Should High-Frequency Ventilation in the Adult Be Abandoned?

Pathophysiology Considerations

Most of our understanding of HFOV has come from small animal models and the neonatal/pediatric population, and these concepts may not apply to the much larger adult lung.²³ Moreover, in some of these animal studies, potential harmful effects of HFOV were observed. For example, in canine models, significant regional differences were seen in alveolar pressure at various locations in the lungs despite having mean alveolar pressure nearly equal to the mean opening airway pressure. This resulted in the observation of distinct areas of both hyperinflation and hypoinflation.²⁴

In an adult sheep model, Sedeek et al²⁶ used adult high-frequency oscillatory ventilator management strategies that were identical to the ones used in early adult clinical trials by Fort et al²¹ and Mehta et al.²⁵ These investigators found that quite large V_T (ie, 6 mL/kg) were being delivered into the proximal airways. The authors believed that the larger endotracheal tubes used in adults may have provided less attenuation of the oscillating pressures in HFOV compared with pediatric size endotracheal tubes.²⁶ As a result, the alveoli could be overdistended and injured from large pressure swings. However, caution must be exercised to avoid overinterpreting these data because, as noted above, considerable damping of the circuit pressure and volume swings occurs in the more distal tracheobronchial tree.

As discussed earlier, improvements in the P_aO2/F_IO2 with HFOV are dependent on providing pressure above the critical opening pressure of the alveoli and maintaining pressures above this threshold to minimize cyclical opening and closing of alveoli that leads to VILI.¹⁰ However, this is not always the case. Papazian et al²⁷ compared HFOV in both the supine and prone position with prone-conventional ventilation targeting V_T of 6 mL/kg in subjects with moderate ARDS. There was no significant improvement in the P_aO2/F_IO2 at 12 h in the supine HFOV arm, whereas both the prone conventional ventilation and prone HFOV groups showed a significant improvement (from 138 ± 58 to 217 ± 110 mm Hg [P <.001] and from 126 ± 40 mm Hg to 227 ± 64 mm Hg [P <.001], respectively) at 12 h. Bronchoalveolar lavage was done before and 12 h after randomization. In both HFOV groups, cytokine interleukin-8 increased significantly from baseline levels as well as compared with the prone conventional ventilation arm. The prone conventional ventilation arm saw a decrease in concentration compared with its baseline level. The neutrophil concentration was lower in the prone conventional ventilation group compared with both HFOV groups. This study suggests that HFOV subjects alveoli to harmful cyclical opening and closing as interleukin-8 is up-regulated when epithelial cells are subjected to cyclical overstretching.

Conventional mechanical ventilation can often achieve improvements in P_aO2/F_IO2 similar to HFOV, depending upon settings. For example, in rabbit lung injury models, utilizing a short lung recruitment maneuver followed by V_T <5 mL/kg and PEEP adjusted to be above the lung's critical closing pressure was found to improve the P_aO2/F_IO2 similar to HFOV. In addition, lung histology in this group showed minimal airway injury.⁹ This was also supported by the work of Vazquez de Anda et al²⁸ in rats where high PEEP and low V_T improved oxygenation but also reduced the influx of alveolar proteins as compared with HFOV.

Right ventricular dysfunction is a known complication of ARDS, and the high ̄P_aw and general absence of spontaneous inspiratory efforts during HFOV can exacerbate this problem further, especially under fluid-depleted conditions. Guervilly et al²⁹ evaluated right-ventricular function using transesophageal echocardiography in adult subjects with moderate to severe ARDS receiving HFOV. At baseline, over half of their subjects had right-ventricular dysfunction (defined by a right-ventricular end-diastolic area to left-ventricular end-diastolic area ratio >0.6), and 25% had right-ventricular failure (defined by right-ventricular end-diastolic area/left-ventricular end-diastolic area >0.9). Increasing the ̄P_aw >5 cm H₂O to improve the P_aO2/F_IO2 was found to worsen the right-ventricular end-diastolic area/left-ventricular end-diastolic area by as much as 40%. Cardiac index was also found to be significantly reduced due to increasing ̄P_aw and worsening right-ventricular function. As a consequence, vasopressor requirements may be higher in HFOV patients. These hemodynamic issues are important, given that death associated with HFOV appears to be more related to end-organ failure and hemodynamic compromise as opposed to failure to oxygenate.³⁰

High-frequency oscillatory ventilation requires synchrony with any existing patient breathing efforts. Spontaneous respiration results in a reduced airway pressure that the ventilator may interpret as a circuit disconnect, subsequently stopping ventilation.²⁶ In addition, inadequate synchrony results in oxygen desaturation, patient discomfort, and barotrauma. To achieve synchrony, heavy sedation and the use of continuous neuromuscular blockade are often required. Indeed, in most of the clinical trials described below, the sedation needs and requirements for neuromuscular blockade were almost always higher in the HFOV group versus the control group.

