Diss Schindler

Stefanie Schindler

Optimization and novel applications of the in vitro

pyrogen test (IPT) using human whole blood

Dissertation

Universität Konstanz

November 2005

Optimization and novel applications of the in vitro

pyrogen test (IPT) using human whole blood

Dissertation

zur Erlangung des akademischen Grades

des Doktors der Naturwissenschaften

an der Universität Konstanz (Fachbereich Biologie)

vorgelegt von

Stefanie Schindler

Tag der mündlichen Prüfung: 23. 01. 2006

Referenten: Prof. Dr. Dr. T. Hartung

Prof. Dr. A. Wendel

PD Dr. Bert-André Zucker

http://www.ub.uni-konstanz.de/kops/volltexte/2006/1948/

LIST OF PUBLICATIONS

List of publications:

Major parts of this thesis are published or submitted for publication:

Schindler, S., and Hartung, T. (2005).

Development, validation and applications of the in vitro pyrogen test (IPT)

based on human whole blood (submitted to J Clin Immunol)

Hoffmann, S., Peterbauer., A., Schindler, S., Fennrich, S., Poole, S., Mistry,

Y., Montag-Lessing, T., Spreitzer, I., Loschner, B., van Aalderen, M., Bos, R.,

Gommer, M., Nibbeling, R., Werner-Felmayer, G., Loitzl, P., Jungi, T., Brcic,

M., Brugger, P., Frey, E., Bowe, G., Casado, J., Coecke, S., de Lange, J.,

Mogster, B., Naess, L. M., Aaberge, I. S., Wendel, A., and Hartung, T., 2005.

International validation of novel pyrogen tests based on human monocytoid

cells. J Immunol Methods 298:161-73

Schindler, S.; Asmus, S., von Aulock, S., Wendel, A., Hartung, T., and

Fennrich, S., 2004. Cryopreservation of human whole blood for pyrogenicity

testing. J Immunol Methods 294: 89-100

Schindler, S., Rosenberg, U., Schlote, D., Panse, K., Kempe, A., Fennrich,

S., and Hartung, T., 2005. Application of the InVitro Pyrogen Test (IPT) based

on cryopreserved human whole blood for lipidic parenterals (Pharmeuropa, in

print)

Schindler, S., Spreitzer, I., Hoffmann, S., Hennes, K., Halder, M.,

Brügger, B., Frey, E., Montag-Lessing, T., Löschner, B., Poole, S., and

Hartung T., 2005. International validation of pyrogen tests based

on cryopreserved human primary blood cells (J Immunol Methods, in print)

LIST OF PUBLICATIONS

Further publications, not integrated into this thesis:

Schindler, S., Bristow, A., Cartmell, T., Hartung, T., and Fennrich, S. 2003.

Comparison of the reactivity of human and rabbit blood towards pyrogenic

stimuli. ALTEX 20: 59-63

Schindler, S., and Hartung, T., 2002. Comparison and validation of novel

pyrogen tests based on the human fever reaction.

Dev Biol 111: 181-6

Mazzotti, F., Beuttler, J., Zeller, R., Fink, U., Schindler, S., Wendel, A.,

Hartung, T. and von Aulock, S. 2006. In vitro Pyrogen Test - a new test

method for solid medical devices. J Biomed Mat Res: Part A, in print

Spreer, A., Gerber, J., Hanssen, M., Schindler, S., Hermann, C., Lange, P.,

Eiffert, H., and Nau, R. 2006. Dexamethasone increases hippocampal

neuronal apoptosis in a rabbit model of Escherichia coli meningitis. Pediatric

Research, in print

ACKNOWLEDGEMENTS

Acknowledgements

The work presented here was carried out between January 2002 and March

2005 at the chair of Biochemical Pharmacology at the University of Konstanz

under the supervision of Prof. Dr. Dr. Thomas Hartung.

I especially want to thank my supervisor Thomas Hartung for his advice, his

confidence, and for the excellent working facilities.

Also special thanks to Dr. Stefan Fennrich for his encouragement and

friendship.

Many thanks go to Prof. Dr. Albrecht Wendel for welcoming me in his

department.

Special thanks to the pyrogen team; Ina Seuffert for the introduction into the

topic, Gregor Pinski, Ilona Kindinger and Silvia Asmus for their friendship and

never tiring support, their valuable ideas and their threats of physical violence

should I ever give up on this project.

I want to thank all my lab colleagues for their continuous help. My special

thanks in this respect goes to Dr. Sonja von Aulock and to Dr. Sebastian

Hoffmann.

I thank all members of the “Lehrstuhl Wendel” for their support, their

contribution to the outstanding working atmosphere and for five wonderful

years.

And last but not least I thank my parents Klaus and Birgit Schindler.

ABBREVIATIONS

Abbreviations

AAMI American Association of Medical

Instrumentation

AWIPT absorb and wash in vitro pyrogen test

BAL bronchoalveolar lavage

BET bacterial endotoxin test

cAMP cyclic adenosinmonophospate

CD cluster of differentiation

CFU colony forming unit

COX cyclooxygenase

CRP C-reactive protein

CV coefficient of variation

DIC disseminated intravascular

coagulation

DL developing laboratory

DMSO dimethyl sulfoxide

ECVAM European Center for the Validation of

Alternative Methods

ELC endotoxin limit concentration

ELISA enzyme-linked immunosorbent assay

EU endotoxin unit

GLP good laboratory practice

HD hemodialysis

HSA human serum albumin

ICE interleukin-1 converting enzyme

IL interleukin

IPT In Vitro Pyrogen Test

IU International Unit

NIBSC National Institute for Biological

Standards and Controls

LAL Limulus Amoebocyte Lysate

LBP LPS-binding protein

ABBREVIATIONS

LoD Limit of Detection

LPS lipopolysaccharide

LTA lipoteichoic acid

LVP large volume parenteral

MID minimum interference dilution

MM-6 Monomac-6

MVD maximum valid dilution

PAMPs pathogen-associated molecular

patterns

PBMCs peripheral blood mononuclear cells

PBS phosphate buffered saline

PEI Paul Ehrlich Institute

PG prostaglandin

POD peroxidase

PPC positive product control

NFКB nuclear factor kappa B

NL naive laboratory

NPC negative product control

NSAID non-steroidal anti-inflammatory drug

OD optical density

OVLT organum vasculosum laminae

terminalis

PM prediction model

PMN polymorphonuclear

PTFE polytetrafluorethylene

RT room temperature

SOP standard operating procedure

SVP small volume parenteral

TLR toll-like receptor

TMB tetramethylbenzoate

TNF Tumor Necrosis Factor

USP United States Pharmacopoeia

WBT whole blood test

ABBREVIATIONS

WHO World Health Organisation

TABLE OF CONTENTS

Table of Contents

1. Introduction 1

1.1. Pyrogens 1

1.1.1. Lipopolysaccharide (endotoxin) 1

1.1.2. Non-endotoxin pyrogens 2

1.2. Traditional pyrogen tests 3

1.2.1. Rabbit pyrogen test 3

1.2. 2. Limulus amoebocyte lysate test (LAL) 4

1.3. Mechanism of fever 4

1.4. Cell-based pyrogen tests 5

1.5. Human whole blood test (IPT) 6

2. Aims of the study 7

3. Development, validation and applications of the in vitro

pyrogen test (IPT) based on human whole blood 8

3.1. Abstract 8

3.2. Introduction 9

3.3. Basic principle of the whole blood test 11

3.4. Comparison of the in vitro reaction of the human whole

blood test to rabbit whole blood 13

3.5. Establishment of the IPT as a test for biologicals 13

3.6. Validation

3.7. Development of the commercially available IPT kit 15

3.8. Special adaptations 18

3.9. Conclusion 28

3.10. Appendix

3.11. Acknowledgements 28

TABLE OF CONTENTS

4. International validation of novel pyrogen tests based

on human monocytoid cells 29

4.1. Abstract 30

4.2. Introduction 31

4.3. Materials and Methods 33

4.4. Results 41

4.5. Discussion 50


5. Cryopreservation of human whole blood for

pyrogenicity testing 53

5.1. Abstract 53



5.4. Results 58

5.5. Discussion 71


6. International validation of pyrogen tests based on

cryopreserved human primary blood cells 74

6.1. Abstract 75



6.4. Results 83

6.5. Discussion 92

TABLE OF CONTENTS

7. Pyrogen testing of lipidic parenterals with a novel

in vitro test 95

7.1. Abstract 96



7.4. Results 101

7.5. Discussion 110

7.6. Conclusion 112

8. Summarizing discussion 113

9. Summary 118

10. Zusammenfassung 119

11. References 120

INTRODUCTION

1

1 Introduction

1.1 Pyrogens

The term “pyrogen” derives from the greek word “pyros” (fire). Pyrogens are

therefore substances that have been recognized to cause fever in the

organism. The relation of bacteria and fever was first recognized by

Semmelweis (1) and Lister (2). The association of fever and intravenous

injection, on the other hand, dates back to the eighteenth century, when van

Haller noticed that the injection of putrid materials caused severe fever

reactions (3). Panum, with the help of Virchow, was the first to state that the

substance responsible was heat-stable, water-soluble, alcohol-insoluble, and

independent of the presence of living bacteria (4). The term “pyrogen” was

apparently used first by Billbroth (5).

At the end of the 19th century, Centanni first reproducibly isolated a

substance from a variety of Gram-negative bacteria which he called

pyrotoxina, which was most probably the first purified endotoxin in history (6).

Injection fevers associated with intravenously applied parenterals were first

systematically investigated by Hort and Penfold in 1912 (7), who injected

them intravenously into the rabbit, measured the fever reaction, and classified

the bacteria into pyrogenic and non-pyrogenic. Basically, these were the first

rabbit pyrogen tests. Seibert then proved that the fever reactions were

caused by filterable, heat-stable pyrogens from Gram-negative bacteria, a

finding which was later confirmed by Rademaker, who already stressed the

importance of avoiding contaminations in parenterals and differentiated

between the terms “sterile” and “pyrogen-free” (8, 9). World War II then

brought the development of large volume parenterals as volume substitution

for injured soldiers. The occurrence of severe fever reactions resulted in a

collaborative study establishing the rabbit pyrogen test (10, 11) and its

incorporation into the US Pharmcopoeia in 1942.

1.1.1. Lipopolysaccharide (endotoxin)

Endotoxin as a component of the cell walls of Gram-negative bacteria is the

INTRODUCTION

2

most potent and the most extensively studied pyrogen. Due to the fact that

Gram-negative bacteria are ubiquitous, contaminations of parenterals with

endotoxin pose a constant threat to the health of patients. Endotoxins are

released from the cell not only after lysis, but are shed constantly from the

living bacterium as well (12). Lipopolysaccharide (LPS) is a highly purified

(protein-free) form of endotoxin. Chemically, they are heat-stable substances

with three distinct regions: the lipid A portion, which has been shown to be

responsible for the pyrogenic activity (13, 14), the core polysaccharide, and

the antigenic O-specific side chain. The biological activities of endotoxins do

not restrict themselves to causing fever and other inflammatory reactions, but

also include complement activation, hypotension, and activation of the

coagulation system, all of which can lead to severe complications, up to

hypovolemic shock, disseminated intravascular coagulation (DIC) and death.

A maximum endotoxin contamination of 50 pg/ml (0.5 ng/kg) was first

published by the Bureau of Drugs in 1980 (15), apparently with no scientific

study having been performed to confirm this very restrictive threshold. In

2005, a study at the Paul-Ehrlich Institute (PEI) in Germany fully confirmed

this limit (16).

