1412 www.ccmjournal.

org June 2013 Volume 41 Number 6

V
olume expansion is the rst-line treatment in many
cases of hemodynamic failure. The main reason to
administer volume, by conventional thinking, is to
increase preload and optimize cardiac output with the sub-
sequent effect being increasing oxygen delivery and reducing
Objectives: During circulatory failure, the ultimate goal of treatments
that increase cardiac output is to reduce tissue hypoxia. This can only
occur if oxygen consumption depends on oxygen delivery. We com-
pared the ability of central venous oxygen saturation and markers of
anaerobic metabolism to predict whether a fuid-induced increase in
oxygen delivery results in an increase in oxygen consumption.
Design: Prospective study.
Setting: ICU.
Patients: Fifty-one patients with an acute circulatory failure (78%
of septic origin).
Measurements: Before and after a volume expansion (500 mL of
saline), we measured cardiac index, O
2
- and CO
2
-derived variables
and lactate.
Main Results: Volume expansion increased cardiac index 15%
in 49% of patients (volume-responders). Oxygen delivery
signifcantly increased in these 25 patients (+32% 16%,
p < 0.0001). An increase in oxygen consumption 15%
concomitantly occurred in 56% of these 25 volume-responders
(+38% 28%). Compared with the volume-responders in whom
oxygen consumption did not increase, the volume-responders in
whom oxygen consumption increased 15% were characterized
by a higher lactate (2.3 1.1 mmol/L vs. 5.5 4.0 mmol/L,
respectively) and a higher ratio of the veno-arterial carbon dioxide
tension difference (P(v a)CO
2
) over the arteriovenous oxygen
content difference (C(a v)O
2
). A fuid-induced increase in oxygen
consumption greater than or equal to 15% was not predicted by
baseline central venous oxygen saturation but by high baseline
lactate and (P(v a)CO
2
/C(a v)O
2
ratio (areas under the receiving
operating characteristics curves: 0.68 0.11, 0.94 0.05, and
0.91 0.06). In volume-nonresponders, volume expansion did
not signifcantly change cardiac index, but the oxygen delivery
decreased due to a hemodilution-induced decrease in hematocrit.
Conclusions: In volume-responders, unlike markers of anaerobic
metabolism, central venous oxygen saturation did not allow the
prediction of whether a fuid-induced increase in oxygen delivery
would result in an increase in oxygen consumption. This suggests
that along with indicators of volume-responsiveness, the indica-
tors of anaerobic metabolism should be considered instead of
central venous oxygen saturation for starting hemodynamic resus-
citation. (Crit Care Med 2013; 41:14121420)
Key Words: fuid challenge; oxygen consumption; shock; tissue
oxygenation; veno-arterial carbon dioxide gap; venous oxygen
saturation
Lactate and Venoarterial Carbon Dioxide
Difference/Arterial-Venous Oxygen Difference
Ratio, but Not Central Venous Oxygen Saturation,
Predict Increase in Oxygen Consumption in Fluid
Responders*
Xavier Monnet, MD, PhD
1,2
; Florence Julien, MD
1,2
; Nora Ait-Hamou, MD
1
; Marie Lequoy, MD
1,2
;
Clment Gosset, MD
1,2
; Mathieu Jozwiak, MD
1,2
; Romain Persichini, MD
1,2
; Nadia Anguel, MD
1,2
;
Christian Richard, MD
1,2
; Jean-Louis Teboul, MD, PhD
1,2
Critical Care Medicine
0090-3493
10.1097/CCM.0b013e318275cece
41
6
00
00
2013
Uma
Clinical Investigation
*See also p. 1570 and 1572.
1
AP-HP, Hpitaux universitaires Paris-Sud, Hpital de Bictre, service de
ranimation mdicale, Le Kremlin-Bictre, France.
2
Univ Paris-Sud, Facult de mdecine Paris-Sud, EA 4533, Le Kremlin-
Bictre, France.
Supplemental digital content is available for this article. Direct URL
citations appear in the printed text and are provided in the HTML and
PDF versions of this article on the journals Web site (http://journals.lww.
com/ccmjournal).
Professors Monnet and Teboul are members of the Medical Advisory
Board of Pulsion Medical Systems. The authors received funding from
Assistance Publique des Hpitaux de Paris. Drs. Monnet and Teboul con-
sulted for Pulsion Medical Systems. The remaining authors have not dis-
closed any potential conficts of interest.
For information regarding this article, E-mail: xavier.monnet@bct.aphp.fr
Copyright 2013 by the Society of Critical Care Medicine and Lippincott
Williams & Wilkins
DOI: 10.1097/CCM.0b013e318275cece
*See also p.XXXX.
Feature Articles
Critical Care Medicine www.ccmjournal.org 1413
tissue hypoxia. For predicting whether uid administration
will possibly increase cardiac output, many predictive indi-
ces have been developed during the last years (1). However,
reliable indices indicating whether an increase in oxygen
delivery might result in a reduction of tissue hypoxia, that
is, indices indicating that an oxygen debt is present (2, 3),
are lacking.
Based on the results of a large-scale study conducted in
patients with early septic shock (4), the Surviving Sepsis
Campaign recommends that the treatment of septic shock
should be guided by the central venous oxygen saturation
(ScvO
2
), the goal being to elevate it above 70% (5). Some
authors also suggested that a low mixed venous oxygen satu-
ration value should be the trigger for starting hemodynamic
resuscitation (6). Nevertheless, whereas several studies now
support the notion of a low ScvO
2
being indicative of substan-
dard oxygen delivery, a value of ScvO
2
greater than 70% has the
potential to be misleading (3). Indeed, during septic shock, due
to impairment of oxygen extraction, ScvO
2
is often elevated
in spite of oxygen delivery/oxygen consumption (DO
2
/VO
2
)
dependence. In this respect, some studies showed that in a
majority of septic shock patients who have been already resus-
citated, ScvO
2
was elevated while tissue hypoxia was still present
(711). Thus, ScvO
2
does not always reect tissue hypoxia and
might not be the best trigger for starting hemodynamic resusci-
tation (12). In the present study, we compared ScvO
2
with alter-
native resuscitation targets in patients with multiple etiologies
of tissue hypoperfusion, the majority with sepsis.
Markers of anaerobic metabolism, such as blood lactate, are
likely to be elevated in case of oxygen debt. Thus, these markers
might be better than ScvO
2
for guiding hemodynamic resuscita-
tion, as recently emphasized (13, 14). In a previous study of our
group, the ratio between the venoarterial carbon dioxide (CO
2
)
tension gradient (P(v a)CO
2
) and the arteriovenous O
2
content
gradient (C(a v)O
2
) was suggested to reect the occurrence of
anaerobic metabolism and to be correlated with lactate (15). In
that study, the ratio was calculated using the mixed venous blood.
A possible advantage of this indicator of anaerobic metabolism
over lactate would be to have a faster response to treatments.
Another advantage would be that, unlike lactate (16), it would be
a pure marker of anaerobic metabolism during sepsis.
The main goal of the present study was to compare the
respective ability of ScvO
2
, (P(v a)CO
2
/C(a v)O
2
ratio cal-
culated from the central venous blood, and lactate to predict
whether a uid-induced increase in cardiac output will reduce
the oxygen debt assessed by an increase in VO
2
.
PATIENTS AND METHODS
Patients
As approved by the institutional review board of our institu-
tion, patients relatives were informed about the study at the
time the patient was included. Possibility was given to them
to refuse patients participation at that time. If not, patients
were informed as soon as their mental status enabled it and
possibility was given to them to refuse to participate. Patients
were prospectively included if they met all the following
criteria:
Presence of a circulatory failure defned by a systolic arterial
pressure less than or equal to 90 mm Hg (or fall of systolic
arterial pressure 50 mm Hg in known hypertensive patients)
and one or more of the following signs: 1) urinary ow less
than or equal to 0.5 mL/kg/min for greater than or equal to 2
hours, 2) tachycardia greater than or equal to 100 beats/min,
or 3) presence of skin mottling.
Need for a fuid challenge, as decided by the attending
physician.
Jugular venous and femoral arterial catheters in place.
Investigators (F.J., N.A.-H., M.L., C.G., M.J., R.P.) checked
that the position of the tip of the venous catheter was in the
upper part of the right atrium on chest radiograph.
Hemodynamic monitoring with a PiCCO
2
(Pulsion Medical
Systems, Munich, Germany) in place.
Measurements and Study Design
At baseline, we measured arterial lactate and hemodynamic
and tissue oxygenation variables. The hemodynamic variables
included heart rate, blood pressure, and cardiac index mea-
sured by transpulmonary thermodilution. Tissue oxygenation
variables were measured after arterial and central venous blood
sampling and analysis (ABL 800ex, Radiometer, Copenhagen,
Denmark). CaO
2
was calculated as follows: CaO
2
= 1.34 SaO
2

