American Journal of Respiratory and Critical Care Medicine

Two papers appearing back to back in 1966 serve as the background for my essay (Cherniack NS, Longobardo GS, Levine OR, Mellins R, and AP Fishman. J Appl Physiol 1966;21:1847 and Longobardo GS, Cherniack NS, and AP Fishman. J Appl Physiol 1966;21:1839). One described experiments that produced periodic breathing (Cheyne–Stokes respiration) in anesthetized dogs. The second offered a mathematical model that explained periodic breathing as an instability in the regulation of respiration using automatic control theory. The papers were based on work I did as a fellow in the Cardiorespiratory Laboratory of Alfred P. Fishman at Columbia's College of Physicians and Surgeons. This was such a mind-expanding experience that I returned as a faculty member 6 years later to collaborate with Al Fishman for another 9 years.

I worked on arcane phenomena that carried the names of Traube–Hering–Mayer waves (cyclic fluctuations in mean blood pressure) and Cheyne–Stokes breathing (cyclic fluctuations in ventilation or periodic breathing). I was encouraged by Fishman's enthusiasm, imagination, criticisms and became deeply impressed by what I believe is his aesthetic sense of scientific writing. He suggested that these cyclical phenomena could be explained by cybernetics (bravo to him for recognizing the value of what was called “automatic control theory” to explain biologic phenomena). For me, they were a turning point in my career and they began a 40-year collaboration with Guy Longobardo. They were the start of an effort that still continues with numbers and equations to describe breathing and its regulation by chemical, neural, and psychological factors.

Regarding those early experiments, we found we could produce transient cycles of hyperventilation and apnea or near apnea by mechanically hyperventilating anesthetized animals to hypocapnic levels and then allowing them to breathe spontaneously. Periodic breathing often but not always occurred when we hyperventilated the animals with air and more frequently when we hyperventilated them with a hypoxic gas mixture. Fishman insisted that we show that it never happened if we used carbon dioxide as the inspired gas. We thought this had to be true and not necessary to test. But in retrospect it was a valuable control.

There had been some earlier mathematical models of Cheyne–Stokes respiration appearing in engineering journals. These models, like the one we produced, were based on the idea that periodic breathing was the result of instability in the chemical feedback system that regulates breathing. In the early models, this instability was thought to be due to either increased sensitivity to carbon dioxide (the controller) or increased circulation time. But these models contained unrealistic values for physiological parameters. Our contribution was to include the effects of hypoxia on breathing in the model.

Both the modeling and the experiments, which went on simultaneously, developed quite slowly, with many episodes of discouragement and elation. In some dogs, we could not produce any periodic breathing, and we attributed this to too-deep anesthesia, perhaps out of wishful thinking. The mathematical model at first produced periodic breathing too easily for our liking. We would modify the model to include new factors, and then it became too difficult to simulate periodic breathing. There seemed to be always additional factors that might conceivably improve the model, but reliable data were hard to obtain. There were and still are no data on the ease of occurrence of periodic breathing. Longobardo frequently made suggestions, like “hypoxia could affect the termination of apnea.” This idea was not accepted then by many respiratory physiologists but turned out to be correct. The mathematics soon became sufficiently complicated that we needed to go downtown to the Engineering School at the main Columbia campus. We worked at night on their 1620 IBM computer, which operated on punch cards. The nighttime sessions would begin with punching data into cards, not always correctly. This led the computer at midnight to sort of vomit in disgust and literally fling the cards from its innards all over the room.

I came to the Fishman laboratory at Columbia on an Infectious Disease fellowship at the University of Illinois where I had used pulmonary function tests to evaluate the effects of antibiotics on bronchiectasis. I found that I had little trouble understanding diffusing capacity, but my preparations of supposedly pure bacterial cultures more often than not grew out yeasts. I followed this with a 2-year tour at the Aerospace Medical laboratory at Wright Patterson Airforce Base in Dayton, Ohio, working on the human centrifuge to study the effects of transverse acceleration on breathing and to test astronauts. This reinforced my fascination with the mysteries of respiration. I was fairly aged for a pulmonary fellow and already had over a dozen publications under my belt. I thought I knew a lot about how to do research. But thinking outside of the box and writing clearly were valuable skills Al tried to teach.