1.1.2. Non-endotoxin pyrogens

Substances that have pyrogenic properties but are not of an endotoxin nature

include enterotoxins (17, 18), exotoxins, (19), viruses (20), peptidoglycan (21-

23) and fungi (24, 25). Since Gram-positive bacteria are as frequent as

Gram-negative bacteria, the pyrogens of the former can be a serious health

hazard as well. A major component of the Gram-positive cell wall is the

peptidoglycan, which consists of β-1,4 linked N-acetyl-D- glucosamin and N-

acetyl muramic acid, and was shown to have pyrogenic properties similar to

those of endotoxin (21). The other prominent pyrogen of Gram-positive

bacteria called lipoteichoic acid (LTA) was successfully purified in an

endotoxin-free and biologically active manner in 2001 (26).

INTRODUCTION

3

1.2. Traditional pyrogen tests

1.2.1. Rabbit pyrogen test

The rabbit pyrogen test has been the gold standard in pyrogen testing since

1942, when it was introduced into the USP (United States Pharmacopoeia).

The rabbit species was chosen by Seibert, who also discovered the

pyrogenic principle (8). In 1941, the need for pyrogen testing of LVP (large

volume parenterals) due to World War II caused the Committee of Revision of

the USP to authorize the first USP collaborative study of pyrogens with

pyrogen filtrates of Pseudomonas aeruginosa. The results of this study led to

the incorporation of the rabbit test in the 12th edition of the USP in 1942. In its

simplest form, the test involves measuring a rise in body temperature for 3

hours following intravenous injection of a test solution into the marginal ear

vein at a volume of not more than 10 ml/kg. Temperature is to be measured

by a clinical thermometer inserted into the rectum of the rabbit at a depth of

not less than 7.5 cm. Rabbit breeds used for testing are New Zealand Whites,

Belgium Whites, Chinchillas and Dutch Belts. Differences in sensitivities of

various strains have been investigated by van Dijck et al (27). Animals of one

single sex are preferred, and there have been reports about male rabbits

being more sensitive to pyrogens than females (28).

The test is positive if the sum of the rises in three rabbits exceeds 2.65 °C.

The rabbit has a labile thermoregulation and tends to give false-positive

results. Also, the very rigid fixation and the handling (injection procedure) can

cause a hyperthermia due to excitement. On the other hand, it has been

reported that the fixation and lack of movement can cause a hypothermia

yielding false-negative results (29). Comparisons between the reactivity of

humans and rabbits in vivo by Greisman 1969 showed that the threshold

towards three endotoxin preparations was comparable, but that the humans

respond more vigorously than the rabbits (30).

INTRODUCTION

4

1.2.1. Limulus amoebocyte lysate test (LAL)

When in contact with the lipid A portion of endotoxin, the amoebocytes from

Limulus polyphemus (horseshoe crab) coagulate due to an enzymatic

reaction (31, 32). In the presence of calcium, the clotting enzyme zymogen is

activated by a serin protease and acts on coagulogen, a clottable protein in

the lysate, producing a smaller clot protein. The clotting can be observed by

turning the tube with the lysate 180° (clot end point LAL) or, in a more

quantitative way, by the turbidimetric LAL, which measures kinetically ranges

of the clotting. The basic principle has been improved on and modified in

many ways (33). A sensitivity of 0.0005 µg/ml was determined by the

developers.

The lysate is prepared by placing the crabs in restraining racks and inserting

a needle through the muscular hinge between the cephalothorax and the

abdominal region. Hemolymph is then drawn from the cardiac chamber into a

container with anticoagulant. After collection, the amoebocytes are

centrifuged and the supernatant is discarded. After 2-3 washing steps, the

cells can be subjected to osmotic shock by adding distilled water and the

intracellular lysate is released. The bled crabs are then thrown back into the

sea, and their survival rate is unknown. In some countries (e.g. Japan) the

crabs are squeezed in a mill.

One of the drawbacks of the LAL is that it only detects endotoxin (34, 35).

The pyrogenic potency of non-endotoxin substances has been recognized

since the 1960s, leaving a safety gap when performing pyrogen tests with the

LAL. Contaminations of drugs with Gram-positive bacteria, fungi or their

fragments/toxins are not an unlikely event.

1.3. Mechanism of fever

The concept of a substance produced in the mammalian organism in

response to pyrogens which is causative in the genesis of fever dates back to

1948 (36). This substance, which was produced by immune cells evoked

fever when injected into healthy rabbits and was then called endogenous (or

INTRODUCTION

5

leukocytic) pyrogen. Dinarello et al. could demonstrate, that this endogenous

pyrogen consisted of two distinct proteins (37), probably pro-Interleukin-1 and

Interleukin-1 (IL-1). Other, similar mediators of fever were found later and

were termed Interleukin-6 (IL-6) and Tumor Necrosis Factor-α (TNF-α).

During a response to pyrogens, they are secreted by a subfraction of the

white blood cells, the monocytes, and are called proinflammatory cytokines. It

is of considerable interest that the receptors recognizing pathogen-associated

molecular patterns (PAMPs) of bacteria, the so-called toll-like receptors (tlr)

shares in its cytoplasmic domain the signaling areas with the IL-1 receptor

(38). Additionally, all pyrogenic cytokines share a common intracellular

pathway which results in the activation of the nuclear factor-κB (NF- κB). The

current understanding of the mechanism of fever in the mammal is that this

transcription factor results in the expression of the enzyme cyclooxygenase-2

(COX-2) which results in prostaglandin (PG) E2 synthesis. Mice deficient in

COX-2 did not develop fevers in response to LPS, IL-1, IL-6 or TNF (39-42).

Specifically one of altogether four PGE2 receptors in the brain, the EPR-3, is

required to develop fever (43, 44), probably via the induction of a second

messenger such as cyclic adenosinmonophosphate (cAMP) (45). That IL-1β

is the most potent fever inducer compared to IL-6 and TNF-α when injected

intravenously into rabbits could be demonstrated (46, 47). These findings

formed the basis for the development of cell-based in vitro assays which are

described in the next chapter.

1.4. Cell-based pyrogen tests

The discovery that white blood cells produce cytokines in a dose-dependent

manner in response to pyrogens led to the development of altogether six in

vitro assays based on primary human blood cells or cell lines. All of them

have the same basic concept of incubating the substance in question at 37°C

with the cells, and, as a second step, measuring the cytokine production (or,

in one case, nitric oxide) by an enzyme-linked immunosorbent assay (ELISA).

Four assays have been successfully validated in an international

collaborative study and are described in detail in the publication of Hoffmann

INTRODUCTION

6

et al. 2005 (48). One of these assays was the human whole blood test (IPT)

whose further development is described here.

1.5. Human whole blood test (IPT)

A new way of measuring pyrogens has been introduced in 1995 by Hartung

and Wendel (49). Basically, fresh heparinized human whole blood is diluted in

physiological saline and brought together with the sample. In the case of

pyrogenicity, the monocytes produce IL-1β in vitro over a period of 10-24

hours at 37°C which can be measured by a specific ELISA the next day. The

test has a detection limit of 0.25 EU/ml and has the advantage that it is

performed with the cells of the relevant species, that is, the human reaction is

tested.

The ELISA (Enzyme-Linked-Immunosorbent Assay) is an assay based on the

reaction of specific antibodies towards an antigen, in this case IL-1β. An

antibody is bound to a microtiter plate with high protein binding capacity; the

pyrogen-stimulated cell supernatant is added to the antibody and the cytokine

binds. After a washing step, a second, labeled detection antibody is added

which also binds to the antigen; the label is in this case biotin, which binds

with high affinity to avidin coupled to POD (horseradish peroxidase). After a

second washing step, substrate, in this case TMB (Tetramethylbenzidine) is

added. The enzymatic reaction of the POD with the TMB changes the color of

the latter from colorless to blue and the antibody-antigen reactions are made

visible.

AIMS OF THE STUDY

7

2 Aims of the study

Pyrogen testing of parenterals has been performed routinely in vivo in the

rabbit since the early 1940s. Recently, a cell-based in vitro alternative has

been developed which aims to replace the rabbit pyrogen test as an

alternative method. The European legislation clearly states that animal testing

is forbidden if there is a viable and validated in vitro alternative available.

Making the human whole blood test (IPT) a standardized and commercially

available alternative to the rabbit was the goal of the following work.

• The first part of this thesis validated the human whole blood test in an

international collaborative study including laboratories from England,

Switzerland, Norway, the Netherlands and Germany and control institutions

such as the Paul-Ehrlich Institute, Germany, and the European Centre for the

Validation of Alternative methods.

• The second part of this thesis standardized the most critical and the

most crucial reagent: the human whole blood. In order to make this highly

varying and perishable component of the assay more reliable and available, a

method for cryopreserving the blood was developed, and a pooling protocol

was found which levels out the interindividual differences of the human

donors.

• As a third step, the whole blood test using the newly developed

cryopreserved blood was validated in an international collaborative study

including three different laboratories.

• The last part extended the application possibilities of the new test

towards testing not hydrophilic, but lipophilic substances in order to avoid

large numbers of animal experiments. The testing for pyrogens in so-called

small volume perenterals, e.g. lipophilic drugs, is obligatory since January

2004 due to a change in European Pharmacopoeia.

DEVELOPMENT, VALIDATION AND APPLICATIONS OF THE IN VITRO

PYROGEN TEST (IPT) BASED ON HUMAN WHOLE BLOOD

8

3 Development, validation and applications of the in vitro

pyrogen test (IPT) based on human whole blood

Stefanie Schindler*, Sonja von Aulock* and Thomas Hartung+ *

* Biochemical Pharmacology, University of Konstanz, Universitätsstr. 10,

D-78457 Konstanz + ECVAM, Institute for Health and Consumer Protection, Joint Research Centre,

European Commission, I-21020 Ispra (VA)

Corresponding author

Thomas Hartung, MD, PhD

ECVAM

Institute for Health and Consumer Protection

Joint Research Centre

European Commission

I-21020 Ispra (VA)

e-mail: [email protected]

Tel: +39-0332-785939

Fax: +39-0332-786297

3.1. Abstract

Microorganisms such as Gram-negative or Gram-positive bacteria, viruses and

fungi contain components that activate the innate immune system. These

components, called pyrogens (Greek: pyros = fire), can occur independently of

viable microorganisms and are a major safety concern in parenterally

administered drugs, since they can cause severe reactions such as fever, organ

failure and shock in the recipient. So far, these drugs have been tested by

injecting them intravenously into rabbits and measuring their fever reaction or

alternatively by the Limulus Amoebocyte Lysate (LAL) test, employing the

coagulation of the hemolymph lysate of Limulus polyphemus. Both tests have

inherent limitations. A new in vitro test based on human whole blood, capable of



9

measuring all pyrogens relevant to the human patient was introduced and

validated recently. This review describes its principle, development, validation

and the wide spectrum of applications, such as for testing of medical devices,

blood products, lipidic parenterals and air quality. This alternative method aims

to replace fully the rabbit pyrogen test.

Key words

In vitro pyrogen test; interleukin-1β; validation study; alternatives to animals

3.2. Introduction

Pyrogens, as fever-inducing substances of microbial origin, can derive from

dead or viable bacteria, viruses or fungi. Therefore, they can occur even in

sterile environments. Contaminations of parenterals with such substances can

induce local or systemic inflammatory reactions in the recipient, intended to

eliminate an invading pathogen, including a rise in body temperature, but also

more severe adverse reactions such as shock, disseminated coagulation, organ

failure and even death. Therefore, the testing of parenterals prior to batch

release is obligatory for manufacturers.

The best-known fever-inducing contaminant is a component of the cell wall of

Gram-negative bacteria, i.e. endotoxin or lipopolysaccharide (LPS). Pyrogenic

components of Gram-positive bacteria are equally important and include

lipoteichoic acid (LTA) (26) and peptidoglycan (21 - 23). Further possible

pyrogenic contaminants are exotoxins (19), enterotoxins (17, 18) , viruses (20),

and fungal components (24, 25).

