Denault - Techno Multi BJA 2022
Denault - Techno Multi BJA 2022
Denault - Techno Multi BJA 2022
doi: 10.1016/j.bja.2022.07.053
Advance Access Publication Date: xxx
Clinical Investigation
CLINICAL INVESTIGATION
Abstract
Background: Portal vein Doppler ultrasound pulsatility measured by transoesophageal echocardiography is a marker of
the haemodynamic impact of venous congestion in cardiac surgery. We investigated whether the presence of abnormal
portal vein flow pulsatility is associated with a longer duration of invasive life support and postoperative complications
in high-risk patients.
Methods: In this multicentre cohort study, pulsed-wave Doppler ultrasound assessments of portal vein flow were per-
formed during anaesthesia before initiation of cardiopulmonary bypass (before CPB) and after separation of cardiopul-
monary bypass (after CPB). Abnormal pulsatility was defined as portal pulsatility fraction (PPF) 50% (PPF50). The primary
outcome was the cumulative time in perioperative organ dysfunction (TPOD) requiring invasive life support during 28
days. Secondary outcomes included major postoperative complications.
Results: 373 patients, 71 (22.0%) had PPF50 before CPB and 77 (24.9%) after CPB. PPF50 was associated with longer duration
of TPOD (median [inter-quartile range]; before CPB: 27 h [11e72] vs 19 h [8.5e42], P¼0.02; after CPB: 27 h [11e61] vs 20 h
1
2 - Denault et al.
[8e42], P¼0.006). After adjusting for confounders, PPF50 before CPB showed significant association with TPOD. PPF50 after
CPB was associated with a higher rate of major postoperative complications (36.4% vs 20.3%, P¼0.006).
Conclusions: Abnormal portal vein flow pulsatility before cardiopulmonary bypass was associated with longer duration
of life support therapy after cardiac surgery in high-risk patients. Abnormal portal vein flow pulsatility after cardio-
pulmonary bypass separation was associated with a higher risk of major postoperative complications although this
association was not independent of other factors.
Clinical trial registration: NCT03656263.
Keywords: Doppler ultrasound; heart failure; hepatic vein; perioperative transoesophageal echocardiography; portal
vein; right ventricular dysfunction; venous congestion
Fig 1. (a) Probe orientation to obtain portal vein pulsed-wave Doppler interrogation using transoesophageal echocardiography (TEE) (see
https://vimeo.com/showcase/5452417). (b) The two-dimensional (2D) image obtained will permit the identification of the portal vein for
Doppler assessment. The portal vein pulsatility fraction is obtained by the following formula: 100[maximal velocityeminimal velocity]/
maximal velocity. Normal portal vein Doppler velocities. (c) Portal vein pulsatility fraction of 0%. (d) Portal vein pulsatility fraction of 50%.
(e) Portal vein pulsatility fraction of 73%. IVC, inferior vena cava; PPF, portal pulsatility fraction. Panels (a) and (b) were obtained using a
CAE Healthcare Vimedix simulator.
CPB (T1) and after CPB separation and thoracic closure (T2) Data collection, follow-up, and outcomes
identical to a previous single-centre study.12 The assessment
Data collection was performed and managed using a stand-
consisted of pulsed-wave Doppler evaluation of portal venous
ardised electronic case report form programmed in the
flow and hepatic venous flow. In patients in whom portal
Research Electronic Data Capture (REDCap) platform14 hosted
Doppler was not successful, splenic vein Doppler was
at the Montreal Heart Institute. Data were entered locally by
performed.
research coordinators/investigators and verified by the central
Doppler evaluation with TOE were performed as described
data manager. A summary of missing data is presented in
previously (Fig 1).3 The portal vein Doppler pulsatility fraction
Supplementary Table 1.
(PPF) was defined as the difference between the maximal (in
We collected preoperative data on baseline characteristics
diastole) and the minimal (in systole) velocities during the
and coexisting conditions and intraoperative cumulative fluid
cardiac cycle, divided by the maximal velocity and expressed
balance, as detailed in Supplementary Data 1. Data collection
as a percentage. Abnormal portal flow pulsatility was defined
in the postoperative period was performed for the first 28 days
as a PPF 50% (PPF50) based on previous studies.9,12 The same
after surgery or until hospital discharge. At the end of surgery,
threshold value was used for the splenic vein pulsatility frac-
CPB separation was categorised as easy if no inotrope or only
tion.7 Inter-observer variability of PPF assessment using TOE
one vasopressor agent was used; difficult if at least two
was previously reported as good (mean difference between
pharmacologic agents from different categories were used
repeated measurements, 5.6%; intra-class correlation coeffi-
including inotrope, vasopressor, or inhaled pulmonary vaso-
cient, 0.824; P<0.001).12 With each set of echocardiographic
dilators; or complex if failure of the first CPB separation
measurements, we recorded haemodynamic parameters as
attempt resulted in a return on CPB from haemodynamic
described in detail in Supplementary Data 1.
