David Berlovitz is a clinician scientist who holds a joint appointment from the University of Melbourne and Austin Health in Melbourne, Australia. He maintains his specialist respiratory physical therapy practice within the Victorian Respiratory Support Service, the statewide provider of domiciliary ventilatory support to over 1,000 Americans who live in their own homes thanks to breathing assistance overnight. While obtaining his Ph.D., Dr. Berlovitz discovered that spinal cord injury causes immediate obstructive sleep apnea. This novel finding alerted the world to the previously unrecognized problem of immediate and severe sleep disordered breathing in tetraplegia. Dr. Berlovitz collaborates internationally and leads a research team examining the causes and treatments of breathing and sleep disorders and neuromuscular disease, especially amyotrophic lateral sclerosis and natural spinal cord injury. His research encompasses respiratory physiology, sleep, health systems research, and clinical trials of therapies and care models. His current research trials focus on improving access and uptake of ventilatory support using ventilator-recorded data and artificial intelligence to optimize noninvasive ventilation. And multicenter clinical trials testing therapies to augment neuroplasticity after spinal cord injury and whether polysomnographic titration of noninvasive ventilation in people with amyotrophic lateral sclerosis modifies usage. For his advances and contributions in medicine, we honor Dr. Berlovitz. Please join me in welcoming Dr. David Berlovitz to present the Margaret Frommer Memorial Lecture. Thank you. Thank you very much, John. That's a lovely introduction. I'd also like to thank the Home-Based Mechanical Ventilation and Neuromuscular section for this privilege. I'd especially like to thank the people living with spinal cord injury who've made all of this possible. And I'd also like to specifically acknowledge Margaret Frommer, who was an extraordinary advocate for disability change and inclusion. And if we can do a fraction of what she did, yeah, my work's done. So this is what I'm speaking about. I don't have any commercial conflicts of interest. And the learning objectives are as written. But before I start, I would like to acknowledge the country on which most of the work was done, that of the Wurundjeri people of the Kulin Nation in Australia. And I'd particularly like to pay my respects to their elders, but also to traditional owners of the lands on which we are now. None of this would be possible, though, without research grant support. And we've had extensive support from a very large number of people over the years. So I feel that's important to acknowledge. So by way of background, in Australia, we don't have RTs. We have respiratory physical therapists. So much of the work that RTs do in North America is included in our physiotherapy degrees. So that's why us respiratory physiotherapists hang around in physiology labs and do sleep studies. And so that's exactly what I was doing clinically. Our unit, as well as having the home-based ventilation service, which John described, also is the statewide service for the spinal injury. So if you have a spinal injury in the state of Victoria, which is about 6.4 million people, you come into our health service. I was working clinically in the respiratory and sleep medicine unit. And all of these spinal patients kept coming in with upper airways that didn't work, not hypoventilation. Their CO2s generally were not raised, particularly if they were acute. If they were chronic, maybe, but not acutely. And so this didn't make any sense. Because as you'll all know, fundamentally, a C5 injury should not affect your upper airway function. Genioglosses should be fine. So I managed to scrounge together some money to enable me to do a PhD. And this is the most disappointing slide you'll ever see. It's five years of my life in one slide. What we did was we conducted a prospective cohort of every person who came into the spinal unit with an acute injury, followed them up to 12 months with full bedside polysomnography, so full channel PSG, and measurements of lung function, medications, all the kind of things you might expect. Based on an apnea prediction index, it's likely that a sort of conservative estimate might be that 10% of the people in the cohort may have had OSA before their injury. And certainly, in fact, one person fell asleep at the wheel and had a single vehicle accident. But as soon as I could measure sleep disorder breathing in people, by two weeks, the prevalence of what was then the cutoff for OSA yes, no, using the Chicago criteria, was 60%. By one month, it was 63%, when it peaked at 83%, so four out of five people at three months, and then got slightly better over the following six months out to a year. But we're still dealing with 60% of people with an incident injury who now have obstructive sleep apnea. It's the only model of acute OSA in humans we know of. An early question we tried to examine was, in the context of the tremendous change that occurs to you after a spinal cord injury, what additional burden might you be living with because of your sleep apnea? So these were data from a population study we did. And I'll just take you through all of these pretty colors. For people who've done any health economic analysis, a health utility is a way