The Oscillations of HFO

A. CHARLES BRYAN

Department of Critical Care Medicine, Research Institute, Hospital for Sick Children, Toronto, Ontario, Canada

	ARTICLE

TOP ARTICLE

There are a number of ways of telling the story of HFO, but the paper that I have chosen to discuss is by Kolton and colleagues (Kolton M, Cattran CB, Kent G, Volgyesi G, Froese AB, Bryan AC. Oxygenation during high frequency ventilation compared with conventional mechanical ventilation in two models of lung injury. Anesth Analg 1982;61:323-332). It is not a very tidy paper, and very few people seem to have read it. Despite this, for me, it was the benchmark between high frequency oscillation (HFO) as a fascinating bioengineering phenomenon for controlling CO₂, and HFO as a method for controlling O₂ and CO₂ and a potential ventilatory therapy.

I first stumbled into HFO by mistake. In 1972, we were measuring the effects of muscle relaxants on lung impedance during anaesthesia using a loudspeaker. While setting the equipment up, for some reason we stuck a CO₂ probe into the circuit. We were then astounded that during a breath-hold CO₂ appeared at the mouthpiece in increasing amounts with each beat of the loudspeaker, eventually reaching end-tidal levels. The trouble was that on the next breath either nothing happened or CO₂ started to come out, but then abruptly stopped. This was incomprehensible. I was a well brought up boy: the Pa_CO2 is inversely related to the alveolar ventilation, which is the tidal volume minus the dead space. The throw of the speaker (the tidal volume) was only a few milliliters and my dead space is about 150 ml; ergo there could be no alveolar ventilation. Furthermore, the phenomenon was not reproducible. What we did not realize was that it is very difficult to breath-hold with an open glottis, particularly when your uvula is flapping around at 15 Hz. So I filed away the records and we got on with our proper work, measuring impedance under anaesthesia. Our interest was briefly revived when Lunkenheimer, in the same year, tried to measure the impedance of the mediastinum with an oscillator and showed that the oscillator dramatically lowered Pa_CO2. The fact that all his animals died was not encouraging, and our interest also died.

Three years later, talking to Katsuyuki Miyasaka, one of our ICU fellows, he asked me what would be next after PEEP"would we be vibrating babies next?" I dug into my files and found the yellowing mingograph record of our early "failed" experiments. A few weeks later he came back with an oscillator that he built from a physiotherapist's mechanical chest percussor and a bellows from a child's toy. We borrowed somebody's pig and made it apneic with scoline and then turned on this Oscillator No. 1 (which still sits in my office), and, to our intense excitement, CO₂ came pouring out, rapidly reaching the previous end-tidal levels. What followed was a period of excited confusion. We realized we had to have a large fresh gas flow to blow the CO₂ away and had to control the mean airway pressure. We blew up an expensive loud speaker (which we christened the `Grateful Dead') and a number of animals developing a low-pass filter to get the DC flow out of the circuit, while keeping the AC flow directed to the lung. But as well as excitement there was a great deal of apprehension. Where had the dead space gone? Fortunately at this point I met Jeff Fredberg who was not at all surprised by our results, and said that facilitated diffusion was a well recognized phenomenon, particularly in the soil science literature. Later I talked to Sol Permut, who told me about Yandell Henderson's smoking experiments in 1915 and Briscoe's studies on the ephemeral Dead Space. We now had a theoretical basis for our incomprehensible data and we were off! We, at this time, were Alison Froesewhose remorseless logic kept us focused; Des Bohn from the ICUwhose initial skepticism disappeared rapidly like Irish mist; and, George Volgyesian ingenious engineer. We had a piston pump built and started a systematic study of the effects of oscillation on beagle dogs; this study was presented at FASEB in 1979 by Des Bohn. This was followed by a deluge of papers exploring the mechanisms and potential of HFO.

We progressed from oscillating my intubated friends, to adults on ventilators, to infants with respiratory distress syndrome; and our results were confirmed by other groups. However, although there was no doubt about control of CO₂, the Pa_O2 was always "iffy" and highly dependent on mean airway pressure. The problem lies in the differences in the shape of the dissociation curves for CO₂ and O₂. Because the dissociation curve for CO₂ is linear, if half the alveoli are shut, hyperventilating the rest will eliminate CO₂. For O₂, the dissociation curve is nonlinear, and no amount of hyperventilation, however intriguing the fluid dynamic regime, is capable of compensating for a shunt, and the very low peak pressure on HFO could not by itself open the shunt. What we had been doing was switching on the oscillator and increasing the mean airway pressure on the inflation limb of the PV curve and not getting much volume recruitment until we were well past the inflection point, which in the washed rabbit model is above 15 cm H₂O. Kolton's solution to this dilemma was to utilize the considerable PV hysteresis that is present in early acute lung disease. The lung is first inflated to total lung capacity with a pressure of about 30 cm H₂O and holding it for about 15 s (because the recruitment is time-dependent) and then sliding the pressure down to some point above closing pressure (which in this model is usually about 10-12 cm H₂O) and oscillating at this pressure.

