Академический Документы
Профессиональный Документы
Культура Документы
Septic Shock
Sari Karlsson, MD* BACKGROUND: Vascular endothelial growth factor (VEGF) levels have been shown to
be elevated in severe sepsis. We investigated the value of VEGF in predicting organ
Ville Pettilä, MD, PhD† dysfunction and hospital mortality in adult patients with severe sepsis.
METHODS: We conducted a prospective observational cohort study in 24 closed
multidisciplinary intensive care units (ICU) in Finland. All ICU admission episodes
Jyrki Tenhunen, MD, PhD* (4500) were screened for severe sepsis from November 1, 2004, to February 28, 2005.
Patients were eligible if they fulfilled the criteria for severe sepsis.
Vesa Lund, MD, PhD‡ RESULTS: Severe sepsis was found in 470 patients. Laboratory samples were
obtained after informed consent from 250 patients at study entry (day 0) and from
Seppo Hovilehto, MD§ 215 patients after 72 h. These samples were compared with samples from 30
healthy individuals. The ICU mortality was 13.2% and hospital mortality 26%.
Esko Ruokonen, MD, PhD储 Median serum VEGF concentrations on day 0 were 423 pg/mL (interquartile range
[IQR] 159 and 858 pg/mL), and after 72 h were 521 pg/mL (IQR 182 and 1092
For the Finnsepsis Study Group pg/mL), which were both higher than in healthy controls (P ⫽ 0.029 and 0.003,
respectively). Low VEGF concentrations were associated with more severe renal
and hematological dysfunction (Sequential Organ Failure Assessment scores 3– 4
compared with scores 0 –2). VEGF concentrations in day 0 and after 72 h were
lower in nonsurvivors (P ⫽ 0.01 and ⬍0.01, respectively) than in survivors, but the
receiver operating characteristic curve analyses of concentrations of VEGF on day
0 and at 72 h revealed areas under the curve of 0.58 and 0.63 (95% confidence limits
0.48 – 0.68 and 0.54 – 0.72, P ⫽ 0.1 and 0.009, respectively).
CONCLUSIONS: VEGF concentrations are increased in patients with severe sepsis.
Low concentrations are associated with hematological and renal dysfunction.
VEGF concentrations were lower in nonsurvivors than in survivors, but did not
adequately predict hospital mortality in patients with severe sepsis.
(Anesth Analg 2008;106:1820 –6)
Vol. 106, No. 6, June 2008 © 2008 International Anesthesia Research Society 1821
Table 1. Comparison of Vascular Endothelial Growth Factor (VEGF) Study Cohort with the Rest of Finnsepsis Study Patients
Other Finnsepsis
VEGF patients patients P
No. of patients 250 220
Age 59.4 (15.6) 59.8 (14.8) 0.81
Male 172 (68.8%) 143 (65%) 0.31
APACHE IIa 23.8 (9.1) 24.6 (9.0) 0.37
SAPS IIb 43.5 (16.8) 46.0 (16.9) 0.16
SOFA on day 1c 8.3 (3.6) 8.5 (3.6) 0.83
SOFAmaxd 10.9 (4.3) 10.4 (4.4) 0.31
Septic shocke 173 (69.2%) 162 (73.6%) 0.29
Infection type
Community acquired 161 (64.4%) 116 (52.7%) 0.035
Hospital acquired 84 (33.6%) 97 (44.1%) 0.035
Source of infection
Pulmonary 102 (40.8%) 96 (43.6%) 0.53
Intraabdominal 82 (32.8%) 68 (30.9%) 0.66
Skin or soft tissue 24 (9.6%) 24 (10.9%) 0.60
Urinary tract 11 (4.4%) 11 (5.0%) 0.76
Other 35 (14.0%) 28 (12.7%) 0.69
Blood culture positive 70 (28.0%) 58 (26.4%) 0.69
Postoperative admission 65 (26.0%) 71 (32.3%) 0.15
ICU mortality 33 (13.2%) 40 (18.2%) 0.14
Hospital mortality 66 (26.4) 67 (30.1) 0.39
Data are presented as absolute values (percentages) or means (standard deviation).
a
Acute Physiology and Chronic Health Evaluation.
b
Simplified Acute Physiology Score.
c
Sequential organ failure assessment score (SOFA) on the day following study entry.
d
Maximum SOFA score.
e
Cardiovascular failure with SOFA score 3– 4.