Classical pyrogen tests

Testing for pyrogens has been a major issue since the appearance of large

volume parenterals in the 1930s. These bore a label claim of being pyrogen-

free as asserted by the rabbit pyrogen test. This drew attention to the need for

an official test procedure for non-pyrogenicity, which was strengthened by the

heavy demand for large volume parenterals in World War II. A collaborative



10

study was initiated to develop the rabbit pyrogen test (10, 11), which led to the

incorporation of the rabbit pyrogen test into the pharmacopoeias. Since then, all

parenterals must be tested for pyrogens. This involves the measurement of the

rabbit’s body temperature after the application of not more than 10 ml/kg

bodyweight of the substance to be tested. The very rigid fixation of the rabbit

and the handling (injection procedure) can cause hyperthermia due to

excitement and therefore lead to false-positive results. On the other hand, it has

been reported that the fixation and lack of movement can cause a hypothermia

yielding false-negative results (29).

In 1964, Levin and Bang published that the hemolymph of the horseshoe crab

Limulus polyphemus coagulates upon contact with endotoxin. This led to the

development of the Limulus amoebocyte lysate (LAL) test, which is employed to

exclude endotoxin contamination in parenteral drugs (31, 32). The Limulus is

collected from beaches, its hemolymph is drawn out by puncture and the

animals are then thrown back into the sea. 10 to 20 percent do not survive the

bleeding procedure (50-52). The mortality associated with collecting, shipping

and handling the animals remains unknown. The LAL has not been able to

replace fully the rabbit test, since it is defined not as a pyrogen test, but as an

endotoxin test, which fails to recognize e.g. Gram-positive or fungal

contaminants, toxoids, or viral antigens. Due to the crucial role of Gram-

negative endotoxin, it was nevertheless possible to substitute most

pharmacopoeial pyrogen testing with a mere endotoxin test. Additionally, the

LAL does not reveal the biological potency of a given endotoxin in the mammal,

which can differ between bacterial strains by a factor of up to 10’000 (53). Most

importantly, however, certain products tested in rabbits cannot be tested in the

LAL, e.g. various biologicals and vaccines, due to interference.

Fever reaction in the mammal

The finding that mammalian immune cells produce an endogenous pyrogen

when in contact with pyrogenic materials dates back to 1948 (36). Bennett et al.

could identify leukocytes as the source of this factor in 1953 (54). The nature of

this substance was further elucidated by Dinarello et al. (37), who identified two



11

distinct proteins, probably the pro- and the mature form of interleukin 1β (IL-1β).

The pyrogenicity of IL-1β, when injected at very low doses into rabbits, was

proven by Dinarello et al. 1991 (46). IL-6 and TNF-α, which were isolated later,

were found to be pyrogenic cytokines as well, though only at higher doses (46,

47).

The current understanding of the mechanism of fever in the mammal, as

reviewed by Dinarello 2004 (55), is that these proinflammatory cytokines bind to

receptors on the blood side of the organum vasculosum laminae terminalis

(OVLT) and initiate the expression of the enzyme cyclooxygenase-2 (COX-2),

which mediates prostaglandin (PG) E2 synthesis. Mice deficient in COX-2 do

not develop fever in response to injection with LPS, IL-1β or IL-6 (39-41).

Specifically one of altogether four PGE2 receptors in the brain, the EPR-3, is

required to develop fever (43), probably via the induction of a second

messenger such as cAMP (45). Thus, the pyrogenic cytokines cause a change

in the set-point of body temperature in the hypothalamus and are therefore the

mediators responsible for initiating the fever reaction. The finding that

monocytes, a subfraction of the white blood cells, secrete proinflammatory

cytokines such as IL-1β upon contact with pyrogenic material was the basis for

the development of the whole blood test as a pyrogen test (49).

3.3. Basic principle of the whole blood test

Blood incubation

The procedure is described in detail by Hoffmann et al. (48). Briefly, freshly

drawn, heparinized human whole blood from a healthy donor is diluted in

physiological, pyrogen-free clinical grade saline and brought together with the

test sample. In response to pyrogens, the monocytes contained in the blood

sample produce proinflammatory cytokines in a dose-dependent manner. The

proinflammatory cytokine IL-1β is measured by ELISA.

ELISA procedure

The IL-1β or IL-6 in the sample is sandwiched between a monoclonal coat

antibody and a polyclonal peroxidase-labeled detection antibody. Unbound



12

material is removed by washing. The peroxidase metabolizes e.g.

tetramethylbenzidine. The reaction is stopped with acid and the optical density

(OD) is measured at 450 nm.

Controls

As an assay control, a dose-response curve of an LPS from E. coli O111: B4 is

performed in parallel in each assay. This LPS is calibrated to the international

WHO reference standard from E. coli O113: H10 (56). The dose-response

curve must contain the concentration 0.5 EU/ml and a negative control. The IL-

1β released in response to the concentration of 0.5 EU/ml must test positive

when compared to a negative control for the experiment to be valid. 0.5 EU/ml

corresponds to 50 pg/ml of the international reference standard and is

considered the threshold endotoxin concentration that causes fever in the most

sensitive rabbit strains. This threshold was confirmed by a study performed at

the Paul-Ehrlich Institute in 2005, which analyzed 171 rabbits (16).

Testing for interference

In order to test for a given substance’s interference with the activity of the

monocytes, samples (pure or diluted) are incubated together with a 0.5 EU/ml

concentration of the LPS dose response curve. The mean OD of the spiked

sample must be within a 50-200% range of the 0.5 EU/ml concentration of the

dose response curve. If this is not the case, the sample has to be diluted until

the interference criteria are met.

Development of the Gram-positive standard lipoteichoic acid (LTA)

LTA from Staphylococcus aureus was first purified in a biologically active and

endotoxin-free quality by Morath et al. (26). Later, the improved purification

procedure was applied to produce LTA from Bacillus subtilis (57). The

successful identification of the purified LTA as a pyrogenic substance, which is

negative in the LAL (57) and therefore represents a pyrogenic principle that is

only recognized by the rabbit pyrogen test and the cell-based assays, led to the



13

inclusion of the Gram-positive standard LTA derived from B. subtilis into the IPT

procedure.

In order to make this method commercially available and replace the rabbit

pyrogen test, the following steps were taken:

3.4. Comparison of the in vitro reaction of human whole blood with

that of rabbit whole blood

Since the human whole blood test (WBT) aims to replace the rabbit pyrogen

test, the sensitivity of both species towards different pyrogenic stimuli was

compared using human and rabbit whole blood. For this, a rabbit whole blood

test was developed which followed the procedure of the human whole blood

incubation in every detail (58). Overall, the IL-1β response of the rabbits

towards different pyrogenic stimuli was comparable to that of humans. In the

case of the Gram-positive stimulus, LTA, the rabbit blood was less sensitive

than human blood, thus confirming the human whole blood test as an equal or

even superior test system to reflect the human response.

3.5. Establishment of the IPT as a test for biologicals

Biologicals, such as protein solutions, cytokines, antibodies, heat shock

proteins, blood coagulation factors and vaccines for intravenous use, pose a

particular problem in pyrogen testing. They can influence the LAL results due to

their characteristics, such as color and viscosity, and they are potentially

immunogenic in the rabbit, causing fever reactions that are independent of

contaminations. In any case, if immunogenic substances are tested, the animals

may only be used once, which results in extremely high costs for the

manufacturers. The IPT does not pose such problems. Some examples of the

application of the IPT for pyrogen testing of such samples are given below.

Control of vaccines

In 2001, new batches of vaccines against early summer meningoencephalitis

were released that caused severe fever reactions in some recipients. Although

they were negative in the LAL, these batches gave a high signal in the WBT.



14

This phenomenon could be shown to be due to the removal of the mercury-

containing additive thiomersal. The additive suppressed the IL-1β response in

the WBT, making it likely that it also suppressed the pyrogenic property of the

attenuated virus in the vaccine in earlier batches (59).

In 2003, Carlin and Viitanen demonstrated that trivalent vaccines (diphtheria,

tetanus and polio), which tested negative in the LAL, were powerful inducers of

IL-6 in 4 out of 8 donors in the whole blood incubation (60). This difference

between the LAL and the WBT could be attributed to the toxoid of

Corynebacterium diphteriae, and, to a lesser extent, to that of Clostridium tetani,

both non-endotoxin pyrogens. (61). Additionally, in both studies, the authors

found pronounced differences in the IL-6 and IL-1β response of different donors

towards the vaccines and their components, although they displayed highly

conserved LPS reactivity. This indicated a more variable interindividual

sensitivity of human donors towards these non-endotoxin stimuli. Nonetheless,

it was demonstrated that pyrogenic reactions towards non-endotoxin stimuli can

be just as vigorous as those towards endotoxin. These results show that the

rabbit pyrogen test cannot be replaced by the LAL for vaccines, but that only the

measurement of the cytokine response of primary human cells, e.g. the WBT,

represents an adequate alternative.

Measurement of albumins

Pyrogenic reactions of human patients after the administration of human serum

albumin, which had tested negative in the rabbit, were observed in 1978 (62). In

this study, the LAL yielded positive results without perceivable patient reactions.

Pool et al. (63) tested 22 batches of human serum albumin (HSA), fibronectin

and stabilized human serum solutions using artificial contaminations of

endotoxin and LTA from B. subtilis. None of these products interfered with the

production of IL-6 by whole blood, whereas one batch of artificially

contaminated albumins tested false-negative in the LAL. Another study using

the WBT performed with albumins, coagulation factor, vaccines and

immunoglobulins indicated a high sensitivity and reliability of the WBT for these

substances (19).



15

A comparison between the rabbit and the human whole blood test for the

detection of pyrogens in albumins was performed by Spreitzer et al. (64) with 29

batches of human serum albumin. The WBT was clearly superior to the rabbit

test, especially at the limit of detection of 5 EU/kg (0.5 EU/10ml/kg), with the

WBT retrieving all 29 spiked samples as positive compared to only 5 positive

rabbit tests and 23 temperature rises, which would have required a repetition of

the test. This limit of detection represents the 0.5 EU/ml pyrogenic threshold.

3.6. Validation

Six cell-based assays, including two variants of the WBT measuring IL-1 and IL-

6, respectively, were validated in an international collaborative study including

laboratories from Austria, Germany, Switzerland, England, Norway and Italy

and the participation of control institutions. The study validated assays such as

the cell line THP-1 with the endpoint TNF-α (65) or with the endpoint neopterin

(66, 67), the cell line Monomac-6 measuring IL-6 (68), isolated peripheral blood

mononuclear cells (PBMCs) with endpoint IL-6, and the human whole blood test

(49), using blinded endotoxin stimuli and altogether 13 intravenously applied

drugs. Sensitivities ranged between 73-96% and specificities between 90-97%.

The WBT measuring IL-1 achieved 73 and 93%, and the WBT measuring IL-6

88.9 and 96.6%, respectively. The development and outcome of this study is

described in detail elsewhere (48, 49, 69- 71)

3.7. Development of the commercially available IPT kit

The established WBT procedure was adapted to materials provided by Charles

River Endosafe and a commercial kit was developed, which was named In Vitro

Pyrogen Test (IPT). This kit contains all the reagents necessary for the

incubation and ELISA procedure except for the human whole blood.

