| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Departments of Physiology and Biophysics, Medicine, and Pediatrics, University of Washington, Seattle, Washington 98195-6522
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
ACKNOWLEDGEMENTS
FOOTNOTES
REFERENCES
ABSTRACT 
Mates, Elisabeth A., Jacob Hildebrandt, J. Craig Jackson,
Peter Tarczy-Hornoch, and Michael P. Hlastala. Shunt and ventilation-perfusion distribution during partial liquid ventilation in
healthy piglets. J. Appl. Physiol.
82(3): 933-942, 1997.
Replacing gas in the lung with
perfluorocarbon fluids (PFC) and periodically ventilating with a gas
[partial liquid ventilation (PLV)] has been shown to
improve oxygenation in models of respiratory distress syndrome. We
hypothesized that the addition of PFC to healthy lungs would result in
shunt, diffusion impairment, and increased ventilation-perfusion
(
A/
) heterogeneity.
Previously, Mates et al. showed that
O2 shunt and arterial-alveolar
CO2 difference increased linearly
with dose in piglets given graded intratracheal doses of PFC (10, 20, and 30 ml/kg followed by mechanical ventilation with 100%
O2) (E. A. Mates, J. C. Jackson, J. Hildebrandt, W. E. Truog, T. A. Standaert, and M. P. Hlastala. In: Oxygen Transport to Tissue
XVI, 1994, p. 427-435). Here we report
A/ distribution in
the same animals, showing a 50% increase in
A/ heterogeneity during PLV independent of PFC dose. Ventilation heterogeneity was the
major factor in this increase, and there was no significant change in
dead space ventilation. We also report on five animals given a single
20 ml/kg dose of PFC and followed for 3 h. They showed an increase in
shunt during PLV but no change in arterial-alveolar CO2 difference.
perfluorochemical liquids; multiple inert gas elimination
technique; gas exchange; ventilation-perfusion heterogeneity
INTRODUCTION LIQUID BREATHING, or the exchange of
O2 and
CO2 in liquid- rather than in
gas-filled lungs, has been investigated for treatment of respiratory
distress syndrome (RDS). Liquid ventilation (LV) with perfluorocarbon
fluids (PFC) has been shown to improve gas exchange and lung mechanics
in animal models of RDS as well as in human trials (2, 5, 14, 25).
Reduction of interfacial tension and opening regions of atelectasis in
the RDS lung may be two of the mechanisms responsible for improved
gas-exchange and increased compliance during LV. The implementation of
full tidal LV (FTLV; the tidal exchange of fresh oxygenated fluid with resident PFC) in hospital intensive care units (ICUs) has proven to be
a significant technical challenge. The long-term management of a
liquid-breathing system that continuously warms and oxygenates inspired
fluid, recycles CO2-laden expired
fluid, and controls liquid-filled lung volumes has been prohibitive
(21, 26). In 1991, Fuhrman et al. (4) described partial LV (PLV), a
hybrid approach in which conventional mechanical ventilators are used to deliver tidal breaths of gas to lungs filled with PFC. The benefits
of LV combined with the ease of implementation with standard ICU
ventilators facilitated the introduction of PLV to the clinical setting. Widespread interest in the clinical application of PLV has led
to a variety of studies exploring the effects of PLV on lung mechanics
and gas exchange in animal models of RDS (2, 13, 14, 30). Relatively
few investigators have addressed the physiology of gas exchange in
healthy animals during PLV (4, 6, 17).
The majority of LV studies in the literature offer modest
information about the changes that occur in gas exchange when PFC are
instilled in the lung (1, 2, 4-7, 12, 20, 27, 30, 35). Arterial
blood-gas values are often the only clues provided to interpret this
complex issue. Studies of gas exchange during FTLV offer some insight
into the physiology of gas exchange through a liquid medium; however,
these results cannot be extrapolated to PLV where a mixture of gas and
PFC resides in alveoli (12, 16, 19, 24). Fuhrman et al. (4) provided
the most comprehensive data on gas exchange in healthy animals,
reporting an alveolar-arterial O2
difference
[(A-a)DO2]
and arterial blood gases that indicated a mild increase in
O2 shunt during PLV. Hernan et al.
(6) looked at cardiovascular and respiratory changes in healthy piglets
during PLV with varying inspired PO2,
showing relative desaturation as the inspired
O2 fraction
(FIO2) drops <0.3. Again, in their study, gas-exchange analysis was limited to arterial blood
gases and O2 saturation.
Considering the fundamental changes that must occur in the transport of
gases in a liquid-filled lung, a quantitative evaluation of
gas-exchange efficiency during LV would be valuable.
