Attenuation of sepsis-related immunoparalysis by continuous veno-

venous hemofiltration in experimental porcine pancreatitis

Emre F. Yekebas, MD; Claus F. Eisenberger, MD; Henning Ohnesorge, MD; Armin Saalmüller, MD;
Holger-Andreas Elsner, MD; Madelaine Engelhardt, MD; Andrea Gillesen, MD; Jan Meins, MD;
Marcel The, MD; Tim Strate, MD; Christoph Busch, MD; Wolfram T. Knoefel, MD; Christian Bloechle, MD;
Jakob R. Izbicki, MD

Objectives: In light of evidence suggesting that hemofiltration
favorably influences septic diseases by removing sepsis media-
tors, the impact of different modalities of continuous veno-venous
hemofiltration (CVVH) on outcome and immunologic derange-
ments in porcine pancreatogenic sepsis was evaluated.
Design: Randomized, controlled intervention trial.
Subjects: Forty-eight minipigs of either sex.
Interventions: Pancreatitis was induced by intraductal injec-
tion of sodium taurocholate (4%, 1 mL/kg body weight [BW]) and
enterokinase (2 U/kg BW). Animals were allocated either to un-
treated controls— group 1— or to one of three treatment
groups— group 2: low-volume CVVH (20 mL/kg BW), no change of
hemofilters; group 3: low-volume CVVH, filters changed every 12
hrs; and group 4: high-volume CVVH (100 mL/kg BW), filters
changed every 12 hrs. Survival represented the major parameter
of the study. Serum cytokine levels, sepsis-related down-regula-
tion of major histocompatibility complex II and CD14 expression
on leukocytes, bacterial translocation, and endotoxemia were
further parameters evaluated in the study.
Measurements and Main Results: High-volume CVVH combined with
periodic filter change was significantly superior compared with less
intensive treatment modalities (low-volume CVVH, no filter change) in
sepsis protection. Long-term survival (>60 hrs) was found in 67% of
group 4 and 33% of group 3 animals (p < .05), whereas in controls and
group 2 no animal survived. CVVH ameliorated the initial serum tumor
necrosis factor-␣ response and prevented sepsis-induced in vitro endo-
toxin hyporesponsiveness. Down-regulation of major histocompatibility
complex II and CD14 expression on monocytes was significantly im-
proved by CVVH. Improved oxidative burst and phagocytosis capacity in
polymorphonuclear leukocytes suggested that leukocyte function was
stabilized by CVVH. Also, CVVH significantly reduced bacterial translo-
cation and endotoxemia.
Conclusions: Hemofiltration reversed sepsis-induced immuno-
paralysis in a porcine model of bile acid–induced pancreatitis.
Implications for human pancreatitis must be validated in prospec-
tive, clinical protocols. (Crit Care Med 2001; 29:1423–1430)
KEY WORDS: swine; physiopathology of pancreatitis; mediators
of inflammation; tumor-necrosis-factor; cytokines; sepsis syn-
drome; immunoparalysis; immunology of monocytes; immunology
of neutrophils; bacterial translocation

S evere pancreatitis leading to sep-
sis-related multiple organ dys-
function is a devastating disease
that induces severe alterations
of the immune system and triggers the
activation of numerous inflammatory cas-
From the Departments of Surgery (EFY, CFE, ME,
AG, JM, MT, TS, CB, WTK, CB, JRI), Anesthesiology
(HO), and Medical Microbiology and Immunology (HAE),
University Hospital Eppendorf, Hamburg, Germany; and
the Federal Research Center for Virus Diseases of
Animals, Tuebingen, Germany (AM).
Supported, in part, by grants of the “Forschungs-
und Studienstiftung der Vereinigung Nordwestdeut-
scher Chirurgen,” “Verein zur Förderung der chirurgis-
chen Forschung am UKE e.V.,” and Hospal Medizin-
technik GmbH, Nürnberg, Germany.
Address requests for reprints to: Jakob R. Izbicki,
MD, Abteilung für Allgemeinchirurgie, Universitätsk-
rankenhaus Eppendorf, Martinistr. 52, D-20246 Ham-
burg, Deutschland. E-mail: izbicki@uke.uni-ham-
burg.de
Copyright © 2001 by Lippincott Williams & Wilkins
cades. The final pathway of pancreatogenic
sepsis shares many immunologic charac-
teristics with other septic diseases, includ-
ing the excessive release of cytokines (1– 4)
and functional leukocyte disorders (5, 6).
Treatments have been suggested aiming
symptomatically at the nonselective elimi-
nation of sepsis mediators thought to be
associated with systemic complications af-
ter the onset of pancreatitis. This includes
continuous veno-venous hemofiltration
(CVVH), reported to be of considerable ben-
efit in the treatment of multiple organ dys-
function secondary to sepsis (7, 8) as well as
to severe pancreatitis (9, 10). This benefit of
CVVH has been ascribed to its potential in
removing sepsis mediators such as cyto-
kines (9, 11, 12), activated complement fac-
tors (7, 9), and platelet activating factor
(13).
In a pilot study, we proved CVVH-
induced improvement of several pancre-
atitis-related derangements in a porcine
model of necrotizing pancreatitis (14).
However, because the severity of the
model lead to an eventually lethal “toxin
shock,” CVVH failed to result in long-
term survival. In contrast to that prelim-
inary study, in the present series an ex-
perimental setting was established that
resulted in sepsis-associated multiple or-
gan failure secondary to the onset of the
disease, thus mimicking the clinical sit-
uation in humans better than previous
models. Based on this new setting, we
examined whether CVVH, performed in
different modalities, would result in de-
finitive survival and provide protection
from sepsis. A postulated loss of efficiency
of filter membranes resulting from their
long-term application (13) was assessed
by periodic change of hemofilters. Vari-
ous filtration rates were applied to eval-
uate whether increasing plasma turnover