1822 Vascular Endothelial Growth Factor in Severe Sepsis ANESTHESIA & ANALGESIA
Table 2. Vascular Endothelial Growth Factor (VEGF) Concentrations in Dysfunction of Different Organ Systems
N SOFAmax 0–2 N SOFAmax 3–4 P
Cardiovascular 57 374 (138 and 849) 172 418 (153 and 828) ns
Respiratory 70 330 (167 and 868) 159 457 (148 and 826) ns
Hematological 168 548 (222 and 917) 61 148 (54 and 361) 0.00
Renal 180 456 (190 and 866) 49 281 (61 and 675) ⬍0.02
Hepatic 96 510 (185 and 883) 5 55 (32 and 320) ⬍0.02
Concentrations (pg/mL) are presented as medians (interquartile range, IQR). The P value refers to VEGF concentrations on day 0 between different severities of organ dysfunction (SOFAmax 0 –2
and SOFAmax 3– 4).
dysfunction was not associated with VEGF concentra- The change in VEGF concentration (␦VEGF ⫽ VEGF
tions (P ⫽ 0.66 and 0.14 and P ⫽ 0.94 and 0.40, 72 h ⫺ VEGF day 0) was positive in 56.7% (n ⫽ 122) and
respectively). In contrast, VEGF concentrations were negative in 41.9% (n ⫽ 90) of the 215 patients in whom
decreased in patients with the most severe degrees of ␦VEGF could be determined. The change in SOFA score
renal, hematological, or liver dysfunction. Accord- (␦SOFA, maximum SOFA score ⫺ SOFA score at admis-
ingly, VEGF levels were associated with the severity sion) is presented in Figure 2. There was no correlation
of overall organ dysfunctions: the patients in the highest with ␦VEGF and change in SOFA scores on the days the
maximum SOFA score quartile (SOFA max score 14 –24) blood samples were taken (P ⫽ 0.24). The onset of severe
had lower VEGF levels on day 0 (165 pg/mL [IQR 54 sepsis before hospital admission could be estimated in
and 466] vs 494 pg/mL [IQR 192 and 904], P ⫽ 0.00) and 149 of 161 patients with community-acquired infections
72 h later (208 pg/mL [IQR 52 and 339] vs 626 pg/mL (⬍12 h, 12–24 h, 24 –36 h, 36 – 48 h, ⬎48 h). A significant
[IQR 262 and 1042], P ⫽ 0.00) than patients with SOFA correlation (⫺0.20 by Spearman, P ⫽ 0.022) between the
max scores between 8 and 13. VEGF concentrations on change in VEGF concentrations and the different onset
day 0 were lower in patients with positive blood cultures times of patients with community-acquired severe sepsis
than in patients with negative blood cultures (median was found. There was also a significant association
274 pg/mL [IQR 82 and 720] vs 495 pg/mL [IQR 184 and between the change in VEGF concentrations and sur-
868], P ⫽ 0.03). There were no differences at 72 h (P ⫽ vival (⫺0.24 by Spearman, P ⫽ 0.016): The survivors had
0.11). No association was found between platelet count a more positive ␦VEGF (median 103 pg/mL, IQR ⫺90
and VEGF levels on day 0 or at 72 h (P ⫽ 0.17 and 0.32, and 413) than nonsurvivors (median ⫺2.0 pg/mL, IQR
respectively). ⫺178.5 to 125.5, P ⫽ 0.037 for the difference).
