1

Title: 1
Neonatal Total Liquid Ventilation: Is Low Frequency Forced Oscillation Technique 2
Suitable for Respiratory Mechanics Assessment? 3
4
Author list: 5
Dominick Boss (1) 6
Alexandre Beaulieu (2) 7
Olivier Avoine (1) 8
Philippe Micheau (2) 9
Jean-Paul Praud (1, 3) 10
Herv Walti (1, 3) 11
12
Affiliations 13
14
(1) Facult de Mdecine et des Sciences de la Sant de lUniversit de Sherbrooke, 15
Dpartement de Pdiatrie, Sherbrooke, Qubec, Canada. 16
17
(2) Facult de Gnie de lUniversit de Sherbrooke, Dpartement de Gnie Mcanique, 18
Sherbrooke, Qubec, Canada. 19
20
(3) Facult de Mdecine et des Sciences de la Sant de lUniversit de Sherbrooke, 21
Dpartement de Physiologie et de Biophysique, Sherbrooke, Qubec, Canada. 22
23
Running head: 24
Respiratory Mechanics in Total Liquid Ventilation 25
26
Contact information 27
28
Dr Herv Walti 29
Facult de Mdecine et des Sciences de la Sant 30
Dpartement de Pdiatrie 31
3001, 12e Avenue Nord 32
Sherbrooke, Qubec 33
J1H 5N4, Canada 34
35
Herve.Walti@USherbrooke.ca 36
ABSTRACT 37
Background. This study aimed to implement low-frequency forced oscillation technique 38
(LFFOT) in neonatal total liquid ventilation (TLV) and to provide the first insight into respiratory 39
impedance under this new modality of ventilation. Method. Thirteen newborn lambs weighing 2.5 40
0.4 kg (mean SD) were premedicated, intubated, anesthetized, and then placed under TLV 41
Articles in PresS. J Appl Physiol (June 10, 2010). doi:10.1152/japplphysiol.01042.2009
Copyright 2010 by the American Physiological Society.
2
using a specially-design liquid ventilator and a perfluorocarbon. The respiratory mechanics 42
measurements protocol was started immediately after TLV initiation. Three blocks of 43
measurements were first performed: one during initial respiratory system adaptation to TLV, 44
followed by two others series during steady state conditions. Lambs were then divided into two 45
groups prior to undergoing another three blocks of measurements: the first group received a 10- 46
min i.v. infusion of salbutamol (1.5 g/kg/min) after continuous infusion of methacholine (9 47
g/kg/min) while the second group of lambs was chest-strapped. Respiratory impedance was 48
measured using serial single-frequency tests at frequencies ranging between 0.05-2 Hz and then 49
fitted with a constant-phase model. 0.2 Hz harmonic test signals were also launched every ten 50
minutes throughout the measurement protocol. Results. Airway resistance and inertance were 51
starkly increased in TLV compared to gas ventilation with a resonant frequency 1.2 Hz. 0.2 Hz 52
resistance and reactance were sensitive to bronchoconstriction and dilation as well as during 53
compliance reduction. Conclusions. We report successful implementation of LFFOT to neonatal 54
total liquid ventilation and present the first insight into respiratory impedance under this new 55
modality of ventilation. We show that LFFOT is an effective tool to track respiratory mechanics 56
under TLV. 57
Key Words: Lung Function Test; Perfluorocarbons; Mechanical ventilation; Sheep; 58
Methacholine Chloride 59
INTRODUCTION 60
The advent of neonatal liquid assisted ventilation (LAV) has concurred with global efforts 61
to perfect strategies of ventilation improving the morbidity and mortality of several clinical 62
conditions such as meconium aspiration and infant respiratory distress syndrome, while 63
minimizing ventilator induced injuries. To date, two type of LAV have been used: partial liquid 64
3
ventilation (PLV) (12) and total liquid ventilation (TLV) (59, 60). The former consists in partly 65
filling the lungs with a liquid, usually a perfluorocarbon (PFC), while a conventional gas 66
ventilator allows gas exchanges. In contrast, during TLV, lungs are completely filled with PFC 67
and a dedicated liquid ventilator (39) ensures the oxygenation of the fluid and the renewal of a 68
tidal volume of liquid. TLV appears to be superior to VLP (49, 60) and offers many advantages 69
over conventional mechanical ventilation. Among others, TLV has anti-inflammatory (41, 48, 52, 70
53), recruits atelectatic zones of the lung (57), homogenizes ventilation (60) and pulmonary blood 71
flow distribution (23), and reduces inflation pressure by increasing lung compliance (60). Thus, 72
TLV improves blood oxygenation and reduces occurrence of baro-/volutrauma (15, 60). 73
Moreover, tidal volumes of liquid allow meconium, exudate, and mucus removal and filtering 74
(58) while ensuring adequate ventilation. 75
76
A total liquid ventilator must resemble other conventional ventilators except that it must 77
be able to conduct ventilation with a perfluorocarbon fluid (8). Various types of liquid ventilators 78
have been developed for conducting animal experiments (2, 21, 22, 28), however our fourth 79
prototype developed at the Universit de Sherbrooke for experimental research on animal models 80
of newborns (Inolivent-4) includes the latest up-to-date devices, findings and control algorithms. 81
In vivo experimental results with this prototype have demonstrated its efficiency in maintaining 82
adequate gas exchange, normal acid-base equilibrium and greater minute ventilation whilst 83
nearing flow limits (38). 84
85
As during conventional mechanical ventilation, knowledge of the respiratory mechanics 86
during TLV is of great interest since it affords valuable insight into the lung state as well as into 87
underlying disease pathophysiology. Similarly, knowledge of the respiratory mechanics helps 88
4
clinicians to optimize ventilation, follow treatment progression, plan timely weaning, and prevent 89
iatrogenic injuries (16, 29, 36). The low-frequency forced oscillation technique (LFFOT) (46) 90
appears particularly well suited for this purpose. Readers are referred to review articles for a 91
complete understanding of the technique (13, 25, 27). Briefly, LFFOT is a non-invasive 92
technique to measure respiratory impedance (Z
RS
), from which mechanical properties of the 93
overall respiratory system (resistance, elastance, and inertance) can be derived. Interestingly, this 94
technique also enables the use of parametric models to discriminate the contribution of both 95
tissues and airways to respiratory mechanics (14, 17, 34). Furthermore, it appears to be more 96
sensitive to subtle mechanical changes than other lung function testing (45) and has already been 97
used successfully in human infants (7, 30, 34, 46) and a preterm ovine model (32, 33), as well as 98
during bronchoprovocation challenges (5, 18). Finally, since it does not require active 99
collaboration of the patient, LFFOT can be easily performed during mechanical ventilation, 100
especially under general anesthesia and curarization (10, 26). 101
102
However, unlike conventional mechanical ventilation, there is so far only scarce 103
information relative to dynamic respiratory mechanics during TLV (1, 22, 57). In the present 104
study, we implement for the first time LFFOT to a total liquid ventilator and test whether this 105
technique enables reliable respiratory mechanics characterization during neonatal total liquid 106
ventilation. For this purpose, we used healthy newborn lambs under steady state ventilation 107
conditions and assessed LFFOT responsiveness to changes in respiratory mechanics using 108
methacholine (MCh) and salbutamol infusion as well as chest bandage. 109
GLOSSARY 110
CPM Constant-phase model 111
f Frequency (Hz) 112
5
f
res
Resonant frequency (Hz) 113
FRF
1
Frequency-response complex function (cmH
2
O / mL) 114
G Tissue damping (cmH
2
Os / mL at 1 rad/s) 115
GV Gas ventilation 116