Hb + 0.003 PaO
2
, where SaO
2
is the oxygen saturation of
arterial blood, Hb the hemoglobin concentration, and PaO
2

the oxygen tension of the arterial blood. CvO
2
was calculated
using the same formula but using the O
2
saturation and ten-
sion of the venous blood. The Do
2
was calculated using the
following formula: DO
2
= cardiac index CaO
2
10. The
Vo
2
was calculated using the formula: VO
2
= 10 cardiac
index C(a v)O
2
. The (P(v a)CO
2
) was calculated as
P(v a)CO
2
= PvCO
2
PaCO
2
. Arterial lactate was measured by
a Modular device (Roche, Meylan, France).
A uid challenge was then intravenously administered
(500 mL of saline over 30 min) (17). Immediately after the
uid challenge, we again measured hemodynamic and tissue
oxygenation variables. The dose of norepinephrine was kept
unchanged during the uid challenge. The arterial lactate was
measured (average + SD) 3.0 1.5 hours later. In the meantime,
resuscitation was continued without protocolized algorithm,
aiming to achieve a lactate value less than or equal to
1.7 mmol/L and a ScvO
2
value greater than or equal to 70%. This
resuscitation included continued volume expansion, blood
transfusion, catecholamine administration, and continuous
veno-venous hemoltration depending on the situation.
Statistical Analysis
Patients in whom the uid challenge induced an increase in
cardiac index greater than or equal to 15% were dened as
volume-responders and the remaining ones as volume-nonre-
sponders (18). By analogy, because VO
2
is proportional to car-
diac index, we dened a clinically signicant increase in VO
2

