Carlill, Duke and Jones [1957] which were conducted using pulsatile flows. the lung acts upon the pulmonary blood vessels. The analysis is based upon

Transcription

1 THE EFFECTS OF THE TRACHEAL PRESSURE UPON FLOW: PRES- SURE RELATIONS IN THE VASCULAR BED OF ISOLATED LUNGS. By JEAN BANISTER and R. W. TORRANCE. From The University Laboratory of Physiology, Oxford. (Received for publication 19th July 1960) In isolated cats' lungs, perfused at steady inflows with a mixture of plasma and dextran, we measured the pulmonary arterial pressure, the left atrial pressure, the tracheal pressure and the inflow of perfusate. We were able to obtain directly or to calculate relations between any pair of these four variables with the remaining pair at constant values. The results are used to analyze how tracheal pressures affect the pulmonary vascular bed and to consider some consequences of the changes in pressure which occur during cardiac and respiratory cycles in the whole animal. INTRODUCTION IN this paper we analyze the way in which the pressure in the air spaces of the lung acts upon the pulmonary blood vessels. The analysis is based upon a study of the relations between the four variables flow, arterial pressure, venous pressure and tracheal pressure. These relations, which are described in detail, were determined under conditions of steady flow and pressure in contrast to the investigations of Carlill, Duke and Jones [1957] which were conducted using pulsatile flows and rhythmic ventilation. Carlill et al. studied the effects of three variablesarterial pressure, venous pressure, and flow-upon the pulmonary vascular bed with some thoroughness; the effects of lung ventilation were treated in less detail. We perfused isolated lungs of the cat under the simplest hydrodynamic conditions that we could devise. The pump which we designed delivered a constant output and so developed steady rather than pulsatile pressures in the blood vessels. The perfusate was a mixture of plasma and dextran, a fluid free from the anomalies in viscosity which the blood cells introduce. Our main observations were made with the tracheal pressure kept constant. By this means we avoided the complicated effects on the blood vessels which, because of the changing dimensions of the lung and fluctuations in alveolar air pressure, occur during rhythmic ventilation. METHODS Perfusion System.-Perfusate was delivered continuously at constant volume by means of a two-phase paraffin pump (fig. la). The pump consists of a vertical cylinder with a piston which is moved either up or down as the threaded collar B 352

2 Pulmonary Hydrodynamics 353 ki -IFIG. la.-paraffin pump. Cross-hatching: paraffin. When the piston plate D moved upwards paraffin was driven out of the cylinder H through E and entered through F; when D moved downwards, E paraffin was driven out through F and entered through E (see fig. lb). D was driven by the pulley S which carried a threaded collar engaging with the piston rod at B. The pulley S in turn was driven by the velodyne, which is not shown. At the end of the pump stroke the horizontal bar C on the piston rod engaged with a vertical bar which reversed the velodyne through switch G. At H paraffin seals prevented air leaking past the glands between the piston rod and the D cylinder. Electrical connection between the pulley S and contact J was made once per revolution and was marked on a kymograph to rf record pump output. FIG. la 2 R f FIG. lb.-block diagram of perfusion system. Cross-hatching: Y ~paraffin. E,1 F: connections to pump. 1, 2: perfusion chambers. R: venous reservoir. P: connection for setting venous pressure. p A: input connection to lungs. V: v outlet connection from lungs. GIG. lb V, 1oultcnetofrmlgs TT T FIG. lc.-perfusion chamber (1). The lower part of the L/ chamber contained perfusate, the upper part paraffin (cross- E/A hatched). When paraffin was withdrawn from E into the paraffin pump, perfusate entered the chamber, past a glass valve, from the venous line (V). When paraffin was driven into the chamber at E, perfusate left the chamber, past a glass valve, to the arterial line (A). The flow of paraffin inor out of the other chamber was always equal but U U opposite in phase. Bubbles entering the system could be removed at three traps (T). FIG. lc