Given adequate ventilation and inspired
PO2, the causes of gas-exchange
inefficiency are traditionally enumerated as shunt, diffusion
limitation, and ventilation-perfusion
( In this paper, we continue an investigation of gas-exchange
characteristics of the PFC-filled lung during PLV (17). The healthy
newborn piglet model was chosen to elucidate the basic physiology of
gas exchange through a liquid medium uncomplicated by the effects of
disease. In a previous paper, Mates et al. (17) documented
increased O2 shunt and
arterial-alveolar CO2 difference [(a-A)DCO2]
during PLV and showed these changes were linearly proportional to the
volume of PFC in the lung. We also attempted to measure
METHODS Animal preparation. Fourteen piglets
7-14 days old and of either sex were sedated with ketamine and
xylazine (24 and 2.75 mg/kg im, respectively). Anesthesia was
maintained with pentobarbital sodium (6 mg · kg
A Harvard single-piston animal ventilator was used to deliver tidal
breaths of 100% O2 to animals in
the supine position. Tidal volume and frequency were set to maintain
PaCO2 < 40 Torr during
gas ventilation at the start of an experiment and were unchanged
thereafter. Airway pressure was continuously monitored, and expiratory
flow and volume were periodically assessed with a pneumotachometer and
signal integrator, respectively. Positive end-expiratory pressure
(PEEP) was applied by immersion of the distal expiratory port to 5 cmH2O. Carotid and pulmonary
arterial (Pa) pressures were recorded, and CO was determined in
triplicate via thermodilution. All protocols were approved by the
University of Washington (Seattle) Animal Care and Use Committee.
PLV. Warmed, non-preoxygenated
perflubron
(C8F17Br;
LiquiVent, Alliance Pharmaceutical, San Diego, CA) was instilled
intratracheally via the side port of the endotracheal tube. Each dose
of PFC was trickled into the airway over 4-5 min during 100%
O2 breathing, delivering one-third
of the dose in each of the right and left lateral and supine positions
in an attempt to distribute evenly the fluids in the lung. Hourly doses
of 2 ml/kg were given to replace evaporative losses of PFC (see
APPENDIX Aa).
A/) mismatch. An
increase in any one of these factors may result in increased
arterial-alveolar differences in PO2
and PCO2. All contribute to venous
admixture, the addition of mixed venous blood to arterial blood,
lowering arterial PO2
(PaO2) and raising arterial
PCO2
(PaCO2). Shunt, or right-toleft blood flow that does not participate in gas exchange in the lung, results in a lower PaO2 for a given
inspired PO2. Shunt does
not typically affect PaCO2 unless the
shunt fraction is very high because
CO2 removal is adequately
controlled by increased ventilation. Increased
A/ heterogeneity is the
main cause of a decreased PaO2 in
disease. Its effect on PaCO2 is variable depending on the degree of ventilatory compensation
present. In the face of increased
A/
heterogeneity with little or no respiratory compensation,
PaCO2 will increase. Diffusion
limitation affects both gases, increasing
PaCO2, decreasing
PaO2, and widening the capillary-alveolar partial pressure differences. During PLV,
where fluid resides in alveoli as well as in small and possibly large airways, any or all of the above mechanisms may be responsible for
widened arterial-alveolar differences
[(a-A)D]. Understanding the
mechanisms responsible for decrements in gas-exchange efficiency during
PLV is important in determining the trade-off between improved lung
inflation and impaired gas exchange. It also provides insight into
gas-exchange physiology as patients are being weaned from PLV in the
ICU.
A/ heterogeneity with
the multiple inert gas elimination technique (MIGET) but found it was
limited due to the high solubility of sulfur hexafluoride
(SF6) in PFC. We concluded that
the presence of PFC in the lung slightly impairs
O2 and
CO2 exchange in healthy lungs, but
we were unable to delineate the mechanisms (i.e., diffusion limitation
vs. A/
heterogeneity vs. shunt) responsible for these changes. In this
publication, we reiterate the previous results (17), adding data for a
total of nine animals given graded doses of PFC. We revisit our
A/ analysis of PLV
using a modified form of MIGET with
SF6 eliminated, providing
information on A/
heterogeneity during PLV with varying doses of PFC in the lung. In
addition, we present a gas-exchange analysis of five control animals
given a single dose of PFC and followed over time.
1 · h1
iv, supplemented with 6-13 mg/kg hourly as needed) or
pentobarbital sodium and ketamine (3 and 5 mg · kg1 · h1
iv, respectively). Pancuronium bromide was administered in 0.2 mg/kg
intravenous doses as needed to prevent respiratory efforts, followed by
3-5 mg/kg iv of pentobarbital sodium to ensure a surgical plane of
anesthesia. Intravenous access was established via the left jugular
vein, and the right carotid artery was cannulated for blood-gas and
arterial pressure measurements (Fig. 1). A
pulmonary arterial thermodilution catheter was placed under fluoroscopy via the right jugular vein for measurement of cardiac output (CO), central venous and wedge pressures, and mixed venous blood gases. The
animals were tracheotomized with a metal Y-shaped cannula with a side
port for the administration of PFC.
Fig. 1.
Schematic of animal setup and instrumentation.
VT, tidal volume; f, breathing
frequency, FIO2, inspired
O2 fraction; PFC, perfluorocarbon
fluid; MIGET, multiple inert gas elimination technique; PEEP, positive
end-expiratory pressure.
[View Larger Version of this Image (41K GIF file)]
|
s is shunt flow (in l/min),
T is total
pulmonary blood flow (in l/min), CaO2 is
arterial O2 content,
O2
is mixed venous O2 content, and
Cc
O2 is
end-capillary O2 content, with the
assumption that end-capillary PO2 is
equivalent to alveolar PO2
(PAO2).
PAO2 during 100%
O2 breathing was estimated as
PB 
PH2O PCO2 PPFC, where
PB is barometric pressure,
PH2O
is water vapor pressure, and
PPFC is PFC vapor pressure.
"Alveolar" during PLV refers to the gas adjac- ent to the fluid
layer and does not necessarily represent gas tensions in resident
PFC. The PAO2 values of the
infused MIGET gases were considered negligible (see
APPENDIX Ab).
