Effect of Breathing Pattern and Level of Ventilation on Pulmonary Fluid Filtration in Dog Lung 1- 3
ZOHEIR BSHOUTY· and MAGDY YOUNES
Introduction
SUMMARY The effect of breathing pattern and level of ventilation on fluid filtration In the lung
We recently demonstrated that the lung under edema forming conditions was studied In an In situ left upper lobe (WL) canine preparation. accumulates edema at a greater rate when WL weight was continuously monitored. In Group 1, rate of edema formation (t..W/t..t) was measured ventilated with larger tidal volumes (1). In seven dogs at two vascular pressurea (35 and 45 mm Hg) while the WL was randomly ventilated This effect was independent of capillary under six conditions. At equlvelent vascular pressures and mean airway pressures (158W) (and hence, mean operating lung volume), Increasing respiratory frequency (f) enhanced t..W/t..t. This was reversed hydrostatic pressure (Ps), which was acwhen minute ventilation (liE) was returned to baseline by reducing tidal volume (VT), even when curately controlled in these experiments. J58W ware matched to baseline. Increasing VT also enhanced t..W/t..t whether VE was Increased (J It was also not due to the concomitant Appl Physlol 1988; 64:1900) or not (present study) and whether J58W was matched to baseline. In increase in mean transpulmonary presGroup 2 t..W/t..t was measured at fixed VT and f while Inspiratory/expiratory time ratio (TI/TE) was sure (Pij», since when Ptp was increased switched from 1:1 to 1:6. Shortening Inspiratory time by Increasing Inspiratory flow rate had no by positive end-expiratory pressure effect on t..W/t..t. Weconclude that Increasing VE, whether by raising VT or f, promotes greater ede(PEEP) to an equivalent extent, while ma formation by mechanisms that are Independent of vascular pressure or operating lung volume. keeping tidal volume (VT) constant, the Increasing VT appears to have an additional adverse effect over and above that of Increased VE. rate of edema formation (~WI~t) AM REV RESPIR DIS 1992; 145:372-376 changed in an opposite direction (1). These observations indicated that the pattern of lung volume cycling around a given mean importantly influences one and above those due to the associated in- left lower lobe (LLL) was removed after inserting a wide cannula into the left atrium or more of the fluid filtration factors that crease in VE. (LA) via the LLL vein and another cannula were not controlled in these experiments, into the LLL artery (pointing proximally). A such as perivascular tissue fluid pressure Methods side arm in the latter cannula was used to in(Pur), reflection coefficient for protein troduce an occlusive device into the level of Mongrel dogs were anesthetized with penpermeability (0), or membrane perme- tobarbital sodium and intubated. The animals the left main pulmonary artery (LPA) (4). The ability (2). left upper lobe (LUL) was then perfused, via were initially ventilated with one Harvard In the previous study (1), when VT was respirator at a f of 10 min? and a VT of 10 the LLL artery cannula, with blood withincreased, respiratory frequency (f) was to 15 ml/kg. Systemic blood pressure was drawn from a femoral artery cannula at a rate not altered. Accordingly, minute venti- monitored through a catheter that was insert- determined by the settings on a Cobe-Stockert lation (VE) was also higher. It is possible ed into the right carotid artery. To achieve sep- roller pump. Because low flow rates were that the observed increase in ~WI ~t was arate ventilation of the right and left lungs, desired in these experiments, narrow tubing not specifically related to an increase in a Kottmier tube was inserted through a tra- was inserted in the pump. The actual flow at different pump settings was obtained by timed VTbut simply to an increase in VE. If that cheostomy and inflated at the carina. The collections in graded containers. LUL inflow right and left lungs were then ventilated using were the case, then increasing VE by inpressure was monitored through an orifice in two Harvard respirators. The ventilator to the creasing f should also result in a greater left lung was equipped with a switch that the LPA catheter located proximal to the oc~WI ~t, whereas increasing VT while clusive device. would change inspiratory/expiratory time rakeeping VE constant (slower deeper tio (Tr/TE) from 1:1 to 1:6. The two ventilaThe two LUL veins were cannulated with breaths) should not affect a WI ~t. tors were equipped with switches that proIn the present study, we assess, at com- duced an electrical signal with each cycle and, parable capillary pressures, the effect of using an oscilloscope, f to the left lung could changes in f with and without changes be adjusted to be an exact multiple of that (Received in originalform December 14, 1990and in revised form August 20, 1991) in VT on ~WI at. Apart from their use- to the right lung. This was done to ensure an orderly and repetitive relation between right fulness in better understanding our earFrom the Respiratory Investigation Unit, lung and LUL expansion, thereby eliminatlier findings (1), the results may be of ing erratic behavior of the experimental LUL University of Manitoba, Winnipeg, Canada. 2 Supported by the Medical ResearchCouncil of practical relevance to the issue of which weight signal (which is also sensitive to right breathing pattern should be adopted lung expansion acting through displacement Canada. 3 Correspondence and requests for reprints when ventilating patients with pulmo- of the mediastinum). should be addressed to Dr. M. Younes,RS-307Renary edema. The results indicate that The remainder of the surgical preparation spiratory Hospital, Health Sciences Center, 810 higher VE, per se, promotes greater ede- was similar to the one described in detail earSherbrook Street, Winnipeg, Manitoba R3A IR8, ma formation. Furthermore, larger VT lier (3). Briefly, a wide thoracotomy was per- Canada. Fellow of the Canadian Heart Foundation. has additional independent effects over formed in the left fifth intercostal space. The 1
4
372
373
BREATHING PATTERN AND FWID RLTRATION
flexible tubing. Both tubes were connected to a Y cannula that was in turn connected to the bottom of a reservoir. The reservoir was constructed with an overflow that drained into the LA via the LLL vein cannula. Inflow pressure to the LUL could be accurately controlled by raising or lowering the reservoir to various levels. Systemic blood pressure (BP) and LPA pressure lines were connected to appropriate pressure transducers and zeroed at the level of the LUL pulmonary veins as they entered the LA. AWI At was determined gravimetrically. Using rubber patches glued to the surface, the LUL was hung from a force transducer (for details of lobe hanging see reference 4). To calibrate the force transducer, known weights were placed on several locations on the LUL. This process was repeated throughout the study, especially when the pattern of ventilation was modified. After a step increase in vascular pressure (raising the reservoir), there was an initial fast phase of weight gain (figure I). When vascular pressure exceeded a critical value, this fast phase was followed by a phase of slow weight gain. The rate of weight gain during this period was constant as long as vascular pressure remained constant (4-6). AWlAt wasderived from the slope of this slow weight gain segment over a period of 3 min (figure I). Blood flow to the LUL was maintained at 20 mllmin throughout the experiment by appropriate adjustment of the pump setting. At this rate the total pressure gradient across the lobe (LPA minus LUL vein) is less than I mm Hg (3). This was confirmed in all experiments
Systemic BP (mmH g)
by momentarily stopping pump flow and confirming that LPA pressure decreased by less than 2 mm Hg as it equilibrated with reservoir pressure. The latter value includes I mm Hg that is related to pressuregradient between LUL veins and reservoir at a flow of 20 mllmin (3). In this preparation, therefore, P, was within 1 mm Hg of the measured LPA pressure and could accordingly be accurately controlled and measured. Airway pressure (Paw) was monitored from the left channel of the Kottmier tube. The "raw" pressure signal was continuously recorded and used to determine end-inspiratory and end-expiratory pressures (EIP and EEP, respectively)with the different breathing patterns. EEP could be altered by submerging the respirator exhalation tube under water to different depths. In addition, the raw Paw signal was processed by a low-pass filter (cutoff frequency 0.014 Hz), providing a continuous record of Paw, which should be similar to mean alveolar pressure (Pal) (1). In this fashion, Pal could be monitored and controlled. The LUL was ventilated with 5010 CO, in 0,. With this gas mixture and given the minimal blood flow to the lobe, alveolar Pco, was nearly constant at physiologiclevels(rv 36 mm Hg) and independent of the level or pattern of ventilation.