Development of cryopreserved blood

Fresh human whole blood is a highly perishable item that cannot be stored

longer than 4 hours at room temperature without loss of sensitivity. Additionally,



16

it is not easily available, a potential hazard due to unrecognized infections (HIV,

hepatitis) and, due to donor differences, cannot be standardized. In order to

overcome these difficulties a procedure was developed to successfully freeze

and store whole blood. The protocol closely followed the method of de Boer,

1981, who had already successfully frozen isolated monocytes (72). Blood from

five healthy donors is mixed with 10% endotoxin-free dimethylsulfoxide (DMSO)

(v/v ratio) and left to stand for 15 minutes. The blood is then pooled (Fig. 1) and

frozen in a computer-controlled freezing process to –120°C. The blood is stored

in the vapor phase of liquid nitrogen and, after thawing, can be used like fresh

blood without any washing steps. The cryopreserved pooled blood renders

highly reproducible results and is at least equal to fresh blood concerning a

wide variety of applications and stimuli (73).

0.5 EU/ml

1.0 EU/ml

0

2500

5000

7500

0.5 EU/ml

1.0 EU/ml

saline control0.5 EU/ml1.0 EU/ml

1 2 3 4 5 6 7 8 9 10 pool 1 pool 2

donor number

IL-1

ββ ββ±± ±±

SD

Fig. 1: Comparison of the reactivity of frozen blood from 10 individual

donors and that of pooled blood from the same donors.

The calculated means of the response of the five individual donors towards the

0.5 EU/ml LPS corresponds to the response of the pooled blood. The higher

response of donor 4 is therefore leveled out.

Pool 1: The blood was pooled before freezing

Pool 2: The blood was pooled after thawing



17

Since blood frozen using the described method could only be stored and

shipped in the vapor phase of liquid nitrogen, a reagent that is not available to

all laboratories, an alternative freezing method was developed by the Paul-

Ehrlich Institute, Langen, Germany. The method is described in detail by

Schindler et al, 2006 (74). The alternative cryopreservation method provided

blood that could be stored (and shipped) at -80°C, therefore making the blood

available for users without liquid nitrogen infrastructure.

Validation of the cryopreserved blood

In an additional validation process, which followed the exact procedure of the

former process described above, both methods of cryopreservation were

validated (74). Furthermore, the IPT incubation steps, which had been developed

and validated in pyrogen-free reaction tubes, had in the meantime been

successfully transferred to the 96-well microtiter plate by reducing the volumes

used and adapting the protocol accordingly. Therefore, the fresh blood incubation

in the microtitre plate was validated as well as the cryopreserved blood both in

the 96-well microtiter plate and in the pyrogen-free reaction tubes. The overall

performance of all approaches was very good, with sensitivities of over 90% and

specificities around 80%. Remarkably, these excellent performance

characteristics were achieved although the spike concentrations chosen were at

or below the defined pyrogenicity threshold of 0.5 EU/ml (48). Indeed, the few

misclassifications only occurred for these borderline cases. Therefore, the IPT

could be improved concerning its availability, its performance and its handling

(Table I).

Test Inter-laboratory

reproducibility

(%)

Sample size:

sensitivity

Sensitivity

(%)

Sample size:

specificity

Specificity

(%)

WBT

Fresh blood

Reaction tubes

DL-NL1: 72.9

DL-NL2: 81.6

NL1-NL2: 70.2

88 72.7 59 93.2



18

IPT

Cryopreserved

blood (-80°C)

DL-NL 1: 86.7

DL-NL 2: 87.5

NL 1-NL 2: 100

77

97.4

45

82.2

IPT

Cryopreserved

blood

(nitrogen)

DL-NL 1: 66.0

DL-NL 2: 63.3

NL 1-NL 2: 83.3

74

82.4

46

89.1

IPT

Fresh blood

plate

DL-NL 1: 88.1

DL-NL 2: 89.7

NL 1-NL 2: 91.5

84

98.8

55

83.6

Table I: Outcome of the validation of the basic WBT procedure using

reaction tubes and fresh blood and of the IPT methods using

cryopreserved or fresh blood in a microtiter plate.

3.8. Special adaptations

Medical devices

Due to manufacturing and handling, medical devices can bear pyrogens on their

surface which, when brought into the human organism may lead to

inflammatory reactions and reduced biocompatibility. Recognizing this problem,

the Medical Device Directive 93/42 EEC states that medical devices must be

designed and manufactured in such a way that they will not compromise the

clinical condition or the safety of the patients. The Association for the

Advancement of Medical Instrumentation (AAMI) stated in 2001 that products

with direct or indirect contact with the circulatory system or the lymph or

products that interact systemically with the body should be tested for pyrogens

(75).

Products in direct (blood bags, needles) and indirect (swabs, gloves) contact to

the blood circulation can have a serious impact on the organism, as

contaminations induce systemic reactions. A severe case of contact dermatitis

due to endotoxin contamination of surgical gloves was described in 1984 by



19

Shmunes and Darby (76). After eight pyrogenic reactions in 69 patients

undergoing heart catheterization, Kure et al. described endotoxin contamination

of extracts of the hospital’s surgeon’s latex gloves, which evoked fever in

rabbits and could be successfully transmitted to cardiac catheters (77). Grötsch

et al. were able to evoke fever reactions in rabbits with an eluate of gloves

found to contain up to 2560 EU (78).

Medical devices pose a particular problem for pyrogen testing, since they

cannot be examined directly with the rabbit or the LAL test. Their diversity with

regard to size, form, material and form of application challenges the existing

assays, demanding individual approaches. In order to judge a possible

contamination, an eluate of the respective material must be either injected into

the rabbit or used in the LAL. However, it is unclear, how well rinsing a medical

device in water can release pyrogens from its surface and the dilution of such

released pyrogens in a large volume of rinsing water reduces the limit of

detection. The alternative of transplanting the questionable material directly into

the rabbit is highly invasive, causing possible reactions not associated with

pyrogenic contaminations but rather with tissue damage and is therefore

questionable in its ethical and scientific implications. The obvious advantage of

the IPT over the classical test methods is that the whole blood comes into direct

contact with the respective device and no preparation of an eluate is required.

This has been demonstrated using aneurysm clips as proof of principle (79).

Additionally, unlike the LAL, the IPT detects all pyrogens relevant to humans,

not only endotoxin.

Testing for the inflammation-inducing potential of implant surfaces for the

judgment of biocompatibility is a relatively new field. In the early 1980s, it was

noted that the monocyte is one of the first cells to arrive at an implant site and

displays manifold functions (for review see 80, 81). Its specific preference for

rough and hydrophobic surfaces differs from that of fibroblasts (82). The role of

cytokine production of the monocytes/macrophages in the early stages of

implant insertion is poorly understood. The fact that some materials are

obviously capable of modulating the cytokine response (83, 84) makes it difficult

to distinguish a genuine pyrogenic contamination from an unspecific activation



20

and poses the problem of adequate negative controls. For this purpose, a

process was developed for the thorough depyrogenation and a device was

developed for the testing of metallic or plastic surfaces with the IPT in order to

gain information about possible inherent activating or inhibiting characteristics of

materials (85). The device was made up of a perforated metal plate pressed

onto the sample surface by screws through a metal frame. The resulting wells

were watertight due to the use of washers. The blood was incubated directly in

the wells of the depyrogenized device contacting the surface to be tested. The

study showed that pyrogenic contaminations on surfaces could be reliably

removed only when heated for 5 h at 300° C. This applied to titanium, titanium

alloy (TiAl6V4) and steel material for implants. Artificial contaminations were

detected in a dose-dependent manner.

Some medical devices are absorbed completely by the body, as are any

contained pyrogenic contaminations. Examples are liposomes and alginate

microcapsules used as drug carriers. The detection of pyrogenic contaminations

in alginate microcapsules is illustrated in Fig. 2.

A B C D E 0 25 100 500 10000

2500

5000

7500

10000

12500

15000

17500

alginate samples LPS (E. coli O-113) (pg/ml)

IL-1

ß (

pg

/ml)

Fig. 2: IL-1β production of fresh blood upon stimulation with different

alginate solution samples (A-E).



21

Cellular therapeutics

Cellular therapeutics are defined as living cells that are transferred into the

intact organism in order to introduce a new or restitute a defective function (86).

This includes a wide variety of cells such as chondrocytes, stem cells, bone

marrow cells, and blood cells such as lymphocytes, erythrocytes, thrombocytes.

The latter pose a particular problem, since they are stored at room temperature

and are therefore easily subject to extensive bacterial growth. Transfusion

reactions may range from shivering, fever and chills all the way to septic shock.

The problem is rather under- than overrated, since numerous clinical events are

not recognized as being transfusion-associated, but are often rather attributed

to the underlying disease. Additionally, medication and immunosuppression

might mask an existing septic/pyrogenic event which likely contributes

significantly to the patient’s overall morbidity. Recently attention has focused on

viral infections, although the incidence of viral contaminations of blood products

is less than 1 in 1.000.000 per unit for HIV in comparison to 1 in 3000 for

bacterial contaminations (87).

Two large studies in France (BACTHEM study, 88) and the USA (BaCon study,

89) revealed that platelets hold a significantly higher risk of bacterial

contamination than red blood cells, irrespective of whether they were single-

donor or pooled preparations. Pathogens associated with bacteremia in the US

study were 59% Gram-positive (mainly skin contaminants such as

staphylococci, streptococci and propionibacteria) and 41 % Gram-negative (coli,

serratia, enterobacter). Gram-negative Yersinia enterocolitica was not found in

that study, although it occurs frequently in transfusion-related sepsis and was

responsible for 7 of the 8 fatalities recorded in the US between 1986 and 91.

Incidences of microbial contamination increased with prolonged storage, and

both studies linked fatalities to the occurrence of Gram-negative bacteria. The

US study also determined endotoxin levels (up to 273,500 EU/ml, according to

LAL). The authors estimated rates of transfusion-transmitted bacterial infections

of 1:100’000 for platelets and 1 in 5 million for red blood cells, with fatalities of 1

in 500.000 and 1 in 8 million, respectively. Overall, Gram-negative bacteria



22

tended to occur more frequently in red blood cells, probably due to the storage

at lower temperatures.

In 2004, a roundtable meeting on bacterial detection took place during the

Annual Congress of the International Society of Blood Transfusion in Edinburgh

to discuss the limitations of current testing methods. Currently, culturing

methods, such as the very sensitive BacT/ALERT method, are considered the

best, though they are very time-consuming (12 h to 7 days). Platelet

concentrates are released on a “negative-to-date” base and recalled if

necessary. The panel reported occurrences in the Netherlands, where platelet

concentrates containing skin bacteria were tested positive only after 48 h. By

this time, the batch had been released and about 50% of the units had already

been transfused. Very similar events were described by Belgian blood centers

(87). Anaerobic bacteria, e.g. Corynebacterium spp., are picked up even later

and there is extra cost involved. However, anaerobic bacteria have been linked

to fatal septic transfusion incidents (90). Altogether, culture methods are

incapable of providing complete safety, and other, especially quicker methods,

are sought.

A method to inactive contaminating bacteria in transfusion products by

photochemical treatment (PCT) (91) has been developed. Still, it must be kept

in mind that although this inactivation may inhibit growth, it will have no

influence on the already existing pyrogenic content. Therefore, the testing of

these cellular products and their suspension materials is an interesting future

challenge for the IPT. Pretesting of clinical grade erythrocytes and

thrombocytes intended for transfusion indicated interference-free retrieval of an

artificial endotoxin spike (Fig. 3) when compared to the saline control.



23

0 0,125 0,25 0,5 0,75 1 2 40.0

0.6

1.2

saline controlthrombocytes

erythrocytes

O113 (EU/ml)

OD

450

Fig. 3: Retrieval of endotoxin spikes in red blood cell and platelet

concentrates

Dialysis

Pyrogenic reactions in hemodialysis patients at the end of a session were first

associated with high bacterial and endotoxin levels by Raij et al. (92) and

Favero et al. (93). Since then, contaminations have been found in the pure

water (94-97), the machines, especially in areas with low circulation or dead

spaces which serve as a reservoir for bacteria (98), filter materials (99) and

bicarbonate concentrates (95).