of generating qualities and dalies, quality-adjusted life years, et cetera, where they range from a perceived health state of zero, which is my health state, my self-rated health state is the same as death, and it can, in fact, go down to below death, where one might feel that their life is not worth living, up to one, being perfect health. And these are the bands from Australian population data, and this is the Australian population average, somewhere between good and very good. When we surveyed all these people with motor and sensory complete injuries versus incomplete and OSA yes or no, you can see that people living with a complete spinal cord injury and sleep apnea were willing to trade off four years of their life for one year of better health. So this is a substantial decrement. Major depression with suicidal ideation, the average health utility for people living with that is about 0.43. So these are people with substantially impaired health status. And of course, that was significant. But the thing I want to highlight is this is the minimum clinically important difference for health status. And so you can see that we've got quite a lot of room to move here, if we can make OSA less bad for people. It's likely to be important for them. Recently, one of an implementation scientist, Marnie Graker, who works with us, has taken all of the published data and estimated the population prevalence. And you can see here, so with an AHI of greater than 5, you can see that the point prevalent estimate is that about 83% of people living with tetraplegia in the community will have an AHI greater than 5. And this is the range of reported norms for people without spinal cord injury, the able body population. If we go up to 15, we're still up around 60%. And you can see now that there's quite even more daylight between the non-injured population. And when we get up to above an AHI of 30, we're still up around a third of people in the zone. In fact, we don't have reliable able body population data in that range. So it's highly prevalent. So for us, one of the next questions was why. What is the reason why having, as I say, maybe a C5 injury gives rise to upper airway dysfunction? And thankfully, there is a phenotyping kind of protocol really founded at Harvard that goes through all of the anatomical and neuromuscular determinants that were known at the time, likely determinants, of upper airway dysfunction. So the airway might be too small. You might have an increased resistance. There might be stickier lining fluid. So in Sjögren's, the reason people have higher prevalence of OSA is because the upper airway lining liquid is really sticky. Their airway is closed, and they stay closed. That wasn't a factor here. We thought it might be because of parotid innervation. There might be a more collapsible airway. It might just be due to smaller lung volumes and lower distal tension on the upper airway from a decreased end expiratory lung volume. And then all the neuromuscular control factors. So there may be an alteration in the genioglossus reflex response to a negative pressure or altered respiratory-related evoked potentials, the response of the cortex to the upper airway stimulus, or changes in arousal threshold. So I also slept like this, but I don't have tetraplegia. This is one of our participants with tetraplegia, with needles in julioglossus, pressure catheters in the back of the nose and the back of the throat, the full shebang, incredibly generous participants in this trial, and full polysomnography. That nose mask is connected to a pneumo-tach to give us flow, and when he was asleep we taped the mouth. Also allowed us to deliver the negative pressure pulses and all those kind of things. So lots of measurements. We brought people in and awake. We looked at the upper airway lining liquid. We measured lung volumes with helium dilution, measured nasal and pharyngeal resistance, which I'll go through. Then we did the brain, so the evoked potential and the way in which junioglossus recruited relative to a drop in upper airway pressure. We tested the negative pressure pulse, and we gave them some food, a bit of pizza and television. Went off to sleep. We measured collapsibility doing dial downs to get PCRT, the critical closing pressure of the upper airway. Measured arousal threshold, again with pressure pulses, and muscle responsiveness, the junioglossus response during sleep. We had nine able-bodied controls and 11 people with tetraplegia who were quite similar in age, a little bit bigger. AHI severity was slightly higher in tetras, but fairly well matched. So the first data I'm going to go through is the nose resistance data. The nose is really vascular, is really, really, really vascular. But after a spinal cord injury, the nature of the interruption to the sympathetic nervous system is important. So any lesion above T1 will interrupt the sympathetics, because they travel down, come out at T1 to T6, and then go back up to the superior cervical ganglion. So if you've got a tetraplegic lesion, you're essentially sympathectomised to varying degrees. So normally, if your sympathetic nervous system is active, the hole in the donut is bigger, because sympathetic tone constricts the vasculature and opens up your nose. But if you're running just on parasympathetic tone, the upper airway is constricted, because the vasculature is