Statistical Analysis Statistical significance between the six conditions in the first group of dogs was tested by two-way repeated-measures analysis of variance (ANOVA)and Tukey's test for multiple comparisons (7). Paired t test was used to detect differences between the two conditions in the second group of dogs. Differences were considered significant at p < 0.05.
Protocol Two groups of dogs were studied. In Group 1 (seven dogs), the LUL was randomly ventilated under six conditions (table 2): C,: VT = 4 ml/kg, f = 10 mirr", EEP = 2.5 mm Hg; C,: VT and EEP were kept equal to C, and f was doubled; C3 : f and EEP were kept
Examples of the response of lung weight to a step increase in vascular pressure (45 mm Hg) with two breathing patterns in the same animal are shown in figure 1. In figure lB, f to the LUL was twice the
.. .
:::f-A----~ -~ t••~1
~.~
-lIlli'•••• I&..
B
equal jo C, and VT was halved to produce same VE as in C,; C4 : VT and fwere kept equal to C3 and EEP was raised to match Paw of C, and C,; C,: VT = 8 mIlkg, f = 5 min", EEP = 2.5 mm Hg; and C.: VT and f were kept equal to C, and EEP was dropped to match Paw of Cj. During each condition, the reservoirwaselevated from a baselineof about 10mm Hg to one of two heights corresponding to vascular inflow pressures of 35 and 45 mm Hg, respectively. Both pressuresweretested with each pattern. The reservoir remained elevated for 4 min at each point and then was returned to baseline. AWlAt at each pressure and for each condition was measured at least twice in each lobe and an average value was obtained. In Group 2 (four dogs), AWlAt was measured under the same VT and f conditions as in C, while TIlTh waschanged from 1:1 to 1:6.
Results
~I
.. 'I
I lPA Pre ssure (mmHg)
20 - ~r-r-fITIjl II I ~. II
4°f
•
•
'I
o
I
I'.
h f~' H ~' ~'
I
"
' "I "
H ~H'''H
H n ,, ·.,! . .. .. , ...
·HH,IH,h IH ,111
..... . . . . ..... . . ... . . . . . . . . .... .
l Ul W e ight
_ IlIlillllllllll ll111lllll.~ _ _
l~~~ I ·,·..············....······........·.....··'" ..',,
lO g e
I
1IIIIIIIIIIIIIIIIIIIlllllllilillilllllidil. II.
L--.-J
1 min
t
5 g
,IJJ.
,INI4iljiljiNi~N,WIW
-----~
,l)JJJllIli I 'I !I' ltll"'"
" 11' 11''11111111 1111
I~r·', '
II "I
l'I'rl'l l jl ,illllli"ijljl;'11jl ,ij'i'ilili jljl l jilijlillllili jl l i;li~jilil\ilil g1',i l l l
•
5g
Fig. 1. Representative examples of the response of lung weight to a step increase in vascular pressure (to 45 mm Hg) with two breathing patterns in the same animal. (A) Respiratory frequency is 10 min-'. (B) Respiratory frequency is doubled while tidal volume is kept the same. BP is blood pressure; LPA is left main pulmonary artery; lUl is left upper lobe; and Paw is airway pressure. Downward deflection in weight tracing signifies weight gain. Toward the end of the slow phase of weight gain, 5 g weight was placed on the lobe to check consistency of weight calibration. Horizontal level is shown by the thin line along the weight tracing. Note the faster rate of weight gain (thick segment) in B compared with A.
374
BSHOUTY AND YOUNES TABLE 1 IlW/l1t WITH DIFFERENT BREATHING PATIERNS*
P, = 35 mm Hg Dog
1 2 3 4 5 6 7 Mean ± SEM
P, = 45 mm Hg
C,
C,
C3
C.
C,
C,
C,
C,
C3
C.
C.
C,
0.56 0.35 0.42 0.39 0.49 0.38 0.52 0.44 0.03
0.73 0.40 0.50 0.49 0.56 0.63 0.72 0.58 0.05
0.67 0.35 0.46 0.47 0.48 0.48 0.40 0.47 0.04
0.52 0.33 0.33 0.23 0.33 0.42 0.42 0.37 0.04
1.09 0.36 0.75 0.83 0.75 0.63 0.72 0.73 0.08
0.94 0.53 0.58 0.58 0.48 0.54 0.80 0.64 0.06
0.67 0.43 0.58 0.52 0.67 0.58 0.79 0.61 0.04
0.94 0.77 0.75 0.63 0.89 0.73 0.96 0.81 0.05
0.86 0.62 0.63 0.54 0.97 0.65 0.71 0.71 0.06
0.78 0.51 0.42 0.46 0.83 0.60 0.67 0.61 0.06
1.25 0.52 1.33 0.83 1.25 0.71 1.20 1.01 0.12
1.04 0.56 1.00 0.68 0.92 0.65 1.45 0.90 0.12
* Individual and mean (:!: SEM) values of rate of edema formation (I1WI M. g/min) in seven dogs obtained at two vascular pressures (P, 35 mm Hg and P, 45 mm Hg) and under the six ventilation conditions described in table 2. Each value is the average of at least two observations.
=
=
TABLE 2 BREATHING PATIERNS AND OVERALL MEANS OF IlW/llt* Condition C, C, C3 C. C. C,
f (min-')
VT (mllkg)
EEP (mmHg)
10 20 20 20 5 5
4 4 2 2 8 8
2.5 2.5 2.5 4.0 2.5 1.0
Peak Paw (mm Hg)
11.3 12.9 6.1 8.7 26.2 19.9
± ± ± ± ± ±
0.48 0.58 0.46 0.62 0.97 1.05
Paw (mm Hg)
4.9 5.2 3.2 5.1 7.9 5.4
± ± ± ± ± ±
0.15 0.21 0.09:1= 0.11 0.15:1= 0.17
Il W/l1t (g/min)
0.52 0.69 0.59 0.49 0.87 0.77
± ± ± ± ± ±
0.03 0.05t 0.05 0.05 0.08:1= 0.07:1=
Definition of abbreviations: C, to C, = various conditions under which each dog was studied; f ~ respiratory frequency; VT tidal volume; EEP end-expiratory airway pressure; Paw airway pressure; Paw mean airway pressure; 11W/l1t rate of edema formation (average of data at two vascular pressures) . • For peak Paw, Paw, and I1W/l1t, values are means:!: SEM (n = 14). Statistical significance was determined by two-way repeated-measures analysis of variance and Tukey's test for muttiple comparisons compared with baseline (C,).