(A-a)DO2
was calculated as PAO minus measured PaO2.
(a-A)DCO2
was determined from PaCO2 and end-tidal
PCO2 values. Exhaled
CO2 was continuously monitored
with a Novametrix model 7000 infrared analyzer situated in-line between
the piglet and the solenoid valve (Fig. 1). The analyzer showed
excellent linear correlation (R = 0.998) with a medical mass spectrometer (Perkin-Elmer model 1100) with
respect to PCO2 in the presence of
both water vapor and perflubron vapor. Alveolar
PCO2 (PACO2) was taken as the
peak of the expired capnogram over 5-10 breaths before blood-gas
sampling. Typically, this was the value at end expiration due to the
usual positive slope of the expirogram. However, in some cases during
PLV, peak PACO2 occurred in
the first 40% of an exhaled breath, and the expirogram had a negative
slope, in which case the highest value of expired PCO2 was chosen. End-tidal expired
PCO2 is often used as an estimate of
"true" PACO2 (alveolar
gas in close proximity to PFC in these experiments) because it is
presumed to represent gas exhaled from distal regions of lung,
minimally diluted by dead space. During PLV, there are regional
alterations in lung mechanics due to the presence of a
low-surface-energy fluid, and it is reasonable to assume increased
heterogeneity of alveolar emptying times and of
PACO2 in the context of
variable distributions of PFC. We have previously discussed the
implications of this with regard to "measuring"
PACO2 at the mouth (18). We
have chosen to use peak expired PCO2 to repre- sent PACO2 for the
purposes of estimating
(a-A)DCO2, although one cannot truly represent it as a single number.
(a-A)DCO2 calculated this way underestimates the difference in alveoli with lower
PACO2 values and actually
represents the minimum (a-A)DCO2
that exists in PFC-filled lungs.
MIGET. This technique, subject to a
number of assumptions, allows one to distinguish shunt, dead space, and
the general pattern of
A/ distribution in
a lung (8, 33). It is based on the elimination of six inert gases of
varying blood solubilities (
; in ml solute · 100 ml
solvent1 · mmHg1)
and blood-to-gas partition coefficients (
= b/g,
where b is the blood solubility
and g is the gas solubility) as
they pass through the lung. Measured retention (R = Pa/
, where is
mixed venous pressure) and excretion (E = PE/,
where PE is expired pressure) of
each of the six gases is compared with a multicompartment mathematical
model that provides estimates of shunt, dead space, and
A/ heterogeneity.
The assumptions underlying this model are that steady-state conditions
exist with respect to gas exchange, that ventilation and perfusion to
gas-exchange compartments are constant, and that there is complete
diffusive equilibrium between end-capillary blood and alveolar gas.
In applying MIGET to PLV, reevaluation of these assumptions is
necessary in light of the presence of PFC in the alveolar space. In a
previous communication, Mates et al. (17) discussed the fact that SF6 is not in a steady
state with respect to gas flux under shunt conditions. Using a
two-compartment model describing inert gas exchange in a PFC-filled
alveolus with perfusion but no ventilation, Mates et al. calculated a
time constant of >3 h for SF6
equilibration between blood and PFC. The next longest equilibration
time was 18 min for ethane (17). In some additional unpublished
calculations with a three-compartment model of a
PFC-filled alveolus with ventilation as well as perfusion, we found
much shorter equilibration times for ventilated units, ranging from 7 s
for acetone to 133 s for SF6
(Mates, Hildebrandt, and Hlastala, unpublished observations). Our
conclusion from these calculations was that
SF6 is unsuitable for MIGET during
PLV due to lack of steady-state conditions, but the remaining five
gases satisfy this condition. MIGET results presented in this paper are
based on five gases, with the lowest solubility gas,
SF6, eliminated. This will affect
the results in two ways: MIGET shunt will be invalid because
SF6 retention is the most
sensitive measure of shunt. Second, overall sensitivity of the modeling
technique will be reduced by elimination of one of the six gases.
A/ predictions will
not be systematically biased by the absence of
SF6 but will be more affected by
noise in the other five data points. The overall effect will be to make
it more difficult to find statistically significant trends in
A/ heterogeneity.
Biological fluctuations in blood flow, ventilation, etc. are additional
sources of error in this technique because MIGET model solutions are
based on the assumption of continuous blood flow and ventilation.
Wagner (31) analyzed 400 duplicate pairs of inert gas
samples from multiple experimental conditions and reported excellent
reproducibility of MIGET dispersion indexes despite physiological
variation within and between experiments. During PLV, an additional
source of variation would be redistribution of the PFC within alveoli.
We do not feel that this is a large source of error within the time
frame of our measurements. Continuous PEEP was applied, which has the
effect of pushing the PFC-air interface into distal airways throughout
the respiratory cycle, reducing back-and-forth mixing of the fluid in
larger airways. Dependent drainage of PFC does occur over the course of
a 3- to 4-h experiment, but it occurs slowly and is not likely to
affect hourly measurements.