4 - Denault et al.
Table 1 Preoperative patient characteristics in relation to portal pulsatility fraction (PPF) before initiation of cardiopulmonary bypass
(T1). Data are presented as number of patients/total (%) unless otherwise specified. Portal vein Doppler ultrasound could not be
performed in 51/373 patients. CABG, coronary artery bypass grafting; EuroSCORE II, European System for Cardiac Operative Risk
Evaluation II; GFR, glomerular filtration rate; IQR, inter-quartile range; sd, standard deviation.
reason or if mechanical circulatory support required to ach- AKI was defined according to the Kidney Disease:
ieve CPB separation. This classification has been previously Improving Global Outcomes (KDIGO) group creatinine
validated.15 criteria.18 Severe AKI corresponded to stage 2 of the KDIGO
The primary outcome studied was the cumulative time of classification and was defined as an increase of more than
persistent organ dysfunction (POD) or death up to 28 days after 100% in serum creatinine, achievement of a serum creatinine
cardiac surgery (TPOD). Based on prior studies,16 participants 354 mmol Le1 with evidence of a minimum increase of >26
were considered in POD when they had one or more of the mmol Le1 from baseline, or the initiation of RRT. Major
following organ or life support interventions ongoing: me- bleeding was defined by the Bleeding Academic Research
chanical ventilation; vasopressor therapy (ongoing need for Consortium as one of the following: perioperative intracranial
vasopressor agents such as norepinephrine, epinephrine, bleeding within 48 h, or reoperation after closure of sternot-
vasopressin, dopamine >5 mg kge1 mine1, or phenylephrine omy for the purpose of controlling bleeding, or transfusion of
>50 mg mine1); mechanical circulatory support (ongoing need 5 units of whole blood of packed red blood cells within a 48-h
for mechanical devices such as intra-aortic balloon pump or period, or chest tube output 2 L within a 24-h period.19
extracorporeal membrane oxygenation (ECMO); acute contin- Delirium was defined per local delirium screening check-
uous or intermittent RRT. Participants who died during the lists.20,21 Stroke was defined as a central neurological deficit
observation period got the TPOD imputed from the time of persisting longer than 72 h. Sternal wound infection or
death until the end of the observation period (24 h for 28 days: mediastinitis was defined per surgeon or attending physician
672 h). diagnosis in the progress notes.
Pre-specified secondary outcomes included the following:
duration of mechanical ventilation, duration of vasopressor
support, use and duration of acute RRT, ICU, and hospital
Statistical analysis
length of stay, complications up to 28 days after surgery A sample size of 360 patients was prespecified to enable a
including AKI,15 surgical re-intervention, major bleeding, detection of a meaningful clinical difference of 12 h in TPOD
stroke, delirium, deep sternal wound infection/mediastinitis, with 90% power. The full details of sample size estimation can
and death. We also included a pre-specified composite be found in Supplementary Data 2.
endpoint of major complications after surgery previously used For the primary analysis, we first assessed using the
in the setting of cardiac surgery17 defined as at least one of the ManneWhitney U-test whether the distribution of TPOD was
following: death, prolonged ventilation (>24 h), stroke, severe different between patients with and without PPF50. Analysis
AKI, deep sternal wound infection, and reoperation for any for portal assessment at T1 and T2 were done separately.
reason. Supplementary analysis using linear regression were
Portal vein Doppler in cardiac surgery - 5
a b
200 170
190 160
180 150
170 140
160
130
Pulsatility fraction at T2
Pulsatility fraction (%)
150
140 120
130 110
120 100
110 90
100
90 80
80 70
70 60
60 50
50 40
40 30
30
20 20
10 10
0 0
T1 T2 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Pulsatility fraction at T1
Fig 2. (a) Distribution of portal pulsatility fraction (PPF) before initiation of cardiopulmonary bypass (CPB) (T1) and after CPB separation (T2).
Boxplots illustrating the distribution at T1 and T2. No significant difference is observed (P¼0.48). (b) Relationship between PPF before
initiation of CPB and after cardiopulmonary bypass separation. The green zone includes patients who had PPF <50% at both assessments
(n¼182/292 [62.3%]); the red zone includes patients with PPF 50% both at T1 and T2 (29/292 [9.9%]); the yellow zone includes patients with
PPF 50% only at T2 (42/292 [14.4%]) and patients with PPF 50% only at T1 (39/292 [13.4%]).
subsequently performed. The dependent variable TPOD was There was no interim analysis during recruitment. Results
log-transformed for this purpose to ensure the normality of were considered statistically significant when P<0.05
the residuals (details in Supplementary Data 2). Adjustments throughout all analyses. All analyses are unadjusted (crude)
were performed for EuroSCORE II and for the duration of CPB. unless specified. Data were analysed with the use of SPSS
These adjustment variables were pre-specified per protocol as version 25 (IBM Corp., Armonk, NY, USA), and R (R Develop-
potential confounders. As a post hoc sensitivity analysis, the ment Core Team, R Foundation for Statistical Computing,
same analysis with PPF as a continuous variable was per- Vienna, Austria) with the use of the nricens24 and predictABEL
formed. Results are presented as coefficients (b) with 95% packages.25
confidence intervals (CIs).