In 1993, the AAMI released recommendations for the quality of treated water

and dialysate, which restricted the content of heterotrophic bacteria to 200 and

2000 cfu/ml, respectively. Studies in Germany (97), Greece (100), the USA

(94), and Canada (101) revealed that even these moderate standards are not

met, which is even more critical considering that a patient with chronic renal

failure receives up to 400 l of dialysis fluid a week. Next to Gram-negative

bacteria, cocci (micrococci, staphylococci and streptococci) were found in the

dialysate of 83, 70, and 10% of the centers, respectively, indicating the

importance of Gram-positive contaminations. That this might indeed be crucial

for judging the pyrogenic exposure of a dialysis patient was assessed by



24

Marion-Ferey et al., who tested scrapings of bacterial biofilms in dialysis tubes

and found a 20-fold higher response in the IPT than in the LAL (102). The

passage of cytokine-inducing substances, not only endotoxins, but exotoxins

and peptidoglycans as well, through the dialysis membrane has been

demonstrated (103- 106).

The chronic exposure of dialysis patients to even low concentrations of

pyrogens is thought to contribute to inflammatory processes in the joints and

bones and therefore to the carpal tunnel syndrome and arthropathy associated

with long-term hemodialysis (107, 108). In 1991, Baz et al. showed that the use

of ultrapure water delays the onset of the carpal tunnel syndrome (109). The

group of Schwalbe (110) showed in a retrospective study that the incidence of

amyloidosis decreased between 1988 and 1996 along with the introduction of

reverse osmosis, a very effective method for purifying water. A connection

between other phenomena, such as malnutrition, poor immune responses and

high incidence of malignant tumors in long-term HD patients with pyrogen

exposure has yet to be established.

In all, the testing for pyrogens in dialysis fluids is a crucial issue for the safety of

the patients. Since the fluids themselves are either highly hyper- or hypotonic, a

variant for testing dialysis fluids in the IPT established the percentages of

diluents and samples that can be tested (own unpublished results). Still, the

problem remains that the patients receive very high volumes of fluid in one

session, and therefore pyrogens must be detected at very low concentrations. A

promising possibility is a modification of the basic IPT protocol, the so-called

adsorb and wash IPT (AWIPT), discussed later, which can concentrate

pyrogens on the surface of albumin-coated macroporous Matisse™beads, thus

enhancing the sensitivity by a factor of 250 (111).

Airborne pyrogens

Inhalable whole or fragments of microorganisms have long been recognized as

causes of airway hyperreactivity. Monday sickness with its typical symptoms

(chest tightness, respiratory distress and coughing) was described as early as

1936 (112). In 1942, rural mattress makers experienced headache, nausea,



25

chills and fever about 6 hours after exposure to low grade cotton. Neal et al.

associated these phenomena with high numbers of a Gram-negative bacterium

in the material (113). Additionally, milder symptoms occurring 8 hours after

exposure could also be evoked by sterilized cotton, which was thought due to

remaining endotoxin. A highly significant (r > 0.95) dose-response relationship

between Gram-negative bacterial count and symptoms of byssinosis such as

chest tightness, airway inflammation and coughs was established by Cinkotai et

al. (114). In the same study, a good correlation existed between symptoms and

mainly Gram-positive bacteria, whereas one to fungal spores could not be

established. Acute bronchoconstriction as well as chronic airway irritation with

bronchitis and decrements in airflow over the work day have been reported for

personnel working in animal confinement buildings (115 - 117). Long term

consequences are of allergic, inflammatory and immunostimulatory nature, e.g.

organic dust toxic syndrome (ODTS) and chronic bronchitis. The LAL test for

these contaminants has the drawback that it does not reflect the biological

potency of a given LPS in the mammal (53) and the LAL test can only be

performed with an eluate of a filter or by impingement, i.e. the air to be tested is

led through pyrogen-free water which is then tested in the LAL. The higher

pyrogen retrieval by impingement when compared to filtration, possibly due to

the incomplete eluation of the sample from the filters, was demonstrated by

Zucker et al. (118).

A new approach of measuring the integral inflammatory activity in air samples in

different environments by IPT was reported by Kindinger et al., 2005 (119). A

defined amount of air is drawn through a filter in a sealable plastic monitor. The

blood incubation is performed directly on the filter inside the monitor, thus

making any handling of the filter unnecessary. When compared to the LAL, a 2-

25 fold higher pyrogenic load was found in the IPT in samples drawn in parallel.

Epidemiological studies will show what levels of exposure to inflammatory

stimuli in the air eventually lead to the above-mentioned lung diseases.



26

Lipidic formulations

In January 2004, pyrogen testing of so-called small volume parenterals (< 15ml)

became obligatory in Europe. This concerns many formulations that had not

been subjected to pyrogen testing before, such as vitamin preparations and

steroids. Many of these are applied intramuscularly or subcutaneously and are

therefore not necessarily of a hydrophilic nature. This poses a completely new

challenge to all methods of pyrogen testing, since a lipophilic substance on the

one hand cannot be injected intravenously into the rabbit due to the danger of

clogging small vessels with lipid drops and severely damaging vital organs and

will, on the other hand, influence the optical density measured as the readout in

the LAL due to the formation of oil droplets. Furthermore, the pyrogenic portion

of LPS, lipid A (for review see Rietschel et al., 1993 (120) has been reported to

be masked by lipoproteins (121) and lipophilic parenterals (122) in the LAL.

Therefore, the IPT procedure was adapted to suit lipophilic substances. As a

first step, interference-free oils, such as sesame oil, were identified by

comparing an LPS dose response curve in these oils with a similar curve done

in saline. Surprisingly, many oils (sesame oil, peanut oil, paraffin, miglyol) were

interference-free while others interfered strongly by suppressing the endotoxin

stimulus added. Oils that proved interference-free were then used as diluents

for interfering end-products. It was possible to dilute their interference to non-

detectable limits with full recovery of an artificial endotoxin spike. From this

minimum valid dilution a possibly detectable endotoxin concentration could be

calculated, which was 20 EU/ml for the respective end-products. Since these

products are applied at a very small volume (1 ml per person), a relatively high

endotoxin concentration can be tolerated. The established protocol leaves a

broad safety margin, especially since the strict criteria for intravenous drugs

were applied to this situation (123).

AWIPT (absorb and wash IPT)

Another interesting development is the so-called absorb and wash IPT

(AWIPT). It uses porous acrylic beads with immobilized albumin, which has a

higher affinity than native plasma albumin to endotoxin (124), to separate the



27

pyrogenic contamination from the sample. These beads were originally

developed as LPS adsorbers (Matissebeads™) to be applied in sepsis patients.

The AWIPT uses this material to collect the endotoxin contained in a sample

after an absorption phase in the substance to be tested. The beads are then

washed in order to remove the unbound material and can then be used directly

in the IPT incubation. It could be shown that this works also for LTA of

Staphylococcus aureus and zymosan, a yeast extract. This procedure has

already brought promising results with substances that interfere with the

standard IPT procedure, i.e. toxic or immunomodulatory drugs (111). Another

possibility is the testing of high-volume parenterals such as dialysis fluids, which

contain endotoxin concentrations below the detection limit of other pyrogen

tests. The beads could be used to concentrate the endotoxin on their surface

from a large sample volume. Using this procedure, the detection limit of the IPT

could be lowered from 0.25 EU/ml of E. coli endotoxin down to 1 x 10-5 EU/ml

(Fig. 4).

0.000 0.001 0.0100

50

100

150

200

250

300

350

400

O-113 LPS [EU/ml]

IL-1

ββ ββ [

pg

/ml]

Fig. 4: Limit of detection in the AWIPT

In all, the further development of the IPT into the modified form of the AWIPT

promises to overcome shortcomings for special applications caused by

interferences of certain drugs or substances with the classical IPT procedure. It

allows lowering of the detection limit, and provides a useful tool for the testing of

toxic or strongly interfering substances, even those that suppress the immune

system and therefore cytokine production.



28

3.9. Conclusions

Pyrogens (fever-inducing substances) from microorganisms can occur as

contaminations of parenterals. Until now, the safety of injectable drugs has

been assessed by the in vivo rabbit pyrogen test and the in vitro Limulus

amoebocyte lysate test (LAL).

The new cell-based in vitro pyrogen test based on fresh or cryopreserved

human whole blood (IPT) has been successfully validated and has proven to be

a reliable and useful tool for a wide spectrum of applications, ensuring patient

safety in many medical fields such as hydrophilic and lipophilic drugs, dialysis

fluids, airborne pyrogens, medical devices and biologicals. It is capable of

measuring all known pyrogens relevant for the human and is highly reliable,

robust and easy to perform.

3.10. Appendix

Abbreviations: AAMI, Association for the Advancement of Medical

Instrumentation; AWIPT, absorb and wash in vitro pyrogen test; cAMP, cyclic

adenosinmonophosphate; DMSO, dimethylsulfoxide; ECVAM, European Centre

for the Validation of Alternative Methods; ELISA, enzyme-linked immunosorbent

assay; ELC, endotoxin limit concentration; EU, endotoxin unit; HD,

hemodialysis; HSA, human serum albumin; IL, interleukin; IPT, in vitro pyrogen

test; LAL, Limulus amoebocyte lysate; LPS, lipopolysaccharide; LTA,

lipoteichoic acid; NIH, National Institutes of Health; NIBSC, National Institute of

Biological Standards and Controls; OD, optical density; OVLT, organum

vasculosum laminae terminalis; PBMCs, peripheral blood mononuclear cells;

PEI, Paul-Ehrlich Institute; PG, prostaglandin; POD, peroxidase; RNA,

ribonucleic acid; TMB, tetramethylbenzidine; TNF, tumor necrosis factor; USP,

United States Pharmacopoeia; WBT whole blood test

3.11. Acknowledgements

The validation study was funded by the European Union [QLRT-1999-00811].

The authors would like to thank the numerous scientific and industrial

supporters who provided sample materials.