3 354 Banister and Torrance rotates engaging with the piston rod. The collar is driven by means of a variable pulley system S. The vertical cylinder is, in effect. divided into two compartments by the piston D (a circular plate fixed to the piston rod). The volume of paraffin moved during one complete excursion of the piston plate is 100 ml. As the piston drives upwards, paraffin is forced through aperture E into perfusion chamber 1 while an equal volume is withdrawn from chamber 2 through F (fig. lb). As it drives downwards, flow of paraffin occurs in the opposite direction, entering the cylinder through E and leaving it through F to flow into chamber 2. The displacement of paraffin from the cylinder into either of the perfusion chambers causes perfusate to flow through the outlet valve while the inlet valve remains closed (fig. 1c). At the same time the removal of paraffin from the other chamber causes it to fill with perfusate. Thus the continuous flow of perfusate into the pulmonary artery is maintained by the alternate filling and emptying of the two perfusion chambers. At the end of each excursion of the piston there is, for a fraction of a second, an interruption of flow when the velodyne, and therefore the whole pumping system, reverses. The inflow could be varied from 5 to 600 ml. per min. by adjusting the speed of piston movement. A common venous reservoir, R in fig. lb, is open to atmospheric or selected pressures at P. It supplies perfusate to both chambers and is connected to the left auricle through one arm of a T-piece. All connections in the system were made of polyrinyl chloride tubing of internal diameter 5 mm. and wall thickness 2-5 mm., so that the walls of the perfusion system are kept effectively rigid. Air traps were sited so that any air bubbles entering the system can easily be removed. Thus the contents of the system are not compressible, and the disturbance of flow when the paraffin pump reversed is kept as small as possible. Perfusate.-The perfusion fluid was made up from plasma and dextran solution. The plasma was obtained by bleeding two cats, a donor and the experimental animal. The dextran solution was Dextraven (Benger's 6 per cent w/v dextran in 0 9 per cent w/v NaCl) modified to contain 20 mm NaHCO3, 5-5 mm glucose, 4 0 mm KCI and 2-0 mm CaCl2. Depending upon the amount of plasma collected, which varied from 45 to 70 ml./kg. body weight, the amount of dextran solution added to make up the final perfusate volume of ml. varied between 250 and 350 ml. The calculated concentration of plasma protein in the perfusate varied from 1 to 2 g./100 ml. (The plasma protein was estimated in six experiments, using the micro-kjeldal method, and averaged 1-5 g./kg. body weight.) The viscosity of the perfusate, relative to H20, varied from 3-7 to 4 0 measured at 35 C. in an Ostwald viscometer (five experiments). Experimental.-The donor cat, anaesthetized with 40 mg./kg. Nembutal (Abbott's veterinary solution) intraperitoneally, was first bled to death. This was done after an intravenous injection of heparin, 2500 i.u./kg. Arterial blood was collected in siliconed vessels during three brief periods of bleeding; an injection of Dextraven 10 ml./kg. i.v. was given 1 min. before each bleeding. The plasma, after separation by centrifuging at 17 C., was put into the apparatus followed by 250 ml. of dextran solution, or more if required: next the perfusion chambers were filled up with paraffin. The perfusate was then circulated by the pump to ensure that the solutions were properly mixed and warmed. The experimental cat was then bled in the same manner. Immediately after death the lungs of the experimental animal were well inflated and its trachea clamped. The thoracic viscera were removed, weighed, and set up for perfusion. The input cannula, connected to the arterial side of the perfusion system, was introduced through the wall of the right ventricle and was tied in a position so that its tip was distal to the valves of the pulmonary artery. The output cannula, connected to the venous side, was tied into the left auricular appendage. The ventricles were tied off. The perfusion was started within 30 min. of the experimental cat's death. The input and output cannulae, from which the lungs were suspended, were

4 Pulmonary Hydrodynamics 355 clamped rigidly in such a way that the dorsal surfaces of the lungs rested lightly on a perforated plastic dish. The lungs were lubricated by a foam obtained by bubbling a mixture of 5 per cent CO2 in 02 through the few mls. of fluid which drained from their external surfaces and from the chambers of the heart. The lungs were thus effectively surrounded by an atmosphere rich in CO2. They were ventilated by a moistened mixture of 5 per cent CO2 in air which was fed rhythmically to the tracheal inlet by a Starling pump in the intervals between the short test periods at steady tracheal pressures. The perfusate CO2 tension was kept up during the earlier stages of the experiment by equilibrating the two fluids of which it was composed with 5 per cent CO2 in 02. These precautions were sufficient to maintain the ph of the perfusate between 7-4 and 7-5 for many hours. Samples of perfusate were withdrawn from time to time and their ph measured with a glass electrode. The lungs were warmed by the circulating perfusate, the temperature of which was maintained by keeping the perfusion chambers and venous reservoir submerged in a large water-bath at 38 C. The temperature of the perfusate as it left the lungs was shown on a thermometer inserted into the outflow tubing close to the auricle. The temperature of the venous fluid inevitably varied with changes in the conditions of flow but not by more than 2 C. at the extremes of flow. Usually it was between 33.5 and C. (Edema.-In early experiments, the extent of cedema was judged by comparing the weight of the thoracic viscera before and after perfusion. Later it was estimated from the gain in weight of the lungs alone; lungs were weighed at the end of an experiment and their weight compared with the control figure of 5-7 g. lung/kg. body weight obtained from healthy donor cats. In good experiments, after perfusion for several hours, the weight remained of this order in spite of the fact that the capillary pressure may have been raised well above the osmotic pressure of the perfusate for many minutes at a time during the experiment. But if cedema once started to form, the increase in weight was almost unlimited. In preliminary experiments, when dextran solution without plasma was used as perfusate cedema developed early. In all of the later experiments when the concentration of plasma protein in the perfusate was 1-2 g. per cent w/v, cedema formation was greatly reduced. There was probably little difference between the colloid osmotic pressures of the two perfusates used and so we suppose that plasma protein influenced the development of cedema not by a simple osmotic action but by some action on the capillary wall. Pres8ure Measurements.-Steady tracheal pressures were measured by a water manometer of 6 mm. bore, read directly. Perfusion pressures were measured either by mercury manometers recording on a kymograph or by water manometers read directly. At constant flows the water manometers were very satisfactory; the verticle tubes of the arterial and venous manometers were mounted parallel against the same scale so that small pressure differences could be seen easily. Perfusion pressures were referred to the opening of the arterial cannula at zero level. The opening of the venous cannula was usually within 1 or 2 mm. of this. The pressure on the venous side could fall below atmospheric pressure because the venous reservoir was about 20 cm. below the level of the mouth of the cannula, and was open to the air. In such cases we assumed that the pressure in the pulmonary veins was atmospheric. The venous pressure could be raised by increasing the air pressure acting on the fluid in the venous reservoir. Lung perfusate volume.-changes in the lung perfusate volume were estimated by connecting the venous reservoir to a small Krogh spirometer writing on a kymograph During such estimations the water manometers were cut out of the system to reduce capacity errors. If estimations of lung perfusate volume were required at a raised venous pressure, a thin rubber balloon was inflated in the venous side to a fixed pressure.