=
=
=
=
=
t p < 0.05.
*P < 0.Q1.
value used in figure lA but VT was the same. Note the faster rate of the slow phase of weight gain in figure 1B compared with figure lA. Individual as well as mean values for AWI At at both vascular pressures (P, = 35 and P, = 45 mm Hg) and during the six conditions are shown in table 1. The overall means and statistical differences are shown in table 2. Increasing VE by raising f (C z) caused a significant (p < 0.05) increase in AWl At to 0.69 ± 0.05 glmin at equivalent operating lung volumes (Paw during C, and C, was 4.9 ± 0.15 and 5.2 ± 0.12 mm Hg, respectively). When VE was returned to baseline by decreasing VT, AWl At decreased to levelsthat were not significantly different from baseline. This drop occurred whether operating lung volume was matched to baseline (C 4 , Paw = 5.1 ± 0.11 mm Hg, AWl At = 0.49 ± 0.05 g/min) or not (C3 , Paw = 3.2 ± 0.09 mm Hg, AWl At = 0.59 ± 0.01 g/min). Increasing VT, while keeping VE constant, was also associated with a significant increase in AWl At (C s , AWI At = 0.87
± 0.08 glmin as opposed to 0.52 ± 0.03 in C" p < 0.01). Adjusting Paw to match C, had little effect on the results; AWI At remained significantly elevated (C 6 , AWI At = 0.77 ± 0.07 glmin, p < 0.01). These findings wereconsistent when tested individually at either microvascular pressure (35 or 45 mm Hg). Mean AW/At (± SEM, standard error of the mean) while ventilating the dogs with a TIlTh of 1:1 was 0.49 ± 0.07 compared with 0.51 ± 0.09 when ventilated with a TIlTh of 1:6. These results werenot different statistically (p = 0.80). Discussion
In the present study, the rate of edema formation was inferred from observing AWI At over a brief period of high vascular pressure (min 1-4 after a step increase in pressure). The advantage of this approach is that it permits several observations to be made in the same animal; there is a limit to cumulative weight gain after which frothing occurs and weight becomes unstable. This, in turn, makes
it possible to assess the effect of multiple variables that can affect the response. Two potential disadvantages need to be considered, however. First, it may be argued that the increase in lobar weight between 1 and 4 min is in part or in total the result of vascular filling in response to a step increase in pressure. We believe that the contribution of vascular congestion to weight gain over this period is negligible. The reasons for this assessment werediscussed in considerable detail in an earlier communication in which this preparation was first used (4). (1) Vascular filling occurs both at low and high step increases in pressure, whereas slow weight gain was observed only when vascular pressure exceeded a critical value (P crit). (2) When vascular pressure is returned to baseline, following a 4-min period of weight gain, lobar weight drops to a level that is invariably higher than that before the step increase in pressure. The magnitude of increase in baseline weight agrees very wellwith the amount of slow weight gain beyond the first minute. (3) There is good agreement between the increase in wet weight of the blood-drained lobe at the end of an experiment and the cumulative increase in force transducer signal. Second, it could also be argued that a 4-min observation, as in our study, is not enough to achieve steady-state conditions in which tissue fluid hydrostatic pressure, protein oncotic pressures, and lymph flow are stable. Under these conditions the observed changes in AWI At would be rather transient and pertain only to brief elevations in vascular pressures. As indicated earlier (4) and in agreement with observations by others in in situ lobes (5) (see isolated ex vivo lobes [6]), the rate of weight gain at pressures above P erit does not change beyond the fourth minute. This was confirmed in independent measurements in six dogs in which elevated pressures were maintained for periods of about 1 h and AWI At in the interval from 1 to 4 min was compared with the rates at 12 to "15 and 40 to 43 min. AWl At in the three periods were (mean ± SEM) 0.54 ± 0.08, 0.53 ± 0.08, and 0.49 ± 0.10 gm/rrtin,respectively. There was no significant difference between the three periods by ANOVA for repeated measures (F = 0.8, p > 0.4). It follows that extending the time of observation beyond 4 min offers no advantage in this preparation while limiting the number of observations that can be made. The present study demonstrates that
375
BRE"THING PATTERN AND FLUID FILTRATION
increasing VE by raising f (Cs) is associated with a significantly higher AWl At. In a previous study (1) we showed that increasing VE by increasing VT promotes greater water accumulation in the lung. The combined results indicate that higher ventilation, achieved by increasing either VT or f, promotes greater edema formation. It is wellestablished that increased ventilation induced by physiologic stimuli (hypoxia [8], hypercapnia [9], and exercise [10]) is associated with an increase in lung lymph flow.The increase in lymph flow in these experiments may have been related to an increase in intrarnicrovascular pressure (due to changes in blood flow or in longitudinal or parallel distribution of resistances), to differences in operating (mean) lung volume, to an increase in filtration area (due to vascular recruitment, for example), or reflex effects produced by regional or systemic changes in gas tensions (Pao., Pacoz, and pH). The main advantage of the present study is that all these confounding variables were controlled. Our results therefore indicate that an increase in rate or depth of breathing exerts an independent influence that promotes greater fluid filtration. There are two studies in which the effect of controlled changes in ventilation, produced by deliberate changes in ventilator settings, on lung lymph flow, was assessed (11, 12). Patterson and colleagues (11) measured lymph flow (QL) in isolated sheep lungs at various respiratory frequencies while keeping VTconstant. QL increased linearly with f. In their study, however, it was not clear whether QL increased because of a primary effect of f on lymphatic transport, redistribution of interstitial fluid, or a primary effect on fluid filtration, with the increase in QL a secondary phenomenon. By showing a greater rate of fluid retention with high f, the present results indicate that f acts primarily on fluid filtration. In fact, the associated increase in QL tends to underestimate the effect of f on the rate of fluid filtration. Martin and coworkers (12) measured lung lymph flow (QL) in an artificially ventilated, open-chested (thoracotomy), normally perfused canine preparation. QL increased when hyperventilation (increased VT) was produced using a hypoxic gas mixture but not in the "control" experiments in which VT was similarly increased but the fraction of inspired oxygen (FIoz) was 0.3. The results of these "control" experiments represent, to our
knowledge, the only data for which an increase in ventilation was not associated with an increase in QL. The reason for this is not clear. Two factors may have mitigated the increase in QL. First, the normoxic hyperventilation was associated with a significant decrease in Pacoz (41.5 to 26.8 mm Hg) and a significant increase in pH (7.38 to 7.48). These changes, acting through local or systemic mechanisms, could have altered the longitudinal or parallel distribution of pulmonary vascular resistance (thereby altering capillary pressure) or could have affected the pumping action of the lymphatics. Second, because perfusion pressure was, on average, only 22 mm Hg, it is possible that some derecruitment may have occurred in the high VT condition during inspiration; neither alveolar nor airway pressure was reported. The rate of fluid filtration from the pulmonary microvasculature (Qv.C> is governed by the Starling equation (2); Qv.c
=
LpA[(P c - Pti.r) a(xc - Xli)]
-
(1)
where. Lp is the filtration constant per unit area (index of permeability), A is the membrane area available for filtration, P, is the hydrostatic pressure within the lumen, Pur is the hydrostatic pressure in the tissue fluid surrounding the vessel, a is the reflection coefficient for proteins, and Xc and Xti are the plasma and tissue fluid oncotic pressures, respectively. In our experiments P, was matched, and there was no reason for Xc to change with breathing pattern. Because Paw (and hence operating lung volume) were also matched (C, versus C 1) , there is no reason to suspect a change in filtration area (A).* Furthermore, since Xti should decrease (8, 10)in association with greater filtration rates (unless permeability is increased), changes in this factor cannot account for the observed increase in
* In our experiments both inflow and outflow pressures were higher than peak airway pressures (and hence peak alveolar pressures) under all conditions (see table 1). The entire lobe was therefore in Zone III perfusion under all conditions and there is no reason to expect a difference in the number of recruited vessels between different experimental conditions. To the extent that Paw equals Pal (I) and pleural pressure was zero. where PaW was matched (for example, C.. C,. C., and C.) Pal and mean transpulmonary pressure (and hence, lung volume) was also matched. We believe that if intravascular pressure, alveolar pressure (extramicrovascular pressure), and alveolar surface area are matched, microvascularsurface area is also likely to be similar.