Outcome Data From Randomized Clinical Trials

Although several small trials of HFOV versus conventional ventilation showed promise,²² recently, 2 large trials have been published that have concluded no benefit and possibly harm with the use of HFOV.

The OSCILLATE and OSCAR trials were 2 large multi-center trials published in 2013 comparing HFOV with conventional mechanical ventilation in moderate to severe ARDS. In the OSCILLATE trial,³¹ subjects of similar severity of ARDS were included (P_aO2/F_IO2 of 113 ± 38 mm Hg in the conventional ventilation group and 113 ± 37 mm Hg in the HFOV group); however, subjects who had been ventilated for >72 h were excluded. The OSCAR trial³² included subjects ventilated up to 7 d with similar ARDS severity; 114 ± 38 mm Hg versus 121 ± 46 mm Hg in conventional ventilation versus HFOV groups, respectively. Rather surprisingly, the OSCILLATE trial was prematurely stopped due to worse outcomes in the HFOV group. The mortality in the HFOV group was higher both for in hospital (47% vs 35%) and at 28 d (40% vs 20%). Although the conventional ventilation group had a higher number of subjects with refractory hypoxemia, the overall percentage of subjects dying due to this was similar between the groups. The OSCAR trial was completed as intended; however, it demonstrated no difference in mortality between the 2 groups (41.7% for HFOV vs 41.1% for conventional ventilation).

Interestingly, the mortality in the conventional ventilation strategy with OSCAR was much higher than the conventional ventilation group in OSCILLATE, despite the fact that the disease severity in the OSCILLATE group was higher (Acute Physiology and Chronic Health Evaluation [APACHE] II score 29 ± 7 vs 21.7 ± 6.1). An explanation for the mortality difference may lie in the manner in which the conventional groups were ventilated. The OSCILLATE conventional group received lower V_T of 7.1 ± 1.8 mL/kg, whereas the OSCAR conventional ventilation subjects had on average higher V_T of 8.3 ± 3.5 mL/kg. This supports the notion that lower V_T ventilation conveys a survival advantage in moderate to severe ARDS and suggests that a significant mortality difference favoring conventional ventilation might have been seen in the OSCAR trial if the conventional ventilation arm used lower V_T. A more recent meta-analysis that included these new trials has now concluded that there is no benefit to HFOV in the adult.³³

So why are the results of these more recent trials at odds with previous animal and pediatric data? Perhaps there are effects of HFOV in the adult lung that have not been fully appreciated in the past. High distending pressures are known to put the lung at risk for injury and cardiac compromise, and thus the very mechanism that improves P_aO2/F_IO2 with HFOV may contribute to harm. Moreover, in ARDS, the lungs are affected in a heterogeneous manner, with areas of damaged and collapsed alveoli intermixed with open alveoli. High intrathoracic pressures may thus lead to further regional lung injury. HFOV settings in adults are also different from those of pediatric patients. Lower frequencies (4–8 Hz in adult vs 8–12 Hz in pediatric patients) and pressure amplitudes of up to 60 cm H₂O are often used in adults.³⁴ This can translate into larger delivered V_T in the adult. In a retrospective analysis of 156 subjects, Mehta et al²⁵ observed that >20% of their subjects developed pneumothoraces on HFOV while noting that barotrauma rates in conventional mechanical ventilation ranged from 7 to 14%. In the OSCILLATE trial, 18% in the HFOV arm developed new barotrauma after being switched from conventional ventilation.³¹