Monnet et al
1414 www.ccmjournal.org June 2013 Volume 41 Number 6
by a VO
2
augmentation 15%. Liver failure was dened by the
Sepsis-related Organ Failure Assessment score (19). Data are
expressed as mean SD or as median (interquartile range), as
appropriate. The areas under receiving operating character-
istics (ROC) curves are expressed as mean (95% condence
interval). The comparison of data before vs. after uid chal-
lenge was performed by paired Student t test or Wilcoxon
paired test as appropriate. The comparison of data among
different groups of patients was performed by a two-tailed
Student t test or a Mann-Whitney U test as appropriate. The
absolute change in variables over time was also analyzed by
performing an analysis of covariance with the change in vari-
able as dependent variable, the group of patients as a factor,
and the baseline value as a covariate. Proportions were com-
pared with the chi-square test. We constructed ROC curves
for testing the ability of ScvO
2
, the lactate level, and the (P(v-
a)CO
2
/C(a-v)O
2
ratio at baseline to predict an increase in
VO
2
greater than or equal to 15% with volume expansion in
volume-responders. The comparison among the areas under
the ROC curves was performed using the HanleyMcNeil test.
Correlations were analyzed using the Spearman coefcient. A
p value of less than 0.05 was considered statistically signi-
cant. The statistical analysis was performed using MedCalc
8.1.0.0 (Mariakerke, Belgium).
RESULTS
Study Population
Fifty-one patients were included in the study. Their charac-
teristics at baseline before volume expansion are summarized
in Table 1. Forty-six patients were mechanically ventilated
(all in the assist control mode). Forty patients suffered from
septic shock. The source of infection in these patients was
pneumonia in 33, urinary tract infection in three, infection
of acute pancreatitis in two, and unknown in two. ScvO
2

value was greater than 70% (78% 5%) in 25 patients. All
of them suffered from septic shock. ScvO
2
value was less than
70% (62% 8%) in the remaining 26 patients. Among the
40 septic shock patients, ScvO
2
value was greater than 70% in
25 at baseline. Norepinephrine was administered in 86% of
patients. Patients did not receive any other catecholamine or
any paralyzing agent.
Effects of Volume Expansion in the Whole Population
Considering the population as a whole, volume expansion
signicantly increased cardiac index (21% 22% paired t test,
p = 0.04) and DO
2
(14% 22% paired t test, p < 0.001). VO
2
did
not change signicantly (10% 24% paired t test, p = .62).
There was a signicant correlation between the baseline lac-
tate and the (P(v a)CO
2
/C(a v)O
2
ratio (r = 0.56, p < 0.0001).
This correlation was tested only at baseline because the second
values of lactate and O
2
- and CO
2
-derived variables were not
measured simultaneously after volume expansion. There was
a signicant correlation between the (P(v a)CO
2
) gradient
and the cardiac index (r = 0.36, p = 0.0002) but not between
(P(v a)CO
2
) and lactate at baseline (p = 0.58). The
uid-induced changes in the (P(v a)CO
2
) gradient and
in the cardiac index were signicantly correlated (r = 0.59,
p < 0.0001).
Differences Between Volume-Responders and
Volume-Nonresponders
Volume expansion increased cardiac index by more than 15%
in 25 volume-responders (49% of the whole population;
Fig. 1). In these volume-responders, cardiac index increased by
39% 17% (paired t test, p < 0.001), DO
2
increased by 32%
16% (paired t test, p < 0.001), and VO
2
increased by 32% 16%
(paired t test, p < 0.001, Supplemental Table 2, Supplemental
Digital Content 1, http://links.lww.com/CCM/A577).
Among these 25 volume-responders, volume expansion
increased VO
2
by more than 15% (38% 28%, paired t test, p <
0.001) in 14 VO
2
-responders, whereas it did not change VO
2
in
11 of the volume-responders (Table 2, Fig. 1).
In the remaining 26 volume-nonresponders, the arterial Hb
decreased by 8% 4% during volume expansion (paired t test,
p < 0.001). Concomitantly, DO
2
decreased by 4% 7% (paired
t test, p < 0.001), and VO
2
decreased by 4% 11% (paired
t test, p < 0.001).
Volume-responders and volume-nonresponders did not
differ in terms of baseline ScvO
2
, lactate, or (P(v a)CO
2
/
C(a v)O
2
ratio (Supplemental Table 1, Supplemental Digital
Content 1, http://links.lww.com/CCM/A577). When consid-
ering both volume-responders and volume-nonresponders,
ScvO
2
, lactate, or the (P(v a)CO
2
/C(a v)O
2
ratio did not pre-
dict whether uid administration would result in an increase
in VO
2
15% (areas under the ROC curves not signicantly dif-
ferent from 0.5).
The mortality rate was 43% in VO
2
-responders and 73% in
VO
2
-nonresponders (chi-square test, p = 0.27).
Prediction of the Response of VO
2
to Volume
Expansion in the Subgroup of Volume-Responders
In the 14 patients who were both volume-responders and
VO
2
-responders, mean ScvO
2
value at baseline was 70% 15%
and did not signicantly change with volume expansion
(paired t test, p = 0.20, Table 2 and Figs. 1 and 2). In four
of these patients, ScvO
2
value at baseline was less than 70%
(47% 4%). Shock was of septic origin in three of them. In
all the remaining patients who were both volume- and VO
2
-
responders, ScvO
2
value at baseline was greater than 70% and
shock was of septic origin. In volume- and VO
2
-responders,
the (P(v a)CO
2
/C(a v)O
2
ratio was 2.3 + 0.8 mm Hg/mL
at baseline and signicantly decreased by 42% 30% with
volume expansion (paired t test, p < 0.01, Table 2). The lactate
value was 5.5 4.0 mmol/L at baseline (Table 2).
In the 11 patients who were volume-responders but VO
2
-
nonresponders, shock had a nonseptic origin. Compared
with patients who were both volume-responders and VO
2
-
responders, these patients were characterized by both lower
baseline (P(v a)CO
2
/C(a v)o
2
and lactate (two-tailed
t tests, p = 0.01 and 0.02, respectively; Table 2 and Fig. 1). ScvO
2