INTERNATIONAL VALIDATION OF NOVEL PYROGEN TESTS BASED ON

HUMAN MONOCYTOID CELLS

29

4 International validation of novel pyrogen tests based on

human monocytoid cells

Sebastian Hoffmanna;h§; Anja Peterbauere*§; Stefanie Schindlera§; Stefan

Fennricha; Stephen Pooleb; Yogesh Mistryb; Thomas Montag-Lessingc; Ingo

Spreitzerc; Bettina Löschnerc; Mirjam van Aalderend; Rogier Bosd; Martin

Gommerd; Ria Nibbelingd; Gabriele Werner-Felmayere; Petra Loitzle; Thomas

Jungif; Marija Brcicf; Peter Brüggerg; Esther Freyg; Gerard Boweh; Juan

Casadoh; Sandra Coeckeh; Jan de Langeh; Bente Mogsteri; Lisbeth M. Næssi;

Ingeborg S. Aabergei; Albrecht Wendela; Thomas Hartunga;h#

a Institute of Biochemical Pharmacology and Steinbeis Center InPuT, University

of Konstanz, Universitätsstrasse 10, D-78457 Konstanz, Germany b NIBSC, National Institute for Biological Standards and Control, Blanche Lane,

South Mimms, Potters Bar, Herts EN6 3QG, England, UK c Paul Ehrlich Institute, Paul-Ehrlich Strasse 51-59, D-63225 Langen, Germany d RIVM, National Institute of Public Health and the Environment, A. van

Leeuwenhoeklaan 9, P.O.Box 1, 3720 BA Bilthoven, The Netherlands e Institute of Medical Chemistry and Biochemistry, Fritz-Pregl-Strasse 3, A-6020

Innsbruck, Austria f Institute of Veterinary Virology, Länggass-Strasse 122, University of Bern, CH-

3012 Bern, Switzerland g Biological Analytics, Novartis Pharma AG, CH-4002 Basel, Switzerland h European Centre for the Validation of Alternative Methods (ECVAM), Institute

for Health & Consumer Protection, European Commission Joint Research

Centre, Via Fermi 1, I-21020 Ispra, Italy i Division of Infectious Disease Control, Norwegian Institute of Public Health,

P.O. Box 4404 Nydalen, NO-0403 Oslo, Norway

* Present address: Ludwig Boltzmann Institute for Experimental and Clinical

Traumatology/Blood Transfusion Service for Upper Austria, Blumauerstr. 3-5,



30

A-4020 Linz, Austria § S. Hoffmann, A. Peterbauer, S. Schindler contributed equally to the work

presented here

4.1. Abstract

Parenteral medicines are required to be tested for pyrogens (fever-causing

agents) in one of two animal-based tests: the rabbit pyrogen test and the

bacterial endotoxin test. Understanding of the human fever reaction has led to

novel non-animal alternative tests based on in vitro activation of human

monocytoid cells in response to pyrogens. Using 13 prototypic drugs, clean or

contaminated with pyrogens, we have validated blindly six novel pyrogen tests

in ten laboratories. Compared with the rabbit test, the new tests have a lower

limit of detection and are more accurate as well as cost and time efficient. In

contrast to the bacterial endotoxin test, all tests are able to detect Gram-

positive pyrogens. The validation process showed that at least four of the tests

meet quality criteria for pyrogen detection. The here validated in vitro pyrogen

tests overcome several shortcomings of animal-based pyrogen tests. Our data

suggest that animal testing could be completely replaced by

these evidence-based pyrogen tests and highlight their potential to further

improve drug safety.

Keywords: Pyrogens; validation study; cytokines; monocytes; alternatives to

animals; cell culture

Abbreviations: BET, bacterial endotoxins test; CI, confidence interval; DL,

developing laboratory; ELC, endotoxin limit concentration; ELISA, enzyme-

# Corresponding author: Prof. Thomas Hartung, MD, PhD; European

Commission, Joint Research Centre; Institute for Health and Consumer

Protection; ECVAM; 21020 Ispra; Italy; Tel.: +39 0332 785939; Fax: +39 0332

786297; e-mail: [email protected]



31

linked immunosorbent assay; EU, endotoxin units; IFNγ, interferon γ; IL,

interleukin; LAL; Limulus amoebocyte lysate; LPS, lipopolysaccharide; LTA,

lipoteichoic acid; LoD, limit of detection; MM6, MONO MAC 6; MVD, maximum

valid dilution; NL, naive laboratory; PBMC, peripheral mononuclear blood cell;

PBS, phosphate buffered saline; TLR, toll-like receptor; TNFα, tumor necrosis

factor α; WBT, whole blood test

4.2. Introduction

Pyrogens, a chemically heterogeneous group of fever-inducing compounds, are

derived from bacteria, viruses, fungi or the host himself.

Monocytes/macrophages react to microbial products during an immune

response by producing endogenous pyrogens such as prostaglandins and the

pro-inflammatory cytokines interleukin-1 (IL-1), interleukin-6 (IL-6) and tumor

necrosis factor-α (TNFα) (132). Depending on the type and amount of pyrogen

challenge and the sensitivity of an individual, life-threatening shock-like

conditions can be provoked. Consequently, to assure the quality and safety of

any pharmaceutical product for parenteral application in humans, pyrogen

testing is mandatory.

Depending on the drug, one of two animal-based pyrogen tests is currently

prescribed by the health authorities and Pharmacopoeias, i.e., for more than

sixty years, the rabbit pyrogen test or the bacterial endotoxins test (BET), often

referred to as Limulus amebocyte lysate test (LAL). For the rabbit pyrogen test,

sterile test substances are injected intravenously into rabbits and any rise in

body temperature is measured. This in vivo test detects various pyrogens but

alone the fact that large numbers of animals are required to identify the rare

pyrogen-containing samples in routine practice argues against its use if valid

alternatives are available. In the past two decades, the declared intention to

refine, reduce and replace animal testing, the 3Rs concept (125) that was

implemented e.g. into European legislation in 1986 (126), has led to a reduction



32

in rabbit pyrogen testing by 80 % by allowing the BET as an in vitro alternative

pyrogen test for many parenteral products.

Bacterial endotoxin comprised largely of lipopolysaccharide (LPS) from the cell

wall of Gram-negative bacteria that stimulates monocytes/macrophages via

interaction with CD14 and toll-like receptor 4 (TLR4) (127) is the pyrogen of

major concern to the pharmaceutical industry due to its ubiquitous sources, its

stability and its high pyrogenicity (128-130). With the BET, endotoxin is

detected by its capacity to coagulate the amoebocyte lysate from the

haemolymph of the American horseshoe crab, Limulus polyphemus, or the

Japanese horseshoe crab, Tachypleus tridentatus, a principle recognized some

40 years ago (31). In the United States, Limulus crabs are generally released

into nature after drawing about 20 % of their blood and therefore most of these

animals survive. However, the procedure still causes mortality of about 30.000

horseshoe crabs per year, which adds to the even more severe threats of the

horseshoe crab population such as its use as bait for fisheries, habitat loss and

pollution (http://www.horseshoecrab.org). As with the rabbit test the general

problem of translation of the test results to the human fever reaction persists.

Moreover, although it is highly sensitive, the failure of the BET to detect non-

endotoxin pyrogens as well as its susceptibility to interference by, for example,

high protein levels of test substances or by glucans impedes full replacement of

the rabbit pyrogen test (53,131). Hence, an estimated 200.000 rabbits per year

are still used for pyrogen testing in the European Union.

A test system that combines the high sensitivity and in vitro performance of the

BET with the wide range of pyrogens detectable by the rabbit pyrogen test is

therefore required in order to close the current testing gap for pyrogens and to

avoid animal-based tests. With this intention and due to improved

understanding of the human fever reaction (132), test systems based on in vitro

activation of human monocytoid cells have been developed. First efforts

date back about 20 years, when peripheral blood mononuclear cells (PBMC)

were used to detect endotoxin by monitoring the release of pyrogenic



33

cytokines (133, 134). Subsequently, a number of different test systems, using

either whole blood, PBMC or the monocytoid cell lines MONO MAC 6 (MM6)

(68) or THP-1 (135) as a source for human monocytes and various read-outs

have been established and were recently reviewed (136). Here, the six most

prominent of these test systems were formally validated with the aim of

developing an evidence-based tool for safer, animal-free and more efficient

pyrogen detection and allowing their regulatory acceptance. Formal validation

of in vitro methods, i.e. the evaluation of reliability and relevance of a method,

was developed by the European Centre for the Validation of Advanced and

Alternative Methods (ECVAM) and is now internationally accepted (137-139).

4.3. Methods

Rabbit pyrogen test

For this study data from 171 rabbits (kindly provided by Dr. U. Lüderitz-Püchel)

accumulated over several years at the Paul Ehrlich Institute, the German

Federal Agency for Sera and Vaccines in Langen, were used for analysis. For

these experiments, Chinchilla Bastards (Charles River) were injected with 0, 5,

10, 15, 20 EU in 1 ml/kg of E. coli LPS (EC5) (140) or EC6 (56) in saline

(corresponding to 0, 0.5, 1.0, 1.5 and 2.0 EU/kg in 10 ml, the largest volume

allowed for injection in rabbits). The fever threshold in rabbits was defined as a

body temperature increase of 0.55 °C during 180 min after injection. This value

represents the mean individual rabbit value at the threshold of 6.6 °C of the EP

when the maximum of twelve animals is tested (141).

In vitro monocyte-based tests

Good laboratory practice concordant Standard Operating Procedures of the

various methods were made available by ECVAM (www.ecvam.jrc.it). The

test systems are summarized by Hartung et al. (70) and detailed in previous

work (49, 66, 67, 69, 142, 143).



34

Reagents and consumables for all methods

The 2nd International WHO Standard for endotoxin (from E. coli O113:H10:

K(-) (94/580), which is identical to FDA/USP standard EC6/Lot G was used as

the standard endotoxin (56). Test materials for validation are specified in the

Results section. All consumables were purchased as sterile and pyrogen-free

and not specified reagents were pro analysis grade.

PBMC-IL6

Blood Collection and preparation of PBMC

Blood donors had to describe themselves as being in good health, not suffering

from any bacterial or viral infections for at least one week prior to the donation

of blood and not to be taking drugs known to influence the production of

cytokines. Using a heparinized (50 µl Fragmin at 10000 IU, Dalteparin,

Pharmacia) syringe, 30 ml blood were collected. Within two hours, PBMCs

were isolated from 20 ml Lymphoprep (Nycomed, Oslo, Norway), 15 ml PBS

and 15 ml of heparinized whole blood by centrifuging at 340 x g for 45 min at

room temperature. The PBMC-layer was washed twice with PBS centrifuging at

340 x g for 15 min. The sediment was suspended with RPMI-C (RPMI 1640,

Life Technologies, Paisley, Scotland) with 10 ml/l human serum AB from

clotted human male whole blood (Sigma), 10 ml/l L-Glutamine

(Life Technologies), 200 mM, and 20 ml/l Penicillin/Streptomycin solution

(Seromed, Vienna, Austria)) after counting in a Neubauer haemocytometer to 1

mio cells/ml. The cells shall be incubated with samples within four hours after

blood withdrawal.

Protocol for PBMC-IL6

In quadruplicate per each of four blood donors, 100 µl of RPMI-C, 50 µl of

samples/controls and 100 µl of gently swirled PBMC were incubated in a 96-

well tissue culture plate (Falcon Microtest, Becton Dickinson Labware) at



35

37°C for 16- 24 hours in an atmosphere of 5% CO2 in humidified air. After

incubation, 50 µl of supernatant from each of the wells was transferred on the

ELISA plate ensuring that cells are not aspirated by angling the assay plate.

ELISA for PBMC-IL6

2.5 µg/ml coating mouse monoclonal anti-IL-6 antibody (Novartis in-house

Clone 16) was added at 200 µl to each well of a 96-well microtitre plate (Nunc-

Immuno 96-well plate MaxiSorp, F96; Life Technologies) at 15 - 25 °C for 16 -

24 hours. The washed plate was coated with 200 µl blocking buffer (24.2 g/l

Tris(hydroxymethyl)aminomethane, 0.2 ml/l Kathon MW/WT (Christ Chemie

AG, Reinach, Switzerland) and 10.0 g/l bovine serum albumine). Plates were

incubated with 200 µg/ml horseradish peroxidase conjugated to sheep anti-IL-6

antibodies (Novartis, in-house) for 2-3 hours at 20-25°C. Shortly before use, 90

ml substrate buffer and 4.5 ml TMB solution (240 mg 3,3',5,5'Tetramethyl-

benzidine in 5 ml acetone, 45 ml ethanol and 0.3 ml Perhydrol (30 % H2O2))

were mixed and 200 µl pipetted into each well. After 10-15 minutes, the

enzyme reaction was stopped by 50 µl of 5.4% H2SO4 per well. The

absorbance was measured at 450 nm using 540-590 nm as reference

wavelength.

WBT-IL1

Blood Collection for WBT-IL1

Blood donors should show no evidence of disease or need of medication during

the last two weeks. Blood was collected into heparinized tubes (Sarstedt S-

MONOVETTE 7.5 ml, 15 IU/ml Li-Heparin) and used within four hours (144).