The association between PPF and haemodynamic/echo-
cardiographic parameters was assessed using a general esti-
Results
mating equation to account for repeated measurements in Among the 380 patients enrolled for the study, 373 patients
each patient. Results are presented as estimate (b) with 95% CI underwent TOE assessment and were included in the study
that can be interpreted as the mean change in PPF for each (Supplementary Fig. 1). The characteristics of included pa-
increment in the studied variable. Finally, as sensitivity ana- tients are presented in Table 1. The majority of patient had
lyses, all previous analyses were repeated while using the multiple surgical procedures planned and most had known
splenic pulsatility fraction as a substitute when PPF was pulmonary hypertension. The most common procedure per-
missing owing to a failed assessment. Apart for these ana- formed was aortic valve surgery in 210/373 (56.3%), followed by
lyses, missing data were not imputed throughout all analyses coronary artery bypass grafting in 181/373 (45.8%). Complete
(complete-case analysis). details about surgical procedures can be found in
As an exploratory analysis, we compared outcomes in Supplementary Table 2.
relationship with four groups summarising portal vein Portal vein flow Doppler assessment to measure PPF was
Doppler at T1 and T2. Pairwise comparisons with Bonferroni successful in 322/373 (86.6%) before initiation of CPB (T1), and
correction between each group were performed when P<0.05 in 308/373 (82.6%) patients after separation from CPB (T2). The
was present across all groups. median PPF at T1 was 35% (inter-quartile range [IQR], 24e47%)
As sensitivity analyses, the association between PPF50 at and 71/322 (22%) had a PPF50. The median PPF at T2 was 35%
T2 and major complications was further assessed using [IQR, 23e50%] and 77/309 (24.9%) had a PPF50. The distribution
multivariable logistic regression with EuroSCORE II, CBP of PPF is presented in Fig 2. Overall, there was no significant
duration, and grade of CPB weaning as adjustment variables. difference between the distribution of PPF at T1 and T2. A
The improvement in prediction resulting from the addition of successful Doppler assessment of at least portal, splenic site,
PPF50 to the baseline model was assessed by comparing or both was obtained in 370/373 (99.2%) during anaesthesia
receiver operating characteristic (ROC) curves using the before CPB (T1) and in 354/373 (94.9%) after separation from
DeLong test22 and determining the continuous net reclassifi- CPB (T2).
cation improvement (NRI)23 with 95% CI generated via the Baseline characteristics according to the presence or
bootstrapping method.24 absence of PPF50 at T1 are presented in Table 1. There was a
6 - Denault et al.
higher proportion of known pulmonary hypertension before PPF50 at T1 and T2, and TPOD (b¼0.425; 95% CI, 0.091e0.760;
surgery in patients with portal pulsatility at T1 (P¼0.014). In- P¼0.013 and b¼0.414; 95% CI, 0.106e0.721; P¼0.009, respec-
formation about intraoperative management and post- tively). After adjustment for EuroSCORE II and the duration of
operative outcomes is presented in Table 2. A slight positive CPB, the association remained present for PPF50 at T1
correlation was observed between intraoperative fluid balance (b¼0.360; 95% CI, 0.038e0.682; P¼0.028) but not with PPF50 at
and the change in PPF between T1 and T2 (r¼0.153, P¼0.009). T2 (b¼0.237; 95% CI, e0.069 to 0.544; P¼0.129) as shown in
Separation from CPB was difficult or complex in 139/373 Table 4. Similar results were obtained when considering PPF as
(37.2%) patients. This occurred more frequently in patients a continuous variable as presented in Supplementary Tables 3
with PPF50 at T1 compared with patients with a PPF <50% and 4.
(54.9% vs 32.7%; P¼0.002). Regarding postoperative complications, PPF50 at T1 was
Haemodynamic variables associated with PPF as a contin- associated with longer vasopressor support, but no other as-
uous variable for T1 and T2 are presented in Table 3. Positive sociations with postoperative complications were found
associations with PPF were observed for CVP and pulmonary (Table 2). At T2, PPF50 was also associated with longer vaso-
artery pressure (systolic, diastolic, mean), whereas negative pressor support but was also associated with longer ICU and
associations were seen with mean arterial pressure and car- hospital stay, and with higher rates of major complications
diac index. The presence of atrial arrhythmia or ventricular including severe AKI after surgery (Table 5).
pacing were positively associated with PPF. Very severe left
ventricular dysfunction (LVEF 20%) and alterations in hepatic
Sensitivity analysis
venous flow were associated with PPF.