Unfortunately, most of the other groups working in this field were promoting HFO as a low pressure strategy. Our only ally was deLemos and the San Antonio group, who were using HFO in a neonatal unit set up for premature baboons, with very positive results. The thinking among neonatologists was that high pressures were extremely dangerous for the lung and the pressures that we were talking about seemed reckless. Most unfortunately, NIH decided to start a multicenter randomized controlled trial on HFO in the treatment of RDS. There was a brisk controversy between the high pressure and low pressure factions, an argument that we lost. The final protocol failed to stress early intervention, as ventilator-induced lung injury occurs very rapidly in infants. It failed to stress volume recruitment maneuvers, and it failed to stress high maintenance mean airway pressures. In defense of the protocol committee, it should be stressed that neither our data nor deLemos's data was published at this time, and in defense of the participants in the trial, most of whom had very little experience with HFO, early intervention with such a bizarre method of ventilation must have been daunting. We thought that we could predict that the trial would have a not very impressive outcome, but were appalled when the trial was stopped because of a totally unexpected high incidence of intraventricular hemorrhage (IVH) in the HFO group.

After the HiFi trial was published HFO appeared dead and we were devastated. At first we were very alarmed by the very high (24%) incidence of IVH in the HFO group. But then, after some vigorous investigations, we became angry because two centers of equal size had, in one, an incidence of 4%, and in the other, 44% (add together and divide by 2 and you get the HiFi "result"). This, along with other discrepancies, convinced us that the outcome depended on the driver, not on the machine.

But HFO was not dead. As in the trade talks, Japan and Europe were reluctant to follow the United States lead and started their own trials. The most extraordinary of these was by Ogawa et al (one of the et al's being Kats Miyasaka), which failed to show any difference between HFO and CMV, mainly because the mortality and morbidity was much lower than anything published in the Western literature (confirming our belief that outcome depends on the driver not the machine). I was appalled that this paper was turned down by two major North American journals and ended up barely noticed in Early Human Development. In North America the San Antonio group, subsequently spreading across the USA, continued to champion HFO. This coincided with the rise of the Sensormedics oscillator. Alex Stenzler, who is the company representative, thoroughly understood the physiological problems and acted as an invaluable courier and supporter of groups working in this field. Now HFO is available and used in most neonatal intensive care units.

I took very little part in this renaissance. I was allergic to RCTs because it was now clear that we were testing the driver, not the machine, and that education was the only solution. This banner was carried by my coworkers, Alison Froese and Des Bohn, and it succeeded. Meanwhile I continued to fiddle with aspects of HFO that continue to excite me. There have been a number of papers trying to decide whether HFO is better than CMV that have yielded ambiguous results. Now the problem is not so much how to use HFO, as "What do you mean by conventional mechanical ventilation?" Adult ventilatory tactics have changed radically: limiting peak pressure, increasing PEEP and (sometimes inadvertently) using recruitment maneuvers. Modestly, I think they have been following the recipe of Kolton. As a result, I am not at all sure whether there is going to be much difference between well managed CMV and well managed HFO.

Reflecting on this story, it depended on serendipity, then on colleagues with a more unobstructed view of the problems and the stamina to survive a devastating RCT. The RCT is ideal for evaluating things like drugs, as randomization practically eliminates further medical input. It is far less effective in testing technologies, where the whim of the physician controls the way in which the technology is applied. This tactical variability has plagued, and will continue to plague, the evaluation of new ventilatory strategies.

This story is not finished. There are a number of studies that still need to be done on older children and adults before we blunder into another premature RCT. A major problem is safe and effective lung volume recruitment maneuvers. There are a number of ways to go, that need to be tested in the laboratory; we are still working on this.

	Footnotes

Correspondence and requests for reprints should be addressed to A. Charles Bryan, Professor Emeritus, Department of Critical Care Medicine, Research Institute, Hospital for Sick Children, 555 University Avenue, Toronto, ON, M5G 1X8 Canada.