value at baseline was 64% 4%, and it signicantly increased to
Feature Articles
Critical Care Medicine www.ccmjournal.org 1415
71% 2% with volume expansion (paired t test, p < 0.01, Table 2).
The (P(v a)CO
2
/C(a v)o
2
ratio was 1.3 0.5 mm Hg/mL at
baseline and did not signicantly change with volume expan-
sion (paired t test, p = 0.58, Table 2). Lactate value was 2.3 1.1
mmol/L at baseline (Table 2). Among the subgroup of volume-
responders, the uid-induced changes in the (P(v a)CO
2
/
C(a v)O
2
ratio (analysis of covariance, p < 0.001) but neither
in Scvo
2
(p = 0.28) nor in lactate (p = 0.41) were signicantly
lower in Vo
2
-nonresponders than in Vo
2
-responders.
Considering the population of volume-responders pooled, the
value of ScvO
2
at baseline did not predict the VO
2
increase 15%
(area under the ROC curve not signicantly different from 0.5).
Nevertheless, among volume-responders whose ScvO
2
value was
less than or equal to 70%, the four who were VO
2
-responders
had a baseline ScvO
2
value ranging between 40% and 50%
(Supplemental Table 2, Supplemental Digital Content 1, http://
links.lww.com/CCM/A577). In addition, all volume-responders
who were VO
2
-nonresponders had a baseline ScvO
2
> 50%. Thus,
in volume-responders, a ScvO
2
value 50% was always associated
with a uid-induced increase in VO
2
15%.
Still considering the population of volume-responders
pooled, a lactate value at baseline 2.7 mmol/L predicted the
VO
2
increase 15% with a sensitivity of 93% (66100) and a
specicity of 82% (4898). A (P(v a)CO
2
/C(a v)O
2
ratio
at baseline 1.8 mm Hg/mL predicted the VO
2
increase 15%
with a sensitivity of 86% (5798) and a specicity of 91% (59
100)(Fig. 2). The area under the ROC curve for the baseline
lactate and (P(v a)CO
2
/C(a v)O
2
ratio were both signi-
cantly larger than the area under the ROC curve for baseline
ScvO
2
(0.91 0.06, 0.94 0.05, and 0.68 0.11; Fig. 2).
Difference Between Patients With Baseline
ScvO
2
> 70% and Patients With Baseline ScvO
2
70%
When data were analyzed depending on the level of ScvO
2
at
baseline (Supplemental Table 2, Supplemental Digital Content
1, http://links.lww.com/CCM/A577), we observed that VO
2

increased 15% in patients with a baseline ScvO
2
> 70%, while
it did not change in patients with a baseline ScvO
2
70%.
Compared with patients with a baseline ScvO
2
> 70%, patients
with a baseline ScvO
2
70% were characterized by a higher
baseline VO
2
(88 36 mL/min/m
2
vs. 143 79 mL/min/m
2
,
respectively, two-tailed t test, p = 0.002), a nonsignicantly lower
baseline DO
2
(450 161 mL/min/m
2
vs. 374 188 mL/min/m
2
,
respectively, two-tailed t test, p = 0.13), and a lower baseline
lactate (5.4 4.2 vs. 2.2 1.4 mmol/L, respectively, two-tailed
t test, p < 0.001).
DISCUSSION
Although this study showed that none of the three studied
variables were predictors of increase in VO
2
in the total popula-
tion studied, the lactate and the (P(v a)CO
2
/C(a v)O
2
ratio
at baseline were good predictors of increase in VO
2
in those
patients whose DO
2
increased with uid administration. This
was not the case for ScvO
2
, except for values 50%.
The Concept of VO
2
/DO
2
Dependence as an Indicator
of Oxygen Debt
VO
2
/DO
2
dependence has been demonstrated to be a marker
of tissue hypoxia (20) and poor outcome (21). As a marker
of anaerobic metabolism, lactate increases in case of VO
2
/DO
2

dependence (22). As a rst result, the present study shows
that calculating the oxygen-derived variables from the central
instead of the mixed venous blood also allows detecting tissue
hypoxia through VO
2
/DO
2
dependence. Indeed, baseline lactate
value was elevated in patients in whom uid increased both DO
2

and VO
2
and lower in patients without VO
2
/DO
2
dependence.
Years ago, the concept of DO
2
/VO
2
dependence during sepsis
was a matter of debate (23). Some argued that a mathematical
coupling was responsible for the DO
2
/VO
2
relationship (24).
Nevertheless, the fact that in our study, lactate was elevated
at baseline and decreased after uid administration does not
support that such a phenomenon is responsible for the VO
2