When considering splenic pulsatility fraction in cases for
Association between portal pulsatility fraction and
which PPF was not available, results were mostly concordant
postoperative outcomes
(Supplementary Tables 5e8) including for multivariable
The median TPOD [IQR] was higher in patients with PPF50 at T1 models regarding the association with TPOD (Supplementary
(27 h [11e72] vs 19 [8.5e42] h, P¼0.022) and at T2 (27 h [14e61] vs Tables 9 and 10).
20 h [8e42], P¼0.006) compared with patients with PPF <50% at In exploratory analysis, we compared outcomes according to
the same moment. In univariable linear regression analysis, a four groups based on PPF both at T1 and T2 (Table 6). Across
positive association with TPOD was present between both groups, the rate of major complications, and the duration of
Table 2 Intraoperative management and postoperative outcomes in relationship with portal pulsatility fraction (PPF) before initiation
of cardiopulmonary bypass (T1). Portal vein Doppler could not be performed in 51/373 patients. Composite of major complications:
defined as at least one of the following: death, prolonged ventilation (>24 h), stroke, severe AKI, deep sternal wound infection and
reoperation for any reason.17 Data are presented as number of patients/total (%) unless otherwise specified. *CPB separation was
categorised as easy if no inotrope and only one vasopressor agent was used; difficult if at least two pharmacologic agents from
different categories were used including inotrope, vasopressor, and inhaled pulmonary vasodilators; or complex if failure of the first
CPB separation attempt resulted in a return on CPB owing to a haemodynamic reason or if mechanical circulatory support was
required to achieve CPB separation. yTwo patients died intraoperatively after re-intervention after chest closure, all postoperative
outcomes are considered missing for this patient except death, major complications and TPOD. CPB, cardiopulmonary bypass; IQR,
inter-quartile range; SD, standard deviation.
Duration of CPB (min), median [IQR] 123 [97e163] 122 [83e167] 0.908
Total crystalloids received (ml), median [IQR] 2000 [1000e3100] 1900 [900e2700] 0.330
Diuresis during surgery (ml kge1 he1), median [IQR] 1.80 [0.99e3.17] 2.09 [1.26e3.09] 0.331
Total i.v. Intake (ml), median [IQR] 3051 [2075e4265] 2830 [2000e4056] 0.567
Total output (ml), median [IQR] 1530 [1043e2240] 1870 [1160e2550] 0.067
Cumulative intraoperative fluid balance (ml), median [IQR] 1371 [330e2540] 1100 [100e1965] 0.152
CPB separation*
Easy 169/251 (67.3%) 32/71 (45.1%) 0.002
Difficult 74/251 (29.5%) 34/71 (47.9%)
Complex 8/251 (3.2%) 5/71 (7.0%)
Postoperative complicationsy
Acute kidney injury (AKI) 62/250 (24.8%) 24/70 (34.3%) 0.114
Severe AKI (Stage 2) 21/250 (8.4%) 9/70 (12.9%) 0.26
Renal replacement therapy 7/250 (2.8%) 2/70 (2.9%) 1.00
Major bleeding 21/250 (8.4%) 2/70 (2.9%) 0.187
Surgical re-intervention 24/250 (9.6%) 6/71 (8.5%) 0.769
Postoperative delirium 43/250 (17.2%) 11/70 (15.7%) 0.858
Stroke 7/250 (2.8%) 3/70 (4.3%) 0.460
Sternal infection or mediastinitis 3/250 (1.2%) 0/70 (0%) 1.00
Death 6/251 (2.4%) 4/71 (5.6%) 0.494
Composite of major complications 58/251 (23.1%) 19/71 (26.8%) 0.524
Duration of vasopressor support (h), median [IQR] 16 [3e38] 23 [9e53] 0.008
Duration of mechanical ventilation (h), median [IQR] 6 [3.5e14] 7 [4e18] 0.10
Duration of ICU stay (h), median [IQR] 65 [32e101] 69 [26e111] 0.72
Duration of hospital stay (days), median [IQR] 8 [6e12] 9 [7e14] 0.24
Time of persistent organ dysfunction (TPOD) (h), median [IQR] 19 [8.5e42] 27 [11e72] 0.022
Portal vein Doppler in cardiac surgery - 7
Table 3 Haemodynamic and echocardiographic variables associated with portal pulsatility fraction as a continuous variable before
initiation of cardiopulmonary bypass (T1) and after cardiopulmonary bypass separation (T2). This analysis includes 630 assessments
including 322 at T1 and 308 at T2. Associations were assessed using generalised estimating equations with the pulsatility fraction as
the dependent variable. Estimate (b) can be interpreted as the mean change in portal venous Doppler velocity for each increment in the
studied variable. AF, atrial fibrillation; CI, confidence intervals; D, hepatic vein peak diastolic velocity; Ref., reference; S, hepatic vein
peak systolic velocity.
