ADVENT-HF investigator T. Douglas Bradley, MD, says any risk of increased mortality linked to the use of adaptive servo-ventilation in people with central sleep apnea and heart failure is device- and algorithm-specific.

Interview by Sree Roy

In 2015, the ResMed-funded SERVE-HF trial resulted in practice-changing guidance due to its shocking finding that a minute ventilation-triggered adaptive servo-ventilation (ASV) device increased the risk of mortality in people with central sleep apnea and symptomatic chronic heart failure with left ventricular ejection fraction less than or equal to 45%.

Now a newly published Canadian Institutes of Health Research- and Philips Respironics-funded study, ADVENT-HF, shows promising results for ASV in people with heart failure. The study used different ASV devices—triggered by falls in peak airflow—than those used in SERVE-HF. ADVENT-HF investigator T. Douglas Bradley, MD, says, “The devices work very differently in terms of their outcomes, their algorithms, and their way of generating expiratory pressure and pressure support.”

Though the safety monitoring committee checked repeatedly, it did not find any safety signals during ADVENT-HF. In fact, the mortality trended lower in the people with central apneas who used the Philips ASV devices, though it did not reach statistical significance.

“It’s not just safe,” Bradley says. “It improved quality of sleep and quality of life.”

Key takeaways from ADVENT-HF include:

• The ASV used in SERVE-HF, ResMed AutoSet CS, relied on a minute ventilation-triggered algorithm. The ADVENT-HF trial utilized peak flow-triggered ASVs, the Philips Respironics BiPAP autoSV Advanced and BiPAP autoSV Advanced System One, which reduce the likelihood of over-ventilation.
• In SERVE-HF, the ASV used was not designed to treat obstructive sleep apnea, resulting in a residual (mainly obstructive) apnea-hypopnea index (AHI) of 6.6. In ADVENT-HF, the peak-flow-triggered ASV maintained a mean AHI below 5 for both central and obstructive sleep apnea.
• ADVENT-HF participants randomized to ASV showed significant improvement in sleep structure, including more N3 and REM sleep, reduced arousal frequency, and better quality of life, which were not observed in SERVE-HF. They also experienced improved symptoms, sleep quality, and daily functioning with ASV in ADVENT-HF, akin to treatment goals for patients with obstructive sleep apnea seen in sleep clinics.
• The differences in findings between ADVENT-HF and SERVE-HF highlight that ASV devices are not generic; their efficacy and safety vary based on manufacturer-specific algorithms.

Bradley discussed ADVENT-HF with Sleep Review over videoconference.

How did the ASV used in ADVENT-HF differ from the ASV in SERVE-HF?

The ASV used in SERVE-HF was the first iteration of ASV.  It was designed to treat central sleep apnea (CSA) in general, including in patients with heart failure. 

Its algorithm had a minimum EPAP [expiratory positive airway pressure] of 5 cm H₂O and a minimum pressure support of 3 cm H₂O. On every inspiration, the minimum pressure generated would be 8 cm H₂O. If the minute ventilation dropped below a certain level, its automated algorithm would kick in with inspiratory pressure support at a particular rate to overcome the central apnea. For that purpose, it worked very well.

However, it did not have an automatic algorithm for obstructive apnea. In SERVE-HF, the residual AHI was 6.6, and the majority of events were obstructive.

The device used in our trial, ADVENT-HF, is triggered by changes in peak flow—rather than minute ventilation, also referred to as peak flow-triggered adaptive servo-ventilation or ASV-PF.

Also, it had different default pressures. The minimum default EPAP was 4 cm  H₂O and the minimum default pressure support was 0, which meant that the minimum pressure on an inspiration would be 4 cm H₂O—half of what the minute ventilation-triggered ASV applied.

That means there’s less likelihood of over-ventilation with the device we used.

Another major difference is the device in ADVENT-HF was designed to treat both central sleep apnea (by altering inspiratory pressure support) and obstructive apnea (by altering the expiratory pressure). In ADVENT-HF over the five-year trial, the mean AHI of patients was under 5 in both the obstructive and the central sleep apnea groups. If you compare ADVENT-HF to SERVE-HF, the average expiratory pressure delivered was about the same (~6 cm  H₂O). However, in ADVENT-HF we delivered 1.6 cm  H₂O less inspiratory pressure, which meant less likelihood of over-ventilation. 

We did not experience the increase in mortality seen in SERVE-HF amongst our heart failure patients with central apnea. In fact, we had a tendency to reduce the mortality rate. (The hazard ratio was 0.74, but it wasn’t significant. The p-value was 0.25.) If we compare the all-cause mortality between central apneics in SERVE-HF and central apneics in ADVENT-HF, there’s a significant difference. (The hazard ratio was 1.28 against minute ventilation-triggered ASV in SERVE-HF and 0.74 in favor of ASV-PF in ADVENT-HF, p-value 0.05.)

A key finding is that ASV did not increase mortality in people with either OSA or CSA, even though the mean duration was one year longer than in SERVE-HF. What are the clinical implications of this?

Along with what you just said, we also found, unlike SERVE-HF, that the device we used (peak flow-triggered ASV) caused a very significant improvement in sleep structure, characterized by a reduction in light stage 1 sleep, increases in deep slow wave (N3) sleep, and increases in rapid eye movement sleep. In addition, the arousal frequency dropped dramatically. 