increase in our study. It was suggested that therapy increasing
DO
2
(e.g., beta-adrenergic agents) might increase VO
2
through
a thermogenic effect (25). Nevertheless, this probably did not
occur in our study because DO
2
was increased by saline. Finally,
TABLE 1. Characteristics of Patients at
Baseline
Age (mean SD, yr) 62 15
Gener (female/male), n 15/36
Simplied acute physiologic score
(mean SD)
64 18
Time from the start of care (median
[IQR], hr)
6.1 [1.210.4]
Type of shock, n (%)
Septic 40 (78)
Drug poisoning 5 (10)
Hypovolemic 3 (6)
Cardiogenic 3 (6)
Patients receiving norepinephrine
at baseline, n (%)
44 (86)
Norepinephrine dose at baseline
(median [IQR], g/kg/min)
0.26 [0.220.47]
Diabetes, n (%) 4 (7)
Cirrhosis, n (%) 1 (2)
Acute liver failure, n (%) 2 (4)
Acute respiratory distress syndrome, n (%) 33 (65)
Tidal volume (mean SD, mL/kg of
predicted body weight)
a
7 2
Respiratory rate (mean SD/min)
a
28 2
Positive end-expiratory pressure
(mean SD, mm Hg)
a
7 2
Mortality at day 28 (%) 53
IQR = interquartile range.
a
In the 46 patients receiving mechanical ventilation.
Monnet et al
1416 www.ccmjournal.org June 2013 Volume 41 Number 6
one cannot exclude that an increase in VO
2
could result from
an increase in nonoxidative oxygen use once tissue hypoxia has
resolved (26).
Patients with ScvO
2
value greater than 70% were character-
ized by a relatively lower VO
2
and higher DO
2
than patients with
a ScvO
2
value less than or equal to 70%. This suggests that the
DO
2
/VO
2
relationship was steeper in patients with ScvO
2
value
>70%, that is, a typical septic pattern of DO
2
/VO
2
dependence.
This typical septic condition could also explain why ScvO
2
was
higher (due to the oxygen extraction impairment) and lactate
was higher (possible nonanaerobic production of lactate) (27)
in these patients. This group of patient is likely characterized
by a greater alteration of oxygen extraction capacities than
patients with a low ScvO
2
value.
Prediction of the VO
2
Increase in Response to the
Fluid-Induced Increase in DO
2
During shock, uid administration aims not only at increasing
cardiac output but also at reducing tissue hypoxia. It was
recently reported that ScvO
2
did not predict whether uid
administration will increase cardiac output or not (28). The
present study further suggests that ScvO
2
cannot predict
whether a DO
2
increase will increase VO
2
.
Indeed, in a large proportion of volume-responders, VO
2