XY
Cross-spectral density function estimate between X and Y 117
H Tissue elastance (cmH
2
Os / mL at 1 rad/s) 118
Im(Z
RS
) Imaginary part of respiratory system impedance (cmH
2
Os / mL) 119
I
RS
Respiratory system inertance (cmH
2
Os
2
/ mL) 120
j Imaginary unit (-1) 121
LFFOT Low frequency forced oscillation technique 122
MCh Methacholine 123
P Indicates pressure signal (subscripts) 124
P
aw
Airway pressure (cmH
2
O) 125
PEEP Positive end-expiratory pressure (cmH
2
O) 126
PEEP
ref
Reference positive end-expiratory pressure (cmH
2
O) 127
PFC Perfluorocarbon fluid 128
R
aw
Airway resistance (cmH
2
Os / mL) 129
Re(Z
RS
) Real part of respiratory system impedance (cmH
2
Os / mL) 130
RN random noise 131
R
RS
Respiratory system resistance (cmH
2
Os / mL) 132
SD Standard deviation 133
TLV Total liquid ventilation 134
V Indicates volume signal (subscripts) 135
V
pump
Volume measured at the piston pump (mL) 136
6
X
RS
Respiratory system reactance (cmH
2
Os / mL) 137
Z
RS
Respiratory system impedance (cmH
2
Os / mL) 138
Z
total
Total impedance (respiratory system + ventilator components) (cmH
2
Os / mL) 139
Fractional exponent 140

2
Coherence function 141
Tissue hysteresivity = G / H 142
Angular frequency (rad/s) 143
144
MATERIALS AND METHODS 145
Experimentation was performed in accordance with the Canadian Council on Animal 146
Care guidelines for the care and use of laboratory animals and was approved by our institutional 147
Ethics Committee for Animal Care and Experimentation. 148
##TABLE1## 149
Animal Preparation 150
Thirteen term and healthy newborn Romanov lambs (six females and seven males) 151
weighing on average 2.5 0.4 kg (SD) and < 5 days of age (see table 1 for lamb characteristics), 152
were premedicated with intramuscular injections of ketamine (10 mg/kg), atropine (0.1 mg/kg) 153
and midazolam (0.1 mg/kg) together with antibiotics (0.05 mg/kg duplocillin and 5 mg/kg 154
gentamicin). Lambs were secured in supine position on an open cot equipped with a warming 155
carpet and a radiant heater to maintain their central temperature at 39.5 0.5
o
C throughout the 156
experiment. Pressure-regulated volume-control ventilation was initiated (Servo 300 ventilator, 157
Siemens-Elema AB, Solna, Sweden) following orotracheal intubation performed using a 5.5 G 158
cuffed endotracheal tube (Mallinckrodt, St. Louis (MO) USA). Ventilation settings were adjusted 159
7
as follows: 55 breaths/min with an inspiratory:expiratory ratio = 1:2; tidal volume = 10 mL/kg; 160
positive end expiratory pressure (PEEP) = 4 cm H
2
O and fraction of inspired oxygen = 1.0. 161
Oxygen saturation was monitored using a pulse oximeter probe placed on the base of the tail 162
(Radical, Masimo, Irvine (CA), USA) and heart rate was recorded using a Hewlett-Packard 163
cardio respiratory monitor (model HP78342A, Palo Alto (CA), USA). Anesthesia was induced 164
with one intraperitoneal loading dose of thiopental (20 mg/kg) and followed by continuous 165
infusion of 2 mg/kg/h via the left jugular vein. Complete paralysis was achieved with i.v. 166
administration of vecuronium bromide (0.1 mg/kg) every two hours and continuous jugular 167
infusion of 5% dextrose was given at 4 mL/kg/h. The right jugular vein was cannulated in turn to 168
provide eventual i.v. access for methacholine and salbutamol delivery during the respiratory 169
mechanics measurement protocol. A 3Fr 7-cm catheter (PV2013L07, PiCCO catheter, Pulsion 170
Medical System, Munich, Germany) was positioned into the femoral artery using a cut-down 171
procedure for continuous monitoring of blood temperature and mean arterial, systolic, and 172
diastolic pressures. The catheter also provided access for arterial blood gas sampling. 173
174
Study design 175
Total liquid ventilation. Lambs were allowed 20 min for recovery and then gradually 176
shifted from conventional mechanical ventilation to a volume-controlled pressure-limited total 177
liquid ventilation using our specially-designed ventilator (38). Transition was made as quickly as 178
possible using 10 mL aliquots of warmed (39
o
C), preoxygenated perfluorodecalin (F2 Chemical, 179
Lancashire, UK) and by increasing gas-ventilator PEEP from 4 to 7 cm H
2
O (40). Total number 180
of PFC aliquots was adjusted to achieve lamb calculated functional residual capacity (25 mL/kg). 181
Gas ventilation was then interrupted and total liquid ventilation was intitiated as followed : 182
volume controlled mode, 5.6 breaths/min with an inspiration:expiration ratio = 1:3; end- 183
8
inspiratory and end-expiratory pause = 0.5 s; tidal volume = 28 mL/kg; PEEP
ref
= 5 cmH
2
O and 184
fraction of inspired oxygen at 0.95. Inspiration was volume-controlled (38) while expiration was 185
pressure-regulated and volume targeted; both were pressure limited and time cycled (39). 186
Additional interventions. Sodium bicarbonate or tromethamine was used to maintain pH > 187
7.25. Crystalloids (bolus of 10 mL/kg lactated Ringers solution) or vasopressor (dopamine 5 - 20 188
g/kgmin) were used as needed to maintain a mean arterial pressure 50 mmHg. The rate of i.v. 189
dextrose infusion was adjusted to maintain glucose blood level at 40-100 mg/dL. Once the 190
protocol completed, the lambs were euthanized with pentobarbital (60 mg/kg) and the lung and 191
thoracic cavity carefully inspected for evidence of perfluorothorax or gross abnormalities. 192
LFFOT equipment. Our total liquid ventilator prototype (38) was used to generate 193
volumetric harmonic oscillations and record pressure and volume signals at specified expiratory 194
or inspiratory pauses. Airway pressure was measured using a stainless steel capillary tube 195
inserted in the endotracheal tube (ET), its capped end being located approximately 1 cm before 196
ET distal end, thus reducing Venturi effect by measuring pressure in a constant section duct. Four 197
equidistant radial holes ensured static pressure measurements by minimizing dynamic pressure 198
effects. The capillary was connected to the pressure sensor (Model 1620, Measurement 199
Specialties inc, Hampton (VA), USA) with 30-cm long perfusion flexible tubing. The sensor was 200
calibrated in vitro using a 1 m high water column, has a precision of 1% and 1.2 kHz 201
bandwidth. Volumetric excitation signals were generated using the ventilator expiratory piston 202
pump while a linear position sensor (CS-250-AD, MTS Sensors, Cary (NC), USA) measured 203
piston displacement, and hence the volume of PFC in the pump, with a precision of 0.15 mL 204
and a bandwidth over 2 kHz. Flow spectra were obtained from Fourier analysis of the position 205
sensor signals as detailed below in the signal-processing section. However, the flexible ventilator 206
tubing adds a resonant behavior to the system and the pressure sensor tubing adds a time delay, 207
9
which must be measured and compensated for in order to retrieve the respiratory impedance Z
RS
208
from the total impedance of the experimental system Z
total
. 209
Respiratory mechanics measurement protocol. Respiratory mechanics measurements were 210
divided in six blocks (B) (figure 1). While the first block was performed immediately after onset 211
of TLV in order to monitor initial adaptation of the respiratory system to TLV, the next two 212
blocks were performed to characterize TLV steady state. These first series of measurements were 213
conducted similarly on every lamb whereas the last three blocks where used to assess LFFOT 214
sensibility to induced changes in respiratory mechanics (detailed below). 215
##FIGURE1## 216
All experimental blocks were started and ended with one random noise signal test for 217
respiratory mechanics assessment between 2 and 4 Hz. Single-frequency signals (9) were 218
launched after the first random noise (24) and over a narrow spectrum of 0.05 to 2 Hz inclusively, 219
beginning with oscillations at 2 Hz and decreasing up to 0.05 Hz. Tests were interspaced every 220
0.2 Hz between 2 and 1 Hz and every 0.1 Hz between 1 and 0.1 Hz with a final test at 0.05 Hz in 221
order to closely characterize low frequency respiratory system impedance. All oscillations were 222
performed at fixed flow amplitude of 7.5 mL/s (i.e., a volume from 11.9 mL at 0.1 Hz down to 223
0.6 mL at 2Hz), except for the 0.05 Hz sine wave, in which amplitude was set at 5 mL/s (15.9 mL 224
volume) in order to avoid airway collapse and other nonlinearities associated with high 225
volumetric oscillation amplitude. 226
Forced oscillations were performed during a 30 s apnea except for the 0.05 Hz sinusoid, 227
which necessitated a 45 s apnea to lower the influence of the transients on the spectra. All signals 228
were launched in pairs and at two different lung volumes: the first at end-expiration, the other at 229
end-inspiration. Lambs were allowed a minimum of 30 s between each pair of tests in order to 230
maintain normal blood gases. Consequently, each block lasted between 60 and 90 min. 0.2 Hz 231
10
tests were repeated every 10-12 min to assess LFFOT and short time variability of respiratory 232
system mechanics, independently of frequency. Arterial blood gas analyses were performed after 233
each 10-min break between the five first blocks of measurements for ventilation recovery. 234
Lambs were then divided in two groups to assess LFFOT sensitivity to induced 235
respiratory mechanics alterations. Sensitivity to compliance reduction was assessed in 5 lambs by 236
gently wrapping the chest using a 10-cm wide elastic bandage (Elastoplast, BSN medical Ltd, 237
Brierfield, UK). LFFOT performance in tracking resistance changes was assessed in 5 other 238
lambs in two steps. Two blocks of measurement (B4 & B5) were first performed during 239
methacholine (MCh)-induced rise of respiratory resistance. Continuous infusion of MCh was 240
started at 6 g/kg/min while preventing MCh-induced decline of mean systemic arterial pressure 241
(54) below 50 mmHg. Dosage was then raised to 9 g/kg/min after hemodynamic stabilization. 242
Respiratory mechanics measurements began 10 min after ventilator peak pressure reached 243
plateau. Then, a last block of measurements (B6) was carried out after 10-min i.v. infusion of 244
salbutamol (1.5 g/kg/min) to assess whether bronchodilation can be tracked by LFFOT. 245
Signal processing. The signal processing methodology followed a typical Welchs 246
overlapped segmented average frequency analysis procedure (6, 56). Pressure and volume signals 247
were digitized (PCI-DAS1602/16, Measurement Computing, Norton (MA), USA (16 bits 248
resolution, 2 kHz sampling rate)), then digitally low-pass filtered (6
th
order Butterworth, -3dB 249
cutoff at 10 Hz) and downsampled at 50 Hz for recording. Each recorded test was then divided 250
into overlapping segments (20 s for single-frequency signal, 10 s for noise, 67% overlap) and 251
Hanning windowed. The frequencyresponse function estimate FRF
1
(f), i.e. the transfer function 252
between pump volume and airway pressure, was computed using 253
11
( )
( )
( )
1