As noted above, the need for sedation and neuromuscular blockade when using HFOV in the adult could also contribute to adverse outcomes. In the Multi-Center Oscillatory Ventilation for Acute Respiratory Distress Syndrome Trial (MOAT),³⁵ all subjects in the HFOV arm were paralyzed as well as 90% of the cohort in the study by Mehta et al.²⁵ In the OSCAR trial, over half in the HFOV group required neuromuscular blockade, which was more than the conventional group, and required paralysis on average half a day longer than the control group.³² In the OSCILLATE trial, significantly more subjects in the HFOV arm required paralysis (83% vs 43%) for an average of 3 d, which is 1 d longer than the control.³¹

In summary, HFOV should provide an ideal mode of ventilation for diffuse lung injury. However, this has not been the case in the adult population. This may be due to poorly understood effects of the HFOV ventilator pattern on injured lung units. It may also reflect cardiovascular effects as well as the need for excessive sedation, vasopressors, and neuromuscular blockade agents. Regardless, because there is no clear survival benefit and a possible higher risk for harm, a strong argument can be made that HFOV should not be used for adults with ARDS.

Conceptual Arguments Remain Attractive and Are Supported By Data

The conceptual lung-protective benefits of HFOV have been supported by numerous animal studies.^19,36 A particularly interesting report is that of Goddon et al,³⁷ who obtained static pressure-volume curves in anesthetized sheep before and after saline-induced lung injury. The authors measured changes in ̄P_aw with oxygenation and compared it with the static pressure-volume curve. They found that the ̄P_aw above the lower inflection point resulted in improvements in oxygenation and reached an optimum level when the ̄P_aw approached the upper inflection point. They did not find any further benefit in oxygenation with going beyond the upper inflection point. Going beyond this point, they concluded, was where alveoli distention occurred. This has been seen in other studies and supported with computed tomographic imaging.^38–40

Data from extensive studies in rabbits by several authors have also been enlightening. Injured rabbit lungs ventilated with HFOV demonstrated that increases in ̄P_aw resulted in linear oxygenation improvements with lung volume expansion. Bronchoalveolar lavage samples from these lung-injured rabbit models were found to have lower concentrations of polymorphonuclear neutrophils and less expression of inflammatory markers, such as tumor necrosis factor-α prostaglandins and platelet-activating factors, compared with conventional mechanical ventilation.^41,42 Perhaps more importantly, a series of rabbit experiments by Froese and Bryan³⁶ consistently showed improved survival with HFOV. As we slowly unravel all of the mechanisms of VILI, the concepts underlying HFOV applications remain attractive.

Positive Results Have Consistently Been Reported in Infant/Pediatric Trials and Small Adult Studies

Multiple trials of HFOV in infant/pediatric populations have consistently shown that HFOV can be safely applied and is generally associated with improved long-term outcomes (ie, bronchopulmonary dysplasia and chronic lung disease).^20,43 Admittedly, a few trials are negative, but only one, the HIFI trial, actually suggested harm from HFOV.²⁰ This trial, however, has been heavily criticized because of its unfamiliar devices and strategies, minimizing rather than optimizing mean airway pressure. A recent meta-analysis has concluded that HFOV use is associated with important clinical end points in neonatal/pediatric subjects.

There is much less experience with HFOV in adults than in infant/pediatric populations because devices powerful enough to provide effective gas exchange in adults took longer to develop. There is no question that HFOV in adult respiratory failure has the ability to improve oxygenation. Multiple retrospective studies have observed this phenomenon.^25,43,44 Mehta et al²⁵ found a 70% improvement in oxygenation with the initiation of HFOV for severe ARDS. Randomized trials also supported this observation. The MOAT trial saw the P_aO2/F_IO2 rapidly improve from 114 ± 37 to 205 ± 61 mm Hg 16 h after initiation compared with 146 ± 56 mm Hg from 111 ± 42 mm Hg in the conventional group.³⁵

Whether these gas exchange improvements translate into improved outcomes remains controversial. Small randomized controlled trials have attempted to answer this question with equivocal results. Until 2013, the MOAT trial³⁵ was the largest randomized controlled trial and demonstrated a 30-d mortality of 37% for HFOV versus 52% for the conventional ventilation arm. However, this difference was considered nonsignificant (P = .10). As noted above, when taken together, a meta-analysis of these small pre-2013 trials suggested a mortality benefit to HFOV.²² A common concern in these trials was that the conventional ventilation comparison group was usually not managed in a true lung-protective management algorithm. Nevertheless, the concept that HFOV could be effective in severe ARDS has some reasonably supportive clinical data.