engorged, the basement membrane gets a little leaky as well. So what we did was, as I said, we had pneumatac and two pressure transducers, and we measured baseline breathing for 10 minutes, then sprayed a topical sympathomimetic into the upper airway, and then made the measurements again. And what we found was that nasal resistance, Laura, who did these as part of her PhD, loved pink. So all of the people with tetraplegia are in pink. People living with tetraplegia had nasal resistances of up to 12 times that of the able-bodied population data. But in both cases, phenylephrine was able to reduce it down quite substantially. And of course, that was significant. Then we moved on to looking at the negative pressure-related pulses. So during sleep, you apply a pulse of negative pressure. And also during wake, you can apply a pulse. And then look at the evoked potentials at the frontal, central, and occipital lobes. And you can see here that, despite the same stimulus, you can see that the pink lines are a bit, the peaks are a bit smaller and a bit delayed. So what the summary data showed was that the latency, in particular, was significantly delayed between tetraplegia and able-bodied. And the size of the peaks was different, but not statistically significant. But one thing we did see, which we actually still don't understand, is we had some paradoxical inhibition of genioglossus in the people with tetraplegia, which is incredibly rare to see in people without a spinal cord injury. I still don't know what it means, but that's what we found. So as I said, I'm sorry I'm running through this physiology, but there's a lot of it. We measured the peak during sleep and really found that the collapsing pressure of the upper airway was no different between those living with tetraplegia and those without tetraplegia. So it's not that the upper airway is more collapsible. And in fact, Abdul Sankari, who from Wayne State, has done some similar work and essentially found exactly the same results. So putting all this together, what it means is that people with tetraplegia and OSA do have a higher nasal resistance and a delayed and far more varied cortical response to an upper airway stimulation, implying that there are some modifications of the reflex pathway, but really we don't have enough data to understand exactly why. And we know from other data that we've got kind of this stepwise increase in dysfunction from OSA tetra to OSA able-bodied controls. Abdul's also done some really nice work looking at CO2 control, which has shown that ventilatory control, particularly at sleep onset, is quite different. People with tetraplegia sit at a different spot on the isometabolic curve and that their CO2 reserve is much smaller. A smaller amount of CO2 change, particularly at sleep onset, will give rise to a greater reduction in ventilatory drive and hence the centrals that you see at sleep onset frequently. Also has done some other work looking at other aspects of peripheral chemosensitivity. So we now wanted to look at shape. So we threw everybody in a 3T scanner. Most of the time on this study was getting clearance for the various bits of metal in people's necks because nobody had done this before. So the poor RA, she probably spent 18 months of her sort of two and a half years on this project dealing with safety clearance. Anyway, we managed to pop people in. This is Shona Ruther, who used to work in MRI with the group. We had three groups. She did this in concert with the people at NeurA in Sydney, in Randwick. And we scanned a number of people. In the Melbourne cohort, we also repeated the phenylephrine experiment to see if there was a structural change that underpinned the nasal resistance change. And in Sydney, Lynn Bilston, who's the engineer there, did some really cool dynamic airway work. So normal kind of imaging and reconstruction, volumetric reconstruction. And what we found was that it's not just the hole in the donut. There's no difference in the velo, oro, or hypopharynx airway area between the two groups. Phenylephrine made a difference. Phenylephrine absolutely increased the size of the velopharynx. But it was an OSA effect, not a spinal cord injury versus not effect. So the engorgement, we know about swelling of the tongue, et cetera, in the presence of repeated sleep apnea events. And we know from other people's work around changes in extra-luminal tissue pressure that you see in uncontrolled OSA. I think this is probably what we're looking at here. So when you do an MRI and you've got a really clever physicist, they can take a planar MRI and generate voxels, so three-dimensional pixels that sit over the top of the tongue. And then you can watch how the tongue deforms as you breathe and look at the movement of the tongue and the dynamic size and shape of the upper airway. And so that's what we have here. This is one of the participants with tetraplegia. And you can see that during inspiration, this area at the back of the tongue gets larger as the tongue moves forward and down and out of the way to allow for respiration. Now, unfortunately, we can't do this in sleep, so this is wakefulness data. You have to apply a vibration to the, you have to clamp the tongue and apply a vibration to it, and apparently that interferes with sleep. What we did find was that rather than what might be anticipated, which is that people were