=
VP
VV
G f
H f
G f
(1) 254
where
VP
( f ) is the averaged cross-spectral density function between airway pressure P
aw
and 255
pump volume V
pump
and
VV
( f ) is the averaged autospectral density function of V
pump
. The total 256
impedance of the respiratory system and the ventilator tubing system, Z
total
( f ), is then given by

257
( )
( )
( )
( )
1
1
2
aw
total
pump
P f
Z f H f
V f j f
= =

(2) 258
with j
2
= -1.. 259
Finally, coherence function (
2
) was computed from the spectral analysis of pressure and flow 260
signals. 261
( )
( )
( ) ( )
2

=
VP
PP VV
G f
f
G f G f
(3) 262
A value of
2

0.95 was used as the criterion for test acceptance (47). This overall process is 263
very similar to typical conventional mechanical ventilation forced oscillation technique analysis 264
(35). 265
266
Parametric identification. Similar blocks of measurements (see figure 1) were ensemble 267
averaged when possible and a seven-parameter inverse model was fitted to the average spectrum 268
to improve fitting performance. Hence, prechallenge blocks were grouped (B2 & B3) as well as 269
blocks performed during MCh infusion (B4 & B5) and on chest-strapped lambs (B4, B5, & B6). 270
The complete system inverse model (equation 4) for which parameters needed to be identified 271
consisted of the four-parameter constant-phase model (CPM),Z
RS
(j) (equation 5) (20), and a 272
three-parameter linear transfer function representing ventilator flexible tubing dynamics, 273
Z
tubing
(j) (equation 6). 274
12
( ) ( ) ( ) =
total RS tubing
Z j Z j Z j (4) 275
where the respiratory system impedance is 276
( )

= + +
RS aw RS
G jH
Z j R j I (5) 277
and the flexible tubing impedance is 278
( )
( )
2
2
2
2

=
+ +
d j n
tubing
n n
Z j e
j j
. (6) 279
In the Z
RS
(j) model, is the angular frequency, R
aw
is the airway resistance, I
RS
the respiratory 280
system inertance, G the tissue damping, H the tissue elastance and = (2/)arctan(G/H) is a 281
fractional exponent. The flexible tubing impedance was represented by an underdamped 282
oscillation of natural pulsation
n
and damping factor (37), and a time delay d associated with 283
pressure wave propagation in all flexible tubing (55). While these parameters gave no 284
physiological information, their estimation was necessary for ventilator dynamics compensation 285
and respiratory impedance Z
RS
(j) estimation (figure 2). 286
287
The seven parameters (i.e. R
aw
, I
RS
, G, H,
n
, and d) were identified by a nonlinear weighted 288
least squares error minimization routine (44), the Matlabs lsqnonlin function (The MathWorks, 289
Natick (MA), USA), using experimental total impedance from equation 2. After a first seven- 290
parameter curve-fit, the ventilator-associated parameters were set constant at the estimated value. 291
The routine was launched a second time over a limited frequency range (0.05-1.75 Hz) to gain 292
more precision on the four CPM parameters. This helped reduce computed parameter confidence 293
intervals, yielding more precise physiological parameter values. 294
295
13
The real part of the respiratory impedance Re(Z
RS
) is also called the respiratory system resistance 296
(R
RS
), and comprises airway resistance and tissue damping parameters. Respiratory system 297
inertance and tissue elastance form respiratory system reactance (X
RS
) which is related to the 298
imaginary part of the impedance Im(Z
RS
) by Im(Z
RS
) = jX
RS.
.