increased while baseline ScvO
2
was high. All these patients were
in septic shock with high lactate, conrming that, during septic
shock, oxygen extraction is impaired (i.e., ScvO
2
is high) (29).
It is in agreement with previous studies showing that ScvO
2
is
greater than 70% despite obvious tissue hypoxia in a large pro-
portion of septic shock patients who have been already resus-
citated (8, 10).
In only few patients in whom the uid-induced increase in
DO
2
increased VO
2
, ScvO
2
was very low at baseline (between 40%
and 50%). In other words, only ScvO
2
values 50% were infor-
mative and allowed to predict that VO
2
would increase with DO
2
.
Nevertheless, such ScvO
2
values were rare in our population in
which septic shock patients were the majority. These patients with
very low ScvO
2
values were similar to those of the study by Rivers
et al. before resuscitation (4) but constituted a minority of our
patients. By contrast, a large proportion of volume-responders
with a baseline ScvO
2
value less than 70% did not respond to uid
by increasing VO
2
. Hence, recommendations to guide resuscitation
Figure 1. Flow diagram showing the original 51 patients dividing according their response to volume expansion in terms of cardiac index and in volume-
responders, in terms of oxygen consumption (VO
2
). The VO
2
-responders (VO
2
-R) and VO
2
-nonresponders (VO
2
-NR) are described with relationship to lac-
tate, the ratio between the veno-arterial gradient in carbon dioxide tension (P(v a)O
2
) over the arteriovenous gradient in oxygen content (C(a v)O
2
)
and central venous saturation (ScvO
2
).
51 paents enrolled
25
volume-responders
26
volume-non-responders
* *
0
1
2
3
4
5
6
7
0
1
1
2
2
3
3
0
10
20
30
40
50
60
70
80
90
lactate ScvO
2
P(v-a)CO
2
C(a-v)O
2 ns
mmHg/mL mmoL/L %
VO
2
-R VO
2
-NR VO
2
-R VO
2
-NR VO
2
-R VO
2
-NR
14
VO
2
-responders
11
VO
2
-non-responders
Feature Articles
Critical Care Medicine www.ccmjournal.org 1417
with ScvO
2
(5) may only apply to a minority of patients (12). The
only interest of ScvO
2
would be to predict a uid-induced increase
in VO
2
with a high sensitivity when it is low.
By contrast to ScvO
2
, markers of anaerobic metabolism pre-
dicted the uid-induced increase in VO
2
provided that patients
were volume-responders. In the whole population of volume-
responders and volume-nonresponders, these markers of anaer-
obic metabolism were unable to predict the effects of volume
expansion on VO
2
. This result is logical because the VO
2
challenge
can be interpreted as a test for detecting tissue hypoxia only if
DO
2
increases during the intervention. In other words, one can-
not expect any predicting value of markers of anaerobic metab-
olism if DO
2
does not increase with uid administration. Thus,
predicting whether volume expansion will eventually increase
VO
2
should follow a two-step approach: rst to detect preload
responsiveness (using proper dynamic tests [1]) and then, if
the test is positive, to predict whether uid administration will
increase VO
2
(using markers of anaerobic metabolism).
Importantly, increasing VO
2
should not be the only thera-
peutic goal. Indeed, if the DO
2
/VO
2
relationship is on its plateau,
TABLE 2. Hemodynamic and Tissue Oxygenation Variables During Volume Expansion in
Volume Responders According to the Response of VO
2
VO
2
Change 15% (n = 14) VO
2
Change < 15% (n = 11)
Before Volume
Expansion
After Volume
Expansion
Before Volume
Expansion
After Volume
Expansion
Heart rate (mean SD, beats/min) 90 15 88 12 116 13
a
116 16
a
Mean arterial pressure (mean SD, mm Hg) 76 11 83 13
b
64 9 77 10
b
Cardiac index (mean SD, L/min/m
2
) 2.7 1.2 3.7 1.5
b
2.7 1.1 3.7 1.4
b
Stroke index (mean SD, mL/m
2
) 31 17 44 21
b
23 9 32 11
b
Dose of norepinephrine (median
[interquartile range], g/kg/min)
0.28 [0.230.40] 0.28 [0.230.40] 0.23 [0.000.40] 0.23 [0.000.40]
Arterial pH (mean SD) 7.32 0.14 7.32 0.14 7.35 0.09 7.34 0.09
Arterial oxygen saturation (mean SD, %) 98 2 98 2 96 3 97 2
Arterial Hb (mean SD, g/dL) 10.5 2.2 9.9 2.0
b
10.5 1.5 9.7 1.2
b
Arterial oxygen content (mean SD, mL) 14.2 2.7 13.7 1.8
b
13.7 1.8 13.0 1.6
b
Arterial carbon dioxide tension
(mean SD, mm Hg)
32 7 31 8 37 9 38 9
Venous pH (mean SD) 7.31 0.18 7.30 0.11 7.30 0.09 7.31 0.10
Central venous oxygen saturation
(mean SD, %)
70 15 71 13 64 4 71 2
b
Venous Hb (mean SD, g/dL) 10.7 2.2 10.0 2.2
b
10.5 1.5 9.9 1.3
b
Central venous oxygen content
(mean SD, mL)
10.1 3.4 9.7 3.2 9.2 1.6 9.6 1.4
b
Central venous carbon dioxide tension
(mean SD, mm Hg)
40 7 36 8
b
43 10 41 9
b
Oxygen delivery (mean SD, mL/min/m
2
) 388 197 502 229
b
382 202 488 234
b
VO
2
(mean SD, mL/min/m
2
) 103 51 135 58
b
124 60 129 62
b
P(v-a)CO
2
(mean SD, mm Hg) 8.5 2.8 4.9 3.1
b
5.9 1.8
a
3.9 1.4
b
C(a-v)O
2
(mean SD, mL) 3.6 1.9 3.4 1.1 4.5 0.8 3.5 0.6
b
P(v-a)CO
2
/C(a-v)O
2
(mean SD, mm Hg/mL) 2.3 0.8 1.3 0.8
b
1.3 0.5
a
1.2 0.5
Before Volume
Expansion
3.5 1.0
Hr Later
Before Volume
Expansion
3.5 1.0 Hr
Later
Lactate (mean SD, mmol/L) 5.5 4.0 3.6 1.7
b
2.3 1.1
a
2.0 1.0
a,b
Hb = hemoglobin; VO
2
= oxygen consumption; P(v a)CO
2
= venoarterial carbondioxide tension difference; C(a v)O
2
: arteriovenous oxygen content difference.
a
p < 0.05 patients with a VO
2
change <15% vs. patients with a VO
2
change 15%.
b
p < 0.05 after vs. before volume expansion.
Monnet et al
1418 www.ccmjournal.org June 2013 Volume 41 Number 6
uid administration might securely move the DO
2
/VO
2
rela-
tionship far from the critical DO
2
value and add a margin of
safety in case of further decreases of VO
2
. Also, increasing DO
2

could improve perfusion and reverse tissue hypoxia in a rela-
tively small vascular bed so that global VO
2
could not change
appreciably.
The (P(v a)CO
2
/C(a v)O
2
Ratio As a Marker of
Global Anaerobic Metabolism
We found a signicant relationship between the
(P(v a)CO
2
/C(a v)O
2
ratio and the lactate at baseline,
conrming previous results of a study that used the pulmonary
artery catheter (15). Under conditions of tissue hypoxia, a
decrease in global VO
2
is associated with a decrease in aerobic CO
2

production and a potential increase in anaerobic production
of CO
2
, which is due to acid buffering (30). Thus, the total CO
2

production (VCO
2
) should be less reduced than VO
2
. According
to the Fick equation, VO
2
is equal to the product of cardiac
output and (C(a v)O
2
). Similarly, VCO
2
is equal to the product
of cardiac output and veno-arterial CO
2
content difference.
Therefore, the VCO
2
/VO
2
ratio is equal to the veno-arterial CO
2

content difference/(C(a v)O
2
) ratio. Over the physiological
range of CO
2
contents, CO
2
tension is linearly related to CO
2

content (31). Therefore, under anaerobic conditions and
provided that the amount of acid to buffer is only related to
anaerobism, the increased VCO
2
/VO
2
ratio should be reected by
the increased (P(v a)CO
2
/C(a v)O
2
.
One of the advantages of (P(v a)CO
2
/C(a v)O
2
over
lactate might be to react faster to short-term hemodynamic
changes. We observed that the (P(v a)CO
2
/C(a v)O
2
ratio
decreased as soon as volume expansion ended in volume-
responders. An important limitation of our study is that we
did not re-assess the (P(v a)CO
2
/C(a v)O
2
ratio at the pre-
cise time when lactate was measured again, precluding to test
whether the trends of both markers are linked.
Relationship Between Changes in ScvO
2
and in
Cardiac Index
We found no signicant relationship between changes in car-
diac index and in ScvO
2
, what is in discrepancy with recent
ndings (32). The latter study included cardiac patients while
we included a majority of septic patients. In this septic popula-
tion, the absence of correlation between the changes in ScvO
2