Crit Care Med 2001 Vol. 29, No. 7 1423

results in better outcome as reported in
clinical trials (15). To examine whether
CVVH favorably influences immunologic
disorders, serum levels of pro- and anti-
inflammatory cytokines and sepsis-
induced down-regulation of the mono-
cyte antigens major histocompatibility
complex (MHC) II and CD14 were mea-
sured. The immunologic screening in-
cluded oxidant burst and phagocytosis ca-
pacity of polymorphonuclear leukocytes
(PMNs), which are known to be impaired
in septic diseases (16, 17). Further, we
assessed whether CVVH prevents endo-
toxin hyporesponsiveness, which is a
common problem in advanced sepsis.
MATERIALS AND METHODS
Anesthesia and Surgical Preparation. The
study was approved by the Animal Care Com-
mittee of the University of Hamburg. Forty-
eight fasted minipigs (body weight [BW]
21–30 kg) were premedicated intramuscularly
with flunitrazepam (0.1 mg/kg) and atropine
(0.06 mg/kg). Adequate anesthetic depth was
achieved by continuous intravenous applica-
tion of propofol (6 mg/kg/hr) and fentanyl (10
␮g/kg/hr). After laparotomy, the pancreatic
duct was cannulated by a 5-Fr umbilical vein
catheter. Before closing the laparotomy, two
silicon drainage tubes were placed in the ab-
dominal cavity for sampling peritoneal secre-
tions for microbiological analysis. After the
instrumentation of the animals by arterial and
pulmonary catheters, mean arterial blood
pressure (MAP), central venous pressure, and
heart rate (HR) were monitored continuously.
Total peripheral resistance (TPR) and cardiac
index (CI) were calculated intermittently.
Hemofiltration. Zero-balanced CVVH was
performed in a predilution mode using a poly-
acrylonitrile membrane (AN 69S, Prisma M
60, Hospal Medizintechnik, Nürnberg, Ger-
many) connected to a continuous blood pump
(Prisma, Hospal Medizintechnik). Ultrafiltrate
rates were 20 mL/hr/kg BW in groups 2 and 3,
and 100 mL/hr/kg in group 4. To provide in-
formation about the elimination of pancreati-
tis-related cytokines (tumor necrosis factor
[TNF]-␣, interleukin [IL]-10, and transform-
ing growth factor [TGF]-␤1), sieving coeffi-
cients (SC) were calculated with the following
equation: SC ⫽ 2 Cf /(Ci ⴙ Co), where Cf, Ci,
and Co, respectively, represent concentration
in hemofiltrate, concentration in “inflow”
(prefilter), and concentration in “outflow”
(postfilter) catheters, respectively.
Induction of Pancreatitis and Biometric
Design. Pancreatitis was induced by pressure-
controlled (⬍20 mm Hg), intraductal infusion
of sodium taurocholate (4%, 1 mL/kg BW,
Sigma Chemical, Deisenhofen, Germany) and
enterokinase (2 U/kg BW, Sigma Chemical).
Control animals (n ⫽ 12, group 1) underwent
the spontaneous course of the disease without
1424
any treatment. In three treatment groups, dif-
ferent modalities of CVVH were applied after a
decline of the total peripheral resistance of
30%. Group 2 animals (n ⫽ 12) underwent
CVVH without changing hemofilters. In
groups 3 and 4 (n ⫽ 12 each), hemofilters
were changed every 12 hrs. Groups 2 and 3
underwent a filtration turnover of 20 mL/kg/
hr; group 4, representing the most intensive
treatment modality, underwent a filtration
turnover of 100 mL/kg/hr. After a maximal
observation period of 60 hrs, animals were
killed.
Definition of Sepsis. Modifying generally
accepted criteria (18, 19), the occurrence of
sepsis was assumed when bacteremia proved
by positive blood cultures was associated with
at least two of the following signs: 1) body
temperature ⬎38.5°C or ⬍36°C in (nonhemo-
filtrated) controls and an increase of ⬎1.5°C
compared with baseline values in treatment
groups, respectively, because preliminary ex-
periments had shown a CVVH-related decrease
of body temperature of 1.4°C (range, 1.1–1.6;
n ⫽ 5); 2) heart rate ⬎120 beats/min; and 3)
altered white blood cell count of ⬎12,000 or
⬍4000 cells/mm3.
Serum Lipase, C-Reactive Protein,
Trypsinogen Activation Peptides, and White
Blood Cell Count. Lipase levels were measured
using a commercial assay kit (Delta Test Assay
for pancreatic lipase, Sigma Chemical). Plas-
matic C-reactive protein (CRP) concentrations
were detected with an enzyme-linked immu-
nosorbent assay (ELISA) as previously de-
scribed (20). Urinary trypsinogen activation
peptide (TAP) measurements were performed
with an ELISA (Biotrin, Sinsheim, Germany).
White blood cell (WBC) counts were per-
formed using a Coulter counter (Coulter Elec-
tronics, Krefeld, Germany).
Cytokines. All cytokines evaluated in this
study were assessed both in serum (pre- and
postfilter) and in hemofiltrate. The concentra-
tions of TNF-␣ (pig TNF-␣, Endogen, Eching,
Germany), and IL-10 (pig IL-10, BioSource,
Camarillo, CA) were measured by swine-
specific ELISA kits. TGF-␤1, known to show
high homology between the human and por-
cine protein (21), was detected using a human
ELISA (Quantikine, R&D Systems, Wiesbaden,
Germany), detecting both human and porcine
TGF-␤1. Preliminary experiments had shown
a cross-reactivity of human and recombinant
porcine TGF-␤1 of 96% (n ⫽ 5).
In Vitro TNF-␣ Production by PMNs. PMNs
were isolated by dextran sedimentation and
Ficoll-Hypaque density gradient sedimenta-
tion as previously described (22). Cell viability
was ⬎95% as assessed by trypan blue exclu-
sion. PMNs were incubated in RPMI 1640 me-
dium (Seromed, Berlin, Germany) supple-
mented with 10% calf serum resulting in a
final count of 2 ⫻ 106 cells/mL. TNF-␣ con-
centrations in supernatants were measured af-
ter exposure of suspensions to lipopolysaccha-
ride (5 ␮g/mL Salmonella abortus equi, Sigma
Chemical) for 4 hrs at 37°C.
Microbiological Evaluation. Processing of
swabs for microbiological analysis and differ-
entiation of isolates was performed using stan-
dard techniques classifying microbiological
samples as positive or negative. The following
subsets were assessed: 1) Gram-negative, en-
dotoxin-containing Enterobacteriaceae/non-
fermentative rods (including Escherichia coli,
Proteus mirabilis, and Klebsiella pneu-
moniae); 2) Gram-positive cocci; and 3) Gram-