Graphically, G refers to the 299
frequency-dependent pattern on R
RS
curve and R
aw
to its mainly frequency-independent part. H 300
mainly influences X
RS
at low frequencies, whereas I
RS
determines its shape for higher f. The 301
discrete frequency that graphically discriminates H and I
RS
contributions is where X
RS
= 0, and is 302
called the resonant frequency (f
res
) (figure 2b). Hysteresivity was calculated as the ratio between 303
tissue damping and elastance: = G / H (11). 304
##FIGURE2## 305
Statistical Methods. Changes in respiratory mechanics parameter following adaptation, 306
MCh and salbutamol infusion, chest-strapping, and between respiratory volumes were computed 307
using Wilcoxon signed-rank tests (SPSS software, v16.0). All data are presented as median and 308
interquartile range (25
th
- 75
th
percentile). Statistical significance was assumed at p 0.05. 309
310
311
RESULTS 312
LFFOT implementation in TLV. LFFOT implementation to our total liquid ventilator 313
Inolivent-4 was straightforward and allowed easy measurement of Z
RS
in healthy newborn lambs. 314
Ventilator piston pump forced oscillations enabled reliable assessment of total system mechanics 315
using single-frequency signal as well as random noise. Respiratory system impedance was 316
measured at very low frequency (f 2 Hz) using single-frequency testing while impedance of 317
ventilator flexible tubing was dominant over 2 Hz and measured by random noise. For the sake of 318
14
simplicity, the reader is reffered to table 2 for results concerning 0.2 Hz single-frequency 319
resistance and reactance (RRS
0.2
and XRS
0.2
) and to table 3 for CPM-derived parameter estimations 320
(R
aw
, I
RS
, G, H) throughout the text. 321
Measurements, performed during several consecutive apneas yielded moderate 322
hypercarbia but no significant hypoxia (see figure 3). Arterial oxygen saturation stayed close to 323
100% but during the methacholine, where saturation decreased down to about 90%. Data are not 324
available after salbutamol infusion since blood gas analyses have been performed following each 325
block of measurements, except after that last one since lambs were euthanized. 326
##FIGURE 3## 327
Variability. Mean short-term intra-individual variability that mirrored both the variation in 328
the respiratory resistance over time and the variability of LFFOT was computed using RRS
0.2
. 329
Interestingly, the coefficient of variation was always lower at end-inspiration than at end- 330
expiration, such that steady state values obtained during baseline for the eleven lambs were 12.4 331
and 18.5% respectively (table 2). The variability was also higher immediately after TLV 332
initiation (21%) and lower after salbutamol-induced bronchodilation (7%), when the resistive part 333
of impedance was lower. Intra-individual variability was otherwise maintained within 10 15 % 334
at end-inspiration and within 16 23 % at end-expiration. 335
##TABLE2## 336
Constant-phase model. During baseline conditions (figure 3), the frequency-independent 337
part of the R
RS
curve, associated with airways resistance, was within 0.3 and 1.2 cmH
2
Os/mL and 338
preceded by a short frequency-dependent segment related to tissue damping. Note that the pattern 339
of the R
RS
curves was very similar amongst lambs. Conversely, the dispersion of the X
RS
curve 340
raised towards higher frequencies, suggesting greater variability of the inertance term. 341
15
Furthermore, there was a significant inverse Pearson correlation between the lamb weight and f
res
, 342
with end-inspiratory r = -0.634 (p = 0.018) and end-expiratory r = -0.541 (p = 0.043). I
RS
also 343
correlated with weight at both respiratory volumes: (r = 0.627, p = 0.019) and (r = 0.619, p = 344
0.021) at end-inspiration and end-expiration respectively. Correlation between R
RS
and weight did 345
not otherwise reach statistical significance. 346
##FIGURE4## 347
Continuous intravenous infusion of methacholine. An example of impedance spectra 348
measured pre- and post-methacholine challenge is presented in figure 5a and 5b. Plots show 349
vertical shift of R
RS
at both respiratory volumes, consistent with an increase in airway resistance. 350
Likewise, RRS
0.2
was statistically higher during MCh infusion (table 2), although absolute values 351
of CPM-derived resistive parameters (R
aw
and G) failed to show a significant increase (table 3). 352
Finally, XRS
0.2
was significantly decreased at end-expiration. Note that conversely to the R
RS
353
curve, graphical modulations of X
RS
were quite variable among lambs and therefore no 354
conclusion could therefore be drawn: if elastance appeared right shifted (at frequencies close to 355
zero) owing to MCh infusion, steepness of the curve of reactance however was not consistent 356
among lambs. 357
##FIGURE5## 358
Intravenous infusion of salbutamol. Again, figure 5a shows a clear downward shift of the 359
X
RS
curve after salbutamol infusion as compared to the resistance plateau during methacholine 360
infusion and before challenge. This difference was however substantially reduced at end- 361
expiration (figure 4b) compared to end-inspiration. Similarly, both RRS
0.2
and XRS
0.2
decreased 362
significantly at end-inspiration but not at end-expiration volume, a finding that concurs with the 363
plots. CPM parameters failed to report a significant decrease in R
aw
and G compared to baseline. 364
##FIGURE6## 365
16
Chest-strapping. Respiratory impedance of lambs with a bandage strapped around theirs 366
chest shows graphically revealed an increase in frequency-dependence of R
RS
at low frequencies 367
(figure 6a and b) and a right shift of the X
RS
curve, consistent with a higher elastance (lower 368
compliance). The later was more convincing at end-expiration (figure 6b) than at end-inspiration. 369
XRS
0.2
was significantly lower at end-expiration as compared to baseline, but no change in RRS
0.2
370
was found at either volumes. CPM tissue parameters G and H were significantly higher than that 371
observed during baseline at end-expiration volume, while only H was significantly increased at 372
end-inspiration. RRS
0.2
and XRS
0.2
, as G and H, were however all statistically different at end- 373
inspiration as compared to end-expiration. 374
##TABLE3## 375
376
377
DISCUSSION 378
Clinical and physiological implications. We report herein the first implementation of 379
LFFOT to total liquid ventilation as well as provide the first insight into respiratory impedance of 380
newborn lambs under this new modality of ventilation. We present spectral impedance within a 381
very low range of frequencies (0.05 - 2 Hz inclusively) and show that a single-frequency signal of 382
0.2 Hz can be a simple and reliable tool to track pulmonary mechanics changes in TLV. 383
Likewise, since TLV prevents the lung from collapsing below functional residual 384
capacity, several consecutive apneas can be induced without marked hypoxemia. Figure 3 shows 385
no significant hypoxemia and moderate hypercarbia throughout the experiment. Blood gas data 386
deteriorated following MCh infusion mainly because of bronchoconstriction. However, 387
theoretical effects of these variations in PaCO
2
and PaO
2
on respiratory mechanics were not 388
investigated in the present study. 389
17
390
Methodological considerations. In spite of providing a good fit on our experimental data, 391
the resistive parameters G and R
aw
were found dependent from each other during the parametric 392
identification process (data not shown). This issue has also been reported by previous 393
investigators (19, 51, 61) on conventional gas ventilation and is therefore unlikely to be 394
attributable to PFC fluid in lungs but to the mathematical structure of Z
RS
. Nevertheless, the 395
dependence between both parameters decreases the sensitivity of the CPM, and excludes reliable 396
partitioning of tissue and airway resistance. This phenomenon may also explain why no 397
significant changes in R
aw
were found during MCh and salbutamol challenge, albeit R
RS
398
modulations were graphically appreciable and noticeable by 0.2 Hz single-frequency signal tests. 399
400
Respiratory mechanics in TLV. Respiratory mechanics of newborn lambs under TLV 401
contrast with that found by other authors working with lambs under gas ventilation (31), mainly 402
by I
RS
predominance of the reactance spectrum under 2 Hz and a roughly 30 to 40-fold increase 403
in R
aw
(ref #39:ASAIO2009). Likewise, we found that f
res
was predominantly below 1.2 Hz with 404
PFC fluid into the lung (see figure 3), comparatively to about 10 Hz with air (31). This left shift 405
of the reactance as well as the increase in I
RS
and R
aw
where to be expected since 39 C 406
perfluorodecalin density (1880 kg/m
3
) and dynamic viscosity (3.4 mPas) are starkly greater than 407
air at the same temperature. These observations also concur with results obtained at low 408
pulmonary volume in partial liquid ventilation (when PFC/air ratio is maximal in the lungs) (43). 409
However, compared this latter type of liquid assisted ventilation, respiratory mechanics is much 410
less dependent of lung volumes in TLV and the clinically relevant range of frequency is more 411
confined to very low frequencies (f 2 Hz instead of f 4 Hz in partial liquid ventilation). 412
18
CPM spectra, as displayed on figure 4, showed some variability amongst lambs. Of note, 413
part of this variability is inherent to the neonatal animal model itself given that respiratory 414
mechanics varies with height, lung maturity, and alveolarisation of each newborn lamb (27, 34). 415
On the other hand, the contribution of the method, especially the deconvolution process used for 416
separating ventilator tubing dynamics from the respiratory system parameters could not be 417
overlooked as being a contributor to the observed variability. 418
419
Adequacy of the constant-phase model to TLV and interpretation. Adequacy of the CPM 420
to explain TLV lung impedance spectra was one of the research hypotheses. While the model 421
could be effectively fitted to TLV experimental data, questions arise regarding its interpretation. 422