and cardiac index during volume expansion might be mainly
explained by the fact that VO
2
/DO
2
dependence was present in a
large proportion of our patients, explaining why ScvO
2
did not
change when a DO
2
and VO
2
increased concomitantly.
The Value of (P(v a)CO
2
) Gradient
In volume-responders in whom VO
2
increased, the (P(v a)
CO
2
) gradient was elevated at baseline and decreased with
uid infusion. Like Valle et al (9), we suggest that an elevated
(P(v a)CO
2
) gradient might be useful to identify patients who
Figure 2. Relationship between oxygen delivery (DO
2
) and oxygen
consumption (VO
2
) in patients with a ratio between veno-arterial
carbon dioxide tension difference over arteriovenous oxygen content
difference (P(v a)CO
2
/C(a v)O
2
18 mm Hg/mL (circles) and
<1.8 mm Hg/mL (squares) before (white markers) and after (black
markers) volume expansion in volume-responders. Data are expressed
as mean SD, n = 25.
40
60
80
100
120
140
150 250 350 450 550 650 750 850
V
O
2

(
m
L
/
m
i
n
/
m
2
)
DO
2
(mL/min/m
2
)
Figure 3. Receiving operating characteristics (ROC) curves showing
the ability of the baseline values of lactate, the ratio of the veno-arte-
rial carbon dioxide tension difference over the arteriovenous oxygen
content difference (P(v a)CO
2
/C(a v)O
2
and the central venous
oxygen saturation (ScvO
2
) to predict an increase in oxygen consump-
tion 15% in response to volume expansion. *p < 0.05 for the areas
under the ROC curves.
Feature Articles
Critical Care Medicine www.ccmjournal.org 1419
remain inadequately resuscitated. Conrming previous results
(9, 15), we observed that the (P(v a)CO
2
) gradient was not
correlated with lactate at baseline. These ndings reinforce the
message that the (P(v a)CO
2
) gradient better reects the abil-
ity of cardiac output to wash out the accumulated CO
2
than the
presence of anaerobic metabolism (33, 34).
Effect of Volume Expansion on DO
2
and VO
2
in Volume
Nonresponders
Importantly, in volume nonresponders, DO
2
decreased due
to a decrease in the Hb level related to hemodilution, as pre-
viously reported (35). Although this did not decrease the
(P(v a)CO
2
/C(a v)O
2
level, likely because the decrease in
DO
2
was not of sufcient magnitude, this suggests that admin-
istering uids in nonpreload-dependent patients is harmful.
This supports that testing volume responsiveness rst should
be an essential step in uid management strategy.
Limitations
A rst limitation of the present study is that we assessed O
2
-
and CO
2
-derived variables from the central venous blood and
not from the mixed venous blood, what might miss changes in
oxygenation of the splanchnic territory. Second, we found that
lactate was a marker of VO
2
/DO
2
dependence. However, lactate
is not a pure marker of anaerobism (13, 16, 27, 36). In our
study, conditions in which lactate is elevated without hypoxia
(e.g., epinephrine administration, liver failure) were absent
or rare, what might limit the generalizability of our ndings.
Third, our patients might not be representative of all septic
shock patients because of their high severity. Finally, our study
was not designed for testing whether increasing DO
2
even when
ScvO
2
is already 70% could inuence prognosis.
In conclusion, in volume-responders, unlike markers of
anaerobic metabolism, ScvO
2
did not allow the prediction of
whether a uid-induced increase in DO
2
would result in an
increase in VO
2
. This suggests that markers of tissue hypoxia
should be included in decisional algorithms rather than ScvO
2