vasopressor support, ICU stay, hospital stay, and TPOD were and the association between PPF50 at T2 and complications
different. Pairwise comparison suggested that patients with PPF remained after adjustment with this variable (aOR¼2.12; 95%
<50% both at T1 and T2 (Group 1) had shorter duration of TPOD CI, 1.17e3.83; P¼0.01). However, a second multivariable model
and vasopressor support compared with patients with PPF50 showed that the association was no longer significant after
both at T1 and T2 (Group 4) (19 h [8e41] vs 36 h [21e73], P¼0$032). adjustment for EuroSCORE II, duration of CPB, and difficulty of
Group 1 also had shorter ICU stay and a lower rate of major CPB weaning (aOR¼1.35; 95% CI, 0.72e2.53; P¼0.35)
postoperative complications compared with patients with PPF50 (Supplementary Table 11). The addition of PPF50 at T2 did not
at T2 only (Group 3) (50 h [27e96] vs 96 h [50e143], P¼0.007 and 36/ result in a significant improvement in the area under the ROC
182 (19.8%) vs 17/42 (40.5%), P¼0.024, respectively). curve (0.727 vs 0.726, DeLong P¼0.083; Supplementary Fig 2)
We investigated whether the association between PPF50 at although it did result in a net reclassification improvement
T2 and major postoperative complications remained signifi- (continuous NRI¼0.316; 95% CI, 0.081e0.561; P¼0.01). In pa-
cant after adjustment with arteriovenous gradient pressure tients with difficult and complex CPB weaning (n¼115), com-
(MAPeCVP). Arteriovenous gradient pressure was not signifi- plications arose in 23/55 (51.1%) of patients with PPF50 at T2
cantly associated with complications in univariable analysis compared with 26/70 (37.1%) of patients without PPF50 at T2
(adjusted odds ratio [aOR]¼1.016; 95% CI, 0.990e1.042; P¼0.23) (P¼0.139).
Table 4 Association between portal pulsatility fraction (PPF) and time with persistent organ dysfunction or death (TPOD). *The TPOD
variable was log transformed to ensure the normality of residuals. Coefficients (b) are presented for each increase of ln (TPOD). CI,
confidence interval; EuroSCORE II, European System for Cardiac Operative Risk Evaluation II; PPF50, portal pulsatility fraction 50%;
TPOD, time of persistent organ dysfunction or death.
Univariable Multivariable
Model 1 (T1)
PPF50 before initiation of cardiopulmonary bypass 0.425 (0.091 to 0.760), P¼0.013 0.360 (0.038 to 0.682), P¼0.028
EuroSCORE II (per 1% increments) 0.058 (0.038 to 0.078), P<0.001 0.045 (0.023 to 0.067), P<0.001
Duration of cardiopulmonary bypass (per 1-h increment) 0.232 (0.110 to 0.355), P<0.001 0.195 (0.066 to 0.325), P¼0.003
Model 2 (T2)
PPF50 after cardiopulmonary bypass separation 0.414 (0.106 to 0.721), P¼0.009 0.237 (e0.069 to 0.544), P¼0.129
EuroSCORE II (per 1% increments) 0.058 (0.038 to 0.078), P<0.001 0.051 (0.027 to 0.075), P<0.001
Duration of cardiopulmonary bypass (per 1-h increment) 0.232 (0.110 to 0.355), P<0.001 0.117 (e0.018 to 0.251), P¼0.09
8 - Denault et al.
Table 5 Postoperative outcomes in relationship with portal pulsatility fraction (PPF) after cardiopulmonary bypass separation (T2).
Portal vein Doppler could not be performed in 64/373 patients. Composite of major complications: defined as at least one of the
following: death, prolonged ventilation (>24 h), stroke, severe acute kidney injury, deep sternal wound infection and reoperation for
any reason.17 Data are presented as number of patients (%) unless otherwise specified. *One patient died intraoperatively after re-
intervention after chest closure, all postoperative outcomes are considered missing for this patient except surgical re-intervention,
death, major complications and TPOD. IQR, inter-quartile range.
Postoperative complications
Acute kidney injury (AKI) 59/231 (25.5%) 27/77 (35.1%) 0.11
Severe AKI (Stage 2) 17/231 (7.4%) 12/77 (15.6%) 0.03
Renal replacement therapy 5/231 (2.2%) 2/77 (2.6%) 1.00
Major bleeding 17/231 (7.4%) 6/77 (7.8%) 0.90
Surgical re-intervention 19/232 (8.2%) 11/77 (14.3%) 0.12
Postoperative delirium 35/231 (15.2%) 16/77 (20.8%) 0.25
Stroke 8/231 (3.5%) 3/77 (3.9%) 1.00
Sternal infection or mediastinitis 2/231 (0.9%) 1/77 (1.3%) 1.00
Postoperative mechanical circulatory support 1/231 (0.4%) 0/77 (0%) 1.00
Death 7/232 (3.0%) 2/77 (2.6%) 1.00
Composite of major complications 47/232 (20.3%) 28/77 (36.4%) 0.006
Duration of vasopressor support (h), median [IQR] 17 [3e38] 27 [11e50] 0.001
Duration of mechanical ventilation (h), median [IQR] 6 [4e14] 9 [4e19] 0.14
Duration of ICU stay (h), median [IQR] 27 [14e61] 51 [27e97] <0.001
Duration of hospital stay (days), median [IQR] 8 [6e11] 9 [7e15] 0.004
Time of persistent organ dysfunction 20 [8e42] 27 [14e61] 0.006
(TPOD) or death (h), median [IQR]
Table 6 Postoperative outcomes in relationship with portal pulsatility fraction (PPF) before initiation of cardiopulmonary bypass (T1)
and after cardiopulmonary bypass separation (T2). Portal vein Doppler could not be performed at both T1 and T2 in 82/373 patients.