moving less, they were in fact moving more. So during wakefulness, we actually had an increased sort of anterior displacement of the tongue in the people living with tetraplegia. So the tongue was better able to get out of the way during wake. What we don't know is if when we lost the wakefulness stimulus, perhaps this is part of the pathogenesis, and that that degree of active control was inhibited during sleep, but that's pure speculation, we don't know. OK, so in terms of the imaging data, we found that OSA amongst those with cervical cord injury is probably not due just to their upper airway being smaller. We do have some evidence that it is acutely sort of perioperatively, particularly because many people get fixed through an anterior approach. They get quite a lot of neck swelling, but that goes away after a couple of weeks. It's surgical, parasurgical. Phenylephrine data does suggest that there's probably a role for a topical vasoconstriction, and we did go on and do a little RCT, which I'll talk about. And the tagged imaging results, what I did mention, there's a wide degree of variability from person to person, but it does suggest that during wakefulness, at least, the tongue can get out of the way. So we've got to treat... At some point, we have to deal with treatment, which is, of course, the elephant in the room. And the reason is that CPAP mask application, when you can't move your arms properly, is a real problem. So one of the alternative treatments... I'll come back to CPAP, but one of the alternative treatments we tested was, as I said, based on this vascular phenomenon, we'd shown that there was much more baseline day-to-day... So all of us have variability in our nasal resistance from day-to-day, but compared to the control arm, people living with tetraplegia had much, much, much higher anterior and posterior nasal resistance when we measured it from day-to-day. So we did a small double-blind crossover study of phenylephrine versus a placebo nasal spray for three days. We felt like three days was as long as we could get away with without rebound, so that was just for nocturnal use, and then swapped people over. And we found no effect of phenylephrine on the AHR, the ODI, or whatever. There was a subsequent meta-analysis, which included this paper, and did suggest that perhaps there's an effect of ODI. And in one of the year-in-review talks yesterday, there's been a recent publication demonstrating that there are now perhaps some agents that might be able to be used intranasally to modify OSA severity. So I think we'll be going back to this with a better agent. There was concern with thinking about an oral appliance in the context of not being able to get your hand in your mouth to flick it out overnight. So we essentially did an uncontrolled safety and feasibility study first to really test whether it was feasible and safe. So in this group of participants, we had AHI and other measures at baseline, and then after post in this case was at maximum protrusion. So we didn't have a set time point. We had two... This was run over both Melbourne and Sydney, and we had dentists involved in both of these studies who went out to people's homes, measured them up, made an oral appliance, and then progressively advanced it over time. And we had reductions in AHI, improvements in attention and information processing, which... So the PASAT, the Paced Auditory Serial Addition Task, is exquisitely sensitive to sleep fragmentation and deprivation. One of our rationales for using neurocognitive test battery in this case was that people living with spinal cord injury have particularly poor return-to-work outcomes after their injury. And so we were interested in things that might also sort of limit their decrement in high-level cognitive performance secondary to their sleep-disordered breathing. And we had improvement in KSS, so a marker of state sleepiness taken at the start of the day. But one interesting thing was that normally it takes about three months to fully bring somebody's jaw forward treating their OSA. In these guys, it took us 12 months. They were incredibly intolerant of the anterior protrusion with much, much more TMJ pain than the dentists we used to see. After spinal cord injury, your motor and sensory strip reorganise, so your humunculus is just kind of from injury up. So there is evidence that there's increased pain perception and dysfunction above the level of the lesion in tetraplegia. So we think this was kind of a hyperesthesia sort of phenomenon. There are other situations where this applies. Which brings me to the trial of CPAP that we did. So CPAP is still the best treatment we've got for obstructive sleep apnea. And if there was incident disease, we wondered whether we could detect and treat. And so this was an 11-site, four-country, randomised control trial looking at people with incident tetraplegia. We screened over 1,800 new cases of cervical spinal cord injury for inclusion into the trial and managed to actually achieve our target sample of 150, with one, of course, dropping out to make the numbers not look pretty. The age is really bimodal. So it tends to be younger, predominantly men, who have injuries, and older. And in fact, over the course of this study, the demographics shifted quite rapidly to older people having sort of low-impact falls and incomplete cord injuries. A third of