5 356 Banister and Torrance SYMBOLS USED PA= pulmonary artery pressure PD =pressure difference, PA-PV referred to the opening of the pulmonary artery can- PA +Pv nula (cm. H20). PM=mean pressure, 2 Pv = venous or left auricular pressure similarly referred F (cm. H20). C =conductance -PV F = flow of perfusion fluid (ml./ min.). R=rssac,PA - P PT= tracheal pressure (cm. H20) R =resistance, F RESULTS The effect of the tracheal pressure (PT) upon the arterial pressure (PA) was first studied with the left auricular pressure (Pv) fixed at atmospheric pressure and the inflow set at a medium rate. Under these conditions the relation between the two pressures was linear over 20 - the range 5-16 cm. H20 and the slope was positive; in some tests the slope be-,,/nf4 came less below a tracheal pressure of cm. H20. Observations from a typical test run are plotted in fig. 2. During E~ -such runs the tracheal pressure was raised 0o and lowered at least twice and the lungs were not allowed to collapse, the pressures tested being above 3-0 cm. H20. The graph shows that the increase in arterial pressure about equalled the applied increase in tracheal pressure; slopes of,,this order were always found in good O S preparations. PT. cms Ho The slope of the curve relating the FIG. 2.-The effect on the arterial pressure arterial pressure to the tracheal pressure (PA) of varying the tracheal pressure (PT) was affected by the inflow rate, increasing in the isolated lungs of a cat (2-2 kg.), perfused at a steady inflow of 40 ml./min. as the flow increased. When this was (F 40) and at venous pressure 0 (Pvo). low (less than about 30 ml./min.), the The line has a slope of slope was less than 1; when it was high (greater than about 150 ml. min.), the slope was greater than 1, on occasions exceeding 1.3. The upper curves of fig. 3, determined at two different inflow rates, illustrate the point. The height of the left auricular pressure (Pv) was an important parameter in fixing the slope of the curve relating arterial pressure to tracheal pressure. The bottom curve of fig. 3, determined at a very high venous pressure (Pv 27 cm. H20), well above highest value of the tracheal pressure tested and at a medium inflow, has a slope much less than either of the two curves

6 Pulmonary Hydrodynamics 357 determined with the venous pressure at atmospheric pressure (Pvo) whether with high or low inflows. An example, from another animal, of the action of high venous pressures in diminishing the effect upon the arterial pressure of raising the tracheal pressure is shown in fig. 4. Here the tests at high venous pressure (Pv 15-1 and 12-6 cm. H20) and low venous pressure (Pvo) were carried out at equal flows (50 ml./min.). That the effect of a raised 25s 25- AX,(R PvPvoFso Fl L / E *i E PT. cms H10 PT. cms HLO FIG. 3 FIG. 4 FIG. 3.-The graphs relate the pressure difference (PD) across the lung vascular bed of a cat (2.8 kg.) to the tracheal pressure (PT) at three different steady flows (F). Upper: F 169 ml./min., venous pressure (Pv) 0, slope Middle: F 27 ml./min., Pv 0, slope Lower: F 68 ml./min., PV 27 cm. 20, slope FIG. 4.-The graphs relate the pressure difference (PD) across the lung vascular bed to tracheal pressure (PT) at a steady inflow of 50 ml./min. Upper: venous pressure (Pv) 0. Lower: Pv 12.6 and 15-1 cm. H20. The points plotted as filled circles were determined before those plotted as open circles. (Same cat as in fig. 2.) venous pressure upon the slope of the curves is fully rev-ersible may be seen in fig. 4; the points plotted as filled circles were obtained before those plotted as open circles. The lower curves of fig. 4 were determined with the venous pressure set close to the upper limits of the tracheal pressures used. An increase in slope is apparent as the tracheal pressure is raised. In other experiments the curves were not always found to be so clearly convex to the tracheal pressure axis under these conditions. As might be expected the effect of the tracheal pressure upon the arterial pressure was markedly changed in preparations which developed cedema. The curve relating the two pressures remained approximately linear but the slope was reduced to 041 or even less once gross cedema had set in.