Composite of major complications: defined as at least one of the following: death, prolonged ventilation (>24 h), stroke, severe acute
kidney injury, deep sternal wound infection and reoperation for any reason.17 Data are presented as number of patients (%) unless
otherwise specified. Pairwise comparison revealed significant differences for the following subgroups: yGroup 1 vs 3, P¼0.024. zGroup 1
vs 4, P¼0.004. xGroup 1 vs 3, P¼0.007. ¶Group 1 vs 4, P¼0.032. IQR, inter-quartile range; PPF, portal pulsatility fraction; TPOD, time of
persistent organ dysfunction or death.
Group 1 PPF Group 2 PPF ≥50% at Group 3 PPF Group 4 PPF ≥50% P-value
<50% at T1 and PPF < 50% at <50% at T1 and at T1 and T2 (n¼29)
T1 and T2 T2 (n¼38) PPF ≥50% at
(n¼182) T2 (n¼42)
Postoperative complications
Acute kidney injury (AKI) 44/182 (24.2%) 11/38 (28.9%) 11/32 (26.2%) 13/29 (44.8%) 0.140
Severe AKI (Stage 2) 12/182 (6.6%) 4/38 (10.5%) 6/32 (14.3%) 5/29 (17.2%) 0.123
Renal replacement therapy 4/182 (2.2%) 1/38 (2.6%) 1/42 (2.4%) 1/29 (3.4%) 0.923
Major bleeding 15/182 (8.2%) 1/38 (2.6%) 4/42 (9.5%) 1/29 (3.4%) 0.585
Surgical re-intervention 15/182 (8.2%) 3/38 (7.7%) 7/42 (16.7%) 3/29 (10.3%) 0.388
Postoperative delirium 27/182 (14.8%) 5/38 (13.2%) 9/42 (21.4%) 6/29 (20.7%) 0.617
Stroke 5/182 (2.7%) 3/38 (7.9%) 2/42 (4.8%) 0/29 (0%) 0.255
Sternal infection or mediastinitis 2/182 (1.1%) 0/38 (0%) 1/42 (2.4%) 0/29 (0%) 0.757
Death 5/182 (2.7%) 2/38 (5.1%) 1/42 (2.4%) 1/29 (3.4%) 0.732
Composite of major complications* 36/182 (19.8%) 9/38 (23.7%) 17/42 (40.5%) 8/29 (27.6%) 0.042y
Duration of vasopressor support 15 [3e37] 19 [4e48] 22 [11e44] 30 [20e61] 0.002z
(h), median [IQR]
Duration of mechanical 6 [4e13] 7 [4e18] 9 [4e20] 10 [5e19] 0.251
ventilation (H), median [IQR]
Duration of ICU stay (h), 50 [27e96] 51 [26e100] 96 [50e143] 97 [45e119] 0.007x
median [IQR]
Duration of hospital stay 8 [6e11] 8 [6e12] 9 [7e17] 11 [8e14] 0.035
(days), median [IQR]
Time of postoperative 19 [8e41] 20 [7e65] 25 [14e57] 36 [21e73] 0.018¶
organ dysfunction
(TPOD) (h), median [IQR]
considering the pragmatic nature of the study, with TOE being Finally, although the association between portal Doppler
performed by the attending anaesthesiologist. It is unknown if pulsatility and adverse outcomes is hypothesised to be related
a more rigorous in-person training would have had produced to the mechanism of venous congestion, causality cannot be
better success rates and inter-operator variability was not definitively demonstrated. It must be noted that there is no
assessed although verification of the quality to the image ob- gold standard to diagnose congestive organ injury to which
tained by the operators was verified before the initiation of portal Doppler assessment could be compared.