filtration. It follows that ventilating the lung with a higher f, while keeping VT constant, either increases permeability or decreases Pu.r or a. Although operating lung volume and volume oscillations per breath were matched in C z and C 1 , the rate of change in lung volume during inspiration (inspiratory airflow) was higher in C z • It is possible that the rate at which the alveolar-capillary membrane is stretched may influence permeability or a. Results of the second group of dogs dismisses this possibility. Decreasing inspiratory time by increasing flow while keeping VT and f constant had no effect on Ii.WIli.t. Alternatively, cycling of lung volume may promote the clearance of perivascular fluid to areas remote from the filtration site through a pumping action. At a given operating lung volume, more pumping cyclesper minute (or greater cycling amplitude, as with large VT) may result in a lower mean Pur. Our results do not permit a distinction between this latter mechanism and a change in permeability or a. Although it is possible to calculate a Kf,c value from the data in table 1 (Ii.WI At at P 2 minus Ii.WIli.t at P 1 divided by 10), we believe that two pressure points are not sufficient to reliably determine slope values. The return of Ii.WIli.t to baseline when VE was returned to baseline by reducing VT, despite the persistence of high f (C J and C 4 ) , further supports the notion that the important variable is the total amount of volume cycling per minute rather than f per se, Although the associated increase in VE with larger tidal volumes must have contributed to the higher AWl At observed in our previous study (1), this is not the entire explanation. In the present study increasing VT continued to cause a highly significant increase in AWl li.t even though VE and inspiratory flow rate were not altered (C, and C6 ) . The possible mechanisms for VTeffects werediscussed in detail earlier (1). These will not be repeated here. We concluded then that this effect is likely related to nonlinearities in the relation between lung volume and one or more of the factors that affect fluid filtration. The present results do not necessitate any change in this conclusion. Our findings have some practical implications. First, ventilation and ventilatory pattern of experimental lobes should be considered when comparing edema formation under different experimental conditions and between different studies.
376
Second, the practice of using large inflation pressures in ventilating patients with pulmonary edema may require reassessment. Third, the increase in ventilation and tidal volume that must occur with physiologic challenges, such as exercise and hypoxia (for example at high altitude), may contribute in susceptible individuals to the development of pulmonary edema under these conditions. It must be pointed out, however, that changes in ventilation and breathing pattern, spontaneously produced by an intact organism or artificially induced in a ventilated patient, need not be associated with the same P, or mean lung volume. These changes could enhance or counteract the effect of VE and VT on edema formation. Finally, it seems that high-frequency ventilation (HFV), during which VT is very small, should be associated with lower rates of edema formation. Because of the extremely high dead space effect (Vn/VT) that is associated with HFV, it is necessary to increase total ventilation to one or more orders of magnitude higher than with conventional ventilation to maintain gas exhange (13). The two changes (high VE but low VT) have op-
BSHOUTY AND YOUNES
posite effects on fluid filtration. This may explain the small and inconsistent effects of HFV on lymph flow reported by different investigators (14-16). Acknowledgment The writers thank D. Lobchuk for her technical assistance. References 1. Bshouty Z, Ali J, YounesM. Effect of tidal volume and PEEP on rate of edema formation in insitu perfused canine lobes. J Appl Physiol 1988; 64(5):1900-7. 2. Taylor AE, Parker J'C, Pulmonary interstitial spaces and lymphatics. In: Fishman AP, ed. Handbook of physiology. The respiratory system. Circulation and non-respiratory functions. Bethesda, MD: Am Physiol Soc 1986; sec. 3. vol. I, chap. 4, 167-230. 3. Hornik LA, Bshouty Z, Light RB, Younes M. Effect of alveolar hypoxiaon pulmonary fluid filtration in in situ dog lungs. J Appl Physiol 1988; 65(1):46-52. 4. YounesM, Bshouty Z, Ali J. Longitudinal distribution of pulmonary vascularresistancewith very high pulmonary blood flow. J Appl Physiol1987; 62:344-58. 5. Drake RE, Smith JH, Gabel JC. Estimation of the filtration coefficient in intact dog lungs. Am J Physiol 1980; 238(7):H430-8. 6. Morriss AW, Drake RE, Gabel JC. Comparison of microvascularfiltration characteristicsin isolated and intact lungs. J Appl Physiol 1980;
48:438-43. 7. Neter J, Wasserman W.Applied linear statistical models. Homewood, IL: R. D. Irwin, Inc., 1974; 474-82. 8. Warren MF, Peterson DK, Drinker CK. The effects of heightened negative pressure in the chest, together with further experiments upon anoxia in increasing the flow of lung lymph. Am J Physiol 1942; 137:641-8. 9. Albelda SM, Hansen-F1aschen JH, Lanken PN, Fishman AP. Effects of increased ventilation on lung lymph flow in unanesthetized sheep. J Appl Physiol 1986; 60(6):2063-70. 10. Coates G, O'Brodovich H, Jeffries AL, and Gray GW. Effects of exercise on lung lymph flow in sheep and goats during normoxia and hypoxia. J Clin Invest 1984; 74:133-41. l l. Patterson GA, Mitzner WA,Sylvester JT. Assessment of fluid balance in isolated sheep lungs. J Appl Physiol 1985; 58:882-91. 12. Martin DJ, Grimbert FA, Baconnier P, Benchetrit G. Effect of acute hypoxia on lung transvascular filtration in anesthetized dogs. Bull Eur Physiopathol Respir 1983; 19:7-11. 13. Froese AB, Bryan AC, High frequency ventilation. Am Rev Respir Dis 1987; 135(6):1363-74). 14. Jefferies AL, Hamilton P, Bryan AC, O'Brodovich H. Effect of high frequency oscillation (HFO) on lung lymph flow. J Appl Physiol 1983; 55:1373-8. 15. Martin D, Rehder K, Parker JC, Taylor AE. High-frequencyventilation: lymph flow,lymph protein flux, and lung water. J Appl Physiol 1984; 57:240-5. 16. Raj JU, Goldberg RB, Bland RD. Vibratory ventilation decreases filtration of fluid in the lungs of newborn lambs. Circ Res 1983; 53:456-63.