7 358 Banister and Torrance In another series of tests we studied the effect of tracheal pressure upon the relation between the arterial and venous pressures. This was done at selected steady flows and various tracheal pressures by repeatedly raising then lowering the venous pressure in steps from 0 to a pressure above the tracheal pressure tested. Fig. 5 shows a set of curves relating the arterial and venous pressures at four different tracheal pressures, with the inflow constant at 54 ml./min. The arterial pressure, at any venous pressure, was 30-25,20 dic C: < 20 Pv. cms H,0 FIG. 5.-The effect on the arterial pressure (PA) in the lungs of a cat (2-2 kg.), perfused at a steady inflow of 54 ml./min., of varying the venous pressure (Pv) at tracheal pressures (PT) of 16, 12, 8 and 4 cm. H20. The lines were fitted by eye. higher with raised tracheal pressures as might be expected from the observations described already. The curves also show that the arterial pressure was not much changed by raising the venous pressure until the venous pressure approached a value within a centimetre or so of the tracheal pressure; slight increases in arterial pressure were obtained however before the venous pressure equalled the tracheal pressure. The results plotted in fig. 5 show clearly that raising the venous pressure, at any tracheal pressure, reduced the pressure drop across the vascular bed, and confirm the observations of Carlill and Duke [1956] which were made on rhythmically ventilated preparations. The effect of the tracheal pressure upon the relation between the variables, inflow rate (F) and arterial pressure (PA), will be described only briefly for two reasons. First, the effects we observed were predictable from the relation of the arterial pressure to the tracheal pressure already described in some detail. Secondly, the flow: pressure curves obtained at steady tracheal

8 Pulmonary Hydrodynamics 359 pressures and steady inflows were convex to the P axis and otherwise similar in form to the curves published by many workers. Flow: pressure curves were shifted to the right along the P axis by a rise in tracheal pressure and the shift was greater at high flows. The shifts were similar in direction but larger than those described by Barer and Nusser [1957]. Carlill et al. [1957] also found that flow: pressure curves were shifted along the P axis in response to a raised tracheal pressure; their fig. 2, however, indicates a slightly reduced shift at high flow rates. Flow : pressure curves were determined in a number of preparations; in good experiments they altered little during several hours. A progressive shift of the curve to the right along the P axis was associated with the development of cedema. Changes in lung perfusate volume were estimated on a few preparations so that the effects of the variables, tracheal pressure, venous pressure and flow on the lung perfusate volume, could be ascertained. Raising the tracheal pressure caused an increase in lung perfusate volume at low rates of inflow but had little effect at high rates. If, however, the venous pressure was raised and kept constant, then increasing the tracheal pressure caused a fall in lung perfusate volume both at high and at low rates of inflow. We observed a striking increase in lung perfusate volume on raising the venous pressure, confirming the results of Carlill and Duke [1956]. We also found that the lung perfusate volume increased when inflow was increased. Certain relations between the variables flow and arterial pressure, venous pressure, and tracheal pressure we were not able to determine directly. For, with the apparatus used, it was difficult to keep the arterial or venous pressures at fixed values if the inflow rate was altered. In earlier experiments we attempted to construct curves to obtain the relations between these four variables by using observations taken from arterial : venous pressure curves determined directly at three tracheal pressures for each of three flow rates; this took so long that changes occurred in the state of the preparation which prevented all the sets of observations being comparable. We therefore attempted to devise a method for deriving additional information from our experimental results so that relations between variables not easily determined experimentally could be examined: in particular the effects of tracheal pressure upon flow through the lungs. Calculation of Conductance(C) : Mean Pressure(PM) Curves The relation between flow and pressure can be conveniently expressed as a conductance. Conductance is defined as the ratio of flow in ml./min. to pressure drop across the vascular bed in cm. H20, and is the reciprocal of resistance. Our experiments indicated that conductance was raised both by an increase in arterial pressure and by an increase in venous pressure and that it was related to the mean of the arterial and venous pressures. The relation between conductance and mean pressure was approximately linear under certain conditions whereas the relation of resistance to mean pressure was approximately hyperbolic. This made conductance more convenient to work with, particularly at low vascular pressures.