the group had motor and sensory complete lesions, half-high, half-low. But as you can see, the average apnea hypopnea index is very high. And this is the money. So the PSAT was our primary outcome measure. It's an awful test. So the way it works is I read out the number two, or the tape reads out a number two, because it was a tape when we started. It was that long ago. And then it reads out a five, and you have to add them in your head and shout out seven. Then you have to forget the seven, forget the two, remember the five, and it reads out five, and you have to shout out ten. And then you have to forget the first five, forget the ten, remember the next five, and add it to the next number. And so it cycles through, and you get to the end, you go, ha, and then it starts again faster. So there was no difference in the improvement in the PSAT between the intervention and the control arm. And the blue dots are those who were adherent at a level greater than four hours a night on average over the study. So you can see we had trouble with adherence, but there's really nothing there. And in fact, the change from baseline in both, the spontaneous recovery in attention and information processing, is at the same level as the difference in MS with and without frontal plaques. So it's a really large spontaneous improvement, over three months, really large spontaneous improvement in neurocognitive function that occurs after isolated spinal cord injury. And we've since had really good data showing that isolated spinal cord injury gives rise to generalised CNS inflammation. So I think this has important implications, in fact, for all of rehabilitation and skill acquisition after spinal cord injury. We may need to wait 12 months before we can really do a lot of our rehab in some cases, because it does get better over time. Anyway, yeah, that didn't work. It did work, but it didn't work. Like most trials, you find something different from what you thought. Excuse me. Okay, what we did find was that if you were able to adhere, then your sleepiness improved by the same amount as it would for somebody who didn't have a spinal cord injury. Most patients in the randomised arm were unable to adhere most weeks of the study, but because we had measurements set every week, we were able to model that effect over time. And if you were able to adhere, by the end of that week, you were less sleepy. So there's a strong relationship, as increasingly is shown, between CPAP and reduction in sleepiness, with everything else being, yeah, maybe, nah, unsure. The other thing we've observed, not from those data, but from clinical data from about... from a year's worth of clinical data in our lab, is if you look at the people with tetraplegia, despite having really severe disease, they need very little pressure to stabilise their upper airways. So, you know, this is the kind of relationship I think a lot of us have in our minds, that as your AHR gets bigger, it's not surprising that sometimes you need more pressure. And that does not hold up in tetraplegia. So while we've previously shown that the upper airway is no more collapsible during sleep in spinal cord injury, it does seem to be more distensible. This might be a lung volume effect. We didn't find any difference in FRC between... in the previous physiology study, between people with and without tetraplegia, but we did find, as you might expect, that total lung capacity was smaller. The compartments are just arranged differently, essentially. So it's likely that a little bit of CPAP does probably deliver a change in end expiratory lung volume, which will increase distal tethering of the trachea and drop the resistance of the upper airway slightly. That's what we hypothesise is going on here. We don't really think it's the upper airway that's more distensible, per se. It might be, but that does provide some kind of biological plausibility. So, with all that in mind, we then moved on to a series of... a program of work that has really been led by Marnie, looking at the... OK, so now what? The burden-benefit kind of question. Is the benefit or the burden of CPAP worth it for people who are able to use it? And from this large mixed-methods study, we were able to show that... ..this equation for every individual strongly influenced their ongoing use. Commonly attributed problems were the same ones that we see in everybody. And, similarly, the determinants of benefit were people who were motivated by the immediate effects. Related to this bit down the bottom, which, again, we see in people without a spinal cord injury living with OSA, that they often don't know how severely impacted they are until after you manage to get them on treatment for a little while. Speaking of which, I need to say, in the RCT, I forgot to mention, we had a run-in period and we only randomised people who were able to adhere for more than four hours on one of three nights. And once they got past four hours, then they proceeded to randomisation. So, we didn't have people who we already knew would be likely to fail in the trial. So, to take it one step further, Mani then actually found a number of units around the world in... Excuse me. ..in Switzerland and Canada and the Netherlands who were doing this really well within their rehabilitation environment. It's hard to get a spinal patient into a sleep lab and it's hard to get a sleep