negative or -positive anaerobes.
Flow Cytometric Analysis—Antibodies
and Conjugates for Immunophenotyping. An-
ti-MHC class II (monoclonal antibody MSA3,
contributed by Dr. Saalmüller, Tuebingen,
Germany) to assess monocyte and T-cell MHC
II expression (T-cells not reported); anti-CD14
(My 4, Coulter Immunotech Diagnostics,
Krefeld, Germany) to assess endotoxin recep-
tor expression on monocytes; and anti-SWC3
(Swine Workshop Cluster 1, 8/1a3, contrib-
uted by Dr. Glatthaar, Reutlingen, Germany),
a specific porcine antigen without a human
CD analog to differentiate SWC3⫺ lympho-
cytes from SWC3⫹ monocytes and granulo-
cytes, were used.
Staining of cells for two-color flow cyto-
metric analysis was performed in a two-step
procedure: 1) incubation of whole peripheral
blood (100 ␮L) with the respective primary
antibodies (50 ␮L); and 2) incubation with the
respective isotype-specific conjugates. Cells
were shielded from light at 4°C before analy-
sis. Analysis was performed on a flow cytom-
eter with a four-decade, 1024-channel, loga-
rithmic amplifier (FACS-Calibur, Becton
Dickinson, Heidelberg, Germany). A mini-
mum of 15,000 events was analyzed for each
sample.
Phagocytosis Activity and Oxidant Burst.
The quantification of phagocytic and respira-
tory burst activity in PMNs to assess pancre-
atitis-related alterations of leukocyte function
was performed using commercially available
specific test kits (PHAGOTEST, PHAGO-
BURST, Orpegen Pharma, Heidelberg, Ger-
many).
For measurement of phagocytosis activity
in PMNs, whole blood aliquots (100 ␮L) were
incubated with a fluorescein isothiocyanate–
stained E. coli suspension (2– 4 ⫻ 106/␮L) for
10 mins at 37°C resulting in a bacteria/
neutrophil ratio of 5:1. Fluorescence-activated
cell-sorter analysis was performed within 60
mins.
The assessment of respiratory burst activ-
ity is based on the measurable oxidation of the
substrate dihydrorhodamine (DHR 123) in the
test reflecting the capability of PMNs to gen-
erate oxygen radicals. Whole blood samples
were activated with 20 ng/mL phorbol myris-
tate acetate or formyl-methionyl-leucyl-
phenylalanine.
Statistics. Data are reported as mean ⫾ SD.
Survival times were calculated by Kaplan-
Meier analysis and compared by the log-rank
test. Differences of baseline values vs. changes
of parameters after pancreatitis were evaluated
Crit Care Med 2001 Vol. 29, No. 7
by analysis of variance for repeated measures.
Differences between treatment groups were
analyzed by analysis of variance, followed by
the Scheffé test to assess significance. Differ-
ences in incidences of bacterial growth were
compared using Fisher’s exact test. Correla-
tion analysis was performed using the Pearson
correlation method. A p value ⬍ .05 was con-
sidered statistically significant.
RESULTS
Clinical Data. After the onset of pan-
creatitis, group 1 (control) animals ex-
hibited an early phase, hyperdynamic re-
sponse characterized by increases in CI,
HR (not demonstrated), and body tem-
perature, and decreases in MAP and TPR
(Table 1). In the late phase of septic
macrocirculatory derangements, a dra-
matic breakdown of the entire macrocir-
culation and a decrease in body temper-
ature occurred. The major reason for this
was a progressive cardiac insufficiency in-
dicated by a decrease in CI. In treatment
groups, the median time interval after
which animals had to be connected to
CVVH according to the experimental pro-
tocol (decrease in TPR of 30%) was 14.2
hrs (range, 11.4 –16.6 hrs). CVVH led in
all treatment groups to a reversal of he-
modynamic impairment that resulted
eventually in significantly prolonged sur-
vival and, in a total of 12 animals (group
3: n ⫽ 4; group 4: n ⫽ 8), in definitive
long-term (60 hrs) survival. Both the ini-
tial elevation of body temperature up to
almost 41°C and the hypothermia in the
late course of experiments were signifi-
cantly ameliorated by CVVH. The change
of hemofilters (group 3) and, notably, ad-
ditional increase of filtration rate (group
4) were distinctly superior in preventing
sepsis-related hemodynamic impairment
compared with group 2. This resulted in a
definitive prevention of sepsis signs by
CVVH in a total of ten animals, seven of
which belonged to group 4 and three to
group 3. No animals from group 2 sur-
vived.
Laboratory Parameters
Lipase, TAP, CRP, and WBC. The ac-
tivities of lipase in blood serum were be-
low 50 U/L before the induction of pan-
creatitis. Slight differences between
groups in the rise of lipase activities up
1500 U/L postinduction were not signifi-
cant. Comparable findings were observed
concerning urinary TAP concentrations
increasing from baseline values ⬍1
nmol/L to peak values ⱕ230 nmol/L
without considerable differences between
the experimental groups (data not
shown).
Before the induction of pancreatitis,
CRP concentrations were ⬍8 mg/L. In
controls, pancreatitis resulted in a 20-
fold increase in CRP concentrations
(baseline: 7.2 ⫾ 4.4 mg/L; 24 hrs postin-
duction: 148 ⫾ 31 mg/L). CVVH weak-
ened distinctly the CRP response. In
group 2, this attenuation was only tran-
sient and CRP levels increased in the later
course of experiments (48 hrs postinduc-
tion: 139 ⫾ 45 mg/L). In groups 3 and 4,
even 60 hrs postinduction, CRP concen-
trations were considerably lower than
those detected 24 hrs postinduction in
controls (group 3: 81 ⫾ 31; group 4: 72 ⫾
28).
WBC counts in controls showed a
characteristic biphasic course with an
early phase increase from 8.2 ⫾ 1.8 ⫻
103/␮L (baseline) to 15.2 ⫾ 4.3 ⫻ 103/␮L
(24 hrs) preceding severe leukopenia (3.4
⫾ 1.5 ⫻ 103/␮L, 36 hrs). Although peak
and nadir values were reached later, the
same tendency was observed in group 2
(baseline: 7.9 ⫾ 1.5 ⫻ 103/␮L; 36 hrs:
14.8 ⫾ 2.0 ⫻ 103/␮L; 48 hrs: 3.0 ⫾ 1.9 ⫻
103/␮L). In animals undergoing filter
change, the increase in circulating WBC
peaked 48 hrs postinduction (group 3:
13.9 ⫾ 2.4 ⫻ 103/␮L; group 4: 13.6 ⫾ 2.6
⫻ 103/␮L). In group 3, the drop in WBC
counts observed at the end of the exper-
iments was significantly attenuated (6.0

