Ventilating the Acutely Injured Lung

H. BARRIE FAIRLEY

Department of Anesthesia, Stanford University School of Medicine, Stanford, California

	ARTICLE

TOP ARTICLE

Rather than a scientific breakthrough, the article I have selected reflects an evolution of thinking on the subject of mechanical ventilation of the lungs over a period of several years. Entitled "Optimum end-expiratory airway pressure in patients with acute pulmonary failure," and authored by Peter M. Suter, H. Barrie Fairley, and Michael D. Isenberg (New Engl J Med 292:284-289, 1975), it showed that the level of positive end-expiratory airway pressure (PEEP) applied during mechanical ventilation could be adjusted in each patient to optimize the balance between recruitment of atelactatic lung and overdistension, and that in this study group this point coincided with optimal pulmonary gas exchange.

I will outline the circumstances leading up to this paper. My own background was as a clinical anesthesiologist with special training in anesthesia for thoracic surgery. After moving to the University of Toronto from London in 1955, I worked with Bigelow's group on induced hypothermia, doing research under his guidance in the laboratory and the cardiac surgery operating rooms of the Toronto General Hospital. It was at that time that a young emigrant British neurologist, Richard Chambers, came to talk to me about how we could better manage his paralyzed patients with respiratory failure. Together we organized and obtained approval for a multidisciplinary unit with its own specially trained nurses and equipment (Can Med Assn J 81:97-101, 1959). In the early years we learned much that has now become routine in the management of patients in intensive care units. I was particularly interested in optimizing mechanical ventilation of the lungs. Initially, blood gas electrodes were not commercially available, and we used the Radford nomogram for setting tidal volume and respiratory rate, but this was derived from subjects with normal lungs. The focus was on maintenance of CO₂ elimination, and it was not until the early 60s that the advent of commercially available oxygen electrodes permitted much needed research in patients with acute lung injury. The group at Harvard led by Myron Laver, Henrik Bendixen, and Henning Pontoppidan were particularly active at that time, and studied both anesthetized patients and those in their Respiratory Intensive Care Unit at the MGH. They showed that small tidal volume ventilation approximating spontaneous respiration permitted a gradual loss of lung volume and an increase in oxygen tension difference across the lung, probably due to right-to-left shunting through regions with poor gas exchange. They postulated that this was due to a lack of normal spontaneous intermittent sighs, and showed that large tidal volume ventilation was necessary to avoid this. We already knew that lobar atelectasis was common in our ventilated paralyzed patients and could correct this by "artificial coughing": the patient was placed in the lateral position with the affected side uppermost, and a manual lung hyperinflation was performed followed by vigorous chest compression and percussion. If one adds to this the experience of regularly reinflating lungs at the end of thoracic surgery and seeing atelactatic areas "pop open," then the concept of regional recruitment with hyperinflation was firmly in place.

We followed the large tidal volume principle aggressively, using ventilators that delivered constant tidal volumes, but an occasional patient whose lungs were ventilated in this way developed diffuse parenchymal damaged, and a few developed pneumothoraces.

I realized that the one-tidal volume-fits-all approach was inappropriate (even though adjusted for body size) and that we had been grossly overdistending some patients' lungs. We proposed to study this in the animal laboratory and Alan Laws, an Australian anesthesiologist/Ph.D. candidate, embarked on a series of studies in rabbits under my direction and that of Charles Bestthe then chairman of physiology. We soon discovered that many of the rabbits we were using had a viral type of pneumonia and decided to abandon that approach, thinking that we needed to use animals with normal lungs to study the effects of overdistension. Maybe so, but I now believe that in fact we had a good model of the clinical situation that we perhaps should have pursued.

Somewhere in the mid-60's we were fortunate to recruit "Charlie" Bryan to our department from his post as Director of the Canadian Air Force experimental physiology unit. His Ph.D. dissertation had been on the distribution of ventilation and perfusion under gravitational forces and he recognized some of our mechanical ventilation problems as variations on this theme. He suggested the use of a positive end-expiratory pressure as a means of optimizing the distribution of regional lung volume but I was reluctant to proceed, protesting that it would decrease cardiac output. Of course I was wrong, but this was the belief following the work of Cournand and his group who had shown that positive pressure ventilation decreased cardiac output when airway pressure did not return to zero in expiration. They had studied a small group in a very different clinical setting, and this preceded the Swan-Ganz catheter and clinical cardiac output measurement.

Shortly thereafter, I moved to the University of California in San Francisco but meanwhile the group in Toronto, now under the guidance of Alan Laws and Charles Bryan, had shown that PEEP was indeed effective in the respiratory failure setting without necessarily lowering cardiac output (Can Anaesth Soc J 16:477-486, 1969). They caught my attention!