While in gas ventilation R
aw
and I
aw
refer to the airways mechanics and G and H to the tissue, 423
these assumptions could be misleading in TLV. For instance, a high density PFC flow in the 424
airways could yield non negligible fluid-structure interactions such as elastic deformations and 425
associated structural damping. Therefore, if the general behavior of the respiratory impedance is 426
similar, further evidence is needed to give the gas-CPM parameters the same meaning in TLV. 427
428
According to the gas constant-phase theory, tissue damping and elastance should be 429
elementary coupled and therefore, changes in G value should always be followed by H, such as 430
hysteresivity = G / H remains constant (11). However, we observed relatively high variability 431
in parameter among lambs within the same experimental block as well as from one block to 432
another (see table 3). This indicates that G could possibly account for other phenomena than pure 433
tissue viscance and therefore its value should be interpreted with precaution. In gas ventilation, 434
the artefactual rise in G has been extensively ascribed to airway inhomogeneities, a phenomenon 435
19
well described by Lutchen and coworkers (4) and particularly apparent during inhalated MCh 436
challenge, when inhomogeneities are probably greater (42, 50, 53). On the other hand, 437
explanation is unlikely to be true in TLV since airway recruitment and uniform distribution of 438
PFC into the respiratory system enhance ventilation homogeneity and uphold alveolar patency 439
even at end-expiration volume (15, 59). 440
Herein, figure 3 shows that the real part of the impedance response, which refers to 441
respiratory system resistance, is only frequency dependent at very low frequencies (typically 0.05 442
and 0.1 Hz) and constant at upper frequencies. Such distribution of the measured impedance real 443
part implies two possible biases on the value of G: i) parameter G in the CPM equation (equation 444
4) is the gain of the frequency-dependence on the real part (inverse of the frequency). However, a 445
good fit is not a guarantee for identification of parameter G of the CPM: if an increase in airway 446
resistances for these low frequency experiments is indeed generated, it will be identified by 447
parameter G. Indeed, it was observed that a too large volume amplitude at low frequencies could 448
generate a nonlinear response of the airways (and eventually collapses, which is equivalent to an 449
increase in airway resistance). Actually, few collapses were observed for the high volume 0.05 450
Hz tests, but the results were always rejected based on the resulting low coherence values. As 451
suggested in (3), airway collapses in TLV induce neither structural changes in the respiratory 452
system nor any noticeable airway injury, so we can expect that subsequent measurements were 453
unaffected by these events. In spite of the fact that heavily nonlinear responses were discarded 454
from our analysis, an undetected nonlinear behavior of the airways could explain the increase in 455
resistance, and consequently create a bias for the value of G. A recommendation could be to 456
perform measurements at the lowest amplitudes in order to ensure a linear response of the 457
airways. ii) The identification of the parameter G, which characterizes the frequency-dependence, 458
is strongly sensitive of the few very low frequency measurements: a potential noisy measurement 459
20
on one recording would strongly affect the identified parameter G (and indirectly the estimated 460
parameter R
aw
). This could explain the large dispersion on the value of G from one experimental 461
block to the other. A recommendation could be to perform more measurements at very low 462
frequencies in order to obtain additional data showing the frequency-dependence of the real part 463
in TLV. 464
465
Single-frequency vs. spectral analysis. Authors are aware that spectral analysis would 466
have required multiple-frequency signal tests (24) which afford a more straightforward approach 467
to track respiratory mechanics and reduce temporal bias. However, as a first step in TLV, we 468
reasoned that is was better to use single-frequency signals at f 2 Hz, in order to closely 469
characterize respiratory mechanics at very low frequencies with high signal-to-noise ratio (SNR 470
20 dB) by avoiding flexible tubing resonance excitation. On the other hand, fixed-frequency tests 471
are easy to implement and fairly understandable by clinicians. RRS
0.2
and XRS
0.2
seem particularly 472
suitable in newborn lambs under TLV since they provide a quick overview of respiratory 473
mechanics and show good responsiveness to bronchoconstriction and bronchodilation challenges, 474
as well as to compliance reduction. In fact, the reduction in compliance using bandage wrapping 475
around the lambs chest induced a significant lowering of XRS
0.2
whereas MCh 476
bronchoconstriction induced a significant increase in RRS
0.2
, that is at both pulmonary volumes. A 477
higher frequency (e.g. 1.0 or 1.5 Hz) reflecting only the frequency-independent resistance 478
(airway resistance) could also be used, for instance, to track airway or endotracheal tube 479
obstruction. 480
481
CONCLUSION 482
21
In summary, we have successfully implemented the low frequency forced oscillation 483
technique to neonatal total liquid ventilation and present the first report of respiratory impedance 484
under this modality of ventilation. The process revealed considerable differences between 485
conventional mechanical ventilation and TLV lung mechanics. Namely, the high viscosity and 486
density of PFC shift frequencies of interest towards lower values and yield considerably higher 487
inertance and resistance values. In addition, we show that a 0.2 Hz single-frequency signal 488
represents an optimal tool to track pulmonary mechanics modulations under diverse 489
physiological circumstances such as bronchoconstriction and dilation as well as compliance 490
reduction. This experiment, the first of a series, is a key element toward enhancing our 491
knowledge of lung dynamics in total liquid ventilation. Further studies will be needed to adapt 492
multiple-frequency signal to TLV in order to improve and enable a more effective assessment of 493
spectral respiratory mechanics at f 2 Hz. Moreover, the physiological explanation of the model 494
parameter meaning, mostly unusual G and values, is yet to be determined. The main long-term 495
objective is to achieve the implementation of LFFOT in future total liquid ventilators ready to be 496
introduced in neonatal intensive care units. Hence, as in gas ventilation, LFFOT is expected to 497
provide insight into treatment progression and help plan optimal weaning procedures in total 498
liquid ventilation. 499
500
501
ACKNOWLEDGMENTS 502
The authors gratefully acknowledge Mrs. Nathalie Carrier for statistical consultations, 503
together with Mr. Raymond Robert and Mrs. Nathalie Samson for their technical assistance. We 504
22
also acknowledge Pulsion Medical Systems for providing the PiCCO device used for 505
hemodynamic monitoring. 506
507
GRANTS 508
This work was supported in part by the Fondation des toiles and the Facult de Mdecine et des 509
Sciences de la Sant of the Universit de Sherbrooke (Perinatal Research Team on Ovines), the 510
Natural Sciences and Engineering Research Council of Canada (NSERC) and the Fonds 511
qubcois de la recherche sur la nature et les technologies (FQRNT). P Micheau, J-P Praud and H 512
Walti are members of the FRSQ-funded Centre de Recherche Clinique tienne-Le Bel of the 513
Centre Hospitalier Universitaire de Sherbrooke. 514
515
DISCLOSURES 516
None declared. 517
23
518
REFERENCES 519
520
1. Alvarez FJ, Gastiasoro E, Rey-Santano MC, Gomez-Solaetxe MA, Publicover NG, 521
and Larrabe JL. Dynamic and quasi-static lung mechanics system for gas-assisted and liquid- 522
assisted ventilation. IEEE Trans Biomed Eng 56: 1938-1948, 2009. 523
2. Baba Y, Taenaka Y, Akagi H, Nakatani T, Masuzawa T, Tatsumi E, Wakisaka Y, 524
Toda K, Eya K, Tsukahara K, and Takano H. A volume-controlled liquid ventilator with 525
pressure-limit mode: imperative expiratory control. Artif Organs 20: 1052-1056, 1996. 526
3. Bagnoli P, Tredici S, Seetharamaiah R, Brant DO, Hewell LA, Johnson K, Bull JL, 527
Costantino ML, and Hirschl RB. Effect of repeated induced airway collapse during total liquid 528
ventilation. ASAIO J 53: 549-555, 2007. 529
4. Bates JH, and Lutchen KR. The interface between measurement and modeling of 530
peripheral lung mechanics. Respir Physiol Neurobiol 148: 153-164, 2005. 531
5. Bayat S, Strengell S, Porra L, Janosi TZ, Petak F, Suhonen H, Suortti P, Hantos Z, 532
Sovijarvi AR, and Habre W. Methacholine and Ovalbumin Challenges Assessed by Forced 533
Oscillations and Synchrotron Lung Imaging. Am J Respir Crit Care Med 2009. 534
6. Bendat JS, and Piersol AG. Engineering Applications of Correlation and Spectral 535
Analysis. New-York: John Wiley & Sons, Inc., 1980, p. 302. 536
7. Beydon N, Davis SD, Lombardi E, Allen JL, Arets HG, Aurora P, Bisgaard H, Davis 537
GM, Ducharme FM, Eigen H, Gappa M, Gaultier C, Gustafsson PM, Hall GL, Hantos Z, 538
Healy MJ, Jones MH, Klug B, Lodrup Carlsen KC, McKenzie SA, Marchal F, Mayer OH, 539
Merkus PJ, Morris MG, Oostveen E, Pillow JJ, Seddon PC, Silverman M, Sly PD, Stocks J, 540
24
Tepper RS, Vilozni D, and Wilson NM. An official American Thoracic Society/European 541
Respiratory Society statement: pulmonary function testing in preschool children. Am J Respir 542
Crit Care Med 175: 1304-1345, 2007. 543
8. Costantino ML, Micheau P, Shaffer TH, Tredici S, and Wolfson MR. Clinical design 544
functions: round table discussions on the bioengineering of liquid ventilators. ASAIO J 55: 206- 545
208, 2009. 546
9. Dubois AB, Brody AW, Lewis DH, and Bugess BF. Oscillation mechanics of lungs and 547
chest in man. J Appl Physiol 8: 587-594, 1956. 548
10. Farre R, Ferrer M, Rotger M, and Navajas D. Servocontrolled generator to measure 549
respiratory impedance from 0.25 to 26 Hz in ventilated patients at different PEEP levels. Eur 550