Arguments to Abandon HFOV Focus Heavily on OSCAR/OSCILLATE, Both Flawed Trials

As noted above, HFOV settings are often counterintuitive; CO₂ clearance is driven largely by the power setting and is actually a function of the inverse of frequency (lower frequencies result in larger V_T with HFOV devices). Moreover, oxygenation depends on high constant distending pressures that can seriously impair cardiac filling and increase right-ventricular afterload. These features demand experienced operators who have a good understanding of pulmonary physiology, cardiopulmonary interactions, and the operational characteristics of the HFOV device.

Unfortunately, in both the OSCAR and OSCILLATE trials described above,^31,32 many centers were involved that had only limited HFOV experience. Although both studies claim to have trained leadership personnel in all sites, the necessary 24/7 coverage by experts was often lacking. OSCAR was further hampered by the use of a new HFOV device that very few had ever operated before. This inexperience problem may have had even more impact in OSCILLATE, where >70% of the population was in shock at study entry and mean airway pressures were suddenly increased in the HFOV group from 20 to >30 cm H₂O. Interestingly, despite this dramatic increase in ̄P_aw, HFOV subjects in OSCILLATE did not have a change in central venous pressure from pre-randomization values suggesting preload impairment. The mixture of high mean airway pressures, poor cardiac function, and inexperienced clinicians probably had serious consequences. Taken together, the inexperience in these 2 trials, coupled with the challenges of new devices and hemodynamically compromised subjects, undoubtedly contributed to many of the adverse outcomes. It is important to remember that a clinical trial of a device requires optimal operation of the device and is thus orders of magnitude more complex to accomplish than a simple drug versus placebo clinical trial.

Both OSCAR and OSCILLATE also recruited subjects for whom HFOV was probably not needed. Bollen et al,⁴⁵ in a review of infant/pediatric HFOV studies done over the last few decades, nicely demonstrated that as conventional lung-protective strategies were being introduced, the benefits to HFOV in clinical trials became less. This would suggest the HFOV probably adds little to effective conventional lung-protective ventilator support. Rather, HFOV should be considered a strategy to extend lung-protective support to sicker patients unable to be safely supported with conventional ventilation. HFOV logically should thus be reserved for those patients not in whom conventional ventilation is working, but in whom conventional ventilation is failing. In both OSCAR and OSCILLATE, the average entry P_aO2/F_IO2 was well over 100, which translates into a P_aO2 of >60 mm Hg with an F_IO2 of <0.6, both in the setting of plateau pressures well below 30 cm H₂O, clearly acceptable parameters for effective lung-protective ventilation. Indeed, in OSCILLATE, the included population was actually more of a hemodynamically compromised population than a respiratory failure population. Exposing patients properly supported safely with conventional ventilation to HFOV is not likely to provide much benefit and, as noted above, could cause harm in the wrong hands (and probably did).

This selection process was even further complicated in OSCILLATE, where 72 subjects were excluded because their physicians felt that HFOV was required and randomization would be unethical. Thus, the very patients these trials should have studied were systematically excluded. Finally, the crossover process confuses interpretation. In OSCILLATE, 31 subjects were switched from conventional ventilation to HFOV, presumably because they were failing conventional ventilation. Because it was an intention-to-treat analysis, any benefit of HFOV in these subjects would have improved conventional ventilation outcomes.

Taken together, these concerns cast serious doubt on the generalizability of OSCAR and OSCILLATE to most ICU settings managing patients with severe ARDS. Abandoning HFOV based solely on these results seems premature.

If HFOV Still Has a Role, When and How Should It Be Used?

For HFOV to be a safe and effective therapy in ARDS, it needs to be used in the right patient (severe hypoxemic respiratory failure, failing conventional ventilation), with the right expertise and device settings, and with the right hemodynamic monitoring/management strategy. Another important consideration is knowing when to stop HFOV as ineffective. In general, observational studies suggest that if HFOV is going to be effective, gas exchange improvements will occur over the first 6–12 h. If this does not occur, the likelihood of HFOV success is low, and alternative rescue strategies, such as extracorporeal membrane oxygenation, should be considered.