ZOHEIR BSHOUTY· and MAGDY YOUNES
Introduction
SUMMARY The effect of breathing pattern and level of ventilation on fluid filtration In the lung
We recently demonstrated that the lung under edema forming conditions was studied In an In situ left upper lobe (WL) canine preparation. accumulates edema at a greater rate when WL weight was continuously monitored. In Group 1, rate of edema formation (t..W/t..t) was measured ventilated with larger tidal volumes (1). In seven dogs at two vascular pressurea (35 and 45 mm Hg) while the WL was randomly ventilated This effect was independent of capillary under six conditions. At equlvelent vascular pressures and mean airway pressures (158W) (and hence, mean operating lung volume), Increasing respiratory frequency (f) enhanced t..W/t..t. This was reversed hydrostatic pressure (Ps), which was acwhen minute ventilation (liE) was returned to baseline by reducing tidal volume (VT), even when curately controlled in these experiments. J58W ware matched to baseline. Increasing VT also enhanced t..W/t..t whether VE was Increased (J It was also not due to the concomitant Appl Physlol 1988; 64:1900) or not (present study) and whether J58W was matched to baseline. In increase in mean transpulmonary presGroup 2 t..W/t..t was measured at fixed VT and f while Inspiratory/expiratory time ratio (TI/TE) was sure (Pij», since when Ptp was increased switched from 1:1 to 1:6. Shortening Inspiratory time by Increasing Inspiratory flow rate had no by positive end-expiratory pressure effect on t..W/t..t. Weconclude that Increasing VE, whether by raising VT or f, promotes greater ede(PEEP) to an equivalent extent, while ma formation by mechanisms that are Independent of vascular pressure or operating lung volume. keeping tidal volume (VT) constant, the Increasing VT appears to have an additional adverse effect over and above that of Increased VE. rate of edema formation (~WI~t) AM REV RESPIR DIS 1992; 145:372-376 changed in an opposite direction (1). These observations indicated that the pattern of lung volume cycling around a given mean importantly influences one and above those due to the associated in- left lower lobe (LLL) was removed after inserting a wide cannula into the left atrium or more of the fluid filtration factors that crease in VE. (LA) via the LLL vein and another cannula were not controlled in these experiments, into the LLL artery (pointing proximally). A such as perivascular tissue fluid pressure Methods side arm in the latter cannula was used to in(Pur), reflection coefficient for protein troduce an occlusive device into the level of Mongrel dogs were anesthetized with penpermeability (0), or membrane perme- tobarbital sodium and intubated. The animals the left main pulmonary artery (LPA) (4). The ability (2). left upper lobe (LUL) was then perfused, via were initially ventilated with one Harvard In the previous study (1), when VT was respirator at a f of 10 min? and a VT of 10 the LLL artery cannula, with blood withincreased, respiratory frequency (f) was to 15 ml/kg. Systemic blood pressure was drawn from a femoral artery cannula at a rate not altered. Accordingly, minute venti- monitored through a catheter that was insert- determined by the settings on a Cobe-Stockert lation (VE) was also higher. It is possible ed into the right carotid artery. To achieve sep- roller pump. Because low flow rates were that the observed increase in ~WI ~t was arate ventilation of the right and left lungs, desired in these experiments, narrow tubing not specifically related to an increase in a Kottmier tube was inserted through a tra- was inserted in the pump. The actual flow at different pump settings was obtained by timed VTbut simply to an increase in VE. If that cheostomy and inflated at the carina. The collections in graded containers. LUL inflow right and left lungs were then ventilated using were the case, then increasing VE by inpressure was monitored through an orifice in two Harvard respirators. The ventilator to the creasing f should also result in a greater left lung was equipped with a switch that the LPA catheter located proximal to the oc~WI ~t, whereas increasing VT while clusive device. would change inspiratory/expiratory time rakeeping VE constant (slower deeper tio (Tr/TE) from 1:1 to 1:6. The two ventilaThe two LUL veins were cannulated with breaths) should not affect a WI ~t. tors were equipped with switches that proIn the present study, we assess, at com- duced an electrical signal with each cycle and, parable capillary pressures, the effect of using an oscilloscope, f to the left lung could changes in f with and without changes be adjusted to be an exact multiple of that (Received in originalform December 14, 1990and in revised form August 20, 1991) in VT on ~WI at. Apart from their use- to the right lung. This was done to ensure an orderly and repetitive relation between right fulness in better understanding our earFrom the Respiratory Investigation Unit, lung and LUL expansion, thereby eliminatlier findings (1), the results may be of ing erratic behavior of the experimental LUL University of Manitoba, Winnipeg, Canada. 2 Supported by the Medical ResearchCouncil of practical relevance to the issue of which weight signal (which is also sensitive to right breathing pattern should be adopted lung expansion acting through displacement Canada. 3 Correspondence and requests for reprints when ventilating patients with pulmo- of the mediastinum). should be addressed to Dr. M. Younes,RS-307Renary edema. The results indicate that The remainder of the surgical preparation spiratory Hospital, Health Sciences Center, 810 higher VE, per se, promotes greater ede- was similar to the one described in detail earSherbrook Street, Winnipeg, Manitoba R3A IR8, ma formation. Furthermore, larger VT lier (3). Briefly, a wide thoracotomy was per- Canada. Fellow of the Canadian Heart Foundation. has additional independent effects over formed in the left fifth intercostal space. The 1
4
372
373
BREATHING PATTERN AND FWID RLTRATION
flexible tubing. Both tubes were connected to a Y cannula that was in turn connected to the bottom of a reservoir. The reservoir was constructed with an overflow that drained into the LA via the LLL vein cannula. Inflow pressure to the LUL could be accurately controlled by raising or lowering the reservoir to various levels. Systemic blood pressure (BP) and LPA pressure lines were connected to appropriate pressure transducers and zeroed at the level of the LUL pulmonary veins as they entered the LA. AWI At was determined gravimetrically. Using rubber patches glued to the surface, the LUL was hung from a force transducer (for details of lobe hanging see reference 4). To calibrate the force transducer, known weights were placed on several locations on the LUL. This process was repeated throughout the study, especially when the pattern of ventilation was modified. After a step increase in vascular pressure (raising the reservoir), there was an initial fast phase of weight gain (figure I). When vascular pressure exceeded a critical value, this fast phase was followed by a phase of slow weight gain. The rate of weight gain during this period was constant as long as vascular pressure remained constant (4-6). AWlAt wasderived from the slope of this slow weight gain segment over a period of 3 min (figure I). Blood flow to the LUL was maintained at 20 mllmin throughout the experiment by appropriate adjustment of the pump setting. At this rate the total pressure gradient across the lobe (LPA minus LUL vein) is less than I mm Hg (3). This was confirmed in all experiments
Systemic BP (mmH g)
by momentarily stopping pump flow and confirming that LPA pressure decreased by less than 2 mm Hg as it equilibrated with reservoir pressure. The latter value includes I mm Hg that is related to pressuregradient between LUL veins and reservoir at a flow of 20 mllmin (3). In this preparation, therefore, P, was within 1 mm Hg of the measured LPA pressure and could accordingly be accurately controlled and measured. Airway pressure (Paw) was monitored from the left channel of the Kottmier tube. The "raw" pressure signal was continuously recorded and used to determine end-inspiratory and end-expiratory pressures (EIP and EEP, respectively)with the different breathing patterns. EEP could be altered by submerging the respirator exhalation tube under water to different depths. In addition, the raw Paw signal was processed by a low-pass filter (cutoff frequency 0.014 Hz), providing a continuous record of Paw, which should be similar to mean alveolar pressure (Pal) (1). In this fashion, Pal could be monitored and controlled. The LUL was ventilated with 5010 CO, in 0,. With this gas mixture and given the minimal blood flow to the lobe, alveolar Pco, was nearly constant at physiologiclevels(rv 36 mm Hg) and independent of the level or pattern of ventilation.
