Professional Documents
Culture Documents
ORIGINAL ARTICLE
Abstract subtracting Ppl (calculated from chest wall and respiratory system
elastance) from the airway plateau pressure.
Rationale: Esophageal manometry is the clinically available
method to estimate pleural pressure, thus enabling calculation of Measurements and Main Results: In pigs and human cadavers,
transpulmonary pressure (PL). However, many concerns make it expiratory and inspiratory PL using Pes closely reflected values in
uncertain in which lung region esophageal manometry reflects local PL. dependent to middle lung (adjacent to the esophagus). Inspiratory PL
estimated from elastance ratio reflected the directly measured
Objectives: To determine the accuracy of esophageal pressure (Pes) nondependent values.
and in which regions esophageal manometry reflects pleural pressure
(Ppl) and PL; to assess whether lung stress in nondependent regions Conclusions: These data support the use of esophageal manometry
can be estimated at end-inspiration from PL. in acute respiratory distress syndrome. Assuming correct calibration,
expiratory PL derived from Pes reflects PL in dependent to middle
Methods: In lung-injured pigs (n = 6) and human cadavers (n = 3), lung, where atelectasis usually predominates; inspiratory PL
Pes was measured across a range of positive end-expiratory pressure, estimated from elastance ratio may indicate the highest level of lung
together with directly measured Ppl in nondependent and dependent stress in nondependent “baby” lung, where it is vulnerable to
pleural regions. All measurements were obtained with minimal ventilator-induced lung injury.
nonstressed volumes in the pleural sensors and esophageal balloons.
Expiratory and inspiratory PL was calculated by subtracting local Ppl Keywords: esophageal manometry; transpulmonary pressure;
or Pes from airway pressure; inspiratory PL was also estimated by acute respiratory distress syndrome
( Received in original form September 5, 2017; accepted in final form January 9, 2018 )
Supported by a RESTRACOMP (Research Training Competition) training award from the Research Institute of the Hospital for Sick Children, Toronto, Ontario,
Canada; São Paulo Research Foundation grant FAPESP #2014/02030-7, São Paulo, Brazil; National Council for Scientific and Technological Development,
Brazil; and Coordination for the Improvement of Higher Level Personnel, Brazil.
Author Contributions: T.Y. designed the study (pigs and cadavers), conducted the study (pigs), analyzed the data (pigs), wrote the manuscript, and revised the manuscript.
M.B.P.A. designed the pleural sensors (pigs and cadavers), designed the study (pigs and cadavers), analyzed data (pigs), revised the manuscript, and organized the
study as a supervisor (pigs). D.L.G. conducted the study (cadavers) and analyzed the data (cadavers). L.C. designed the study (cadavers) and revised the manuscript.
C.A.S.L. prepared the pleural sensors (pigs and cadavers) and conducted the study (pigs). R.R., C.C.A.M., and S.G. conducted the study (pigs). E.L.V.C. analyzed
the data (pigs) and revised the manuscript. P.F.G.C. performed surgical procedures (pigs). E.C. designed the study (cadavers), conducted the study (cadavers), and
revised the manuscript. J.-C.M.R. designed the study (cadavers), conducted the study (cadavers), analyzed the data (cadavers), revised the manuscript, and organized
the study as a supervisor (cadavers). L.B. designed the study (cadavers) and revised the manuscript. B.P.K. designed the study (pigs) and revised the manuscript.
Correspondence and requests for reprints should be addressed to Takeshi Yoshida, M.D., Ph.D., Keenan Research Centre, Li Ka Shing Knowledge Institute,
St. Michael’s Hospital, 30 Bond Street, Toronto, ON, M5B 1W8 Canada. E-mail: takeshiyoshida@hp-icu.med.osaka-u.ac.jp.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.