Diffusion equilibrium between blood and alveolar gas may be hampered by
the presence of PFC as we predict it might for
O2 and
CO2. This would have the effect of
increasing the inert gas (a-A)D
area that, as we discuss below, is the variable we have chosen to
predict A/
heterogeneity. In an as yet unpublished study, we examined
the potential role of diffusion limitation in creating
(a-A)D values for
O2,
CO2, and the inert gases (Mates, Hildebrandt, and Hlastala, unpublished observations). Briefly, this
model consists of a spherical gas-exchange unit the size of a terminal
sac with capillary perfusion in the outer shell, an evenly distributed
shell of PFC, and an inner sphere of alveolar gas. Inert and
respiratory gases were modeled as diffusing from the capillary bed to
alveolar gas along radial lines. Capillary-to-alveolar gas partial
pressure gradients were calculated for a variety of PFC thicknesses and
alveolar gas volumes. Inert gas partial pressure gradients were found
to be significantly lower than those of
O2 and
CO2 and <5% of the driving
pressure over the range of PFC thickness estimated for the evenly
distributed 30 ml/kg dose of PFC in ventilated gas-exchange units.
Inert gas (a-A)D area should not
be affected by diffusion disequilibrium with PFC present as long as gas
penetrates the unit (i.e., it is ventilated). If little or no gas is
present in a gas-exchange unit, R and E of the gases will approximate
shunt, as would be appropriate.
Graphic presentation of
A/ heterogeneity as
determined by MIGET has been expressed in a variety of ways. The most
recognized format as well as the most intuitively satisfying is the
presentation of continuous distributions of alveolar ventilation
(A) and blood flow () vs.
A/ as described by
Wagner et al. (33) (Fig.
2A).
This approach uses a 50-compartment model with mathematical smoothing
functions to generate a solution describing shunt, dead space, and a
distribution of A/
that may have up to three modes. One of the main criticisms of this
analysis is that the contours are not unique solutions and are often
overinterpreted (10, 11). Hlastala and Robertson (10)
described a simplified approach to analysis that computes the inert gas
(a-A)D area from a
four-compartment model fit to measured R and E data (Fig.
2B). The four compartments consist
of shunt, dead space, and two intermediate A/ compartments
that are chosen by the algorithm as a best fit to measured data points.
It is a simpler solution and avoids the pitfalls of the transformation
mathematics of Wagner et al. (33). Inert gas
(a-A)D area is the area between
the measured R and E curves and the predicted R and E curves from the
four-compartment homogeneous lung model (Fig.
2C).
(a-A)D area increases with
increased heterogeneity as the measured R and E values deviate from the ideal R and E values. The R and E components of the
(a-A)D area reflect perfusion
heterogeneity and ventilation heterogeneity, respectively. Hlastala and
Robertson (10) also found that the peak of the
(a-A)D area curve shifts left on
the solubility curve with predominantly low
A/, shifts right
along that axis with predominantly high
A/, and is centered
around a solubility of 1 in healthy lungs. We have chosen to use the
method of Hlastala and Robertson to analyze inert gas
data gathered during PLV. The four-compartment fit was felt to be a
more robust technique for measuring changes in heterogeneity in the
face of elimination of SF6. The
measurement of one fewer inert gases would only exacerbate the problem
of nonunique solutions in the technique of Wagner et al.
(33). We emphasize that the inert gas
(a-A)D area is not a measure of
diffusion limitation in the lung; rather it measures the goodness of
fit of a four-compartment homogeneous lung model to the experimental R
and E values.
A) and
blood flow () vs. ventilation-perfusion
(A/).
B: Raw (measured) retention (Rmeas) and excretion
(Emeas) curves (dotted lines)
and ideal homogeneous curves
(Rhomo and
Ehomo, respectively; solid lines) vs. solubility (lambda = solubility of blood/solubility of gas) from a
4-compartment model fit of Hlastala et al. (10). R = expired
pressure/mixed venous pressure
(PE/);
E = arterial pressure (Pa)/;
VD/VT,
dead space ventilation. Hatched and crosshatched areas, areas between
measured and ideal R and E curves showing origin of inert gas
arterial-alveolar difference
[(a-A)D] area. C:
(a-A)D area vs. lambda. Hatched
and crosshatched areas (correspond to areas in
B), R and E components corresponding
to blood flow and ventilation heterogeneity, respectively.
The theory and methodology of MIGET have been discussed at length elsewhere (8, 10, 31-33). The average solubilities of MIGET gases in blood and perflubron are listed in Table 1. Solubilities of the gases in pig blood and perflubron were measured with the double-extraction technique of Wagner et al. (32), and the average values of six trials are presented. There are no chemical reactions between the MIGET gases and PFC, and we did not measure a net loss of inert gases during PLV. Mass balance as determined by the ratio of expired inert gases to arteriovenous inert gas difference (
E · E/(1 R) · · ,
where E is expired
ventilation) was within 2% of 1 (perfect mass balance)
for 90% of all gases and all runs. Heated lines and gas-collection
chambers prevented loss of inert gases due to condensation of PFC or
water vapor.