Table 1. Clinical outcome parameters

Parameter Group No. Baseline 12 hrs 24 hrs 36 hrs 48 hrs 60 hrs

Survival rate, % (survivors/ 1 100 100 83 (10/2) 33 (4/8) 0 (0/12) NC

nonsurvivors)
2 100 100 100 (12/0) 67 (8/4) 25 (3/9) 0 (0/12)a
3 100 100 92 (11/1) 92 (11/1) 50 (6/6) 33 (4/8)a,b
4 100 100 100 (12/0) 100 (12/0) 75 (9/3) 67 (8/4)a,b,c
MAP, mm Hg 1 97 ⫾ 12 72 ⫾ 14d 44 ⫾ 21d 36 ⫾ 18 NC NC
2 96 ⫾ 14 68 ⫾ 16d 86 ⫾ 18a 59 ⫾ 17d 40 ⫾ 19d NC
3 101 ⫾ 13 74 ⫾ 13d 85 ⫾ 17a,e 74 ⫾ 16e 62 ⫾ 17b,d 47 ⫾ 18d
4 99 ⫾ 15 71 ⫾ 15d 92 ⫾ 15a 91 ⫾ 18b,c 80 ⫾ 18b,c,d 66 ⫾ 18c,d
CI, L/min䡠m2 1 4.5 ⫾ 1.1 6.2 ⫾ 1.5d 9.8 ⫾ 2.8d 2.8 ⫾ 2.4 NC NC
2 4.2 ⫾ 1.5 6.0 ⫾ 1.4d 6.3 ⫾ 1.7a,d 11.8 ⫾ 2.8d 3.1 ⫾ 2.8 NC
3 4.4 ⫾ 1.2 5.8 ⫾ 1.6d 6.6 ⫾ 1.9a 5.9 ⫾ 2.5b,e 6.0 ⫾ 3.5b,e 7.6 ⫾ 3.7d
4 4.8 ⫾ 1.3 6.5 ⫾ 1.7d 5.9 ⫾ 1.9a,b,e 5.6 ⫾ 2.1b 5.4 ⫾ 2.6b 5.8 ⫾ 3.0c
TPR, dyn䡠sec䡠cm⫺5 1 1875 ⫾ 366 1485 ⫾ 429e 695 ⫾ 291d 574 ⫾ 203d NC NC
2 2025 ⫾ 214 1428 ⫾ 345e 1529 ⫾ 351a,e 820 ⫾ 385d 673 ⫾ 350d NC
3 1942 ⫾ 295 1549 ⫾ 262e 1673 ⫾ 273a 1495 ⫾ 462a,b,e 1148 ⫾ 429b,e 749 ⫾ 376d
4 1815 ⫾ 392 1334 ⫾ 385d 1592 ⫾ 304a 1631 ⫾ 481a,b 1521 ⫾ 458b,c,e 1111 ⫾ 317c,d
BT, °C 1 37.3 ⫾ 0.9 38.4 ⫾ 1.6d 40.9 ⫾ 1.4d 35.4 ⫾ 1.5 NC NC
2 37.0 ⫾ 0.7 38.6 ⫾ 1.2e 37.8 ⫾ 1.5a 39.3 ⫾ 2.1d 35.2 ⫾ 1.4d NC
3 37.5 ⫾ 0.9 38.9 ⫾ 1.3e 37.5 ⫾ 1.7a 37.7 ⫾ 1.3b 37.8 ⫾ 2.0b 37.4 ⫾ 2.9
4 37.2 ⫾ 0.6 38.3 ⫾ 0.9e 37.1 ⫾ 1.4a 37.2 ⫾ 1.5b 37.3 ⫾ 1.5b 38.0 ⫾ 1.9

Survival rates, mean arterial pressure (MAP), cardiac index (CI), total peripheral resistance (TPR), and body temperature (BT) after pancreatitis (mean ⫾
SD). Group 1: controls; group 2: continuous veno-venous hemofiltration (CVVH) without change of hemofilters; group 3: low-volume CVVH (20 mL/kg body
weight [BW]/hr) with change of hemofilters every 12 hrs; group 4: high-volume CVVH (100 mL/kg BW) with change of hemofilters every 12 hrs; NC, not
calculated (no survival).
a
p ⬍ .05 vs. the respective value in controls; bp ⬍ .05 vs. the respective value in group 2; cp ⬍ .05 vs. the respective difference in group 3; dp ⬍ .01
and ep ⬍ .05 vs. baseline values, respectively.