9 360 Banister and Torrance Typical conductance: mean pressure curves, derived from a series of observations on one preparation, are shown in fig. 6 for two steady tracheal pressures. To obtain the curves the arterial pressure was varied by changing the flow at low venous pressures; other points were obtained by varying the venous pressure at two steady flows. At low mean pressures the curves are approximately straight lines, but at the higher pressures they become concave - 2C z u PT4 0 PT12 5 MEAN PRESSURE cms H,O.I IS 20 FiG. 6.--The effect of tracheal pressure (PT) upon the relation between Conductance (ml./min./cm. H20 PD) and mean pressure (PA + Pv/2) in the lungs of a cat (2-1 kg.). Points were obtained at PT values of 4 and 12 cm. H20; those plotted as open circles were derived from flow: arterial pressure (PA) observations at a venous pressure (Pv) zero; those plotted as filled circles from PA: PV curves at inflow 70 ml./min.; those plotted as crosses from PA: Pv curves at inflow 140 ml./min.: curve PT8 was interpolated using the linear relation between PA and PT. to the P axis. The curve at the low tracheal pressure (PT4) extrapolates to cut the mean pressure axis very close to the origin; at the high tracheal pressure (PT12) the curve extrapolates to well to the right of the origin. Conductance: mean pressure curves can be constructed for intermediate tracheal pressures by using the linear relation already established between arterial and tracheal pressures at constant flow and venous pressure. The middle curve of fig. 6 for the intermediate tracheal pressure (PT8) was interpolated in this way. Conductance: mean pressure curves enable us to calculate the flow to be expected at any set of values for the pressure variables PA, Pv, and PT. From the conductance: mean pressure curve for a given tracheal pressure the conductance at the required mean pressure is read off and multiplied by

10 Pulmonary Hydrodynamics the pressure drop (PA - PV) to get flow. We derived in this way a series of curves from fig. 6 to illustrate how flow is related to each of the pressure variables separately with the remaining pair at selected values. This series is presented in figs Flow: arterial pressure curves are displayed at low (fig. 7a) and high (fig. 7b) tracheal pressures, and show how strikingly flow is reduced at the raised tracheal pressure. They illustrate also the increasing shift to the right caused by the raised tracheal pressure at the higher flows. The influence of the venous pressure on the two sets of curves should also be noted. Raising the venous pressure shifts the curve to the right by an amount which is always less than the rise in venous pressure and is markedly affected by the height of the tracheal pressure. The shifts to the right caused by increases in venous pressure in fig. 7b are clearly less than those in fig. 7a. At low venous and tracheal pressures the formula F = KPA2 seems a reasonable approximation because the conductance: mean pressure curve is approximately linear under such conditions and extrapolates to cut the pressure axis close to the origin. Flow: venous pressure curves are shown in fig. 8 at low and high tracheal pressures. The flow is not much altered by increasing the venous pressure when this is low in relation to the arterial pressure. When the tracheal pressure is high the effects on flow of increases in venous pressure are delayed. Flow: tracheal pressure curves are shown in fig. 9. They are all negative in slope and convex to the PT axis. So at all the pairs of arterial and venous pressures examined, there is a marked reduction in flow when the tracheal pressure is raised. An interesting point is shown up by the bottom two curves in the figure. Small flows still occur when the tracheal pressure is equal to the pulmonary arterial pressure. DISCUSSION 361 The results which we describe in this paper were obtained under conditions very different from those used by most previous workers. Our technique permitted accurate measurements of arterial and tracheal pressures at steady states of flow and inflation of the lung. We hope to show that observations made under these conditions furnish a reliable basis for a discussion of the effects of tracheal pressure upon some aspects of pulmonary hydro dynamics. That the tracheal pressure has a striking action upon the flow: pressure characteristics of the pulmonary vessels is clear from our experimental results. As no single hypothesis can be found to account for them all, we conclude that inflation of the lungs by positive pressure must act on the vessels of the pulmonary circuit in a number of different ways. Four of the possible ways in which the pulmonary vascular bed, perfused at constant steady inflow, might be affected by a state of lung inflation will be considered. First.-The increased pressure in the air-spaces of the lung might be applied to sets of collapsible blood vessels whose response would be determined by their internal pressure. Such vessels would collapse and flow would