Respir J 8: 1222-1227, 1995. 551
11. Fredberg JJ, and Stamenovic D. On the imperfect elasticity of lung tissue. J Appl 552
Physiol 67: 2408-2419, 1989. 553
12. Fuhrman BP, Paczan PR, and DeFrancisis M. Perfluorocarbon-associated gas 554
exchange. Crit Care Med 19: 712-722, 1991. 555
13. Goldman MD. Clinical application of forced oscillation. Pulm Pharmacol Ther 14: 341- 556
350, 2001. 557
14. Goldman MD, Saadeh C, and Ross D. Clinical applications of forced oscillation to 558
assess peripheral airway function. Respir Physiol Neurobiol 148: 179-194, 2005. 559
15. Greenspan JS, Wolfson MR, and Shaffer TH. Liquid ventilation. Semin Perinatol 24: 560
396-405, 2000. 561
25
16. Grinnan DC, and Truwit JD. Clinical review: respiratory mechanics in spontaneous and 562
assisted ventilation. Crit Care 9: 472-484, 2005. 563
17. Hall GL, Hantos Z, Petak F, Wildhaber JH, Tiller K, Burton PR, and Sly PD. 564
Airway and respiratory tissue mechanics in normal infants. Am J Respir Crit Care Med 162: 565
1397-1402, 2000. 566
18. Hall GL, Hantos Z, Wildhaber JH, Petak F, and Sly PD. Methacholine responsiveness 567
in infants assessed with low frequency forced oscillation and forced expiration techniques. 568
Thorax 56: 42-47, 2001. 569
19. Hantos Z, Collins RA, Turner DJ, Janosi TZ, and Sly PD. Tracking of airway and 570
tissue mechanics during TLC maneuvers in mice. J Appl Physiol 95: 1695-1705, 2003. 571
20. Hantos Z, Daroczy B, Suki B, Nagy S, and Fredberg JJ. Input impedance and 572
peripheral inhomogeneity of dog lungs. J Appl Physiol 72: 168-178, 1992. 573
21. Heckman JL, Hoffman J, Shaffer TH, and Wolfson MR. Software for real-time 574
control of a tidal liquid ventilator. Biomed Instrum Technol 33: 268-276, 1999. 575
22. Larrabe JL, Alvarez FJ, Cuesta EG, Valls-i-Soler A, Alfonso LF, Arnaiz A, 576
Fernandez MB, Loureiro B, Publicover NG, Roman L, Casla JA, and Gomez MA. 577
Development of a time-cycled volume-controlled pressure-limited respirator and lung mechanics 578
system for total liquid ventilation. IEEE Trans Biomed Eng 48: 1134-1144, 2001. 579
23. Lowe CA, and Shaffer TH. Pulmonary vascular resistance in the fluorocarbon-filled 580
lung. J Appl Physiol 60: 154-159, 1986. 581
24. Michaelson ED, Grassman ED, and Peters WR. Pulmonary mechanics by spectral 582
analysis of forced random noise. J Clin Invest 56: 1210-1230, 1975. 583
26
25. Navajas D, and Farre R. Oscillation mechanics. Eur Respir Monogr 12: 112-140, 1999. 584
26. Navajas D, Farre R, Canet J, Rotger M, and Sanchis J. Respiratory input impedance 585
in anesthetized paralyzed patients. J Appl Physiol 69: 1372-1379, 1990. 586
27. Oostveen E, MacLeod D, Lorino H, Farre R, Hantos Z, Desager K, Marchal F, and 587
Measurements ETFoRI. The forced oscillation technique in clinical practice: methodology, 588
recommendations and future developments. Eur Respir J 22: 1026-1041, 2003. 589
28. Parker JC, Sakla A, Donovan FM, Beam D, Chekuri A, Al-Khatib M, Hamm CR, 590
and Eyal FG. A microprocessor-controlled tracheal insufflation-assisted total liquid ventilation 591
system. Med Biol Eng Comput 2009. 592
29. Pelosi P, and Gattinoni L. Respiratory mechanics in ARDS: a siren for physicians? 593
Intensive Care Med 26: 653-656, 2000. 594
30. Petak F, Hayden MJ, Hantos Z, and Sly PD. Volume dependence of respiratory 595
impedance in infants. Am J Respir Crit Care Med 156: 1172-1177, 1997. 596
31. Pillow JJ, Hall GL, Willet KE, Jobe AH, Hantos Z, and Sly PD. Effects of gestation 597
and antenatal steroid on airway and tissue mechanics in newborn lambs. Am J Respir Crit Care 598
Med 163: 1158-1163, 2001. 599
32. Pillow JJ, Jobe AH, Collins RA, Hantos Z, Ikegami M, Moss TJ, Newnham JP, 600
Willet KE, and Sly PD. Variability in preterm lamb lung mechanics after intra-amniotic 601
endotoxin is associated with changes in surfactant pool size and morphometry. Am J Physiol 602
Lung Cell Mol Physiol 287: L992-998, 2004. 603
27
33. Pillow JJ, Sly PD, and Hantos Z. Monitoring of lung volume recruitment and 604
derecruitment using oscillatory mechanics during high-frequency oscillatory ventilation in the 605
preterm lamb. Pediatr crit care med 5: 172-180, 2004. 606
34. Pillow JJ, Stocks J, Sly PD, and Hantos Z. Partitioning of airway and parenchymal 607
mechanics in unsedated newborn infants. Pediatr Res 58: 1210-1215, 2005. 608
35. Polak A, Wysoczanski D, and Mroczka J. A multi-method approach to measruement of 609
respiratory system mechanis. Metrol Meas Syst 13: 3-18, 2006. 610
36. Polese G, Serra A, and Rossi A. Respiratory mechanics in the intensive care unit. Eur 611
Respir Monogr 31: 195-206, 2005. 612
37. Robert R. Modlisation numrique et stratgies de commande du dbit expiratoire pour 613
viter le collapsus des voies respiratoires en ventilation liquidienne totale. In: Gnie Mcanique. 614
Sherbrooke: Universit de Sherbrooke, 2007. 615
38. Robert R, Micheau P, Avoine O, Beaudry B, Beaulieu A, and Walti H. A Regulator 616
for Pressure Controlled Liquid Ventilation. IEEE Trans Biomed Eng 2009. 617
39. Robert R, Micheau P, Cyr S, Lesur O, Praud JP, and Walti H. A prototype of 618
volume-controlled tidal liquid ventilator using independent piston pumps. ASAIO J 52: 638-645, 619
2006. 620
40. Robert R, Micheau P, and Walti H. Optimal expiratory volume profile in tidal liquid 621
ventilation under steady state conditions, based on a symmetrical lung model. Asaio J 55: 63-72, 622
2009. 623
28
41. Rossman JE, Caty MG, Rich GA, Karamanoukian HL, and Azizkhan RG. 624
Neutrophil activation and chemotaxis after in vitro treatment with perfluorocarbon. J Pediatr 625
Surg 31: 1147-1151, 1996. 626
42. Sakai H, Ingenito EP, Mora R, Abbay S, Cavalcante FS, Lutchen KR, and Suki B. 627
Hysteresivity of the lung and tissue strip in the normal rat: effects of heterogeneities. J Appl 628
Physiol 91: 737-747, 2001. 629
43. Schmalisch G, Schmidt M, Proquitte H, Foitzik B, Rudiger M, and Wauer RR. 630
Measurement of changes in respiratory mechanics during partial liquid ventilation using jet 631
pulses. Crit Care Med 31: 1435-1441, 2003. 632
44. Seber GAF, and Wild CJ. Nonlinear Regression. New-York: John Wiley & Sons, Inc., 633
1989. 634
45. Skloot G, Goldman M, Fischler D, Goldman C, Schechter C, Levin S, and Teirstein 635
A. Respiratory symptoms and physiologic assessment of ironworkers at the World Trade Center 636
disaster site. Chest 125: 1248-1255, 2004. 637
46. Sly PD, Hayden MJ, Petak F, and Hantos Z. Measurement of low-frequency 638
respiratory impedance in infants. Am J Respir Crit Care Med 154: 161-166, 1996. 639
47. Smith H, Reinhold P, and Goldman M. Forced oscillation technique and impulse 640
oscillometry. Eur Respir Monogr 72-105, 2005. 641
48. Smith TM, Steinhorn DM, Thusu K, Fuhrman BP, and Dandona P. A liquid 642
perfluorochemical decreases the in vitro production of reactive oxygen species by alveolar 643
macrophages. Crit Care Med 23: 1533-1539, 1995. 644
29
49. Tarczy-Hornoch P, Hildebrandt J, and Jackson JC. Gravitational effects on volume 645
distribution in a model of partial and total liquid ventilation. Respir Physiol 120: 125-138, 2000. 646
50. Tgavalekos NT, Tawhai M, Harris RS, Musch G, Vidal-Melo M, Venegas JG, and 647
Lutchen KR. Identifying airways responsible for heterogeneous ventilation and mechanical 648
dysfunction in asthma: an image functional modeling approach. J Appl Physiol 99: 2388-2397, 649
2005. 650
51. Thamrin C, Janosi TZ, Collins RA, Sly PD, and Hantos Z. Sensitivity analysis of 651
respiratory parameter estimates in the constant-phase model. Ann Biomed Eng 32: 815-822, 2004. 652
52. Thomassen MJ, Buhrow LT, and Wiedemann HP. Perflubron decreases inflammatory 653
cytokine production by human alveolar macrophages. Crit Care Med 25: 2045-2047, 1997. 654
53. Varani J, Hirschl RB, Dame M, and Johnson K. Perfluorocarbon protects lung 655
epithelial cells from neutrophil-mediated injury in an in vitro model of liquid ventilation therapy. 656
Shock 6: 339-344, 1996. 657
54. Wagner EM, and Mitzner WA. Contribution of pulmonary versus systemic perfusion of 658
airway smooth muscle. J Appl Physiol 78: 403-409, 1995. 659
55. Wang YYL, Chang C, Cheng J, Hsiu H, and Wang W. Pressure wave propagation in 660
arteries, amodel with radial dilatation for simulating the behavior of a real artery. IEEE Eng Med 661
Biol Mag 16: 51-54, 1997. 662
56. Welch P. A Method Based on Time Averaging Over Short, Modified Periodograms. 663
IEEE Transactions on Audio Electroacoustics AU-15: 70-77, 1967. 664
30
57. Wolfson MR, Greenspan JS, Deoras KS, Rubenstein SD, and Shaffer TH. 665
Comparison of gas and liquid ventilation: clinical, physiological, and histological correlates. J 666
Appl Physiol 72: 1024-1031, 1992. 667
58. Wolfson MR, Hirschl RB, Jackson JC, Gauvin F, Foley DS, Lamm WJ, Gaughan J, 668
and Shaffer TH. Multicenter comparative study of conventional mechanical gas ventilation to 669
tidal liquid ventilation in oleic acid injured sheep. ASAIO J 54: 256-269, 2008. 670
59. Wolfson MR, and Shaffer TH. Liquid ventilation: an adjunct for respiratory 671
management. Paediatr Anaesth 14: 15-23, 2004. 672
60. Wolfson MR, and Shaffer TH. Pulmonary applications of perfluorochemical liquids: 673
ventilation and beyond. Paediatr Respir Rev 6: 117-127, 2005. 674
61. Yuan H, Suki B, and Lutchen KR. Sensitivity analysis for evaluating nonlinear models 675
of lung mechanics. Ann Biomed Eng 26: 230-241, 1998. 676
677
678
679
680
681
682
683
684
685
686
687
31
688
689
690
691
692
32
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
Table depicting sex and weight distribution within both groups. * indicates that the lambs were 714
not considered in the results since lamb 2 had perfluorothorax and lamb 6 had idiopathic 715
hemothorax (prior to total liquid ventilation initiation). ** Impedance of chest-strapped lamb 11 716
was not counted due to a loosening of the bandage during the experiment. M, male ; F, female ; 717
MCh, Methacholine 718
719
720
721
Table 1. Lamb characteristics
Lambs Groups Sex
Weight
(kg)
1 Chest-strapped M 2.85
2* Chest-strapped M 2.50
3 Chest-strapped F 2.06
4 Chest-strapped M 1.90
5 Chest-strapped F 2.34
11** Chest-strapped F 3.09
13 Chest-strapped F 2.45