Crit Care Med 2001 Vol. 29, No. 7 1425

⫾ 4.3 ⫻ 103/␮L); in group 4, WBC counts
remained elevated (11.9 ⫾ 4.3 ⫻ 103/␮L,
60 hrs postinduction). The subgroup of
nonseptic animals in groups 3 and 4 (n ⫽
10) had significantly higher WBC counts
than the total of 38 septic animals at the
last period of individual observation (11.2
⫾ 4.0 vs. 3.7 ⫾ 3.3 ⫻ 103/␮L; p ⬍ .01).
TNF-␣. Pancreatitis resulted in a tre-
mendous increase of TNF-␣ concentra-
tions, peaking 24 hrs postinduction in
controls. The subgroup of four control
animals that were still alive 36 hrs postin-
duction showed a reversal of the serum
TNF-␣ response (Fig. 1A). CVVH signifi-
cantly attenuated the increase of TNF-␣
levels in serum. In group 2, this effect
was temporarily limited, resulting in ele-
vated TNF-␣ levels 36 hrs postinduction.
Changing hemofilters (group 3) and,
Figure 1. Plasma tumor necrosis factor (TNF)-␣ levels (A) and in vitro TNF-␣ generation by endotoxin-
stimulated polymorphonuclear leukocytes (PMNs) obtained at different points of time (B). Definitive
protection from overwhelming TNF-␣ response was only achieved by intensified continuous veno-
venous hemofiltration (CVVH). Attenuation of endotoxin hyporesponsiveness was most effective in
animals undergoing high-volume CVVH. LPS, lipopolysaccharide. *p ⬍ .01 vs. 0 hrs; #p ⬍ .05 vs.
control; §p ⬍ .05 vs. group 2; $p ⬍ .05 vs. group 3.
1426
more effective, increasing the filtration
rate (group 4), led to an arrest of TNF-␣
response (⬍200 pg/mL) that remained, in
contrast to group 2, definitively sup-
pressed compared with peak values in
controls. Furthermore, the attenuation of
TNF-␣ response in treatment groups was
reciprocally associated with the removal
of TNF-␣ in hemofiltrate (R ⫽ ⫺0.83, p
⫽ .004). Highest hemofiltrate levels of
TNF-␣, and thereby highest sieving coef-
ficients, were found in group 4 (data not
shown). The late-phase decrease in
TNF-␣ serum concentrations in controls
and, although in a delayed fashion, in
group 2 was consistent with in vitro ex-
periments showing progressive decrease
in TNF-␣ generation from isolated, endo-
toxin-challenged PMNs in these groups
(Fig. 1B). Taking into account the results
of endotoxin measurements (see below),
the final decrease in TNF-␣ generation in
groups 1 and 2 suggested endotoxin hy-
poresponsiveness. This was either atten-
uated or even completely prevented in
animals undergoing filter change in
groups 3 and 4.
TGF-␤1 and IL-10. Serum concentra-
tions of TGF-␤1 (11 ⫾ 7 to 19 ⫾ 8 pg/mL
preinduction) peaked 24 hrs postinduc-
tion and ranged from 846 ⫾ 238 (group
2) to 1102 ⫾ 415 pg/mL (group 4). In the
further course, TGF-␤1 serum levels de-
creased gradually. Despite considerable
removal in hemofiltrate indicated by high
sieving coefficients (0.51– 0.59) and sig-
nificant gradients of pre- and postfilter
concentrations, TGF-␤1 levels in CVVH-
treated groups tended to be rather ele-
vated when compared with untreated
controls (not significant). IL-10 levels in
controls showed an overwhelming pri-
mary increase, but decreased in the later
course of experiments (Fig. 2). Sieving
coefficients of IL-10 were comparable
with those of TGF-␤1 (range, 0.48 – 0.55,
24 hrs postinduction). Interestingly, with
more effective IL-10 removal in hemofil-
trate (group 4 ⬎ group 3 ⬎ group 2),
serum IL-10 concentrations increased. In
the late experimental course, long-term
survivors protected from sepsis (n ⫽ 10)
had significantly higher IL-10 levels com-
pared with final values assessed in the
total of 38 septic animals (134 ⫾ 41 vs. 62
⫾ 30 pg/mL; p ⬍ .01).
Microbiology and Endotoxemia. None
of blood and ascites samples collected im-
mediately after instrumentation showed
bacterial growth. The near 100% inci-
dence of positive blood and ascites cul-
tures 24 hrs postinduction in controls
was significantly weakened by CVVH, re-
sulting in positive cultures of 43% of
ascites cultures (15/35) and 31% of blood
cultures (11/35) (p ⬍ .001 vs. controls for
each). High-volume CVVH reduced bac-
teremia better compared with groups 2
and 3 (data not shown). The microbiolog-
ical differentiation revealed a broad spec-
trum of bacteria with slightly elevated
prevalence (not significant) for E. coli,
Morganella morganii, Pseudomonas
aeruginosa, and Gram-negative anaer-
obes (i.e., Bacteroides fragilis). Bacteria
found in positive blood cultures were
without exception simultaneously
present in ascites, too.
CVVH succeeded in significantly sup-
pressing the considerable increase of en-
dotoxin levels (Fig. 3). Both the periodic
change of hemofilters and the increase of
Crit Care Med 2001 Vol. 29, No. 7
 Phagocytosis and Oxidative Burst Ac-
tivity in PMNs. PMNs obtained from con-
trols showed initially a marked priming
response (maximum at 24 hrs) for phor-
bol myristate acetate–stimulated oxida-
tive burst activity when compared with
baseline PMNs (Fig. 5A). In contrast,
PMNs obtained at 36 hrs postinduction
demonstrated a breakdown of respiratory
burst orchestrated by decreased phagocy-
tosis of fluorescein isothiocyanate–
stained E. coli. Comparable findings were
observed in group 2 animals, although
both the initial enhancement and end-
stage breakdown of oxidative burst and
Figure 2. Plasma interleukin (IL)-10 concentrations. Continuous veno-venous hemofiltration (CVVH) PMN-related phagocytosis were delayed.
resulted in significantly increased IL-10 levels that peaked 24 hrs postinduction. Intensified CVVH with In animals undergoing intensified CVVH
filtrate turnover of 100 mL/kg body weight/hr arrested IL-10 levels at these levels, whereas those in (groups 3 and 4), this pattern of func-
animals undergoing low filtrate turnover, notably when filters were not changed, decreased in the tional alterations in PMNs was not ob-
further course of the observation period. *p ⬍ .01 vs. 0 hrs; #p ⬍ .05 vs. control; §p ⬍ .05 vs. group served. The change of hemofilters and,
2; $p ⬍ .05 vs. group 3. more effective, the increase of filtration
rate succeeded, in distinct contrast to
group 2, in attenuating the deterioration
of PMN function in the later course of
experiments.