A succession of papers followed from various centers, including our own group in San Francisco but, both in the clinical and the research settings, a single level of PEEP was commonly selected for study. Again the one-size-fits-all approach seemed inappropriate in terms of the balance between regional lung recruitment and overdistention and of the pulmonary gas exchange, but we lacked a way to determine the optimum level of PEEP for each patient. This was in an era preceding noninvasive continuous monitoring of gas exchange at the bedside.

I was looking for a project for a new research fellow, Peter Suter, an internist/intensivist from Geneva who would become the dean of the medical school there. He and I decided to study this problem of optimizing lung volume and gas exchange in the respiratory failure patient. By now, there was acceptance that a positive end-expiratory pressure was the way to maintain lung volume, but we needed better insight into the determination of how much to apply. We postulated that the best compromise between recruitment and overdistension would be reached by ventilating a patient's lungs on the optimal (steep) part of the pressure volume curve, but did not know how this would relate to pulmonary gas exchange. We also realized of course that these whole lung measurements reflected the status of a wide range of regional abnormality as well as that of the normal regions.

Two special tools were involved in this study. First we needed to collect expired gas samples uncontaminated from the inspiratory phase, to measure mixed expired CO₂ (for physiologic dead space) and nitrogen washout (for distribution of ventilation). The ventilator nonreturn valves were not absolute and mixing occurred at the switch between the two phases of respiration. Fortuitously, our most popular ventilator at the time (Ohio 250) used electrical circuitry to switch between inspiration and expiration and we designed a replacement valve that used solenoids activated from the ventilator's circuit board to open and close the inspiratory and expiratory ports. At each change between inspiration-expiration-inspiration, both valves closed for a fraction of a second before the next one opened. Second, in order to measure FRC by helium dilution we devised a special bag-in-a-box interface, which could be interposed by valve switching between the ventilator and the patient.

Suter, aided by an anesthesiology research fellow Michael Isenberg, was the principal patient-recruiter and data collector for this study and it is a tribute to his considerable abilities that we were able to collect a complete data set from sufficient patients, since so few met the criteria for inclusion. I can still remember the feeling almost of excitement when we met to review the results after each study. After the first few it seemed that we were not going to find a consistent pattern. Each patient had a (different) level of PEEP up to which lung-thorax compliance progressively improved and above which a decrease in compliance indicated overdistension but, although there was an initial improvement in gas exchange and an eventual decrease in oxygen transport and an increase in physiological dead space, this was not at any consistent PEEP level. The question was how to "put it all together." Up to this time our work and that of most others describing the physiological effects of PEEP grouped the data from all patients according to the actual PEEP level in cm H₂O. After several weeks of data collection, it dawned on us that a more appropriate reference point was that level of PEEP, which gave the best lung inflation in each patient, be it zero or 15 cm H₂O for example. The measure of this was the lung-thorax compliance, and so we examined what happened to all the other variables at this baseline maximum compliance level and at PEEP levels above and below this value that we called "best PEEP."

It turned out that, in our study group, this "best PEEP" coincided with the best oxygen transport, mixed venous oxygen tension, physiologic and alveolar dead space fractions, and it correlated inversely with the initial FRC (as a percent of predicted). We were delighted and even suggested in the article that, in the absence of direct measures of gas exchange, adjustment of PEEP to the level that produced the greatest compliance would be likely to provide the best values. The paper reads as though we were exploring this possibility from the start, but the reality was that the hypothesis that was ultimately tested and reported was not developed until we had collected data from several patients. Claude Lenfant, Director of the NIH Heart and Lung Institute, wrote an accompanying positive editorial review but, perhaps clairvoyantly, pointed out that confirmatory studies were required. Of course we now recognize that, as is so often the case, this is a much more complex situation than our paper would suggest. The coincidence of optimum compliance and gas exchange that we found was to some extent just that, although the directional changes in cardiac output, dead space and oxygen exchange, that occur with sequential increases in end-expiratory pressure, are correct.

This paper was but one of the many contributions from various research groups in the development of our general understanding of pulmonary gas exchange during mechanical ventilation, but it does reflect the great interest that so many of us had in improving our insight in this area. Those were exciting times and I am glad I was there! What did we learn? (1) That there is an optimal level of mechanical distension for each diseased lung, at end-expiration and end-inspiration. Studies of pulmonary gas exchange related to the use of end-expiratory pressure are potentially flawed unless one normalizes that pressure in terms of each patient's lung-thorax pressure-volume curveor, better, lung pressure-volume curve. (2) That one should not be too quick to accept results that confirm a preconceived idea; in this case that optimizing lung-thorax mechanics would necessarily optimize pulmonary gas exchange. (3) Most importantly that one should keep asking questions and not permit oneself to be bound by unsubstantiated clinical dicta, as exemplified by my nonacceptance of Charles Bryan's correct prediction as to the utility of PEEP.

	Footnotes

Correspondence and requests for reprints should be addressed to H. Barrie Fairley, M.D., Department of Anesthesia, Stanford University School of Medicine, Stanford, CA 94305. E-mail: b.fairley{at}att.net