Protocol for WBT-IL1

In this order and in quadruplicates per single blood donor, 1000 µl saline, 100

µl sample/control and 100 µl blood were added to pyrogen-free reaction tubes

(Greiner Bio-one tubes, 1.2 ml (polystyrene) or 1.5 ml (polypropylene),



36

Frickenhausen, Germany). Closed tubes were mixed gently, inverted once or

twice and then incubated in an incubator or a heating block at 37°C ± 1°C for

10-24 hours. The incubation tubes were mixed thoroughly by inverting them.

Incubations were centrifuged for 2 minutes at 10.000 g and the clear

supernatant, taking aliquots of ≥ 150 µl, was used for the ELISA (ENDOSAFE-

IPT, Charles-River Endosafe, Charleston, USA) following the manufacturer’s

procedure.

WBT-IL6

Blood Collection for WBT-IL6

Blood donors were selected as described for PBMC-IL6. 30 ml blood were

drawn and immediately transferred into a 50 ml sterile centrifuge tube

containing 300 IU heparin (Fragmin, Pharmacia, diluted 1/10 with saline). The

closed tubes were inverted slowly five times to ensure thorough mixing without

vortexing and used within four hours (174).

Protocol for WBT-IL6

In quadruplicate per each of four blood donors, 50 µl of saline, 50 µl of gently

mixed blood, 50 µl of samples/controls and 100 µl of saline were incubated in a

96-well tissue culture plate (Falcon Microtest, Becton Dickinson Labware) at

37°C for 16-24 hours in a humid atmosphere of 5% CO2. After incubation, 50 µl

of supernatant from each of the wells was transferred on the ELISA plate

ensuring that cells are not aspirated by angling the assay plate. The

same IL-6 ELISA as for PBMC-IL6 was used.

MM6-IL6

Cell culture for MM6-IL6

The human monocytoid cell line MonoMac-6 was obtained from Prof. H.W.L.

Ziegler-Heitbrock (Institute for Immunology, University of Munich, Munich,



37

Germany). Frozen cells from liquid nitrogen were thawed on ice. Cells were

transferred to a 50 ml centrifuge tube, 10 ml RPMI (+4°C) (e.g.

Life Technologies) added and then centrifuged at 100 x g for 5 min at +4°C.

Afterwards the cells were resuspended in 10 ml RPMI-M (containing 10% ml

heat-inactivated low-pyrogen foetal calf serum, 2 mM L-Glutamine, 0.1 mM

MEM non-essential amino acid, 0.23 IU/ml Bovine insulin, 1 mM Oxaloacetic

acid, 1 mM Sodium pyruvate, 20 mM HEPES). After a wash step, cells were

transferred to a 25 cm2 tissue culture flask and incubated at 37°C, with 5% CO2

and high humidity. The number of viable cells was determined by Trypan blue

exclusion using a haematocytometer. The cells were passaged with 2 x 105

cells/ml twice a week.

Protocol for MM6-IL6

To pre-incubate the cells for a test, 30-50 ml of cell suspension were

centrifuged at 100 x g for 8 min at room temperature and resuspended in

RPMI-C (as RPMI-M, but only 2% heat-inactivated foetal calf serum) at a final

concentration of 4x105 cells/ml. The cells were incubated approximately 24

hours at 37°C, 5% CO2 and high humidity. Cells were washed and counted as

above, diluting to 2.5 x 106 viable cells/ml, just prior to addition to the culture

plate. In quadruplicates, 50 µl of samples/controls, 100 µl of RPMI-C and 100 µl

of gently swirled MM6 were incubated in 96-wells tissue culture plates at 37°C

for 16-24 hours with 5% CO2 and humidified air. After incubation, 50 µl of

supernatant from each of the wells was transferred on the

ELISA plate ensuring that cells are not aspirated by angling the assay plate.

The same IL-6 ELISA as for PBMC-IL6 was used.

THP-Neo

Cell culture for THP-Neo

THP-1 cells were obtained from the American Type Culture Collection (ATCC,

TIB-202). 6 x 106 cells were seeded in 60 ml medium (RPMI 1640

supplemented with 10 % (v/v) FCS (high-quality lots with the lowest endotoxin



38

content available (< 30 pg/ml) were chosen, e.g. Biochrom, Berlin, Germany) in

75 cm2 culture flasks. Flasks were incubated in upright position at 37° C with

5% CO2 and humidified air. On the fourth day of culture, further 30 to 60 ml

(depending on the culture doubling time) of culture medium were added and

cells were incubated for another three days. If cells from freshly thawed stocks

are used, they have to be grown for two to three weeks in order to ensure that

they divide properly before using them for tests. Furthermore, cells should not

be kept in culture for more than four months but new cultures should be started

from frozen stocks at regular intervals. Cells were counted with a

hemocytometer and cell viability by trypan blue exclusion was ≥ 90%. Tubes

with 2.5 x 107 cells (for one plate) were centrifuged at 400 x g and 20° C for 7

min and resuspended in 20 ml medium, 2 mM L-glutamine and 50 µM 2-

mercaptoethanol.

Protocol for THP-Neo

100 µl IFNγ (human, recombinant, endotoxin content < 0.1 EU/mg;

Gammaferon 50, Rentschler Biotechnologie, Laupheim, Germany) stock

solution (6250 U in 100 µl medium, 110 µl aliquots) were added to 20 ml of cell

suspension and mixed well. 200 µl/well of mixed cell suspension were added to

a 96-well cell culture microtiter plate. After incubation for 30 min, 50 µl of

vortexed samples/controls (in quadruplicate) were added and put on an orbital

plate shaker for 2 min at room temperature and 500 rpm. After 18-22

hours of incubation, 150 µl of supernatant were collected and frozen and/or

directly processed with the neopterin ELISA (Elitest Screening, Brahms

Diagnostica, Berlin, Germany) according to the manufacturer’s protocol.



39

THP-TNF

Protocol for THP-TNF

THP-1 cells (obtained from ATCC or ECACC) were used. Subclones from this

cell line prepared in-house showed a higher sensitivity towards LPS.

Cells were cultured in RPMI (1% L-glutamine, 1% HEPES, 1%

Penicillin/streptomycin solution, 1% Sodium pyruvate, all from Biochrom (Berlin,

Germany), 1% nonessential aminoacids for MEM, 0.4% MEM vitamin solution,

0.5% β-mercaproethanol (10 mM), all from Invitrogen (Basle, Switzerland), and

12% heat-inactivated low-pyrogen FCS in 6-well plates or T25 flasks at 37°C in

a humidified 5% CO2 incubator. They were passaged once weekly. When new

cells are required for an assay, cells from a cryovial were thawed two to three

weeks before use. For the last passage prior to the test, terminal differentiation

was induced by cultivating the cells in the presence of sterile-filtered calcitriol

(1,25-dihydroxy vitamin D3, Sigma or Hoffmann-La Roche, Basle, Switzerland)

(10 µg/ml) for 44-48 hours. Cells were collected, centrifuged and resuspended

in culture medium containing calcitriol (final concentration 100 ng/ml). They

were counted and adjusted to 1 to 1.25x106 cells/ml. Cells were cultured for 44-

48 hours in T25 flasks. Then, terminally differentiated cells were harvested and

counted using a haematocytometer and trypan blue. Cells were diluted to

1.25x106 cells/ml and 200 µl of suspension were dispensed into each well of

the above 96-well cell culture plate containing already 50 µl of sample/control in

quadruplicates. Plates were incubated for 16-24 hours at 37°C and 5%CO2.

TNFα ELISA for THP-TNF

Non-sterile plates Dynex PF microtiter ‘flat bottom’ styrene 96-well plates

(Dynex Tech., Worthing, UK) were rinsed extensively with pyrogen-free PBS.

The plates were coated with 1 µg/ml monoclonal antibody 101-4 against human

TNFα (a generous gift from Dr. T Meager, Division of Immunobiology, NIBSC,

UK) at 100 µl/well and 4 ºC overnight. 50 µl of sample/control (in

quadruplicates) or duplicates of TNFα standards (250, 62.5, 15.6, 3.9, 0.98,

0.24, 0 U/ml, NIBSC) were added for 16-24 hours at 37°C and 5% CO2. An



40

aliquot of the detecting antibody (biotinylated goat-anti-human TNF-α from the

Duoset kit, R&D) was diluted 180-fold, using dilution buffer (0.1 % bovine

serum albumin, 0.1% Tween 20, in 20 mM Tris, 100 mM NaCl, pH 7.2-7.4). 100

µl were dispensed to each well for two hours at room temperatue. After

washing, 100 µl Streptavidin-peroxidase conjugate (R&D) was added for 20

min. After washing, 100 µl of TMB (Sigma) were dispensed and incubated in

the dark before reading at 650 nm. Incubation time was chosen so that 250

U/ml TNFα value had an OD > 1.5.

Data analysis

The rabbit fever reaction was modeled by regression techniques applied to the

logarithmically transformed data. The within- and between-laboratory

reproducibility were assessed comparing the resulting classifications by means

of simple matching, i.e. the proportions of identically classified samples, as a

measure of similarity. In case of the within-laboratory reproducibility, where

three independent but identical runs were performed, the mean similarity was

calculated.

A one-sided t-test, assuming hazard and thus designed to proof safety of a

tested compound, was employed as a so-called prediction model (PM) to

dichotomize the test results into a classification of either ‘pyrogenic’ or ‘non-

pyrogenic’. The t-test compares the data of a given sample against the data of

the standard positive control of 0.5 EU/ml, which is performed in parallel. It is

calculated with the log-transformed data and a local significance level of 1%

was chosen in order to increase safety. If this test resulted in a significant

p-value, i.e. smaller than 1 %, then the considered sample was classified as

non-pyrogenic, and as pyrogenic otherwise. This means that a negative sample

had to be significantly lower than 0.5 EU/ml. The levels of contaminations

chosen were 0, 0.25, 0.5 (twice) and 1 EU/ml. According to the rabbit model, 0

and 0.25 EU/ml were considered as non-pyrogenic samples and 0.5 and 1

EU/ml as pyrogenic samples. Having thus defined the reference standard, i.e.



41

the ‘true’ contamination level, we calculated via 2x2-contingency tables the

performance parameters sensitivity, i.e. the probability of a correct positive

classification, and specificity, i.e. the probability of a correct negative

classification. Confidence intervals for these parameters were calculated with

the Clopper and Pearson method based on the F distribution (145).

4.4. Results

The limit of endotoxin detection in rabbits

Employing regression techniques, the temperature data from 171 rabbits could

be modeled by the equation y = 0.217 * (EU + 1)0.508, where y is the expected

temperature increase for a given concentration EU/ml (Fig. 1). This approach

was recently described in more detail and further exploited (16). The model

indicated that 50 % of the animals develop fever, i.e. showing a 0.55 °C rise of

body temperature within 180 min after injection, in response to 5.22 EU per kg

body weight of endotoxin with a 95 %-confidence interval of 4.24 to 6.21 EU/ml.

Only at 20 EU per kg of body weight, all animals showed an increase in

temperature of 0.55 °C or more. We deduced from these data that a sample

concentration of 0.5 EU/ml represents the required limit of detection (LoD) that

alternative pyrogen tests must meet. This assumption takes into account the

fact that the largest volume allowed for injection into rabbits is 10 ml per kg,

corresponding to 0.5 EU/ml for injections at 10 ml/kg.

Thus, the concentration of 0.5 EU/ml was defined as the threshold between

pyrogenic and non-pyrogenic samples.