Am J Respir Crit Care Med Vol 197, Iss 8, pp 1018–1026, Apr 15, 2018
Copyright © 2018 by the American Thoracic Society
Originally Published in Press as DOI: 10.1164/rccm.201709-1806OC on January 11, 2018
Internet address: www.atsjournals.org
1018 American Journal of Respiratory and Critical Care Medicine Volume 197 Number 8 | April 15 2018
ORIGINAL ARTICLE
Computed tomography. Computed evaluated by a one-way ANOVA followed approaching midway between nondependent
tomography (CT) scans were used to by Tukey pairwise multiple comparisons. and dependent Ppl at higher levels of
estimate the superimposed pressure PL,es or PL,ER and PL,ND or PL,D were PEEP (Figure E2A). At lower levels of
between two pleural sensors, as previously compared using regression analysis and PEEP, PL,es approximated to PL,D: PL,D
described (18). Bland-Altman analysis. Differences were PL,es , PL,ND, whereas at higher levels of
considered significant if P , 0.05. PEEP, PL,es was intermediate between PL,ND
Human Cadaver Study and PL,D: PL,D , PL,es , PL,ND (Figure 2A).
There was a good correlation between PL,
Preparation. Cadavers were immersed in Results D and PL,es (R = 0.94, P , 0.01, slope =
2
the special fluid to preserve their elasticity, 1.13, y-intercept = 0.55 cm H2O; Figure E3,
according to the Thiel method (22). Pleural Pressure Gradient in Normal
left panel). Bland-Altman analysis revealed
Cadavers were intubated, suctioned, and and Injured Lungs
that PL,es served a good surrogate for
connected to a mechanical ventilator. An In pigs, the vertical Ppl gradient was
significantly greater (at all PEEP levels) in expiratory PL in the dependent lung regions
esophageal balloon catheter (NutriVent) (i.e., mean difference, 0.9 cm H2O), but PL,
was inflated with minimal nonstressed injured versus normal lungs (Figure E1); at
PEEP of 10 cm H2O, the vertical gradient es was systematically higher than PL,D as
volume (Figure E6) and its position PEEP was increased (Figure E3, right
validated (as above). The respiratory was 1.8-fold greater in injured versus normal
lungs (Figure 1). This is similar to the values panel).
mechanics in these cadavers are In human cadavers, Pes at end-
comparable to the values in patients with of superimposed pressure (1.7-fold) from
nondependent to dependent lung estimated expiration was approximately midway
lung injury (23). between actual (measured) Ppl in
Ppl measurements. Regional Ppl was using CT images (Figure 1). The vertical Ppl
gradient was correlated with superimposed nondependent and dependent lung; this
measured using surgically inserted pleural
pressure between two sensors (R2 = 0.81, relationship was consistent at all levels of
sensors (as above) in nondependent and
P , 0.01, slope = 1.37, y-intercept = 21.07 PEEP (from 5–15 cm H2O; Figure E2B).
dependent pleural spaces; pressures were
cm H2O). The baseline characteristics The PL,es was midrange between PL,ND and
recorded simultaneously with Pes, across a
(before thoracotomy) of the human cadavers PL,D, and this relationship was maintained
range of PEEP levels (decreasing from 15 to
(n = 3) are described in Table E1, and (at at all values of PEEP (Figure 2B).
5 cm H2O). PEEP levels were lower here to
PEEP 10 cm H2O) the vertical Ppl gradient Representative (serial) CT images in
avoid barotrauma in cadavers; they are a
scarce resource. During this procedure, low was 10.0 6 3.1 cm H2O (Figure 1). a pig confirmed that the position of the
esophagus was close to the dependent sensor
VT ventilation used volume-controlled
ventilation (VT 6 ml/kg, respiratory rate of Expiratory Transpulmonary Pressure at lower levels of PEEP (Figure 2A), whereas
10/min, pause 0.3 s). Calculated from Pes at higher levels of PEEP it was in the
In pigs with injured lungs, the end-expiratory midzone (i.e., distant from the dependent
Definitions and Calculations Pes, nondependent Ppl, and dependent sensor) (Figure 2A). In contrast to the pigs,
1. Superimposed pressure (18) = lung Ppl all decreased in parallel as PEEP the position of the esophagus is closer to
density 3 lung height, where lung was decreased. End-expiratory Pes closely midthorax (in the anterior-posterior
density = 1 2 (CT numbers/21,000), reflected dependent Ppl; it was always dimension) in a patient with ARDS
compared before and after lung injury. slightly less than dependent Ppl, (Figure 2B).