DISCUSSION
Significant evidence has accumulated
from clinical studies suggesting that sec-
ondary multiple organ failure resulting
from septic complications is the cardinal
cause of mortality in severe pancreatitis.
Progress in the field of intensive care
therapy has led to the suggestion that
patients who seem to be at risk to develop
severe, systemic complications after the
onset of pancreatitis might profit from
Figure 3. Plasma endotoxin concentrations. Endotoxin levels surged during the first 24 hrs of the “early” application of CVVH (10). This
study, reaching peak levels by 36 hrs. Increase of endotoxin levels was significantly attenuated by implies a hypothetical indication of CVVH
continuous veno-venous hemofiltration (CVVH). Definitive prevention was only achieved by CVVH irrespective of the presence of acute renal
performed in high-volume modality. *p ⬍ .01 vs. 0 hrs; #p ⬍ .05 vs. control; §p ⬍ .05 vs. group 2; $p failure. The impetus of early application
⬍ .05 vs. group 3. of CVVH in nonanuric patients is thereby
to eliminate overwhelmingly released in-
flammatory mediators supposed to initi-
filtration rate weakened endotoxemia at genic sepsis on MHC class II antigen ex- ate a sequela of pathophysiologic de-
significantly lower levels compared with pression on monocytes are shown in rangements that finally results in sepsis-
group 2. Moreover, maximal endotoxin Figure 4. CVVH was highly effective in induced acute renal failure, uremia, and
levels in the entirety of ten nonseptic delaying or even preventing MHC class II multiple organ failure.
animals from groups 3 and 4 were signif- down-regulation. CD14 down-regulation In the present study, we investigated
icantly lower than those assessed in the on monocytes occurred in a similar fash- whether CVVH, performed in different
total of 38 septic animals (41 ⫾ 18 vs. 79 ion (data not shown). In controls, CD14 modalities, would prevent sepsis-induced
⫾ 27 pg/mL; p ⬍ .05). Preliminary exper- expression decreased from 68% (baseline) derangements following pancreatitis. Be-
iments (n ⫽ 4) performed to explore to 34% (36 hrs postinduction). This ten- cause a model of experimental pancreato-
whether endotoxin would be filtered or dency, which was only delayed in group 2, genic sepsis does not exist, we first estab-
adsorbed by the filter membranes re- was almost completely prevented in lished a new pancreatitis model sharing
vealed that the recovery of exogenous en- group 4 animals (baseline: 65%; 60 hrs many parallels with severe human pan-
dotoxin (5 ␮g/mL) was ⬎90%. Also, en- postinduction: 54%). Moreover, a signif- creatitis, e.g., the nonbacterial onset of
dotoxin was not detectable in any of the icant, reciprocal correlation of CD14 the disease preceding secondary bacterial
hemofiltrate samples. down-regulation on monocytes and the translocation and endotoxemia, a hyper-
Changes in Leukocyte Immunophe- degree of endotoxemia was found (R ⫽ dynamic circulatory response, and char-
notypes. The adverse effects of pancreato- ⫺0.82, p ⫽ .007). acteristic laboratory and immunologic

Crit Care Med 2001 Vol. 29, No. 7 1427

 C
ontinuous veno-
venous hemofil-
tration distinctly
improved outcome and de-
layed occurrence of sepsis in
experimental pancreato-
genic sepsis.

Figure 4. Major histocompatibility complex (MHC) II antigen expression in monocytes assessed by flow
cytometry. A total of 15,000 cells were analyzed for each sample. Down-regulation of MHC II
direct correlation between cumulative ul-
expression ⬍40% compared with baseline values indicating severe immunoparalysis occurred in
controls and, although being significantly delayed, in group 2 animals undergoing continuous
trafiltrate volume and survival rate in pa-
veno-venous hemofiltration (CVVH) without filter change. This was significantly improved when filters tients with sepsis-associated acute renal
were periodically changed. In animals undergoing high-volume CVVH (group 4), MHC II expression failure (15), we report significantly im-
was arrested at approximately control levels. *p ⬍ .05 vs. 0 hrs; #p ⬍ .05 vs. control; §p ⬍ .05 vs. group proved outcome and even considerable
2. long-term survival in animals undergo-
ing high-volume CVVH representing the
most intensified treatment modality per-
formed in this study. Furthermore, our
data provide evidence for a linkage be-
tween duration of filter application and
efficiency of CVVH. Although it has been
previously hypothesized that there may
be a need for periodic changes of hemo-
filters to prevent progressive loss of effi-
cacy of CVVH (13), this assumption has
not been proven in clinical or experimen-
tal studies to date. The significant im-
provement of outcome in animals under-
going filter change identifies the
prevention of filter “fouling,” in addition
to a sufficient filtration rate, as a second,
mandatory requirement for the efficacy of
CVVH.
The observation that both the eleva-
tion of CRP concentrations in plasma and
alterations of WBC counts were signifi-
cantly suppressed by CVVH is of great
importance because it reflects the capa-
bility of CVVH in attenuating the sys-
temic inflammatory response in the
course of pancreatitis. With regard to leu-
kocyte function, this attenuation was
Figure 5. Oxidative burst (A) assessed by oxidation of dihydrorhodamine 123 to rhodamine 123 by eventually orchestrated by improvement
polymorphonuclear leukocytes (PMNs) activated with 20 ng/mL phorbol myristate acetate (PMA) and of both oxidative burst and phagocytosis
phagocytosis capacity (B) for opsonized, fluorescein isothiocyanate–stained Escherichia coli (E. coli) capacity in PMNs. In contrast, CVVH did
(2– 4 ⫻ 106/␮L) in PMNs detected by flow cytometry (15,000 cells). The biphasic pattern of oxidative not have any impact on urinary TAP lev-
burst in response to PMA in controls and group 2 was attenuated by intensified continuous veno- els assayed as a parameter for pancreatic
venous hemofiltration (CVVH). Comparably, phagocytosis was arrested near baseline levels in animals necrosis and, therefore, being helpful in
undergoing high-volume CVVH, whereas low-volume CVVH was less effective, notably when filters re-
predicting the local severity of acute pan-
mained unchanged. *p ⬍ .05 vs. 0 hrs; #p ⬍ .05 vs. control; §p ⬍ .05 vs. group 2; $p ⬍ .05 vs. group 3.
creatitis (23, 24).
Previous studies suggest that the local
features found frequently in clinical sep- icantly more effective prevention of sep- pancreatic damage triggers the release of
sis. sis-related hemodynamic deterioration both proinflammatory (TNF-␣) and anti-
Of particular interest is the finding than does low-volume CVVH. In accor- inflammatory (IL-10 and TGF-␤) cyto-
that high-volume CVVH results in signif- dance with clinical studies suggesting a kines in the course of the disease (25).

