories of drugs are related to prolonged ICU admission and to diaphragm atrophy. Some limitations include that we did not include patients immediately at the onset of mechanical ventilation, and that the trial was not designed to assess the impact of such a strategy on diaphragm structure, function, and ICU outcomes. We really wanted to study the feasibility first before moving on to larger trials. And I think this trial really shows that this method is feasible and is highly effective. Future directions. I think it's important that we try to validate more feasible methods to measure lung stress and diaphragm effort at the bedside. I think it's not realistic to expect every patient to receive a double balloon catheter and a dedicated personal computer or ventilator to calculate lung stress and diaphragm effort. I think finding more feasible methods to assess lung stress and effort would really improve the feasibility of this lung and diaphragm protective approach to ventilation. Next up, I think we need to do larger trials to assess the impact of a lung and diaphragm protective approach on diaphragm structure and function. And finally, large multicentric trials powered on ICU outcomes are required to show whether indeed diaphragm effort is causally related to ICU outcomes or whether it was just a confounder, just a marker for disease severity. I think that's really important right now and that those trials should be conducted as soon as possible. So to summarize, I've talked about the following points. First of all, attaining diaphragm effort in the physiological range might be able to improve the outcome of ICU patients by preventing the development of diaphragm weakness, thereby improving the weaning phase of mechanical ventilation. Furthermore, a proof-of-concept RCT that we have conducted shows that simply using the support to titrate diaphragm effort is feasible and is very successful in most of the subjects and has very little effect on average on volume, pressures, and biomarkers for lung injury. So it has little effect on the currently best available markers for lung protective ventilation. But larger multicentric trials are required to assess the impact of this strategy on patient outcomes. If anyone has any questions, I'd be happy to answer them by email. My email address is listed over here. I would love to hear your comments and ideas on the manuscript. I invite you all to read the manuscript, of course. Furthermore, I'd like to thank the committee for inviting me to give this presentation and for publishing the manuscript. I'd like to thank you all for listening.

lung and diaphragm protective ventilation

ICU patients

diaphragm weakness

mechanical ventilation

randomized controlled trial

diaphragm effort