as indicators for targeting hemodynamic resuscitation.
ACKNOWLEDGMENT
We thank Mrs Armelle Arnoux, from the Unit de Recherche
Clinique of Bictre hospital, for her help in statistical analysis.
REFERENCES
1. Marik PE, Monnet X, Teboul JL: Hemodynamic parameters to guide
fuid therapy. Ann Intensive Care 2011; 1:1
2. Vincent JL, De Backer D: Oxygen transport-the oxygen delivery con-
troversy. Intensive Care Med 2004; 30:19901996
3. van Beest P, Wietasch G, Scheeren T, et al: Clinical review: Use of
venous oxygen saturations as a goal - a yet unfnished puzzle. Crit
Care 2011; 15:232
4. Rivers E, Nguyen B, Havstad S, et al; Early Goal-Directed Therapy
Collaborative Group: Early goal-directed therapy in the treatment of
severe sepsis and septic shock. N Engl J Med 2001; 345:13681377
5. Dellinger RP, Levy MM, Rhodes A, et al: Surviving Sepsis Campaign:
International guidelines for management of severe sepsis and septic
shock: 2012. Crit Care Med 2013; 41:580637
6. Pinsky MR, Vincent JL: Let us use the pulmonary artery catheter cor-
rectly and only when we need it. Crit Care Med 2005; 33:11191122
7. Perz S, Uhlig T, Kohl M, et al: Low and supranormal central venous
oxygen saturation and markers of tissue hypoxia in cardiac surgery
patients: A prospective observational study. Intensive Care Med
2011; 37:5259
8. Shapiro NI, Howell MD, Talmor D, et al: Implementation and out-
comes of the Multiple Urgent Sepsis Therapies (MUST) protocol. Crit
Care Med 2006; 34:10251032
9. Valle F, Vallet B, Mathe O, et al: Central venous-to-arterial carbon
dioxide difference: An additional target for goal-directed therapy in
septic shock? Intensive Care Med 2008; 34:22182225
10. van Beest PA, Hofstra JJ, Schultz MJ, et al: The incidence of low
venous oxygen saturation on admission to the intensive care unit: A
multi-center observational study in The Netherlands. Crit Care 2008;
12:R33
11. Grissom CK, Morris AH, Lanken PN, et al; National Institutes of
Health/National Heart, Lung and Blood Institute Acute Respiratory
Distress: Association of physical examination with pulmonary artery
catheter parameters in acute lung injury. Crit Care Med 2009;
37:27202726
12. Perel A: Bench-to-bedside review: The initial hemodynamic resuscita-
tion of the septic patient according to Surviving Sepsis Campaign
guidelinesdoes one size ft all? Crit Care 2008; 12:223
13. Jansen TC, van Bommel J, Schoonderbeek FJ, et al; LACTATE study
group: Early lactate-guided therapy in intensive care unit patients: A
multicenter, open-label, randomized controlled trial. Am J Respir Crit
Care Med 2010; 182:752761
14. Jones AE, Shapiro NI, Trzeciak S, et al; Emergency Medicine Shock
Research Network (EMShockNet) Investigators: Lactate clearance vs
central venous oxygen saturation as goals of early sepsis therapy: A
randomized clinical trial. JAMA 2010; 303:739746
15. Mekontso-Dessap A, Castelain V, Anguel N, et al: Combination of
venoarterial PCO
2
difference with arteriovenous O
2
content differ-
ence to detect anaerobic metabolism in patients. Intensive Care Med
2002; 28:272277
16. Gutierrez G, Wulf ME: Lactic acidosis in sepsis: A commentary. Inten-
sive Care Med 1996; 22:616
17. Vincent JL, Weil MH: Fluid challenge revisited. Crit Care Med 2006;
34:13331337
18. Marik PE, Cavallazzi R, Vasu T, et al: Dynamic changes in arterial
waveform derived variables and fuid responsiveness in mechanically
ventilated patients: A systematic review of the literature. Crit Care
Med 2009; 37:26422647
19. Bakker J, Gris P, Coffernils M, et al: Serial blood lactate levels can
predict the development of multiple organ failure following septic
shock. Am J Surg 1996; 171:221226
20. Schumacker PT, Cain SM: The concept of a critical oxygen delivery.
Intensive Care Med 1987; 13:223229
21. Bihari D, Smithies M, Gimson A, et al: The effects of vasodilation with
prostacyclin on oxygen delivery and uptake in critically ill patients. N
Engl J Med 1987; 317:397403
22. Reuse C, Vincent JL, Pinsky MR: Measurements of right ventricular
volumes during fuid challenge. Chest 1990; 98:14501454
23. Squara P: Matching total body oxygen consumption and delivery: A
crucial objective? Intensive Care Med 2004; 30:21702179
24. Phang PT, Cunningham KF, Ronco JJ, et al: Mathematical coupling
explains dependence of oxygen consumption on oxygen delivery in
ARDS. Am J Respir Crit Care Med 1994; 150:318323
25. Teboul JL, Graini L, Boujdaria R, et al: Cardiac index vs oxygen-
derived parameters for rational use of dobutamine in patients with
congestive heart failure. Chest 1993; 103:8185
26. McCord JM: Oxygen-derived free radicals in postischemic tissue
injury. N Engl J Med 1985; 312:159163
27. Levy B: Lactate and shock state: The metabolic view. Curr Opin Crit
Care 2006; 12:315321
28. Velissaris D, Pierrakos C, Scolletta S, et al: High mixed venous oxy-
gen saturation levels do not exclude fuid responsiveness in critically
ill septic patients. Crit Care 2011; 15:R177
Monnet et al
1420 www.ccmjournal.org June 2013 Volume 41 Number 6
29. Astiz ME, Rackow EC, Kaufman B, et al: Relationship of oxygen
delivery and mixed venous oxygenation to lactic acidosis in patients
with sepsis and acute myocardial infarction. Crit Care Med 1988;
16:655658
30. Randall HM Jr, Cohen JJ: Anaerobic CO2 production by dog kidney
in vitro. Am J Physiol 1966; 211:493505
31. West JB: Gas transport to the periphery:how gases are moved to the
peripheral tissues. In: Book. West JBs (Eds). Baltimore, Williams and
Wilkins, 1990, pp 6985
32. Giraud R, Siegenthaler N, Gayet-Ageron A, et al: ScvO(2) as a marker
to defne fuid responsiveness. J Trauma 2011; 70:802807
33. Bakker J, Vincent JL, Gris P, et al: Veno-arterial carbon dioxide gradi-
ent in human septic shock. Chest 1992; 101:509515
34. Lamia B, Monnet X, Teboul JL: Meaning of arterio-venous PCO
2

difference in circulatory shock. Minerva Anestesiol 2006;
72:597604
35. Haupt MT, Gilbert EM, Carlson RW: Fluid loading increases oxygen
consumption in septic patients with lactic acidosis. Am Rev Respir
Dis 1985; 131:912916
36. James JH, Luchette FA, McCarter FD, et al: Lactate is an unreli-
able indicator of tissue hypoxia in injury or sepsis. Lancet 1999;
354:505508