6* MCh/Salbutamol M 3.04
7 MCh/Salbutamol F 2.54
8 MCh/Salbutamol F 2.00
9 MCh/Salbutamol M 2.10
10 MCh/Salbutamol M 2.06
12 MCh/Salbutamol M 3.01
33
722
TABLE 2. RESISTANCE AND REACTANCE OF THE RESPIRATORY SYSTEM AS MEASURED USING A 0.2 HZ SINGLE-
FREQUENCY SIGNAL
End-inspiration End-expiration
RRS
0.2
XRS
0.2
RRS
0.2
XRS
0.2

Median (IQR) Median (IQR) CV Median IQR Median IQR CV
(cmH
2
O s/mL) (cmH
2
O s/mL) % (cmH
2
O s/mL) (cmH
2
O s/mL) %
Adapt. 0.75
*
(0.43-0.82)

0.21

0.44-

0.16) 20.7 0.87

(0.58-1.19)

0.23 (

0.31-

0.19) 20.7
BL 0.60 (0.41-0.81)

0.17 (

0.26-

0.13) 12.4 0.62 (0.44-0.86)

0.21 (

0.28-

0.13) 18.5

BL 0.73 (0.60-0.75)

0.22 (

0.33-

0.11) 13.4 0.62 (0.61-0.86)

0.21 (

0.31-

0.12) 19.0
MCh 0.93
*
(0.78-0.99)

0.28 (

0.38-

0.15) 11.4 1.26

(0.95-1.37)

0.26

0.53-

0.20) 16.4
Salbut. 0.43

(0.35-0.52)

0.15

0.20-

0.09) 6.5 0.66 (0.64-0.82)

0.17 (

0.35-

0.12) 6.9

BL 0.41 (0.36-0.50)

0.16 (

0.22-

0.12) 11.2 0.53 (0.42-0.74)

0.24 (

0.26-

0.13) 17.9
Bandage 0.42
*
(0.40-0.79)

0.22
*
(

0.25-

0.17) 15.5 0.74

(0.58-1.11)

0.43

0.53-

0.24) 22.6
723
724
725
726
Table illustrating group median respiratory system resistance (RRS
0.2
) and reactance (XRS
0.2
) at 0.2 727
Hz for each experimental block. Data are presented as median with their respective interquartile 728
range (IQR) = 25
th
- 75
th
. Coefficient of variation (CV) provides an insight into mean intra- 729
individual variability within blocks and was computed using RRS
0.2
. Measurements have been 730
performed: during initial respiratory system adaptation to total liquid ventilation (Adapt.), during 731
continuous methacholine infusion (MCh) then after a 10-minutes salbutamol infusion (Salbut.), 732
and while lambs were chest-strapped (Bandage). Baseline steady state conditions (BL) values are 733
provided for each of the three conditions.

denotes statistical difference from the respective 734

baseline, while

is the comparison between methacholine and salbutamol, and

*
represents end- 735
inspiration values statistically different from end-expiration values. p 0.05 736
737
738
34
739
TABLE 3. LOW FREQUENCY FORCED OSCILLATION TECHNIQUE RESPIRATORY MECHANICS DURING
NEONATAL TOTAL LIQUID VENTILATION
BASELINE CHALLENGES

End-
inspiration
End-
expiration
End-
inspiration
End-
expiration

Baseline (n = 11) Adaptation (n = 11)
R
aw
(cmH
2
O s/mL) 0.37 (0.26-0.53) 0.40 (0.32-0.55) 0.40 (0.29-0.64) 0.47 (0.23-0.86)
I
RS
(cmH
2
O s
2
/mL) 0.06 (0.02-0.07) 0.06 (0.02-0.09) 0.05
*
(0.01-0.08) 0.04 (0.01-0.07)
G (cmH
2
O /mL) 0.25 (0.05-0.34) 0.17 (0.11-0.44) 0.15 (0.04-0.26) 0.18 (0.02-0.33)
H (cmH
2
O /mL) 0.25 (0.18-0.33) 0.28 (0.18-0.29) 0.26 (0.19-0.43) 0.31 (0.20-0.35)
0.80 (0.25-1.15) 0.85 (0.32-1.46) 0.61 (0.26-0.86) 0.62 (0.15-1.67)