Statistical Analysis Statistical significance between the six conditions in the first group of dogs was tested by two-way repeated-measures analysis of variance (ANOVA)and Tukey's test for multiple comparisons (7). Paired t test was used to detect differences between the two conditions in the second group of dogs. Differences were considered significant at p < 0.05.
Protocol Two groups of dogs were studied. In Group 1 (seven dogs), the LUL was randomly ventilated under six conditions (table 2): C,: VT = 4 ml/kg, f = 10 mirr", EEP = 2.5 mm Hg; C,: VT and EEP were kept equal to C, and f was doubled; C3 : f and EEP were kept
Examples of the response of lung weight to a step increase in vascular pressure (45 mm Hg) with two breathing patterns in the same animal are shown in figure 1. In figure lB, f to the LUL was twice the
.. .
:::f-A----~ -~ t••~1
~.~
-lIlli'•••• I&..
B
equal jo C, and VT was halved to produce same VE as in C,; C4 : VT and fwere kept equal to C3 and EEP was raised to match Paw of C, and C,; C,: VT = 8 mIlkg, f = 5 min", EEP = 2.5 mm Hg; and C.: VT and f were kept equal to C, and EEP was dropped to match Paw of Cj. During each condition, the reservoirwaselevated from a baselineof about 10mm Hg to one of two heights corresponding to vascular inflow pressures of 35 and 45 mm Hg, respectively. Both pressuresweretested with each pattern. The reservoir remained elevated for 4 min at each point and then was returned to baseline. AWlAt at each pressure and for each condition was measured at least twice in each lobe and an average value was obtained. In Group 2 (four dogs), AWlAt was measured under the same VT and f conditions as in C, while TIlTh waschanged from 1:1 to 1:6.
Results
~I
.. 'I
I lPA Pre ssure (mmHg)
20 - ~r-r-fITIjl II I ~. II
4°f
•
•
'I
o
I
I'.
h f~' H ~' ~'
I
"
' "I "
H ~H'''H
H n ,, ·.,! . .. .. , ...
·HH,IH,h IH ,111
..... . . . . ..... . . ... . . . . . . . . .... .
l Ul W e ight
_ IlIlillllllllll ll111lllll.~ _ _
l~~~ I ·,·..············....······........·.....··'" ..',,
lO g e
I
1IIIIIIIIIIIIIIIIIIIlllllllilillilllllidil. II.
L--.-J
1 min
t
5 g
,IJJ.
,INI4iljiljiNi~N,WIW
-----~
,l)JJJllIli I 'I !I' ltll"'"
" 11' 11''11111111 1111
I~r·', '
II "I
l'I'rl'l l jl ,illllli"ijljl;'11jl ,ij'i'ilili jljl l jilijlillllili jl l i;li~jilil\ilil g1',i l l l
•
5g
Fig. 1. Representative examples of the response of lung weight to a step increase in vascular pressure (to 45 mm Hg) with two breathing patterns in the same animal. (A) Respiratory frequency is 10 min-'. (B) Respiratory frequency is doubled while tidal volume is kept the same. BP is blood pressure; LPA is left main pulmonary artery; lUl is left upper lobe; and Paw is airway pressure. Downward deflection in weight tracing signifies weight gain. Toward the end of the slow phase of weight gain, 5 g weight was placed on the lobe to check consistency of weight calibration. Horizontal level is shown by the thin line along the weight tracing. Note the faster rate of weight gain (thick segment) in B compared with A.
374
BSHOUTY AND YOUNES TABLE 1 IlW/l1t WITH DIFFERENT BREATHING PATIERNS*
P, = 35 mm Hg Dog
1 2 3 4 5 6 7 Mean ± SEM
P, = 45 mm Hg
C,
C,
C3
C.
C,
C,
C,
C,
C3
C.
C.
C,
0.56 0.35 0.42 0.39 0.49 0.38 0.52 0.44 0.03
0.73 0.40 0.50 0.49 0.56 0.63 0.72 0.58 0.05
0.67 0.35 0.46 0.47 0.48 0.48 0.40 0.47 0.04
0.52 0.33 0.33 0.23 0.33 0.42 0.42 0.37 0.04
1.09 0.36 0.75 0.83 0.75 0.63 0.72 0.73 0.08
0.94 0.53 0.58 0.58 0.48 0.54 0.80 0.64 0.06
0.67 0.43 0.58 0.52 0.67 0.58 0.79 0.61 0.04
0.94 0.77 0.75 0.63 0.89 0.73 0.96 0.81 0.05
0.86 0.62 0.63 0.54 0.97 0.65 0.71 0.71 0.06
0.78 0.51 0.42 0.46 0.83 0.60 0.67 0.61 0.06
1.25 0.52 1.33 0.83 1.25 0.71 1.20 1.01 0.12
1.04 0.56 1.00 0.68 0.92 0.65 1.45 0.90 0.12
* Individual and mean (:!: SEM) values of rate of edema formation (I1WI M. g/min) in seven dogs obtained at two vascular pressures (P, 35 mm Hg and P, 45 mm Hg) and under the six ventilation conditions described in table 2. Each value is the average of at least two observations.
=
=
TABLE 2 BREATHING PATIERNS AND OVERALL MEANS OF IlW/llt* Condition C, C, C3 C. C. C,
f (min-')
VT (mllkg)
EEP (mmHg)
10 20 20 20 5 5
4 4 2 2 8 8
2.5 2.5 2.5 4.0 2.5 1.0
Peak Paw (mm Hg)
11.3 12.9 6.1 8.7 26.2 19.9
± ± ± ± ± ±
0.48 0.58 0.46 0.62 0.97 1.05
Paw (mm Hg)
4.9 5.2 3.2 5.1 7.9 5.4
± ± ± ± ± ±
0.15 0.21 0.09:1= 0.11 0.15:1= 0.17
Il W/l1t (g/min)
0.52 0.69 0.59 0.49 0.87 0.77
± ± ± ± ± ±
0.03 0.05t 0.05 0.05 0.08:1= 0.07:1=
Definition of abbreviations: C, to C, = various conditions under which each dog was studied; f ~ respiratory frequency; VT tidal volume; EEP end-expiratory airway pressure; Paw airway pressure; Paw mean airway pressure; 11W/l1t rate of edema formation (average of data at two vascular pressures) . • For peak Paw, Paw, and I1W/l1t, values are means:!: SEM (n = 14). Statistical significance was determined by two-way repeated-measures analysis of variance and Tukey's test for muttiple comparisons compared with baseline (C,).