|
||||||||||||||||||||||||||||||||||||||||
RESULTS
Of 21 animals studied, 4 died due to puncture of the right ventricle during pulmonary artery catheter placement, 1 was excluded due to improper placement of the pulmonary artery catheter, and 2 were excluded for incomplete protocols, leaving 14 animals from which data are reported for the 2 protocols. The mean weight of nine animals in the graded-dose protocol was 2.54 ± 0.6 kg and of five animals in the single-dose protocol was 3.05 ± 0.8 kg. Cardiovascular status was stable throughout all experiments, with no statistically significant changes in Pa or CO; however, there was a small decline in pulmonary arterial wedge pressure in the single-dose protocol (Table 2). Mean peak proximal airway pressure was 19.5 cmH2O during the graded-dose protocol and 21 cmH2O during the single-dose protocol and did not change from GV to PLV in either case. Average tidal volume and frequency were 17.1 ± 2.9 ml/kg and 20 ± 4 breaths/min, respectively, for the graded-dose protocol and 16.0 ± 2.3 ml/kg and 24 ± 4 breaths/min, respectively, the single-dose protocol. Average PEEP values were 4.6 and 5.2 cmH2O for the graded- and single-dose protocols, respectively.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gas exchange in piglets given graded doses of
PFC. Shunt and
(a-A)DCO2
data from eight of the nine animals in this group were previously
reported (17) and are repeated here for ease of comparison with the additional gas-exchange analysis presented here.
PaO2 decreased
significantly compared with baseline GV throughout PLV (P < 0.05) but was well maintained
at >300 Torr in all animals (Table 3).
PaO2 and mixed venous
PO2
(
) were unchanged from GV to
PLV.
(A-a)DO2
increased throughout PLV (Fig. 3), showing
a statistically significant change with PFC dose by RM-ANOVA
(P < 0.05). Drainage of PFC from the
lung to return to the 20 ml/kg dose produced an
(A-a)DO2
value insignificantly different from values obtained at the 20 ml/kg
dose earlier in experiments. Berggren shunt also showed a rise with PFC
dose (Fig. 3) and a statistically significant change with the
progressive addition of PFC by RM-ANOVA
(P < 0.05). As with
(A-a)DO2,
drainage of PFC from the lung resulted in
O2 shunt insignificantly different from that of the first 20 ml/kg dose.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
] and O2 shunt (
) vs. PFC
dose. Right arrows, progressive
addition of PFC; left arrows, drainage
of 10 ml/kg of PFC from lung. Values are means ± SE;
n = 9 piglets. Significant
rise in
(A-a)DO2
and shunt with dose by repeated-measures analysis of variance
(RM-ANOVA), P < 0.05.
PaCO2 and
rose
significantly during PLV (P < 0.05),
arterial pH fell with the onset of PLV
(P < 0.05), and
PaCO2 was unchanged. Despite the rise
with PLV, PaCO2 remained within
physiological limits (<45 Torr) throughout all experiments.
(a-A)DCO2
also demonstrated a statistically significant increase throughout PLV
(Fig. 4) and showed a significant dose response by RM-ANOVA (P < 0.05).
Drainage of 10 ml/kg of PFC after the 30 ml/kg dose resulted in an
a-ADCO2
not statistically different from that of the first 20 ml/kg dose.
MIGET inert gas (a-A)D area, a
unitless index of
A/
heterogeneity, increased roughly 50% during PLV (Fig.
5). There was a statistically significant
change with dose by RM-ANOVA (P < 0.05), but the changes do not appear to be linear with dose. There was
a significant rise in the E component of the inert gas
(a-A)D area throughout PLV
(P < 0.05) but no significant change
in the R component by RM-ANOVA (Fig. 5). There were no significant
differences in any MIGET parameters between the 20 ml/kg dose and
drainage of 10 ml/kg of PFC from the lung after the 30 ml/kg dose. The (a-A)D area vs. solubility curve
was symmetrical around a solubility of 1 for most experiments, with no
systematic shifts along the abscissa. Dead space ventilation
(VD/VT)
as determined by MIGET was 0.31 ± 0.06 during GV and was unchanged
throughout PLV by RM-ANOVA. The mean sum of squares for all MIGET
solutions reported (9 experiments with 5 runs each) was 0.5469 ± 0.5782. Of the 45 runs, 35 had a residual sum of squares < 1.0, 6 were <10.0, 2 were <20.0, and 2 were eliminated due to inert gas
sampling errors.
A/ heterogeneity) vs.
PFC dose. , Inert gas (a-A)D
area; ×, E component (comp) of inert gas
(a-A)D area; , R comp of
inert gas (a-A)D area.
Right arrows, progressive addition of
PFC; left arrows, drainage of 10 ml/kg of PFC from lung. Values are means ± SE;
n = 9 piglets. Inert gas
(a-A)D area increases
significantly over baseline during partial liquid ventilation (PLV) and
changes with dose (P < 0.05 by
RM-ANOVA) but not in a predictable manner. E component of
(a-A)D area increases significantly from baseline, but there are no statistically significant trends in R component.
Gas exchange in piglets given a single dose of
PFC. Similar to the graded-dose protocol,
PaO2 decreased significantly from baseline GV throughout PLV (Table 3).
was slightly but
insignificantly higher during PLV compared with GV. In contrast to the
graded-dose scheme, PaCO2 did not rise significantly over baseline during PLV.
PaCO2 was unchanged from pre- to
post-PLV and did not change with PLV or time.
(A-a)DO2 increased significantly with the addition of PFC as did
O2 shunt (P < 0.05), but there were no
systematic trends with time in either parameter (Fig.
6).
(a-A)DCO2
increased slightly but insignificantly from GV to PLV (Fig.
7) and did not vary over 3 h of PLV.