2. Vertical Ppl gradient = Ppl,D 2 Ppl,ND.
3. PL = Paw 2 Ppl, which was calculated at
end-inspiration, and at end-expiration, 14
Vertical pleural pressure gradient
as follows: Superimposed pressure
d Using esophageal pressure (8), PL,es =
12
Paw 2 Pes. 10
d Using direct measurement of Ppl in
*
(cmH2O)
1020 American Journal of Respiratory and Critical Care Medicine Volume 197 Number 8 | April 15 2018
ORIGINAL ARTICLE
15 ‡
*+ corresponding to PL,es just above 0 cm H2O).
*+‡
Pressure (cmH2O)
+‡
*
* +‡
10 +‡ * +‡ ‡
‡
Inspiratory Transpulmonary Pressure
‡ *
+ * +‡ *+ ‡
Calculated from Esophageal Pressure
+ *+ * ‡ ‡
5
* ‡
‡
‡ In the pig model with injured lungs, Pes at
‡
‡ ‡ end-inspiration closely reflected dependent
‡ ‡
‡ ‡ Ppl at lower levels of PEEP (Figure E4A). At
‡ ‡
0 ‡ Non-dependent sensor higher levels of PEEP, Pes was midway
‡
Esophageal between nondependent Ppl and dependent
Dependent sensor Ppl (Figure E4A). At lower levels of PEEP,
–5 PL,es approached PL,D, such that: PL,D PL,
4 6 8 10 12 14 16 18 20 22 24
es , PL,ND, whereas at higher levels of PEEP,
PEEP (cmH2O)
PL,es was intermediate: PL,D , PL,es , PL,ND
(Figure 4A).
B PEEP12 In human cadavers, Pes at end-
inspiration was approximately midway
between Ppl in nondependent and
dependent lung (at PEEP ranging from
15
5–15 cm H 2 O; Figure E4B). The P L,es
was midrange between PL, ND and P L, D,
and this relationship, PL, D , P L,es , P L,
10
ND , was maintained at all values of PEEP
Expiratory Transpulmonary
(Figure 4B).
Pressure (cmH2O)
15 60
Esophageal
corroborates the CT findings that
Atelectasis 50 distribution of aeration follows this
gradient (i.e., more atelectasis at more
Atelectasis (%)
10 40
dependent lung) after injury. In injured
30
lungs with an increased vertical Ppl
gradient, lung-protective strategy
5 20 targeting global parameters (e.g., the
limitation of VT, Pplat, and PEEP) cannot
10 protect each region at the same time
(7–10).
0 0 Previous studies reflect these limitations.
24 22 20 18 16 14 12 10 8 6 4 For example, a lung-protective strategy
PEEP (cmH2O) (i.e., Pplat , 30 cm H2O; VT, 6 ml/kg) was
–5 not sufficient to limit tidal hyperinflation—
and inflammation—in nondependent
Figure 3. The relationship of expiratory transpulmonary pressures calculated from esophageal
“baby” lung regions (7). Also, setting PEEP
pressure (Pes) and the amount of lung collapse in lung-injured pigs. Lung collapse (estimated by
electrical impedance tomography) was progressively increased as positive end-expiratory pressure in the absence of understanding the impact
(PEEP) was decreased. Expiratory transpulmonary pressure was calculated using absolute Pes on local distension, such as with use of a
(i.e., PEEP minus expiratory Pes). The minimum transpulmonary pressure needed to prevent PEEP/FIO2 table, may increase the risk of
atelectasis (i.e., maintain ,1% collapse as measured by electrical impedance tomography) was under recruiting dependent “atelectatic”
4.6 6 2.2 cm H2O (at PEEP 16 cm H2O). lung (8, 9). Thus, to reduce inspiratory
stress and minimize atelectasis, clinicians
might separately target inspiratory PL
dependent lung regions (i.e., mean difference, esophageal manometry was used in less measured in nondependent “baby” lung
7.3 cm H2O) (Figure 6B , right panel). than 1% of patients with ARDS (25); a and expiratory PL measured in dependent
major reason may be that its validity has “atelectatic” lung. These hypothetical aims
been questioned, as follows. would be especially important in the setting
Discussion First, in the supine position, a gradient of lung injury, where the vertical Ppl
of Ppl exists (higher in dorsal, lower in gradient is increased.