11 362 Banister and Torrance E U PT P 300- PVI2 "~ PVI o PTI2 PPvo PVS Pa PA. cms. HILO PA. cms H,O FIG. 7a FIG. 7b FIGS. 7a and 7b.-Curves derived from fig. 6 relating flow (F) and pulmonary arterial pressure (PA) at venous pressures (Pv) 0, 8, 12, and 16 cm. H20. Fig. 7a at tracheal pressure (PT) 4 cm. H20; fig. 7b at tracheal pressure 12 cm. H PA20 PT4 C.200L 200-\20\ PA6 E U- 100 X PA1I2 PAI6 0 S I PV c m s. H10 Pv.emsHLO. FIG. 8a FIG. 8b FIGS. 8a and 8b.-Curves derived from fig. 6 relating flow (F) and venous pressure (Pv) at pulmonary arterial pressures (PA) of 28, 24, 20, 16, 12 and 8cm. H20. Fig. 8a at tracheal pressure (PT) 4 cm. H20; fig. 8b at tracheal pressure 12 cm. H PA \\ 2 NZ FIG. 9.-Curves derived from fig. 6 relating flow (F) and PA20rX2 tracheal pressure (PT) at pulmonary arterial E pressures (PA) and 12 cm. H20 and venous pressures (Pv) 0, 8,.12 cm. H20. LL loaa ~ PT. cms. H,O.

12 Pulmonary Hydrodynamics 363 cease when the pressure within them was some value perhaps equal to or just less than the external pressure applied; but, when the internal pressure built up to exceed this value the vessels would distend and flow would then occur quite freely. These vessels behave like sluices; they behave like the arterial resistance of a Starling heart-lung preparation or like the vessels beneath a sphygmo manometer cuff. It is not difficult to think of alveolar capillaries so acting. Secondly.-The increased volume of the lung might lengthen some at least of the vessels and so tend to increase their resistance to flow. However the lengthening would almost certainly change the diameter of the vessels and might also alter their distensibility. So the change in resistance would depend upon the elasticity of the elements which form the vessel wall and their orientation within it. Thirdly.-The increased arterial pressure produced by the increased tracheal pressure might, as a secondary effect, lower the resistance of the vascular bed. Lowering of resistance will occur either if the increased pressure acts in vessels not subjected to the full tracheal pressure change or if the rise of arterial pressure in a set of vessels is greater than the increase in tracheal pressure applied because of lengthening and sluice effects in vessels further downstream. Fourthly.-The mean height of vessels, relative to the arterial cannula, would tend to be greater during a state of inflation of the lungs than during a state of deflation. This should reduce the " effective " or trans-mural pressure in the vessels and so increase their resistance. Our experimental results relating the arterial pressure to the tracheal pressure can be interpreted in terms of these four possibilities. The predictions of the sluice hypothesis (1) are fitted by two sets of results. First, a rise in tracheal pressure caused an approximately equal rise in arterial pressure, so long as the inflow rate was moderate and the venous pressure low. Secondly, the arterial pressure was affected by the venous pressure only when the venous pressure approached the tracheal pressure. Two other sets of results however are not compatible with this hypothesis alone. These are (a) the effect of inflow rates, and (b) the effect of high venous pressure, upon the slope of the curve relating the arterial and tracheal pressures. (a) At high flow rates the slope of the curve relating the two pressures was greater than unity. This fact can be accounted for if there is a constant sluice action together with a lengthening, by inflation, of vessels upstream from the sluice vessels (hypothesis 2). For a simple lengthening of vessels should increase the pressure drop across them more at high flows though it increases their resistance in the same proportion at all flows. Another factor contributing to a slope steeper than unity might be a rise in the mean height of vessels in relation to the arterial cannula (hypothesis 4); the rise should affect the slope equally at high or at low flows. At low flow rates the slope of the curve relating the two pressures could be much less than unity. Two explanations are possible. First, if the resistance of some pathways through the lungs were independent of the