) and O2 shunt () vs. time
during PLV with 20 ml/kg of PFC in lungs. Time
0 corresponds to gas ventilation, and time > 0 corresponds to PLV. Values are means ± SE;
n = 5 piglets. Significant rise in
both parameters during PLV compared with gas ventilation,
P < 0.05.
DISCUSSION
The aim of this study was to determine the effect of a PFC-filled lung on gas-exchange efficiency in the healthy animal. In applying the standard tools of gas-exchange analysis to the study of PLV, it is important to carefully examine the assumptions under which these analytic tools were derived and determine whether they were satisfied in the case of PLV. It is also important to note that the large body of information about gas exchange during FTLV is not directly applicable to PLV due to fundamental differences in ventilation (i.e., fluid vs. gas).
Shunt during PLV. As discussed in our previous publication (17), Berggren shunt increased in all animals during PLV and did so in a dose-dependent manner. In this publication, we show (A-a)DO2 alongside O2 shunt (Figs. 3 and 6) because the (A-a)DO2 represents the fundamental disturbance that occurred due to the presence of PFC. O2 shunt is calculated from the Berggren equation, which assumes complete equilibration of O2 tension across alveolar membranes in ventilated areas. It is a measure of blood flow past unventilated regions, attributing all (A-a)DO2 to blood flow past collapsed air space or anatomically unventilated areas (i.e., bronchial circulation). During PLV, one must consider a second mechanism of increased (A-a)DO2 (and the derived parameter shunt): diffusion limitation in PFC that can potentially lead to a partial pressure difference between alveolar gas and capillary blood. The presence of liquid in alveoli presents a significant diffusion barrier to O2, with a five order of magnitude reduction in diffusion coefficients of O2 and CO2 in PFC (9, 29). There is probably an alveolar-to-arterial gradient of O2 due to diffusion limitation in ventilated alveoli, a situation different from the purely gas-filled lung. It must be kept in mind that what we call shunt in gas-filled gas-ventilated lungs must be thought of as a combination shunt and diffusion limitation in lungs partially filled with liquid fluorocarbon.
To be thorough, we must also examine the underlying assumption that an
FIO2 of 1.0 eliminates
A/ heterogeneity as a
cause of increased
(A-a)DO2
during PLV. West (34) explains that high
FIO2 abolishes
(A-a)DO2
due to A/ heterogeneity by greatly increasing driving pressure and eliminating diluent gas
N2. During PLV, inspired
O2 penetrates the lower airways, probably at the level of the terminal alveolar sac, mixing with resident gas and PFC. PAO2
during PLV approximates atmospheric pressure minus
PH2O,
PACO2, and
PPFC (10.5 Torr for perflubron), typically >600 Torr, which is more than enough to eliminate
(A-a)DO2 due to A/ heterogeneity
even in the presence of PFC. Recent work by Hernan et al. (6) in
healthy piglets illustrates the dependence of adequate oxygenation on
FIO2, noting a significant
desaturation at an FIO2
of 0.3. At this inspired PO2,
A/ heterogeneity
probably plays a large role in
(A-a)DO2.
In this study, we found that shunt increased in proportion to the volume of PFC in the lung. Our time-control trials confirmed the fact that the dose dependence of shunt seen here was not confounded by time-dependent changes during prolonged periods of PLV (Fig. 6). We hypothesize that the relationship between PFC volume and shunt is due to creation of shunt regions by the pooling of excess PFC and/or an increased diffusion limitation throughout the lung. Ongoing work in the area of mathematical modeling of gas diffusion in PFC-filled alveoli in our laboratory may help determine whether diffusion limitation is a significant contributor to (A-a)DO2 or a minor component. These data seem to suggest that lower volumes of fluid (i.e., less than the FRC doses used in recent clinical trials) provide a gas-exchange advantage. It must be kept in mind, however, that we are comparing PLV and GV in piglets with a healthy baseline. In subjects with lung injury, it has been shown many times that shunt improves dramatically from GV to PLV with an FRC dose of fluid in the lung (2, 13, 14). Injured lungs contain collapsed or edematous regions that would presumably open and participate in gas exchange with the addition of low-surface-energy liquid. Our data may apply to the situation in which patients whose lung function is improving are weaned from PLV. We speculate that the optimum PFC dose will fall as lung disease improves.
(a-A)DCO2 during PLV. In healthy lungs, (a-A)DCO2 is close to zero. As shown in our previous work (17) and here, (a-A)DCO2 increases >300% from GV to PLV in animals given graded doses of PFC. The mechanism(s) underlying these increases can be attributed toA/ heterogeneity and/or diffusion limitation. West (34) describes increased
(a-A)DCO2 in the gas-filled lung as due primarily to an increased alveolar dead
space that reduces PACO2.
However, in the case where ventilation is constant and respiratory
compensation is not possible (as with mechanical ventilation), changes
in
(a-A)DCO2
are due to worsened A/
heterogeneity. We did not measure
VD/VT
by the CO2 method in these studies
because the underlying assumption of complete arterial-alveolar
equilibration is probably invalid. VD/VT
determined by MIGET showed no change from GV to PLV. As discussed with
O2, diffusion limitation is also
an important mechanism of
(a-A)DCO2,
where it would not be considered otherwise.