Vol. 106, No. 6, June 2008 © 2008 International Anesthesia Research Society 1823
DISCUSSION studies.24 Interestingly, in the present study, VEGF
The main finding of this study was that, although levels were very low in five patients with severe
VEGF concentrations are increased in severe sepsis, hepatic failure. Yano et al. found that heart, liver, and
low VEGF levels are associated with more severe kidney were major sources of sepsis-induced VEGF
forms of organ dysfunction and mortality. Signifi- production in animal models of sepsis.17 One can
cantly lower VEGF concentrations in nonsurvivors do therefore speculate that existing septic organ failure
not, however, predict hospital mortality. Low circulat- may result in reduced VEGF response. Similar results
ing VEGF levels are associated with hematological have been reported by Grad et al., who showed that
and renal dysfunction, suggesting that VEGF produc- low VEGF concentrations were associated with an
tion in severe sepsis may be disturbed. increased occurrence of complications (sepsis, ARDS,
Previous studies have suggested that VEGF is an and multiple organ failure) in polytrauma and burn
important mediator of inflammation and that high patients.3
VEGF concentrations correlate with the severity of Our investigation has a few limitations. The
organ dysfunction.4,17,24 Our results are not in agree- samples were taken only at two time points and
ment with these previous results: low serum VEGF patients must have been at different phases in the
concentrations were associated with high maximum course of sepsis. Onset of sepsis can only be estimated,
SOFA scores and poor outcome. VEGF increases en- and T0 was the day severe sepsis was diagnosed, and
dothelial permeability.22 Indeed, it has been shown the patient was enrolled in the study. The half-life of
that VEGF is significantly elevated in ARDS patients VEGF has been reported to be short: 33.7 ⫾ 13 min
compared with ventilated patients without ARDS.25 based on pharmacokinetic studies with recombinant
Mura et al. have shown in animal acute lung injury VEGF.27 Nonetheless, previous studies have shown
that the VEGF expression may depend on the stage of increased VEGF levels up to 29 days, indicating sus-
the disease. The acute inflammatory response may tained VEGF production in patients with severe sep-
first release VEGF into the alveolar space and increase sis.17 Our VEGF concentrations were measured in
vascular permeability, but later epithelial injury may serum samples. Plasma samples have been preferred
reduce the expression of VEGF and its receptors and by some investigators because platelet-mediated se-
lead to cell death.15 We did not find any relation cretion of VEGF during the clotting process could
between the severity of septic lung injury (severe interfere with the results.28,29 However, it has been
oxygenation impairment with a SOFA score 3– 4) and shown recently that VEGF concentrations were corre-
high serum VEGF concentrations. lated between plasma and serum in paired samples in
A different time pattern may partly explain differ- otherwise healthy women having controlled ovarian
ences between the studies.4,17,24 Yano et al. found that hyperstimulation. Serum values were higher than
peak VEGF concentrations occur in the first 24 h and plasma values, the factor being about six-fold (Spear-
VEGF levels remain elevated up to several days.17 In mann correlation coefficient 0.61, P ⬍ 0.005).30 It is
an earlier study by van der Flier et al.,24 18 patients possible that in severe sepsis serum and plasma levels
with severe sepsis had a peak VEGF concentration on are not directly correlated, which may explain differ-
the first day after severe sepsis criteria were met. In ent results seen in previous studies with smaller
our study, the unselected cohort of severe sepsis patient groups. It is also possible that VEGF release
patients was larger than in previous studies, and more from platelets may have influenced our results. On the
than half of our patients (56.7%) had further increas- other hand, one of our main findings was low serum
ing VEGF concentrations over the first 72 h. VEGF concentrations in nonsurvivors with identical
Severe sepsis is a continuum of dynamic processes, sampling and sample processing. Although VEGF
from systemic inflammation response syndrome to concentrations were not associated with platelet
sepsis, severe sepsis, and septic shock. Even though all counts in our study, it is possible that low VEGF
patients included in this study fulfilled the criteria of concentrations can reflect more diminished release
severe sepsis, the severity of illness varies widely. In than disturbed production in patients with the most
our opinion, there are insufficient data to define which severe forms of sepsis. The VEGF analyses were made
time interval should be used for the analysis of VEGF; in all studies with the same ELISA method (Quan-
for this study, three days were used. tikine威), which detects the soluble VEGF121 and
Low VEGF concentrations were associated with VEGF 165 isoforms.
hematological and renal dysfunction in our study. It In conclusion, VEGF levels are elevated in severe
has been shown that VEGF production correlates with sepsis, but are lower in nonsurvivors. Low VEGF
platelet count in cancer patients after chemotherapy- concentrations are associated with severe hemato-
induced thrombocytopenia, and that a platelet re- logical and renal dysfunction, possibly indicating
bound is followed by a peak in VEGF production.26 disturbed VEGF production during severe sepsis.