Our study confirmed the accuracy of the ventral), and this gradient is far greater in
absolute values of Pes, if calibrated properly. injured lungs (17, 18); thus, it is not Application of the Findings
Also, the two different estimates of PL possible for any single reading of Pes to It has been recognized that two different
(calculated from Pes or from the elastance simultaneously represent Ppl in dependent methods (i.e., calculation from directly
ratio) can reasonably reflect PL for clinical lung and in nondependent lung. measured Pes [8] or elastance ratio of chest
purposes but not in the same lung regions. Second, local factors may be important. wall to respiratory system [10, 14–16])
First, PL calculated from Pes (either For example, the mass of the mediastinum allow the estimation of PL, which we named
inspiratory or expiratory) accurately appears to increase—and falsely elevate— PL,es and PL,ER, respectively. The differences
reflected PL in the lung regions adjacent to Pes by direct compression (26, 27). Also, obtained by these two methods were
the esophageal balloon (i.e., dependent to incorrect volumes of air in the esophageal believed to indicate the inaccuracy
middle lung), if correctly calibrated. Thus, balloon can result in over- or underestimation (unreliability) of either (or both) as
Pes at end-expiration is potentially useful to of the surrounding Ppl (28). measures of PL. Our findings suggest
guide PEEP settings to counter atelectasis in Third, two different estimates of Ppl instead that the two values, PL,es and PL,ER,
dependent to middle lung regions. Second, (and thus PL)—both derived from accurately represent the local PL in
inspiratory PL calculated from the elastance esophageal manometry—are widely used. different locations: the dependent to
ratio (PL,ER) reasonably reflects local PL One estimate of PL is based on measured middle and the nondependent lung,
in the nondependent “baby” lung. PL,ER Pes (termed PL,es) (8), and the other is respectively. Far from diminishing the
indicates the highest level of inspiratory based on elastance ratio of chest wall to utility of esophageal manometry because
lung stress and may represent a novel and respiratory system (termed PL,ER) (14–16). of potential inaccuracy, this raises the
testable clinical target to reduce ventilator- These two estimates of PL have been shown possibility of “individualizing” lung
induced lung injury. to yield quite different results (12, 13), and protection by using expiratory PL, as well
this discrepancy raised concerns about as inspiratory PL, both derived from
Use of Pes in Patients with ARDS the validity of esophageal manometry in esophageal manometry.
Pes is a surrogate for Ppl, and its reflecting local distending pressure. Pes at end-expiration accurately
measurement has enhanced our knowledge reflects Ppl in the lung regions adjacent
about the mechanical properties of the lung Pleural Pressure Gradient to the esophageal balloon (i.e., middle to
and chest wall (11, 24). Although widely The current data demonstrate that the dependent lung) in which atelectasis usually
used in research, a large epidemiologic vertical gradient of Ppl increased (1.7 times) predominates. Our findings are compatible
study covering 50 countries reported that after induction of lung injury; this is with previous pioneering findings showing
1022 American Journal of Respiratory and Critical Care Medicine Volume 197 Number 8 | April 15 2018
ORIGINAL ARTICLE
20
Esophageal Dependent sensor
* +‡ 6.2 6 1.3 cm H2O at PEEP of 4 cm H2O,
+‡
pigs; 2.2 6 1.8 cm H2O vs. 13.1 6 5.0 cm
Pressure (cmH2O)
*+‡
*+ ‡ +‡
* +‡
‡ H2O at PEEP of 5 cm H2O, cadavers; Figure
15 *+‡
+‡ ‡
*+ +‡
+‡ ‡ E2) This explains why the estimate of Ppl
*+ *+ *+ ‡
+
+
‡ calculated from elastance ratio reflects the
10 *+ *+ ‡
+ ‡ value in nondependent lung but not other
‡ ‡
lung regions.