enrolment. Nevertheless, we showed that in situations where
the portal vein is difficult to assess, the splenic vein can likely
Conclusions
be used as an alternative. This results in an increase in the
success rate of the technique to 95e99% with the results In conclusion, abnormal portal vein flow pulsatility before
remaining consistent in sensitivity analyses. Secondly, cardiopulmonary bypass was associated with longer duration
although the information for portal Doppler assessment was of life support therapy after cardiac surgery in high-risk
not supposed to be used to change the management of the patients. Abnormal portal vein flow pulsatility after cardio-
patient, we cannot exclude the fact that the absence of pulmonary bypass separation was associated with a higher
blinding of the anaesthesiologist in charge of the care during risk of major postoperative complications, although this as-
surgery may have had an influence on the postoperative out- sociation is not independent from other clinical factors. It re-
comes. However, it must be noted that this information was mains to be determined if interventions aiming to correct
not transmitted to the attending intensivist and is therefore portal pulsatility fraction in the setting of cardiac surgery
unlikely to have affected postoperative management or improves outcomes.
outcome assessment. Furthermore, intraoperative fluid man-
agement including intake and output did not appear to differ
significantly on the basis of the portal Doppler assessment at
Authors’ contributions
the start of surgery. Finally, as a pragmatic study, we did not Conception and design: AD, EJC, WBS
mandate the installation of a pulmonary artery catheter nor Acquisition and analysis of data: AD, EJC, WBS
collected other detailed haemodynamic information such as Drafting of the article: AD, EJC, WBS
details regarding the use of different vasoactive agents, or All authors contributed to the interpretation of data and
detailed assessment of RV function. However, we previously revising the manuscript critically for important content.
provided a detailed report on haemodynamic and echocar- All authors gave final approval of the version to be published.
diographic parameters including markers of right ventricular All authors agree to be accountable for all aspects of the work
function in relationship with PPF.12 in ensuring that questions related to the accuracy or integrity
10 - Denault et al.
of any part of the work are appropriately investigated 10. Beaubien-Souligny W, Eljaiek R, Fortier A, et al. The as-
and resolved. sociation between pulsatile portal flow and acute kidney
injury after cardiac surgery: a retrospective cohort study.
Declarations of interest J Cardiothorac Vasc Anesth 2018; 32: 1780e7
11. Benkreira A, Beaubien-Souligny W, Mailhot T, et al. Portal
WB-S has received speaker fees from Baxter in 2021. AD is
hypertension is associated with congestive encephalopa-
speaker for Masimo, consultant for CAE Healthcare and
thy and delirium after cardiac surgery. Can J Cardiol 2019;
received a grant form Edwards and Masimo. None of those
35: 1134e41
disclosures are related to any aspect of the current study.
12. Eljaiek R, Cavayas YA, Rodrigue E, et al. High post-
The other authors have no conflicts to declare.
operative portal venous flow pulsatility indicates right
Funding ventricular dysfunction and predicts complications in
cardiac surgery patients. Br J Anaesth 2019; 122: 206e14
The Canadian Anesthesia Research Foundation, Toronto, ON, 13. Beaubien-Souligny W, Rola P, Haycock K, et al. Quanti-
Canada. WB-S’s work is supported by the KRESCENT Program fying systemic congestion with Point-Of-Care ultrasound:
of the Kidney Foundation of Canada and the Fonds de Recherche development of the venous excess ultrasound grading
du Quebec en Sante (Clinical Research Scholar 1). The funding system. Ultrasound J 2020; 12: 16
sources had no involvement in any aspects of the study, 14. Harris PA, Taylor R, Minor BL, et al. The REDCap con-
design, writing of the report, or the decision to submit the sortium: building an international community of software
paper for publication. platform partners. J Biomed Inform 2019; 95, 103208
15. Denault AY, Bussieres JS, Arellano R, et al. A multicentre
Acknowledgements randomized-controlled trial of inhaled milrinone in high-
The authors thank Hilary Grocott and the Canadian Peri- risk cardiac surgical patients. Can J Anesth 2016; 63:
operative Anesthesia Clinical Trials Group. 1140e53
16. Stoppe C, McDonald B, Benstoem C, et al. Evaluation of
Appendix A. Supplementary data persistent organ dysfunction plus death as a novel com-
posite outcome in cardiac surgical patients. J Cardiothorac
Supplementary data to this article can be found online at
Vasc Anesth 2016; 30: 30e8
https://doi.org/10.1016/j.bja.2022.07.053.
17. Shahian DM, O’Brien SM, Filardo G, et al. The Society of
Thoracic Surgeons 2008 cardiac surgery risk models: part
References
1dcoronary artery bypass grafting surgery. Ann Thorac
1. Stein A, de Souza LV, Belettini CR, et al. Fluid overload and Surg 2009; 88: S2e22
changes in serum creatinine after cardiac surgery: pre- 18. Kidney Disease: Improving Global Outcomes (KDIGO)
dictors of mortality and longer intensive care stay. A Acute Kidney Injury Work Group. KDIGO clinical practice
prospective cohort study. Crit Care 2012; 16: R99 guideline for acute kidney injury. Kidney Int Suppl 2012; 2:
2. Lopez MG, Shotwell MS, Morse J, et al. Intraoperative venous 1e138
congestion and acute kidney injury in cardiac surgery: an 19. Mehran R, Rao SV, Bhatt DL, et al. Standardized bleeding
observational cohort study. Br J Anaesth 2021; 126: 599e607 definitions for cardiovascular clinical trials: a consensus
3. Toraman F, Evrenkaya S, Yuce M, et al. Highly positive intra- report from the Bleeding Academic Research Consortium.
operative fluid balance during cardiac surgery is associated Circulation 2011; 123: 2736e47
with adverse outcome. Perfusion 2004; 19: 85e91 20. Bergeron N, Dubois MJ, Dumont M, Dial S, Skrobik Y.