Baseline (n = 5) Methacholine (n = 5)
R
aw
(cmH
2
O s/mL) 0.49 (0.29-0.52) 0.40 (0.32-0.51) 0.64 (0.27-0.68) 0.53 (0.25-0.66)
I
RS
(cmH
2
O s
2
/mL) 0.07 (0.02-0.09) 0.06 (0.02-0.09) 0.04 (0.01-0.09) 0.05 (0.01-0.08)
G (cmH
2
O /mL) 0.34 (0.16-0.51) 0.44 (0.24-0.57) 0.36 (0.19-0.73) 0.74 (0.30-1.18)
H (cmH
2
O /mL) 0.30 (0.21-0.36) 0.29 (0.24-0.33) 0.34 (0.23-0.39) 0.34 (0.26-0.63)
1.15 (0.58-1.51) 1.46 (0.74-1.96) 0.98 (0.68-2.62) 1.85 (1.17-2.06)

Baseline (n = 5) Salbutamol (n = 5)
R
aw
(cmH
2
O s/mL) 0.49 (0.29-0.52) 0.40 (0.32-0.51) 0.30 (0.13-0.38) 0.34 (0.15-0.49)
I
RS
(cmH
2
O s
2
/mL) 0.07 (0.02-0.09) 0.06 (0.02-0.09) 0.06 (0.02-0.07) 0.03 (0.02-0.06)
G (cmH
2
O /mL) 0.34 (0.16-0.51) 0.44 (0.24-0.57) 0.09
*
(0.09-0.38) 0.30

(0.20-0.55)
H (cmH
2
O /mL) 0.30 (0.21-0.36) 0.29 (0.24-0.33) 0.21

(0.19-0.27) 0.37 (0.28-0.51)

1.15 (0.58-1.51) 1.46 (0.74-1.96) 0.48

(0.38-1.75) 0.92 (0.60-1.19)

Baseline (n = 5) Chest-strapped (n = 5)
R
aw
(cmH
2
O s/mL) 0.30 (0.25-0.65) 0.33 (0.28-0.74) 0.29 (0.09-0.60) 0.28 (0.00-0.58)
I
RS
(cmH
2
O s
2
/mL) 0.05 (0.01-0.06) 0.05 (0.01-0.07) 0.05 (0.02-0.07) 0.05 (0.02-0.07)
G (cmH
2
O /mL) 0.06 (0.05-0.19) 0.13 (0.07-0.21) 0.14
*
(0.07-0.52) 0.53

(0.14-0.96)
H (cmH
2
O /mL) 0.21 (0.17-0.25) 0.25 (0.20-0.32) 0.25
*
(0.20-0.30) 0.40

(0.26-0.67)
0.26 (0.23-0.87) 0.56 (0.24-0.83) 0.81

(0.27-1.65) 0.83 (0.55-1.72)

740
741
742
743
Values are presented as median with interquartile range (IQR) = 25
th
- 75
th
for parameters airway 744
resistance (R
aw
), respiratory system inertance (I
RS
), tissue damping (G), tissue elastance (H) and 745
hysteresivity () as provided by constant-phase model fitting on respiratory impedance, between 746
0.05 and 2 Hz. Parameter values are given for end-inspiration and end-expiration volumes and 747
with statistically significant differences between both indicated by *.

indicates a significant 748

difference between parameters for each condition and its respective baseline and

between 749
methacholine and salbutamol. p 0.05 750
35
Figure 1 - Sequential arrangement of the six blocks (B) of measurements over time. Each block 751
was comprised of a complete frequency spectrum. Following performance of B1 (adaptation) and 752
B2 and B3 (baseline) in every lamb at the beginning of the experiment, lambs were separated in 753
two groups: one group receiving methacholine (MCh) for two blocks (B4 and B5) followed by 754
salbutamol (B6) and second group in which lambs were chest-strapped (B4 to B6). 755
756
757
Figure 2 a) shows typical experimental total impedance (Z
total
) distribution over frequency with 758
the line representing the seven-parameter inverse model fit on the data. Data from lamb 11, 759
baseline, end-inspiratory spectrum, fitting error = 0.033 b) shows the respiratory system 760
impedance (Z
RS
) spectrum, given by the constant-phase model fit, after retrieval from the Z
total
. 761
Dashed lines and crosses represent the imaginary part of impedance Im(Z
RS
); solid lines and dots 762
represent the real part of impedance Re(Z
RS
). Constant-phase model parameters are graphically 763
represented with G, the tissue damping; R
aw
, airway resistance, and H, the elastance separated 764
from I
RS
, the respiratory system inertance by f
res
, the resonant frequency. 765
766
Figure 3 - The figure shows averaged partial pressure of carbon dioxyde (PaCO
2
), of oxygen 767
(PaO
2
), and arterial saturation in oxygen (% SaO
2
) for all the lambs. 768
769
770
Figure 4 End-inspiration respiratory impedance (Z
RS
) curves of lambs at baseline conditions (n 771
= 11). R
RS
, respiratory system resistance; X
RS
, respiratory system reactance. The zero crossing 772
point on the reactance curves represents the resonant frequency. 773
774
775
776
Figure 5 Plots exemplifying typical changes (in one lamb) in respiratory system impedance 777
(Z
RS
) during continuous infusion of methacholine (MCh) (dark black lines) and after a 10-minutes 778
36
infusion of salbutamol (dashed grey lines) as compared to baseline conditions Z
RS
conditions 779
(thin dark lines) at both end-inspiration (a) and end-expiration (b) volumes. 780
781
782
783
Figure 6 Plots exemplifying typical changes in respiratory system impedance (Z
RS
) following 784
chest strapping of one lamb at both end-inspiration (a) and end-expiration (b) volumes. Thin line: 785
Z
RS
at baseline conditions; Dark line: Z
RS
on chest-strapped lamb. 786
787
788
789
790
791
792
B1 B2 B3
B4 B5 B6
B4 B5 B6
Adaptation Baseline
MCh Salbutamol
or
Bandage
0 1 2 3 4
2
1.5
1
0.5
0
0.5
1
1.5
2
2.5
Frequency(Hz)
Z
t
o
t

(
c
m
H
2
0

m
L
)
A

Re(Z
tot
)
Im(Z
tot
)
95%CIs
0 1 2 3 4
2
1.5
1
0.5
0
0.5
1
1.5
2
2.5
Frequency(Hz)
Z
R
S

(
c
m
H
2
0

m
L
)
B

Re(Z
RS
)
Im(Z
RS
)
95%CIs
I
RS
R
aw
H
G
0 0.5 1 1.5 2
2
1
0
1
2
3
Frequency(Hz)

Z
R
S

(
c
m
H
2
0

m
L
)

Re(Z
RS
)
Im(Z
RS
)
0 0.5 1 1.5 2
0.8
0.4
0
0.4
0.8
1.2
1.6
2
Frequency(Hz)
I
m
p
e
d
a
n
c
e

(
c
m
H
2
0

s
/
m
L
)
A
Lamb#8
MCh/Salubutamol
0 0.5 1 1.5 2
0.8
0.4
0
0.4
0.8
1.2
1.6
2
Frequency(Hz)
I
m
p
e
d
a
n
c
e

(
c
m
H
2
0

s
/
m
L
)
B
Lamb#8
MCh/Salubutamol
0 0.5 1 1.5 2
0.8
0.4
0
0.4
0.8
1.2
1.6
2
Frequency (Hz)
I
m
p
e
d
a
n
c
e

(
c
m
H
2
0

s
/
m
L
)
A
Lamb #1
Cheststrapped
0 0.5 1 1.5 2
0.8
0.4
0
0.4
0.8
1.2
1.6
2
Frequency (Hz)
I
m
p
e
d
a
n
c
e

(
c
m
H
2
0

s
/
m
L
)
B
Lamb #1
Cheststrapped