1428 Crit Care Med 2001 Vol. 29, No. 7

TNF-␣ is known to be of crucial impor-
tance in mediating systemic effects of
septic diseases, e.g., necrotizing pancre-
atitis. It was of striking interest that the
potential of CVVH in decreasing TNF-␣
levels was temporarily limited when only
one filter was used, whereas periodic fil-
ter change as well as the increase in
plasma turnover prevented excessive
TNF-␣ levels more effectively. Also, the
drop in TNF-␣ concentrations to near
control levels in advanced sepsis was ab-
sent in nonseptic animals, most of which
were treated by intensified, high-volume
CVVH. In contrast, sepsis prevention was
associated with an arrest of TNF-␣ re-
sponse at elevated levels. This seemed to
be paradoxic, as the final, sepsis-related
TNF-␣ decrease was orchestrated by sig-
nificantly higher endotoxin levels than
those in nonseptic animals. As endotoxin
is known to be one of the major stimula-
tors of TNF-␣ generation, the dissocia-
tion of TNF-␣ decrease and exorbitant
endotoxin levels therefore suggested en-
dotoxin hyporesponsiveness in advanced
sepsis. This assumption was strengthened
by in vitro experiments showing sepsis-
induced, disturbed TNF-␣ generation by
PMNs subjected to exogenous endotoxin.
CVVH attenuated or completely pre-
vented disturbed TNF-␣ generation in en-
dotoxin-stimulated PMNs.
Moreover, CVVH had considerable ef-
fects on serum levels of the anti-inflamma-
tory cytokines IL-10 and TGF-␤. Numerous
reports underline the role of anti-inflam-
matory cytokines, notably of IL-10, in both
sepsis and pancreatitis (25–28). In the
present series, IL-10 and TGF-␤ levels mea-
sured in controls and CVVH-treated ani-
mals developing sepsis showed a character-
istic biphasic course: after an
overwhelming primary response, serum
levels approximated baseline levels in the
later course of sepsis. In contrast, animals
protected from sepsis by CVVH showed nei-
ther this exorbitant primary increase nor a
final drop in IL-10 and TGF-␤ levels. Also,
in nonseptic animals, IL-10 and TGF-␤
were removed from hemofiltrate in higher
concentrations than in septic animals.
Hence, we assume that CVVH restrained
up-regulated anti-inflammatory cytokine
response as long as it prevented sepsis.
Further evidence that CVVH improved
immunologic alterations was obtained
from the screening of monocyte activa-
tion markers. We particularly focused on
the expression of the MHC II antigen
corresponding to human leukocyte anti-
gen-DR and the endotoxin receptor
Crit Care Med 2001 Vol. 29, No. 7
CD14. CVVH delayed, in instances of sep-
sis protection over the whole observation
period, and even definitively suppressed
down-regulation of MHC II known to oc-
cur in fatal, sepsis-induced immunopa-
ralysis (29 –32). In accord with studies
reporting a reduction of CD14 receptor
density during sepsis (33), CVVH also at-
tenuated CD14 down-regulation as long
as it prevented sepsis. Most interestingly,
the latter finding was obviously the result
of the close, reciprocal association of
CD14 down-regulation with the extent of
endotoxemia.
Finally, CVVH obviously improved the
mucosa barrier and thereby restrained both
endotoxemia and bacteremia. This was
caused by mechanisms different from filtra-
tion/adsorption processes as evidenced,
first, by the finding that in no hemofiltrate
sample endotoxin was detectable. Further-
more, preliminary experiments had re-
vealed that the filters used in our setting
are not able to remove endotoxin. The find-
ing that bacteria assessed in positive blood
cultures were, without exception, also
present in ascites suggests that CVVH, by
improving splanchnic perfusion, protects
the bloodstream from bacterial invasion
from the abdominal cavity.
In summary, CVVH distinctly im-
proved outcome and delayed occur-
rence of sepsis in experimental pancre-
atogenic sepsis. Intensifying CVVH by
periodically changing hemofilters and
increasing filtration rate resulted in
considerable sepsis protection and
long-term survival. These findings
identify both the integrity of filter
membranes and a sufficient filtration
rate to be essential for the efficiency of
CVVH. These results beg further exam-
ination in a prospective, clinical trial.
However, as most patients recover
spontaneously without interventional
and/or surgical treatments, the “pro-
phylactic” application of CVVH in all
patients suffering from pancreatitis
cannot be recommended. Evidence for
necrosis and clinical signs of a septic
course of the disease might be suitable
tools to discriminate patients who
might profit from CVVH from those in
whom it would represent overtreat-
ment.
REFERENCES
1. Curley P, Nestor M, Collins K, et al: De-
creased interleukin-2 production in murine
acute pancreatitis: potential for immuno-
modulation. Gastroenterology 1996; 110:
583–588
2. Norman J, Franz M, Messina J, et al: Inter-
leukin-1 receptor antagonist decreases sever-
ity of acute experimental pancreatitis.
Surgery 1994; 117:648 – 655
3. Heath DL, Cruickshank DH, Gudgeon M, et
al: Role of interleukin-6 in mediating the
acute phase protein response and potential as
an early means in acute pancreatitis.
Pancreas 1993; 66:41– 45