However, the platelet counts for the time points at Clinically, VEGF does not seem to be a useful tool
which the samples were taken did not correlate with for the prediction of an unfavorable outcome in the
VEFG levels in our patients or in one of the earlier critical care setting.
1824 Vascular Endothelial Growth Factor in Severe Sepsis ANESTHESIA & ANALGESIA
Appendix. Participants of the Finnsepsis Study
Hospital Investigator
Satakunta Central Hospital Hospital Dr. Vesa Lund
Savonlinna Central Hospital Dr. Markku Suvela
Central Finland Central Hospital Dr. Raili Laru-Sompa
Mikkeli Central Hospital Dr. Heikki Laine
Pohjois-Karjala Central Hospital Dr. Sari Karlsson
Seinäjoki Central Hospital Dr. Kari Saarinen
Etela-Karjala Central Hospital Dr. Seppo Hovilehto
Päijät-Häme Central Hospital Dr. Pekka Loisa
Kainuu Central Hospital Dr. Tuula Korhonen
Vaasa Central Hospital Dr. Pentti Kairi
Kanta-Häme Central Hospital Dr. Ari Alaspää
Lappi Central Hospital Dr. Outi Kiviniemi
Keski-Pohjanmaa Central Hospital Dr. Tadeusz Kaminski
Kymenlaakso Central Hospital Dr. Jussi Pentti, Dr. Seija Alila
Helsinki University Hospital Dr. Ville Pettilä, Dr. Marjut Varpula, Dr. Marja Hynninen
Helsinki University Hospital (Jorvi) Dr. Tero Varpula
Helsinki University Hospital (Peijas) Dr. Rita Linko
Tampere University Hospital Dr. Esko Ruokonen, Dr. Pertti Arvola
Kuopio University Hospital Dr. Ilkka Parviainen
Oulu University Hospital Dr. Tero Ala-Kokko, Dr. Jouko Laurila
Länsi-Pohja Central Hospital Dr. Jorma Heikkinen
Vol. 106, No. 6, June 2008 © 2008 International Anesthesia Research Society 1825
24. van der Flier M, van Leeuwen HJ, van Kessel KP, Kimpen JL, 28. Webb NJ, Bottomley MJ, Watson CJ, Brenchley PE. Vascular
Hoepelman AI, Geelen SP. Plasma vascular endothelial growth endothelial growth factor (VEGF) is released from platelets
factor in severe sepsis. Shock 2005;23:35– 8 during blood clotting: implications for measurement of circulating
25. Thickett DR, Amstrong L, Christie SJ, Millar AB. Vascular VEGF levels in clinical disease. Clin Sci (Lond) 1998;94:395–404
endothelial growth factor may contribute to increased vascular 29. Kusumanto YH, Dam WA, Hospers GAP, Meijer C, Mulder NH.
permeability in acute respiratory distress syndrome. Am J Platelets and granulocytes, in particular the neutrophils, form
Respir Crit Care Med 2001;164:1601–5 important compartments for circulating vascular endothelial
26. Verheul HMW, Hoekman K, Luykx-de Bakker S, Eekman CA, growth factor. Angiogenesis 2003;6:283–7
Folman CC, Broxterman HJ, Pinedo HM. Platelet: transporter of 30. Manau D, Fábregues F, Peñarrubia J, Creus Montserrat Creus,
vascular endothelial growth factor. Clin Cancer Res 1997;3:
Carmona F, Casals G, Jiménez W, Balasch J. Vascular endothe-
2187–90
lial growth factor levels in serum and plasma from patients
27. Eppler SM, Combs DJ, Henry TD, Lopez JJ, Ellis SG, Yi JH,
Annex BH, McCluskey ER, Zioncheck TF. A target-mediated undergoing controlled ovarian hyperstimulation for IVF. Hum
model to describe the pharmacokinetics and hemodynamic Reprod 2007;22:669 –75
effects of recombinant human vascular endothelial growth
factor in humans. Clin Pharmacol Ther 2002;72:20 –32
1826 Vascular Endothelial Growth Factor in Severe Sepsis ANESTHESIA & ANALGESIA