13 364 Banister and Torrance tracheal pressure and if those pathways lay in parallel with others exhibiting a sluice-like response to the action of the tracheal pressure, then it can be shown that the slope of the curve would be less than unity. And the slope would approach zero as the proportion of vessels by-passing the sluice vessels increased. As it has been claimed that there are large capillaries into which flow is diverted on inflation of the lungs [Olkon and Joannides, 1930] and arterio-venous anastomoses may be present [see Daly, 1958], vessels of these kinds could be the pathways concerned. On the other hand the sluice vessels themselves may become inaccessible to the action of the tracheal pressure in some parts of the lung, due to the development of cedematous patches or to bronchial obstruction; effects of this kind cannot be excluded in experiments on perfused lungs. Secondly, any reduction in resistance as a result of the rise in arterial pressure produced by inflation would be most marked at low flows; for the resistance of the pulmonary vessels is most sensitive to changes in intravascular pressures when these are low. Effects such as these presumably occur at all flows, but at high flows they would be masked by the factors causing slopes greater than unity. (b) At high venous pressures, well above the tracheal pressure, the slope of the curve relating the arterial and tracheal pressures was well below unity. The sluice vessels should remain distended during steady inflations at high venous pressures. The increase in resistance which occurred under these conditions however might be attributed to lengthening of the blood vessels, as discussed later. The slope of the arterial-tracheal pressure curves increased as the tracheal pressure approached the venous pressure. We had expected that there should be a tracheal pressure for any venous pressure at which the sluice mechanism would begin to act, and, at this value, the slope of the curve should become suddenly steeper. Bu no sudden alteration in slope was found either when arterial: tracheal pressures at intermediate venous pressures were determined directly (fig. 5) or when they were derived from arterial: venous pressure curves. In this treatment of the sluice hypothesis it is suggested that the pressure required to open the sluices is related to the tracheal pressure and so far we have regarded the multitude of sluices scattered throughout the lung as being concentrated at one level relative to the cannulae. At any venous pressure, however, the pressure within vessels above the cannulee will be less than that in the vessels below. So, as the tracheal pressure is raised with the venous pressure at a moderate level, the sluice vessels should collapse first in the higher parts of the lung and later in the lower parts. As the highest part of the lungs may be more than 5 cm. above the lowest the slope of the arterial: tracheal pressure curve should change gradually over a 0-5 cm. H20 range of tracheal pressures. Thus we were able to predict, on the basis of the hypotheses discussed, that the arterial: tracheal pressure curve should be convex to the tracheal pressure axis when the venous pressure falls within the range of tracheal pressures used.

14 Pulmonary Hydrodynamics The curve relating the arterial and tracheal pressure should have a slope of unity if the sluice hypothesis were valid and slopes close to this value were recorded under conditions of moderate inflow rates and low venous pressures. Under other sets of conditions slopes different from unity were obtained. Low slopes were found when the tracheal pressure was less than the venous pressure and when the inflow rates were very low. All slopes recorded at any flow always decreased if the lungs developed cedema. Slopes greater than unity were obtained if the inflow rate was high. We have listed various factors thought to contribute to the final shape of the curve but the relative significance of each has not been fully assessed. Conductance: mean pressure curves provide a further basis for the analysis of the effects of the tracheal pressure on the pulmonary circulation. If the tracheal pressure affected the conductance of the pulmonary circuit solely by compressing all significant vessels then conductance should remain unaltered if the tracheal, arterial and venous pressure are raised by the same amount. The conductance : mean pressure curve should merely be shifted along the pressure axis by an amount equal to the change in the tracheal pressure. Fig. 6 shows however that a shift of this simple kind does not occur. The extent of the shift depends upon the initial mean pressure and is less than the change in tracheal pressure at low mean pressures and is greater than this at high mean pressures. A shift of the conductance: mean pressure curve along the mean pressure axis which is greater than the tracheal pressure change implies that factors other than compression of vessels per se must be increasing the resistance to flow (reducing the conductance). Such shifts are pronounced at high mean pressures i.e. conditions under which the venous pressure is well above the tracheal pressure tested. Lengthening of the blood vessels could be an important cause of the increase in resistance in these conditions. Some workers have studied the effects of inflation upon the pulmonary circulation by applying a pressure below atmospheric to the pleural surfaces. This is equivalent to raising the tracheal, arterial and venous pressures by the same amount whilst applying a constant pressure to the pleural surfaces. At physiological flow rates and mean pressures, such a method of inflation should decrease conductance, although at low flows it might increase conductance. The haemodynamic effects of pneumothorax in the cat can now be considered. In unilateral pneumothorax the same arterial and venous pressures will be applied to each lung and the mean alveolar pressure is also likely to be the same. However, the intra-pleural pressure will be less negative on the side of the pneumothorax so the conductance of the lung vessels on this side will be increased. Hence the immediate haemodynamic effect of the pneumothorax should be to increase the fraction of the heart output going to the affected lung. This effect may be modified later if ventilation of the affected lung is reduced and vasoconstriction develops, and the discussion is not relevant to cases of pneumothorax in which complete collapse of a lung occurs. VOL. XLV, NO