lab to think having a spinal patient is a great idea. So, we need other care models. And Mani was able to look at these and synthesise this into a map of what equalled success. And I'm just going to highlight this sort of flow diagram, which is slightly different from what we do, perhaps, in non-injured patients and was associated with better overall adherence, was that treatment was very much driven by symptoms. And an AHI of 7, even 15, would be tolerated by the clinicians if the person felt like they were getting better. So, it's a very symptom-driven approach. It's about making it less bad. You know, perfect is the enemy of good, all that kind of stuff. And then, lastly, the last kind of thing I want to touch on is... ..there's been debate in the literature around whether this is obstructive sleep apnea or whether it's, in fact, central sleep apnea. There's a large amount of debate. The guys from Abdul and Safwan and myself have, you know, a lot of arguments about this. And that's good. That's what science is. So, we tried to get all of our data together and have a look. So, we've got all of the data from our population study, the randomised control trial, a study which I haven't talked about, where we developed an ambulatory care prediction model similar to the OSA-50 that looks at the likelihood of a sleep disorder breathing in a tetraplegic population, and all of our clinical data, and put it all together to look at this kind of question. And this has just been published online in Sleep in the last couple of weeks. So, from that, from 606 full polysomnography in people with tetraplegic spinal cord injury... It's my first time using a graphical abstract. I'm really happy. It's very bright and colourful. We found that the prevalence of central sleep apnea is somewhere between 4% and 8%. And that 4% to 8% is whether we're looking at... So, pretty much half and half. Predominant CSA or CSA with OSA, but 75% of people had obstructive sleep apnea and 17% of people did not. 9 to 18 times more prevalent. And a lot of the explanation, I think, is here. This is across the hypopnea index, the obstructive apnea index, the central apnea and the mixed apnea index for degrees of severity. And as you can see, as you would expect in the non-injured population, as severity increases, the hypopneas go away and they get replaced by apneas, pretty much. Obstructive apneas. Whereas the central apnea index is pretty stable and is almost certainly due, I think, to that exquisite chemical control at sleep-wake and the instability associated with that. And that is real and important and it's not the main clinical problem, I'd suggest. So, we have a lot of findings and a lot of butts. LAUGHTER So, I haven't talked about periodic leg movements or melatonin, but both of them go wrong, too. PLMs are highly prevalent and your pathway from the back of your eye to your pineal goes with the sympathetics. So, if you've got a cervical injury, you've got no circulating melatonin. At all. So, we've done an RCT of using three milligrams two hours before bed. Probably has a sedative effect. Needs to do it a bit more. People actually got this really weird hangover, perhaps because they've been living in a low melatonin... This is what we hypothesise. They've been living in a low melatonin environment, they've lost receptor density, but perhaps increased receptor affinity, like you see in other changes after spinal cord injury. Acutely, OSA definitely impairs attention and information processing straight after the injury, but it gets better over time and CPAP doesn't modify that improvement. OSA reduces quality adjusted life years by a good third in people with spinal cord injury. It's really highly prevalent and it is a major, major treatment challenge. It's not just because the upper airway is smaller, more collapsible or it don't move right. There are some reflex or compensatory kind of failure issues. It's harder to think what one might do about that, but that doesn't mean it's not potentially a target for therapy. The upper airway definitely has a higher resistance, but it can be reduced and we need to do some more work on the agents that are likely to make a difference. The upper airway is not a perfect Starling resistor, but it kind of is. So if you can modify upstream resistance, you will decrease the collapsibility, the tendency towards collapsibility. Oral appliances appear safe. It takes a long time and we need to do some more research in that place. As I said, melatonin might be good. But I think the real money at this point is around models of care, generalisable models of care that disrupt the way in which we think about looking after people with spinal cord injury, but I think even more so people living with large amounts of physical disability. They need tailored solutions. So like in the spirit of Margaret, I'm going to kind of leave you with that thought. This is a call to arms to just put our big girl and boy pants on and fix this. So I feel like I'm here. You know, not entirely sure which end I'm coming out yet. And I need to thank all of these wonderful people who contributed so much to this research. Thanks very much. Thank you.

Dr. David Berlowitz

respiratory physical therapy

sleep disorders

spinal cord injuries

obstructive sleep apnea

neuromuscular diseases

continuous positive airway pressure

noninvasive ventilation