4. Gross V, Andreesen R, Leser HG, et al: Inter-
leukin-8 and neutrophil activation in acute
pancreatitis. Eur J Clin Invest 1992; 22:
200 –203
5. Richter A, Nebe T, Kattermann R, et al: Im-
mune paralysis in acute pancreatitis—
HLA-DR antigen expression on CD14⫹DR⫹
monocytes. Langenbecks Arch Chir 1996;
381:38 – 41
6. Widdison AL, Karanjia ND, Alvarez C, et al:
Reticuloendothelial function and efficacy of
levamisole for the treatment of pancreatic
infection in acute necrotizing pancreatitis.
Am J Surg 1992; 163:100 –104
7. Gotloib L, Barzilay E, Shustak A, et al: He-
mofiltration in septic ARDS. The artificial
kidney as an artificial endocrine lung.
Resuscitation 1986; 13:123–132
8. Gomez A, Wang R, Unruh H, et al: Hemofil-
tration reverses left ventricular dysfunction
during sepsis in dogs. Anesthesiology 1990;
73:671– 685
9. Blinzler L, Hausser J, Bödeker H, et al: Con-
servative treatment of severe necrotizing
pancreatitis using early continuous veno-
venous hemofiltration. Contrib Nephrol
1991; 93:234 –236
10. Gebhardt C, Bödeker H, Blinzler L, et al:
Wandel in der Therapie der schweren akuten
Pankreatitis. Chirurg 1994; 65:33– 41
11. Lee PA, Matson JR, Pryor RW, et al: Con-
tinuos arteriovenous hemofiltration therapy
for Staphylococcus aureus-induced septice-
mia in immature swine. Crit Care Med 1993;
21:914 –924
12. Bellomo R, Tipping P, Boyce N: Continuous
veno-venous hemofiltration with dialysis re-
moves cytokines from the circulation of sep-
tic patients. Crit Care Med 1993; 21:522–526
13. Ronco C, Tetta C, Lupi A, et al: Removal of
platelet-activating factor in experimental
continuous arteriovenous hemofiltration.
Crit Care Med 1995; 23:99 –107
14. Yekebas EF, Treede H, Knoefel WT, et al:
Influence of zero-balanced hemofiltration on
the course of severe experimental pancreati-
tis in pigs. Ann Surg 1999; 229:514 –522
15. Storck M, Hartl WH, Zimmerer E, et al:
Comparison of pump-driven and spontane-
ous continuous haemofiltration in postoper-
ative acute renal failure. Lancet 1991; 337:
452– 455
16. Kono K, Serikawa T, Matsumoto Y: Influence
of surgical stress on monocytes and compli-
cations of infection in patients with esopha-
geal cancer—monocyte HLA-DR antigen ex-
1429
 pression and respiratory burst capacity.
J Surg Res 1995; 58:275–280
17. Windsor ACJ, Walsh CJ, Mullen PG, et al:
Tumor necrosis factor-␣ blockade prevents
neutrophil CD18 receptor upregulation and
attenuates acute lung injury in porcine sep-
sis without inhibition of neutrophil oxygen
radical generation. J Clin Invest 1993; 91:
1459 –1468
18. American College of Chest Physicians/
Society of Critical Care Medicine Consensus
Conference: Definition for sepsis and organ
failure and guidelines for the use of innova-
tive therapies in sepsis. Crit Care Med 1992;
1992:20:864 – 874
19. Bone RC: Toward an epidemiology and nat-
ural history of SIRS (systemic inflammatory
response syndrome). JAMA 1992; 268:
3452–3455
20. Schroedl W: Diagnostische Verwendbarkeit
ausgewählter immunologischer Parameter
in Seren von Schweinen und Pferden. Doc-
toral thesis. University of Leipzig, Leipzig,
Germany, 1994.
21. Murtaugh MP: Porcine cytokines. Vet Immu-
nol Immunopathol 1994; 43:37– 44
1430
22. Boyum A: Isolation of mononuclear cells and
granulocytes from human blood. Scand
J Clin Lab Invest 1968; 97:77– 89
23. Karanjia ND, Widdison AL, Jehanli A, et al:
Assay of trypsinogen activation in the cat
experimental model of acute pancreatitis.
Pancreas 1993; 8:189 –195
24. Gudgeon AM, Heath DI, Hurley P, et al:
Trypsinogen activation peptides assay in the
early prediction of severity of acute pancre-
atitis. Lancet 1990; 335:4 – 8
25. Kusske AM, Rongione AJ, Ashley SW, et al:
Interleukin 10 (IL-10) prevents death in le-
thal necrotizing pancreatitis in mice. Sur-
gery 1996; 120:284 –288
26. Bean AGD, Freiberg RA, Andrade S, et al:
Interleukin 10 protects mice against staphy-
lococcal enterotoxin B-induced lethal shock.
Infect Immun 1993; 61:4937– 4939
27. Howard M, Muchamel T, Andrade S, et al:
Interleukin 10 protects mice from lethal
endotoxaemia. J Exp Med 1993; 177:
1205–1208
28. Van Laethem JV, Marchant A, Delvaux A, et
al: Interleukin 10 prevents necrosis in mu-
rine experimental pancreatitis. Gastroenter-
ology 1995; 108:1917–1922
29. Hershman MJ, Cheadle WG, Wellhausen SR,
et al: Monocyte HLA-DR antigen expression
characterizes clinical outcome in the trauma
patient. Br J Surg 1990; 77:204 –207
30. Birkenmaier C, Hong YS, Horn JK: Modula-
tion of the endotoxin receptor (CD14) in sep-
tic patients. J Trauma 1991; 32:473– 479
31. Lin RY, Astiz ME, Saxon JC, et al: Altered
leukocyte immunophenotypes in septic
shock. Studies of HLA-DR, CD11b, CD14,
and IL-2R expression. Chest 1993; 104:
847– 853
32. Volk HD, Reinke P, Krausch D, et al: Mono-
cyte deactivation—Rationale for a new ther-
apeutic strategy in sepsis. Intensive Care
Med 1996; 22:474 – 481
33. Larsen NE, Sullivan R: Interaction between
endotoxin and human monocytes: Charac-
teristics of the binding of 3H-labeled lipo-
polysaccharide and 51C-labeled lipid A before
and after the induction of endotoxin toler-
ance. Proc Natl Acad Sci U S A 1984; 81:
3491–3495
Crit Care Med 2001 Vol. 29, No. 7