15 366 Banister and Torrance The effect of bronchial obstruction can be considered similarly. The alveolar pressure in the obstructed portion of the lung will differ from that within the remainder of the tissue while the same arterial and venous pressures will be applied throughout the lung and the intra-pleural pressure will be approximately constant. Hence, at the peak of inspiration the portion of lung distal to the obstruction should receive a greater fraction of the heart output than it does at expiration. Attempts have been made to apply the concept of critical closing pressure to the pulmonary circulation. Our results suggest that flow through the alveolar capillaries will cease if the pressure within them falls to some value less than the tracheal pressure (though related to it). We think it inappropriate to describe such a collapse of the vessel wall as "critical" closure in the sense that Burton [1950] uses the term. The term might be more applicable to channels in parallel with the alveolar capillaries. The curves, derived from fig. 6, relating flow and tracheal pressure (fig. 9) provide confirmation of some points discussed already. The curves, plotted at constant arterial pressure, should be straight lines with an intercept on the tracheal pressure axis equal to the arterial pressure if the tracheal pressure affected the pulmonary vascular bed only by having a sluice action upon the alveolar capillaries and these could not be by-passed. But the curves are all convex to the tracheal pressure axis and some flow still occurs when the arterial pressure equals the tracheal pressure. The presence of channels in parallel with the alveolar capillaries might account for these results. Some observations, obtained from a different set of experiments, suggest that a small flow occurs in blood-perfused lungs when the tracheal pressure exceeds the pulmonary arterial pressure, as well as under the special conditions of the present set. We occluded the pulmonary artery, with the tracheal pressure held constant, in cats with widely opened chests and natural circulation. The pressure distal to the occlusion fell at first rapidly and then more slowly until it reached a level below the tracheal pressure. This indicated that there were alternative channels not subjected to alveolar compression through which the blood was escaping. The level at which flattening out occurred was always greater at higher tracheal pressures. The interpretation of results on flow and closing pressure is complicated by the different level of vessels within the lung. In a lax organ of a density similar to blood, differences in height of vessels should be compensated by the differences in interstitial pressures. No such factor is likely in the air-filled lung, where the intra-alveolar pressure may be regarded as an important component of the interstitial pressure. The changes which we describe in lung perfusate volume caused by raising the tracheal pressure are not difficult to explain. At low venous pressures inflation of the lungs raises the arterial pressure as we have shown. It increases the capacity of the pulmonary artery; Macklin [1946] using latex proved this to be so in the dog's lung. An increase in the lung perfusate volume would be therefore expected. At high venous pressures though the capillaries are well distended, and a raised tracheal pressure, would by

16 Pulmonary Hydrodynamics collapsing them, tend to lower the volume of perfusate in the lungs. The overall change would depend upon the height of the venous pressure. Our results are consistent with Macklin's observations that the pulmonary artery capacity is increased by inflating the lungs, but not with his deduction that this procedure lowers the vascular resistance. Our results show that resistance to flow through cat lungs is raised by inflation and they fail to confirm the observations of Burton and Patel [1958] on the excised lungs of rabbits. They found that as the tracheal pressure is raised through the physiological range, the pulmonary vascular resistance falls. At higher ranges of pressure however it increases. Moreover, their theoretical treatment of the general problem does not account for the slopes greater than unity which we found in arterial pressure: tracheal pressure curves under certain conditions. At steady states of inflation and inflow to the lungs, increases in tracheal pressure reduce the conductance of the pulmonary blood vessels. That these results, obtained under the simplest possible hydrodynamic conditions may be applied to the pulmonary circulation in vivo is suggested by a few observations which were recorded under conditions of pulsatile arterial pressures. A sine wave of frequency 2/sec. was imposed upon the arterial pressure at constant mean flow, venous and tracheal pressures. At physiological pulse and tracheal pressures the conductance was little changed by the application of pulsatile pressures. However, when the tracheal pressure was high and the pulse pressure was so high that the " diastolic " pressure fell to atmospheric pressure, the conductance was found to increase. It is in such conditions that a sluice actioni would be most likely to occur during each cycle of the pulmonary arterial pressure in vivo. ACKNOWLEDGMENTS We wish to thank Mr. Frank O'Connor for his valuable technical assistance and cheerful co-operation throughout this work, and Mr. Tom Wright for building the paraffin pump. 367 REFERENCES BARER, G. R. and NUSSER, E. (1957). "Pulmonary blood flow in the cat. The effect of positive pressure respiration", J. Physiol. 138, BURTON, A. C. (1950). "On the physical equilibrium of small blood vessels", Amer. J. Physiol. 164, BURTON, A. C. and PATEL, D. J. (1958). "Effect on pulmonary vascular resistance of inflation of the rabbit lungs", J. Appl. Physiol. 12, CARLILL, S. D. and DUKE, H. N. (1956). "Pulmonary vascular changes in response to variations in left auricular pressure", J. Physiol. 133, CARLILL, S. D., DUKE, H. N. and JONES, M. (1957). "Some observations on pulmonary haemodynamics in the cat", J. Physiol. 136, DAT.Y, I. DE B. (1958). "Intrinsic mechanisms of the lung", Quart. J. exp. Physiol. 43, MACKLIN, C. C. (1946). "Evidences of increase in the capacity of the pulmonary arteries and veins of dogs, cats and rabbits during inflation of the freshly excised lung", Rev. Canad. Biol. 5, OLKON, D. M. and JOANNIDES, M. (1930). "Capillaroscopic appearance of the pulmonary alveoli in the living dog", Anat. Rec. 45,