5 Because nondependent “baby” lung
regions will be the region most susceptible
0 to stress during inspiration when the
4 6 8 10 12 14 16 18 20 22 24 ventilation is shifted to nondependent lung
PEEP (cmH2O) regions because of dependent atelectasis
(1–3), its local PL (i.e., PL,ER) may be a
B more specific marker of risk from
30 Non-dependent sensor Elastance ratio
ventilation-induced lung injury than local
PL in the middle to dependent lung regions
Inspiratory Transpulmonary
A
30 5
Difference in Inspiratory PL
(Esophageal) (cmH2O)
20
Inspiratory PL
–5
–10
10
–15
0 –20
0 10 20 30 –10 0 10 20 30
Inspiratory PL Mean Inspiratory PL
(Nondependent sensor) (cmH2O) (Esophageal + Nondependent sensor)/2 (cmH2O)
5
Difference in Inspiratory PL
(Elastance Ratio) (cmH2O)
20
Inspiratory PL
–5
10
–10
0 –15
0 10 20 30 –10 0 10 20 30
Inspiratory PL Mean Inspiratory PL
(Nondependent sensor) (cmH2O) (Elastance Ratio + Nondependent sensor)/2 (cmH2O)
Figure 5. The correlation between inspiratory transpulmonary pressure calculated from nondependent sensor versus (A) esophageal pressure and
(B) elastance ratio. Each correlation between inspiratory transpulmonary pressure (PL) calculated from nondependent sensor versus (A) esophageal
pressure (Pes) and (B) elastance ratio was shown in the left panel. The black dots represent the data from lung-injured pigs and the white dots
represent the data from human cadavers. Black lines represent the linear regression line. The corresponding Bland-Altman analyses are shown in the
right panel. Black lines and thin gray lines represent the mean 6 2 SD of the differences between inspiratory PL calculated from Pes (A) and elastance
ratio (B) versus nondependent sensor. (A) There was a good correlation between inspiratory PL calculated from nondependent sensor versus Pes.
However, PL calculated from Pes underestimated (z4.4 cm H2O) inspiratory PL in the nondependent lung regions. (B) There was a good correlation
between inspiratory PL calculated from nondependent sensor versus elastance ratio. Also PL calculated from elastance ratio served a good surrogate
for inspiratory PL in the nondependent lung regions (i.e., mean difference, 0.8 cm H2O).
Third, the accuracy of direct Ppl testing indicated that the “dynamic” injured (porcine) lungs open, using pixel
measurements is key. We confirmed that changes in Ppl and Pes were reliable, and compliance-based collapse in EIT. Thus, we
there were no major pneumothoraces using confirmation that the vertical Ppl gradient need to confirm how much PL,es will be
EIT (six of six pigs), CT scans (four of six (from our pleural sensors) was equal to necessary to keep injured lungs open in
pigs), and ultrasound and X-ray (all superimposed pressure between two patients with ARDS in CT.