=
=
=
=
=
t p < 0.05.
*P < 0.Q1.
value used in figure lA but VT was the same. Note the faster rate of the slow phase of weight gain in figure 1B compared with figure lA. Individual as well as mean values for AWI At at both vascular pressures (P, = 35 and P, = 45 mm Hg) and during the six conditions are shown in table 1. The overall means and statistical differences are shown in table 2. Increasing VE by raising f (C z) caused a significant (p < 0.05) increase in AWl At to 0.69 ± 0.05 glmin at equivalent operating lung volumes (Paw during C, and C, was 4.9 ± 0.15 and 5.2 ± 0.12 mm Hg, respectively). When VE was returned to baseline by decreasing VT, AWl At decreased to levelsthat were not significantly different from baseline. This drop occurred whether operating lung volume was matched to baseline (C 4 , Paw = 5.1 ± 0.11 mm Hg, AWl At = 0.49 ± 0.05 g/min) or not (C3 , Paw = 3.2 ± 0.09 mm Hg, AWl At = 0.59 ± 0.01 g/min). Increasing VT, while keeping VE constant, was also associated with a significant increase in AWl At (C s , AWI At = 0.87
± 0.08 glmin as opposed to 0.52 ± 0.03 in C" p < 0.01). Adjusting Paw to match C, had little effect on the results; AWI At remained significantly elevated (C 6 , AWI At = 0.77 ± 0.07 glmin, p < 0.01). These findings wereconsistent when tested individually at either microvascular pressure (35 or 45 mm Hg). Mean AW/At (± SEM, standard error of the mean) while ventilating the dogs with a TIlTh of 1:1 was 0.49 ± 0.07 compared with 0.51 ± 0.09 when ventilated with a TIlTh of 1:6. These results werenot different statistically (p = 0.80). Discussion
In the present study, the rate of edema formation was inferred from observing AWI At over a brief period of high vascular pressure (min 1-4 after a step increase in pressure). The advantage of this approach is that it permits several observations to be made in the same animal; there is a limit to cumulative weight gain after which frothing occurs and weight becomes unstable. This, in turn, makes
it possible to assess the effect of multiple variables that can affect the response. Two potential disadvantages need to be considered, however. First, it may be argued that the increase in lobar weight between 1 and 4 min is in part or in total the result of vascular filling in response to a step increase in pressure. We believe that the contribution of vascular congestion to weight gain over this period is negligible. The reasons for this assessment werediscussed in considerable detail in an earlier communication in which this preparation was first used (4). (1) Vascular filling occurs both at low and high step increases in pressure, whereas slow weight gain was observed only when vascular pressure exceeded a critical value (P crit). (2) When vascular pressure is returned to baseline, following a 4-min period of weight gain, lobar weight drops to a level that is invariably higher than that before the step increase in pressure. The magnitude of increase in baseline weight agrees very wellwith the amount of slow weight gain beyond the first minute. (3) There is good agreement between the increase in wet weight of the blood-drained lobe at the end of an experiment and the cumulative increase in force transducer signal. Second, it could also be argued that a 4-min observation, as in our study, is not enough to achieve steady-state conditions in which tissue fluid hydrostatic pressure, protein oncotic pressures, and lymph flow are stable. Under these conditions the observed changes in AWI At would be rather transient and pertain only to brief elevations in vascular pressures. As indicated earlier (4) and in agreement with observations by others in in situ lobes (5) (see isolated ex vivo lobes [6]), the rate of weight gain at pressures above P erit does not change beyond the fourth minute. This was confirmed in independent measurements in six dogs in which elevated pressures were maintained for periods of about 1 h and AWI At in the interval from 1 to 4 min was compared with the rates at 12 to "15 and 40 to 43 min. AWl At in the three periods were (mean ± SEM) 0.54 ± 0.08, 0.53 ± 0.08, and 0.49 ± 0.10 gm/rrtin,respectively. There was no significant difference between the three periods by ANOVA for repeated measures (F = 0.8, p > 0.4). It follows that extending the time of observation beyond 4 min offers no advantage in this preparation while limiting the number of observations that can be made. The present study demonstrates that
375
BRE"THING PATTERN AND FLUID FILTRATION
increasing VE by raising f (Cs) is associated with a significantly higher AWl At. In a previous study (1) we showed that increasing VE by increasing VT promotes greater water accumulation in the lung. The combined results indicate that higher ventilation, achieved by increasing either VT or f, promotes greater edema formation. It is wellestablished that increased ventilation induced by physiologic stimuli (hypoxia [8], hypercapnia [9], and exercise [10]) is associated with an increase in lung lymph flow.The increase in lymph flow in these experiments may have been related to an increase in intrarnicrovascular pressure (due to changes in blood flow or in longitudinal or parallel distribution of resistances), to differences in operating (mean) lung volume, to an increase in filtration area (due to vascular recruitment, for example), or reflex effects produced by regional or systemic changes in gas tensions (Pao., Pacoz, and pH). The main advantage of the present study is that all these confounding variables were controlled. Our results therefore indicate that an increase in rate or depth of breathing exerts an independent influence that promotes greater fluid filtration. There are two studies in which the effect of controlled changes in ventilation, produced by deliberate changes in ventilator settings, on lung lymph flow, was assessed (11, 12). Patterson and colleagues (11) measured lymph flow (QL) in isolated sheep lungs at various respiratory frequencies while keeping VTconstant. QL increased linearly with f. In their study, however, it was not clear whether QL increased because of a primary effect of f on lymphatic transport, redistribution of interstitial fluid, or a primary effect on fluid filtration, with the increase in QL a secondary phenomenon. By showing a greater rate of fluid retention with high f, the present results indicate that f acts primarily on fluid filtration. In fact, the associated increase in QL tends to underestimate the effect of f on the rate of fluid filtration. Martin and coworkers (12) measured lung lymph flow (QL) in an artificially ventilated, open-chested (thoracotomy), normally perfused canine preparation. QL increased when hyperventilation (increased VT) was produced using a hypoxic gas mixture but not in the "control" experiments in which VT was similarly increased but the fraction of inspired oxygen (FIoz) was 0.3. The results of these "control" experiments represent, to our
knowledge, the only data for which an increase in ventilation was not associated with an increase in QL. The reason for this is not clear. Two factors may have mitigated the increase in QL. First, the normoxic hyperventilation was associated with a significant decrease in Pacoz (41.5 to 26.8 mm Hg) and a significant increase in pH (7.38 to 7.48). These changes, acting through local or systemic mechanisms, could have altered the longitudinal or parallel distribution of pulmonary vascular resistance (thereby altering capillary pressure) or could have affected the pumping action of the lymphatics. Second, because perfusion pressure was, on average, only 22 mm Hg, it is possible that some derecruitment may have occurred in the high VT condition during inspiration; neither alveolar nor airway pressure was reported. The rate of fluid filtration from the pulmonary microvasculature (Qv.C> is governed by the Starling equation (2); Qv.c
=
LpA[(P c - Pti.r) a(xc - Xli)]
-
(1)
where. Lp is the filtration constant per unit area (index of permeability), A is the membrane area available for filtration, P, is the hydrostatic pressure within the lumen, Pur is the hydrostatic pressure in the tissue fluid surrounding the vessel, a is the reflection coefficient for proteins, and Xc and Xti are the plasma and tissue fluid oncotic pressures, respectively. In our experiments P, was matched, and there was no reason for Xc to change with breathing pattern. Because Paw (and hence operating lung volume) were also matched (C, versus C 1) , there is no reason to suspect a change in filtration area (A).* Furthermore, since Xti should decrease (8, 10)in association with greater filtration rates (unless permeability is increased), changes in this factor cannot account for the observed increase in
* In our experiments both inflow and outflow pressures were higher than peak airway pressures (and hence peak alveolar pressures) under all conditions (see table 1). The entire lobe was therefore in Zone III perfusion under all conditions and there is no reason to expect a difference in the number of recruited vessels between different experimental conditions. To the extent that Paw equals Pal (I) and pleural pressure was zero. where PaW was matched (for example, C.. C,. C., and C.) Pal and mean transpulmonary pressure (and hence, lung volume) was also matched. We believe that if intravascular pressure, alveolar pressure (extramicrovascular pressure), and alveolar surface area are matched, microvascularsurface area is also likely to be similar.