In the graded-dose protocol, overall
A/ heterogeneity
increased 40-50% over baseline during PLV, independent of PFC
dose (Fig. 5). Some of the increase in
(a-A)DCO2
is attributable to worsening
A/ heterogeneity, but
it is impossible to calibrate the percent change in
(a-A)DCO2
due to this mechanism. As discussed in
A/
heterogeneity during PLV, increased
A/ heterogeneity was
global and not attributable to a relative increase in low or high
A/. As with
O2 shunt, we are unable to
determine the degree to which diffusion disequilibrium affects
(a-A)DCO2
with traditional techniques.
(a-A)DCO2
did not change significantly from GV to PLV in animals given a single
20 ml/kg dose of PFC. In addition, the absolute value of
(a-A)DCO2
was not comparable to the
(a-A)DCO2
measured in the graded-dose protocol at the 20 ml/kg dose. The
explanation for differences in this measure between protocols is
unclear. It may be that delivering a single large dose of PFC over a
short period of time allows for a more even distribution of the fluid and reduces A/
heterogeneity and thereby
(a-A)DCO2.
It is unfortunate that technical difficulties precluded accurate MIGET determination of A/
heterogeneity in these studies.
A/
heterogeneity during PLV.
A/ heterogeneity
increased during PLV in healthy piglets and the changes did not
correlate with PFC volume. Furthermore, all of the increase in inert
gas (a-A)D area was attributable
to increased ventilation heterogeneity as measured by the E component
of the (a-A)D area. There were no trends in the data suggesting predominance of high or low
A/ during
PLV. MIGET analysis did not show any change in
VD/VT
from GV to PLV at any dose supporting the conclusion that
VD/VT
is not the principle cause of measured increases in
(a-A)DCO2.
The potential mechanisms of
A/ heterogeneity during
PLV are numerous. The presence of a low-surface-energy fluid in alveoli alters local surface and interfacial tensions, alveolar-capillary configuration, and possibly local capillary blood flow (15). Ventilation distribution is altered as fluid in small airways and local
compliance changes affect resistance and compliance, resulting in
increased heterogeneity of ventilation time constants (18, 28).
Maneuvers such as increasing PEEP to push fluid menisci into distal
airways throughout ventilation may help decrease ventilation
heterogeneity by minimizing back-and-forth movement of the fluid.
Additional studies investigating the influence of respiratory rate,
PEEP, mean airway pressure, etc. on
A/
heterogeneity during PLV would be helpful. This study does not attempt
to characterize the mechanisms responsible for heterogeneity but rather
points out that in the healthy lung it is an important cause of an
increased (a-A)DCO2
and would cause a significant
(A-a)DO2
if the FIO2 was
significantly <1.
As discussed in METHODS, the validity of MIGET is limited
by its underlying assumptions, as are all modeling methodologies. The
assumption that there is a complete diffusive equilibrium between
end-capillary blood and alveolar gas is of particular concern during
PLV because there may exist a diffusion limitation in fluid-filled
alveoli. The effect of diffusion limitation would be to increase R (R = Pa/) and decrease E (E = gas
pressure/) of all inert gases. MIGET
results from these studies indicate a negligible role of diffusion
limitation in creating error. If inert gas elimination in the
liquid-filled lung were affected by diffusion limitation, one would
expect the MIGET to show increased shunt and to overestimate dead space
as the difference between the R and E curves widens. Shunt and dead
space were not consistently or disproportionately elevated in these
studies, decreasing our suspicion of error due to diffusion limitation.
In addition, the sum of squares values for all experiments indicated a
good fit between model predictions and measured inert gas R and E.
Summary. A moderate decrease in
gas-exchange efficiency was noted in healthy piglets during PLV and can
be explained by increased shunt and
A/ heterogeneity. The
effect of A/
heterogeneity on oxygenation is minimal if 100%
O2 breathing is used during PLV
but may play a role if lower
FIO2 values are used. Furthermore, there may be a component of shunt due to diffusion limitation in the liquid-filled alveolus that we are currently unable
to quantitate.
(a-A)DCO2
increased with graded doses of PFC but not significantly with a single
dose of PFC. The reason for this discrepancy is unclear, but an
increase in A/
heterogeneity correlates with the rise in
(a-A)DCO2
in the graded-dose protocol. There may also be a component of
(a-A)DCO2
due to diffusion limitation.
The changes in gas exchange described here were mild, with blood gases
well within the range of normal. It must be emphasized that ventilator
settings were not adjusted to optimize gas exchange on the transition
to PLV. This was done to compare GV and PLV with the same ventilatory
parameters. One would not expect gas exchange to improve on filling
healthy lungs with liquid. In fact, it is remarkable that blood-gas
tensions within the normal range were achieved with an FRC volume of
liquid in the lung. Adjustments to ventilator settings such as
increasing the mean airway pressure or increasing the ventilator rate
during PLV would most likely result in blood-gas values no different
from during GV. These observations in healthy animals have relevance to
the clinical management of patients with RDS who will, during recovery,
reach a phase when more PFC will result in worse rather than improved gas exchange.
ACKNOWLEDGEMENTS
Thanks to Wayne Lamm for thorough, precise, and comprehensive technical assistance and to David Frazer and Dr. T. A. Standaert for technical support. Drs. Alan Hodson and William E. Truog were of particular help in data interpretation.
FOOTNOTES
Address for reprint requests: M. P. Hlastala, Division of Pulmonary and Critical Care Medicine, Box 356522, Univ. of Washington, Seattle, WA 98195-6522.
Received 1 May 1996; accepted in final form 28 October 1996.