I initially worked on Fishman's idea that there might be an independent vasomotor control of the pulmonary circulation, and that this could lead to cyclic changes in pulmonary artery pressure. Instead, I managed to produce cyclic changes in the systemic circulation, which if large enough, also produced cycles in pulmonary artery pressure. I needed to explain the blood pressure waves. Fishman introduced me to Guy Longobardo who was then in the Engineering School at Columbia. Guy gave one of the first courses in automatic control theory for mechanical engineers. This was our first collaboration and led to a simple mathematical model of blood pressure control, which emphasized the importance of the arterial chemoreceptors. I was quite receptive to working with engineers because I had quite an admiration for mathematics. As a child, I had wanted to build bridges, but my inability to accurately add long columns of numbers in the third grade discouraged me from pursuing a career that would involve a lot of computations.

After we finished the blood pressure experiments, I tried to find something similar that would allow Longobardo and me to continue to work together. Sometimes when the blood pressure waves occurred, I could see periodic cyclic fluctuations in breathing that looked a lot like Cheyne–Stokes respiration. These observations led to the project on the mechanisms of periodic breathing.

The two papers on periodic breathing were published more than five years after we had finished the first drafts. The delay arose in part because applications of control theory to breathing were not common and we needed to reassure ourselves that the ideas were correct. But even more important was the often frustrating experience of trying to refine the writing so that the concepts of automatic control came through even without the mathematics.

By the time the papers were published, I had returned to the University of Illinois. I had also completed experiments on the changes in carbon dioxide stores after controlled changes in ventilation to see if they agreed with the predictions in the periodic breathing model. We sent a draft of an expanded model of the carbon dioxide stores to Leon Fahri who wrote back saying he thought it was okay. But we needed further reassurance. Longobardo who lived then in New York (and still does) came in December to visit me in the Chicago suburbs. We decided to take the model of periodic breathing to Fred Grodins, the guru of respiratory mathematical models and chairman of the Department of Physiology at Northwestern in Chicago. We took the train downtown in a blinding snowstorm, trudging through knee-deep snow in our galoshes to Northwestern. Grodins greeted us in his laboratory, which was a large room bare of everything except a desk and a 1620 IBM computer, read our paper, and pronounced it “all right.”

Although most of the mathematically minded understood the model, they suspected that periodic breathing was like the Lochness monster and might not really exist. Once in Boston, during another blinding snowstorm, arriving just before the meeting ended, I had to defend the reality of periodic breathing (I thought quite successfully) to a meeting of biomedical engineers. However, the nonmathematicians continued to have difficulties with the automatic control concepts and for years we have been devising better ways to explain them.

Things did get better, though. I was able to do more convincing experiments on periodic breathing in cats at Curt von Euler's laboratory at the Karolinska Institute in Stockholm (Cherniack NS, von Euler C, Homma I, FF Kao. Respir Physiol 1979;37:185). The experiments frequently lasted all day and all night, with Curt there all the time working with us. Curt, Ikuo Homma, Fred Kao, who was visiting Curt but had been my physiology professor in medical school, and I met every afternoon in the kitchen of the Karolinska to have tea and discuss science and philosophy both eastern and western. Curt had broad general interests, was quick to pick up on new ideas, and loved to develop theories that would tie observations together. He suggested we use his servorespirator, an early sort of proportional pressure support ventilator, to increase controller gains and produce periodic breathing. One day while von Euler ventilated me on the servorespirator, I dozed off and began to breathe periodically, encouraging us to believe that our ideas did apply to humans and perhaps to sleep.

Periodic breathing, at about the same time, became more clinically relevant. When we first got involved with periodic breathing, it was considered a strange type of breathing seen only in critically ill patients and occasionally in mountain climbers. It then became much better known as a cause of recurrent central apneas in sleep. The possible role of control system instability in sleep apnea also became better known in general. Many others have since undertaken mathematical modeling of Cheyne–Stokes respiration and breathing during sleep. We have also continued to develop this model, adding reflexes, upper airways, neural drives, and circulatory effects.

Whatever successes Longobardo and I have achieved resulted because what we do is truly interdisciplinary (i.e., based on ignorance of each other's field of expertise). After 40 years, I still have not become a mathematician and Longobardo has not become a physiologist. But that has not stopped me from arguing with Longobardo about mathematics or he arguing physiology with me. This has constantly forced both of us to re-examine the basic assumptions in our thinking and has kept our thinking fresh. Longobardo has convinced me that you do not really have the answers unless you can express your ideas quantitatively and you can put reasonable numbers into your explanations.

Correspondence and requests for reprints should be addressed to Neil S. Cherniack, Department of Medicine, New Jersey Medical School, University of Medicine and Dentistry of New Jersey, Newark, NJ. E-mail:


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record