filtration. It follows that ventilating the lung with a higher f, while keeping VT constant, either increases permeability or decreases Pu.r or a. Although operating lung volume and volume oscillations per breath were matched in C z and C 1 , the rate of change in lung volume during inspiration (inspiratory airflow) was higher in C z • It is possible that the rate at which the alveolar-capillary membrane is stretched may influence permeability or a. Results of the second group of dogs dismisses this possibility. Decreasing inspiratory time by increasing flow while keeping VT and f constant had no effect on Ii.WIli.t. Alternatively, cycling of lung volume may promote the clearance of perivascular fluid to areas remote from the filtration site through a pumping action. At a given operating lung volume, more pumping cyclesper minute (or greater cycling amplitude, as with large VT) may result in a lower mean Pur. Our results do not permit a distinction between this latter mechanism and a change in permeability or a. Although it is possible to calculate a Kf,c value from the data in table 1 (Ii.WI At at P 2 minus Ii.WIli.t at P 1 divided by 10), we believe that two pressure points are not sufficient to reliably determine slope values. The return of Ii.WIli.t to baseline when VE was returned to baseline by reducing VT, despite the persistence of high f (C J and C 4 ) , further supports the notion that the important variable is the total amount of volume cycling per minute rather than f per se, Although the associated increase in VE with larger tidal volumes must have contributed to the higher AWl At observed in our previous study (1), this is not the entire explanation. In the present study increasing VT continued to cause a highly significant increase in AWl li.t even though VE and inspiratory flow rate were not altered (C, and C6 ) . The possible mechanisms for VTeffects werediscussed in detail earlier (1). These will not be repeated here. We concluded then that this effect is likely related to nonlinearities in the relation between lung volume and one or more of the factors that affect fluid filtration. The present results do not necessitate any change in this conclusion. Our findings have some practical implications. First, ventilation and ventilatory pattern of experimental lobes should be considered when comparing edema formation under different experimental conditions and between different studies.
376
Second, the practice of using large inflation pressures in ventilating patients with pulmonary edema may require reassessment. Third, the increase in ventilation and tidal volume that must occur with physiologic challenges, such as exercise and hypoxia (for example at high altitude), may contribute in susceptible individuals to the development of pulmonary edema under these conditions. It must be pointed out, however, that changes in ventilation and breathing pattern, spontaneously produced by an intact organism or artificially induced in a ventilated patient, need not be associated with the same P, or mean lung volume. These changes could enhance or counteract the effect of VE and VT on edema formation. Finally, it seems that high-frequency ventilation (HFV), during which VT is very small, should be associated with lower rates of edema formation. Because of the extremely high dead space effect (Vn/VT) that is associated with HFV, it is necessary to increase total ventilation to one or more orders of magnitude higher than with conventional ventilation to maintain gas exhange (13). The two changes (high VE but low VT) have op-
BSHOUTY AND YOUNES
posite effects on fluid filtration. This may explain the small and inconsistent effects of HFV on lymph flow reported by different investigators (14-16). Acknowledgment The writers thank D. Lobchuk for her technical assistance. References 1. Bshouty Z, Ali J, YounesM. Effect of tidal volume and PEEP on rate of edema formation in insitu perfused canine lobes. J Appl Physiol 1988; 64(5):1900-7. 2. Taylor AE, Parker J'C, Pulmonary interstitial spaces and lymphatics. In: Fishman AP, ed. Handbook of physiology. The respiratory system. Circulation and non-respiratory functions. Bethesda, MD: Am Physiol Soc 1986; sec. 3. vol. I, chap. 4, 167-230. 3. Hornik LA, Bshouty Z, Light RB, Younes M. Effect of alveolar hypoxiaon pulmonary fluid filtration in in situ dog lungs. J Appl Physiol 1988; 65(1):46-52. 4. YounesM, Bshouty Z, Ali J. Longitudinal distribution of pulmonary vascularresistancewith very high pulmonary blood flow. J Appl Physiol1987; 62:344-58. 5. Drake RE, Smith JH, Gabel JC. Estimation of the filtration coefficient in intact dog lungs. Am J Physiol 1980; 238(7):H430-8. 6. Morriss AW, Drake RE, Gabel JC. Comparison of microvascularfiltration characteristicsin isolated and intact lungs. J Appl Physiol 1980;
48:438-43. 7. Neter J, Wasserman W.Applied linear statistical models. Homewood, IL: R. D. Irwin, Inc., 1974; 474-82. 8. Warren MF, Peterson DK, Drinker CK. The effects of heightened negative pressure in the chest, together with further experiments upon anoxia in increasing the flow of lung lymph. Am J Physiol 1942; 137:641-8. 9. Albelda SM, Hansen-F1aschen JH, Lanken PN, Fishman AP. Effects of increased ventilation on lung lymph flow in unanesthetized sheep. J Appl Physiol 1986; 60(6):2063-70. 10. Coates G, O'Brodovich H, Jeffries AL, and Gray GW. Effects of exercise on lung lymph flow in sheep and goats during normoxia and hypoxia. J Clin Invest 1984; 74:133-41. l l. Patterson GA, Mitzner WA,Sylvester JT. Assessment of fluid balance in isolated sheep lungs. J Appl Physiol 1985; 58:882-91. 12. Martin DJ, Grimbert FA, Baconnier P, Benchetrit G. Effect of acute hypoxia on lung transvascular filtration in anesthetized dogs. Bull Eur Physiopathol Respir 1983; 19:7-11. 13. Froese AB, Bryan AC, High frequency ventilation. Am Rev Respir Dis 1987; 135(6):1363-74). 14. Jefferies AL, Hamilton P, Bryan AC, O'Brodovich H. Effect of high frequency oscillation (HFO) on lung lymph flow. J Appl Physiol 1983; 55:1373-8. 15. Martin D, Rehder K, Parker JC, Taylor AE. High-frequencyventilation: lymph flow,lymph protein flux, and lung water. J Appl Physiol 1984; 57:240-5. 16. Raj JU, Goldberg RB, Bland RD. Vibratory ventilation decreases filtration of fluid in the lungs of newborn lambs. Circ Res 1983; 53:456-63.