APPENDIX Aa
Estimation of evaporative losses of
PFC. To arrive at the rate of evaporative loss of PFC
from the lung, a rough estimate of PFC excretion is made, assuming that
body temperature is 37°C and the exhaled gas
(E; in ml/h) is saturated at
PPFC = 10.5 Torr. The volume
flow rate of PFC vapor exhaled
(PFC) is
|
|
PFC · 22,400 ml1 · mol1.
Assuming a ventilatory rate of 20 breaths/min, a tidal volume of 15 ml/kg, and a dead space of 3 ml/kg (volume that is not saturated with
PFC)
|
|
|
|
3
mol · kg1 · h1.
Finally, with the molar weight of the PFC = 499 g/mol and
the density of the PFC = 1.9 g/ml, this becomes
|
APPENDIX Ab
Alveolar partial pressure of inert
gases. The assumption that the alveolar partial
pressure of inert gases is negligible was verified by calculating the
partial pressure of acetone in the MIGET perfusate (5% dextrose in
water). Partial pressures in the perfusate will always be greater than
the partial pressures in blood, where the gases are diluted, and in
alveolar space, where inert gases are excreted. Acetone concentration
in the perfusate is 1 ml acetone/250 ml dextrose solution. The of
acetone in water is ~40 ml · 100 ml1 · mmHg1.
With Henry's law (Cx = x · Px),
where C is concentration and x is gas species, the partial
pressure of acetone in the perfusate is
|
|
REFERENCES
| 1. |
Clark, L. C.,
and
F. Gollan.
Survival of mammals breathing organic liquids equilibrated with oxygen at atmospheric pressure.
Science
152:
1755-1756,
1966.
|
| 2. |
Curtis, S. E.,
J. T. Peek,
and
D. R. Kelly.
Partial liquid breathing with perflubron improves arterial oxygenation in acute canine lung injury.
J. Appl. Physiol.
75:
2696-2702,
1993.
|
| 3. | Dill, D. B., H. T. Edwards, and W. V. Consolazio. Blood as a physicochemical system. XI. Man at rest. J. Biol. Chem. 635-648, 1937. |
| 4. | Fuhrman, B. P., P. R. Paczan, and M. DeFrancisis. Perfluorocarbon-associated gas exchange. Crit. Care Med. 19: 712-722, 1991. [Medline] |
| 5. | Greenspan, J. S., M. R. Wolfson, S. D. Rubenstein, and T. H. Shaffer. Liquid ventilation of human preterm neonates. J. Pediatr. 117: 106-111, 1990. [Medline] |
| 6. | Hernan, L. J., B. P. Fuhrman, M. C. Papo, D. M. Steinhorn, C. L. Leach, N. Salman, P. R. Paczan, and B. Kahn. Cardiorespiratory effects of perfluorocarbon-associated gas exchange at reduced oxygen concentrations. Crit. Care Med. 23: 553-559, 1995. [Medline] |
| 7. | Hirschl, R. B., B. Grover, M. McCracken, M. R. Wolfson, T. H. Shaffer, and R. H. Bartlett. Oxygen consumption and carbon dioxide production during liquid ventilation. J. Pediatr. Surg. 28: 513-518, 1993. [Medline] |
| 8. |
Hlastala, M. P.
Multiple inert gas elimination technique.
J. Appl. Physiol.
56:
1-7,
1984.
|
| 9. | Hlastala, M. P. Diffusion-capacity heterogeneity. In: Handbook of Physiology. The Respiratory System. Gas Exchange. Bethesda, MD: Am. Physiol. Soc., 1987. sect. 3, vol. IV, chapt. 12, p. 217-232. |
| 10. |
Hlastala, M. P.,
and
H. T. Robertson.
Inert gas elimination characteristics of the normal and abnormal lung.
J. Appl. Physiol.
44:
258-266,
1978.
|
| 11. | Jaliwala, S. A., R. E. Mates, and F. J. Klocke. An efficient optimization technique for recovering ventilation perfusion distributions from inert gas data. J. Clin. Invest. 55: 188-192, 1975. |
| 12. | Koen, P. A., M. R. Wolfson, and T. H. Shaffer. Fluorocarbon ventilation: maximal expiratory flows and CO2 elimination. Pediatr. Res. 24: 291-296, 1988. [Medline] |
| 13. | Leach, C. L., B. P. Fuhrman, F. C. Morin, and M. G. Rath. Perfluorocarbon-associated gas exchange (partial liquid ventilation) in respiratory distress syndrome: a prospective, randomized, controlled study. Crit. Care Med. 21: 1270-1278, 1993. [Medline] |
| 14. | Leach, C. L., B. Holm, F. C. Morin III, and B. P. Fuhrman. Partial liquid ventilation in premature lambs with respiratory distress syndrome: efficacy and compatability with exogenous surfactant. J. Pediatr. 126: 412-420, 1995. [Medline] |
| 15. |
Lowe, C. A.,
and
T. H. Shaffer.
Pulmonary vascular resistance in the fluorocarbon-filled lung.
J. Appl. Physiol.
60:
154-159,
1986.
|
| 16. |
Matalon, S. V.,
and
L. E. Farhi.
Ventilation-perfusion lines and gas exchange in liquid breathing: theory.
J. Appl. Physiol.
49:
262-269,
1980.
|
| 17. | Mates, E. A., J. C. Jack |
Torr
Torr