This was accompanied by improvements in quality of life, as assessed by the Minnesota Living with Heart Failure Questionnaire, the New York Heart Association class (both of which manifest cardiac symptoms), as well as the Epworth Sleepiness Score, which diminished significantly.

None of these findings were found in SERVE-HF.

What is the clinical significance of that? In our everyday practice as sleep physicians, we see patients with obstructive apnea very frequently. And what do we treat them for? We treat them because they’re sleepy, because their sleep is disrupted, because they snore, because they wake up with headaches, because they fall asleep in meetings or in uncomfortable situations, or they have driving accidents.

We’re not treating them to save lives. We’re not treating to reduce cardiovascular disease since, so far, randomized trial evidence does not support this. We’re treating them to improve their symptoms and their quality of life.

So if we make that analogy in the heart failure population, we’re doing exactly the same thing.

We don’t have evidence that treating these people with obstructive or central apnea reduces mortality or hospital admissions, but we have evidence that peak flow-triggered ASV does improve sleep quality and quality of life, exactly as you would expect in the usual patient with obstructive sleep apnea.

What this says is: If you have heart failure and you’re very fatigued and you’re short of breath and you’re not sleeping well, this device will make you feel better.

But the ADVENT-HF trial was underpowered to determine whether ASV had a significant effect on the primary composite endpoint or mortality overall for people with CSA, so I feel like the trial leaves the SERVE-HF elephant standing in the room. Can you comment on that?

Yes, we didn’t reach the predicted sample size and were underpowered to assess the primary outcome. The SERVE-HF safety notice meant we had to stop recruiting central apneics in our trial, go to our data and safety monitoring and ethics committees, and change the consent form to include the information from SERVE-HF, then re-consent all the patients. That took about nine months. In Germany and France, the authorities disallowed the recruitment of patients with central apnea into the trial, even though the device we used was different. Then COVID came along, and it was almost impossible to recruit anybody.

However, in terms of the safety and the elephant in the room, the device is definitely safe. And why can I say that? Because when we got the grant to run this trial from the Canadian Institutes of Health Research, it insisted we have our data and safety monitoring committee monitor our patients’ safety every six months, which is unusual, but it worked in our favor.

Every six months, we had to send outcomes data, including hospitalizations, deaths, etc, to this committee. They looked at the data unblinded, and they stratified the patients into OSA and CSA. Every six months, they told us to “continue the trial as per protocol.” In other words, they did not see a safety signal. So we had about 10 different looks at the data, and on none of those occasions was there any evidence of a safety signal amongst patients on ASV.

At the end of the trial, the hazard ratio for the central apneics was in favor of ASV 0.74. In other words, a 26% reduction in overall mortality, although it wasn’t statistically significant.

Moreover, it’s not just safe: it improves quality of life. Qualitatively, the improvements in quality of life were greater in the patients with central apnea than those with obstructive apnea, which may seem counterintuitive, but that’s what we found. And the use of ASV was also greater. The central apneics at the end of five years averaged 4 hours a night, whereas the obstructives averaged 3.3 hours a night.

This trial used the Philips Respironics BiPAP autoSV Advanced and BiPAP autoSV Advanced System Ones. Considering the different findings in ADVENT-HF versus SERVE-HF, does that mean each manufacturer needs to do a study analyzing its specific ASV device on mortality? 

I’m glad you asked that question because I believe for the treatment of sleep-disordered breathing in patients with heart failure, the answer is yes. What we found is that the ASV is not a generic device. It does work differently and according to which manufacturer makes it and which algorithms they’re using. 

Because one trial, SERVE-HF, showed an increase in mortality, whereas our trial did not (and showed improvements in quality of life), it means that devices are different and therefore cannot be treated as generic devices. This means also that any iteration of a positive airway pressure device to come on the market, if you want to test it in terms of its safety and efficacy for reducing, say, morbidity and mortality in heart failure or stroke or any other underlying serious condition, you’re going to have to do a randomized trial to prove that that’s the case.

Of course, that’s very difficult. There are now several neutral studies like ISAACC, RICCADSA, and SAVE that demonstrated treating OSA with positive airway pressure had no effect on morbidity or mortality, no matter which group they were studying.

My guess is we’re not going to see any large-scale trials like ours, SERVE-HF, or SAVE in the near future because the positive airway pressure companies don’t have the appetite or resources to do this in view of the results we’ve had so far. This means that either these trials are not going to get done, or other interventions that are being developed—such as phrenic nerve pacing, hypoglossal nerve stimulation, oral appliances, etc, and the new drugs that are being developed to treat apnea—may have to be tested in such trials. But if this is done, we’re not going to know the results for many years.

Is there anything else you want to add?

One very important issue is the FDA’s proscription of Philips marketing any of its positive airway pressure devices in the United States for the foreseeable future until certain factors are met. For sleep physicians and patients, a reduction in the competition is detrimental to patients’ outcomes. By prohibiting sales of Philips’ devices, the FDA has essentially given one or two companies almost a monopoly.

The devices that Philips makes are stll sold in Canada and Europe. If I was a sleep physician in the United States, I would be lobbying to say, “This is not right unless the FDA has some solid evidence that these devices are actually causing harm.”

Indeed, in Ontario, a study by Tetyana Kendzerska compared the Philips device with ResMed and Fisher & Paykel devices. She tracked cancer rates over several years and found no difference whatsoever among the three different devices. The issue of an “increased cancer risk” from the Philips device from the disintegrated material was not borne out by these data.