cadavers). We used several approaches to sensors (estimated from CT scan) suggested Fifth, it is important to stress that PL,ER
ensure that the pleural sensors were not validity of the “static” Ppl. reflects the highest level of static lung stress
associated with measurement artifact Fourth, we found that a value of PL,es but cannot predict the amount of cyclic
(online supplement). Successful occlusion just above zero was not enough to keep parenchymal deformation at the ventilated
1024 American Journal of Respiratory and Critical Care Medicine Volume 197 Number 8 | April 15 2018
ORIGINAL ARTICLE
A
30 10
Difference in Inspiratory PL
20
(Esophageal) (cmH2O)
Inspiratory PL
0
10
–5
0
–10
–10 –15
–10 0 10 20 30 –10 0 10 20 30
Inspiratory PL Mean Inspiratory PL
(Dependent sensor) (cmH2O) (Esophageal + Dependent sensor)/2 (cmH2O)
10
Difference in Inspiratory PL
(Elastance Ratio) (cmH2O)
20
Inspiratory PL
5
10
0
0
–5
–10 –10
–10 0 10 20 30 –10 0 10 20 30
Inspiratory PL Mean Inspiratory PL
(Dependent sensor) (cmH2O) (Elastance Ratio + Dependent sensor)/2 (cmH2O)
Figure 6. The correlation between inspiratory transpulmonary pressure calculated from dependent sensor versus (A) esophageal pressure and (B)
elastance ratio. Each correlation between inspiratory transpulmonary pressure (PL) calculated from dependent sensor versus (A) esophageal pressure
(Pes) and (B) elastance ratio was shown in the left panel. The solid dots represent the data from lung-injured pigs, and the open dots represent the
data from human cadavers. Black lines represent the linear regression line. The corresponding Bland-Altman analyses were shown in the right panel.
Black lines and thin gray lines represent the mean 6 2 SD of the differences between inspiratory PL calculated from Pes (A) and elastance ratio (B) versus
dependent sensor. (A) There was a good correlation between inspiratory PL calculated from dependent sensor versus Pes. PL calculated from Pes
reflected values nearby the dependent lung regions (i.e., mean difference, 2.1 cm H2O). (B) There was a good correlation between inspiratory PL calculated
from dependent sensor versus elastance ratio. However, PL calculated from elastance ratio overestimated (z7.3 cm H2O) inspiratory PL in the dependent
lung regions.
“baby” lung like driving pressure does, under passive conditions but not under atelectasis and improve survival
which is considered as a major determinant active condition. (NCT01681225). In addition, it suggests
of ventilator-induced lung injury (31). Thus, that estimation of “local” PL in nondependent
we need to confirm that the ventilatory Future Directions (i.e., vulnerable) lung might be possible;
strategy incorporating the limitation of PL, These data add strong evidence for the this would represent a novel and testable
ER at end-inspiration and maintenance potential utility of esophageal manometry target to reduce ventilator-induced lung
of positive PL,es at end-expiration in ARDS. The current study supports injury, and prospective study is needed
would help to limit dynamic lung stress. the rationale for an ongoing trial targeting to assess the plausibility of this
Furthermore, we evaluated this concept Pes at the end-expiration to reduce concept.
Conclusions region most vulnerable to ventilator- statistical advice; Gilles Bronchti, M.D., Ph.D.,
These data further support the validity of induced lung injury. n and Cardiac Arrest and Ventilation International
esophageal manometry. Pes accurately Association for Research, Anatomy Laboratory,
Université du Québec à Trois-Rivières, for the
reflected Ppl in the lung regions adjacent to Author disclosures are available with the text support of Dr. Bronchti’s laboratory and making
the esophageal balloon (i.e., dependent and of this article at www.atsjournals.org.
Thiel cadavers available for this protocol; and Mr.
middle lung) in expiration and inspiration. Almir Ferrer and Mrs. Diana Rosales, Research &
In addition, inspiratory PL calculated Acknowledgment: The authors thank Fabiana Development Department, Medtronic, São
from elastance ratio (PL,ER) corresponds Madotto, Ph.D., School of Medicine and Surgery, Paulo, Brazil, for the development and design of
to the local PL in nondependent lung, the University of Milan-Bicocca, Manza, Italy, for pleural sensors using three-dimensional printers.
1026 American Journal of Respiratory and Critical Care Medicine Volume 197 Number 8 | April 15 2018