4. Pradeep A, Rajagopalam S, Kolli HK, et al. High volumes of Intensive care delirium screening checklist: evaluation of
intravenous fluid during cardiac surgery are associated a new screening tool. Intensive Care Med 2001; 27: 859e64
with increased mortality. HSR Proc Intensive Care Cardiovasc 21. Ely EW, Inouye SK, Bernard GR, et al. Delirium in me-
Anesth 2010; 2: 287e96 chanically ventilated patients: validity and reliability of
5. Silva Jr JM, de Oliveira AM, Nogueira FA, et al. The effect of the confusion assessment method for the intensive care
excess fluid balance on the mortality rate of surgical patients: unit (CAM-ICU). JAMA 2001; 286: 2703e10
a multicenter prospective study. Crit Care 2013; 17: R288 22. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing
6. Hassinger AB, Wald EL, Goodman DM. Early postoperative the areas under two or more correlated receiver operating
fluid overload precedes acute kidney injury and is asso- characteristic curves: a nonparametric approach. Bio-
ciated with higher morbidity in pediatric cardiac surgery metrics 1988; 44: 837e45
patients. Pediatr Crit Care Med 2014; 15: 131e8 23. Pencina MJ, D’Agostino Sr RB, Steyerberg EW. Extensions
7. Denault AY, Beaubien-Souligny W, Elmi-Sarabi M, et al. of net reclassification improvement calculations to mea-
Clinical significance of portal hypertension diagnosed sure usefulness of new biomarkers. Stat Med 2011; 30:
with bedside ultrasound after cardiac surgery. Anesth 11e21
Analg 2017; 124: 1109e15 24. Inoue E. Nricens: NRI for risk prediction Models with Time to
8. Bolognesi M, Quaglio C, Bombonato G, et al. Splenic Event and binary response data. R package version 1.6. 2018
Doppler impedance indices estimate splenic congestion in may 30. Vienna: the R Foundation. Available from: https://
patients with right-sided or congestive heart failure. Ul- CRAN.R-project.org/package¼nricens. [Accessed date 30
trasound Med Biol 2012; 38: 21e7 May 2021]
9. Beaubien-Souligny W, Benkreira A, Robillard P, et al. Al- 25. Kundu S, Aulchenko YS, van Duijn CM, Janssens AC.
terations in portal vein flow and intrarenal venous flow PredictABEL: an R package for the assessment of risk
are associated with acute kidney injury after cardiac prediction models. Eur J Epidemiol 2011; 26: 261e4
surgery: a prospective observational cohort study. J Am 26. Ting PC, Wu VC, Liao CC, et al. Preoperative right ven-
Heart Assoc 2018; 7, e009961 tricular dysfunction indicates high vasoactive support
Portal vein Doppler in cardiac surgery - 11
needed after cardiac surgery. J Cardiothorac Vasc Anesth milrinone and epoprostenol in severe right ventricular
2019; 33: 686e93 failure: a report of 2 cases. A A Case Rep 2017; 9: 219e23
27. Lella LK, Sales VL, Goldsmith Y, et al. Reduced right ven- 30. Tremblay JA, Couture EJ, Albert M, et al. Noninvasive
tricular function predicts long-term cardiac re- administration of inhaled nitric oxide and its hemody-
hospitalization after cardiac surgery. PLoS One 2015; 10, namic effects in patients with acute right ventricular
e0132808 dysfunction. J Cardiothorac Vasc Anesth 2019; 33: 642e7
28. Leening MJ, Vedder MM, Witteman JC, Pencina MJ, 31. Couture EJ, Tremblay JA, Elmi-Sarabi M, Lamarche Y,
Steyerberg EW. Net reclassification improvement: Denault AY. Noninvasive administration of inhaled epo-
computation, interpretation, and controversies: a litera- prostenolandinhaledmilrinoneinextubated,spontaneously
ture review and clinician’s guide. Ann Intern Med 2014; 160: breathing patients with right ventricular failure and portal
122e31 hypertension: a report of 2 cases. A A Pract 2019; 12: 208e11
29. Tremblay JA, Beaubien-Souligny W, Elmi-Sarabi M, 32. Rola P, Miralles-Aguiar F, Argaiz E, et al. Clinical applica-
Desjardins G, Denault AY. Point-of-care ultrasonography tions of the venous excess ultrasound (VExUS) score:
to assess portal vein pulsatility and the effect of inhaled conceptual review and case series. Ultrasound J 2021; 13: 32