42

0 5 10 15 200.0

0.5

1.0

1.5

2.0

endotoxin units/ml per kg bodyweight

tem

pera

ture

incr

ease

[°C

]

Fig. 1. Temperature increase of 171 rabbits upon endotoxin injection with

a fitted regression line

The maximum temperature increase in ºC within 180 minutes after endotoxin

injection of 171 rabbits is presented. The mean temperature increase, modeled

with regression techniques, is indicated by the dotted line.

Prevalidation of the novel in vitro pyrogen tests

Before prevalidation, the test-developing laboratories that took part in the study

compiled standard operating procedures for the alternative tests. This required

an intensive phase of test optimization and standardization in order to allow the

transfer of the tests. A standard curve of endotoxin in saline including the 0.5

EU/ml concentration as the threshold for pyrogenicity was included in all tests.

Only if the 0.5 EU/ml endotoxin standard was detectable, did the test run qualify

for analysis. Before prevalidation was started, the naive laboratories proved

evidence of successful transfer of the respective test systems (data not shown).

Prevalidation was then carried out with twelve

blinded samples. These consisted of three drugs spiked with either pyrogen-

free saline (clinical grade 0.9 % NaCl) or with reference endotoxin. Two

negative, i.e. pyrogen-free samples, and two LPS-containing, i.e. pyrogenic

samples (0.5 EU/ml and 1.0 EU/ml sample concentration, respectively) were



43

tested. The concentration of 0.5 EU/ml was the limit of detection defined for the

rabbit pyrogen test (see above). The drugs used were Gelafundin, a volume-

replacement therapy for transfusion with high protein (gelatine) content (B.

Braun Melsungen AG, Melsungen, Germany), Jonosteril, an electrolyte

infusion (Fresenius AG, Bad Homburg, Germany) and Haemate, a factor VIII

preparation (Aventis Behring GmbH, Marburg, Germany). In addition, a positive

control (0.5 EU/ml LPS in saline) and a negative control (endotoxin-free saline)

were included. Each test was performed three times in the respective

developing laboratory (DL) as well as in two naive laboratories (NL).

Test

System

Readout

Ref.

Within-

laboratory

reproduci-

bility (%)

Between-

laboratory

reproduci-

bility (%)

Sensi-

tivity

(%)

Speci-

ficity

(%)

WBT-

IL6

whole

blood

IL-6

136

DL: 83.3

NL1: 94.4

NL2: 100

DL-NL1: 72.2

DL-NL2: 72.2

NL1-NL2: 96.3

72.2

92.6

WBT-

IL1

whole

blood

IL-1β

49

DL: 88. 9

NL1: 95.8

NL2: 94.4

DL-NL1: 91.7

DL-NL2: 76.8

NL1-NL2: 67.8

72.0

100.0

PBMC-

IL6

PBMC

IL-6

143

DL: 94.4

NL1: 100

NL2: 94.4

DL-NL1: 80.6

DL-NL2: 86.1

NL1-NL2: 88.9

87.0

98.1

MM6-

IL6

MM6

(68)

IL-6

136

DL: 100

NL1: 94.4

NL2: 94.4

DL-NL1: 97.2

DL-NL2: 88.9

NL1-NL2: 86.1

72.2

100.0

THP-

TNF

THP-1

clone

TNFα

65

DL: 94.4

NL1: 83.3

NL2: 55.5

DL-NL1: 90.7

DL-NL2: 67.6

NL1-NL2: 65.7

66.7

88.9

THP-

Neo

THP-1

parental

(135)

neo-

pterin

65

DL: 100

NL1: 94.4

NL2: 77.7

DL-NL1: 97.2

DL-NL2: 50.0

NL1-NL2: 51.8

88.9

72.2

Table 1: Novel pyrogen tests and their performance in prevalidation



44

Protocols for all methods are listed in Poole et al. (136) and in the Methods

section. All tests include dilution of the sample by 1:5 with the exception of the

IPT-IL-1 test that requires a 1:12 dilution of the sample. The WBT-IL6 and the

PBMC-IL6 tests combine data from three respectively four blood-donors per

run, the WBT-IL1 from one donor per run. Samples and controls were tested in

quadruplicate in each of the tests. DL denotes developing laboratory, NL1 and

NL2 the two naive laboratories. The sample size analyzed for sensitivity and

specificity was 108 for all tests besides WBT-IL1 (100 samples). Sensitivity

describes the probability to correctly classify positive samples and specificity

describes the probability to correctly classify negative samples.

Table 1 summarizes the six novel test systems used, their major

characteristics, their performance regarding reproducibility, which was

assessed before the blinding code was broken, as well as sensitivity and

specificity. As can be seen, the predictive capabilities of the various tests were

encouraging, particularly in the light of the restricted stability of endotoxin

spikes at the borderline concentration of 0.5 EU/ml. Although all tests were

successfully transferred to the naive laboratories during the preparatory phase

of prevalidation, this optimal performance could not be maintained for the two

test systems using THP-1 cells, as is reflected by the comparatively low

between-laboratory reproducibility between the developing laboratory and one

of the naive laboratories for each. The lower specificity of the THP-Neo test

was entirely caused by misclassification in NL2. Furthermore, prevalidation also

revealed that, despite preceding interference testing and diluting of the drugs

accordingly, interference/recovery problems persisted in some cases, as is

reflected by the values for sensitivity.

Validation phase

For the validation phase,10 drugs with five blinded spikes each (0 (i.e. pyrogen-

free), 0.25, 0.5 (twice) and 1 EU/ml) were tested, again in three laboratories,

i.e. the DL of a test and the two NLs, respectively. To avoid the possibility that



45

different dilutions of the drugs were tested depending on their different

interference with different test systems, all drugs were tested at their maximum

valid dilution (MVD), thus adopting the rationale of the pharmacopoeial BET

reference (limit) test. The MVD is calculated from the endotoxin limit

concentration (ELC in EU/ml) defined for a drug by the European

Pharmacopoeia (146), divided by the threshold of pyrogenicity as the limit of

detection (LoD), i.e. 0.5 EU/ml. Drugs, sources, ELCs and MVDs (= ELCs/LoD,

where LoD=0.5) are summarized in Table 2.

Drug Source Agent Indication ELC

(EU/ml)

MVD

(-fold)

Glucose 5

% (w/v)

Eifelfango

GmbH

glucose nutrition 35 70

Ethanol

13 % (w/v)

B.Braun AG ethanol diluent 17.5 35

MCP Hexal AG metoclo-

pramid

antiemetic 175 350

Orasthin Aventis

Pharma GmbH

oxytocin initiation of

delivery

350 700

Binotal

Aventis

Pharma GmbH

ampicillin antibiotic 70 140

Fenistil

Novartis

Consumer

Health GmbH

dimetinden-

maleat

antiallergic 87.5 175

Sostril

GlaxoSmithKli

ne GmbH

ranitidine antiacidic 70 140

Beloc

Astra Zeneca

GmbH

metoprolol

tartrate

heart

dysfunction

70 140

Drug A* 0.9 % NaCl 17.5 35

Drug B* 0.9 % NaCl 35 70

Table 2: Test substances for the validation phase



46

* Drugs were selected by a selection committee which excluded the developing

laboratories and included experts. Drugs A and B which were saline only were

included as further controls using notional ELCs.

Drugs were obtained from Eifelfango GmbH (Bad Neuenahr-Ahrweiler,

Germany), B. Braun AG (Melsungen, Germany), Hexal AG (Holzkirchen,

Germany), Aventis GmbH (Bad Soden, Germany), Novartis GmbH (München,

Germany), GlaxoSmithKline GmbH (München, Germany) and

Astra Zeneca GmbH (Wedel, Germany). ELCs of drugs were calculated

according to European Pharmacopoeia (146).

While the tests using whole blood, PBMC and MM6 cells performed well in all

three test laboratories in terms of reproducibility (Table 3), technical problems

with the two tests using THP-1 cells were obvious. For the THP-TNF test this

was caused by a batch of TNFα-ELISA plates sent out to the two NLs that did

not satisfy the quality criteria with regard to detection limit when used with cells.

For the THP-Neo test, the technical problems in NL2 persisted such that the

quality criteria defined in the SOP were not met. The tests could not be

repeated due to the limited time frame of validation and for logistical reasons.

Therefore, for the THP-TNF assay only the data from the DL and for the THP-

Neo assay only the data from the DL and from NL1 could be analyzed.

Sensitivity and specificity were 76.7 % and 78.9 % for the THP-TNF assay

(sample size = 40) and 93.3 % and 47.5 % for the THP-Neo assay (sample size

= 100). The data for the other four tests are summarized in Table 3. Almost all

misclassifications, either false negatives or false positives, occurred around or

at the defined classification threshold, i.e. for the contaminations of 0.25 and

0.5 EU/ml. Confidence intervals (CI) with a significance level of 5 % were

calculated for sensitivity and specificity. By focusing on the lower bounds of CI

(Fig. 2), a worst-case scenario can be conducted by which the likelihood of

underestimation of pyrogen content is maximized and thus possible negative

consequences for health can be estimated.



47

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.0 0.1 0.2 0.3 0.4 0.5

1-specificity

sen

sit

ivit

y

MM6-IL6

PBMC-IL6

WBT-IL6

WBT-IL1

rabbit

Fig. 2. Sensitivity and specificity of four in vitro assays in the validation

study and modeled rabbit test performance with 95%-confidence intervals

The sensitivity and specificity resulting from the pre-defined prediction model

and considering samples with 0 and 0.25 EU/ml as non-pyrogenic and with 0.5

and 1 EU/ml as pyrogenic are presented with their corresponding 95 %

confidence intervals for four validated tests. Similarly, the respective

parameters were calculated with the rabbit model. As performance improves

towards the upper left of the graph, all validated tests outperform the rabbit test.

The lower predictive capability of the WBT-IL1 test as compared to the WBT-

IL6 and the PBMC-IL6 test can be explained by the one-donor approach used

for the WBT-IL1 test and the multiple-donor approach used for the other tests

that is more conservative and laborious, but decreases the probability for false-

negative classification. For the THP-TNF assay, the lower bounds of CI for

sensitivity and specificity were 60.6 % and 55.2 %, respectively. For the THP-

Neo assay, the respective lower bounds were 78.0 % and 38.7 %.



48

Applying this kind of analysis to the in vivo assay employing the regression

model based on the data from the rabbit pyrogen test yields a sensitivity of 57.8

% and a specificity of 88.3 % (Table 3) with confidence intervals also presented

in Fig. 2. Thus, the novel pyrogen tests listed in Table 3 show parameters of

performance outperforming the rabbit pyrogen test.

Test Between-

laboratory

reproducibility

Sample

size:

sensitivity#

Sensitivity

(%)

Sample

size:

specificity

Specificity

(%)

WBT-IL6

DL-NL1: 85.4

DL-NL2: 85.4

NL1-NL2:92.0

89 88.9 59 96.6

WBT-IL1

DL-NL1: 72.9

DL-NL2: 81.6

NL1-NL2:70.2

88 72.7 59 93.2

PBMC-IL6

DL-NL1: 84.0

DL-NL2: 86.0

NL1-NL2: 90.0

90 92.2 60 95.0

MM6-IL6

DL-NL1: 90.0

DL-NL2: 89.6

NL1-NL2: 83.3

89 95.5 59 89.8

Rabbit† - - 57.9 - 88.3

Table 3: Validation of the predictive capability of novel pyrogen tests

# sample sizes are reduced by outlier exclusion defined in the study protocol † parameters calculated by the fitted regression model

An additional analysis, which could be conducted with the available data,

supports this conclusion. According to their SOPs, the four systems included an

uncontaminated negative control, i.e. saline, and another positive control of 1

EU/ml. For each of these two controls we adapted the prediction model



49

described above: First, we compared the blinded samples against the

response of 1 EU/ml control. Therefore, we constructed a modified prediction

model using the 1 EU/ml control response instead of the positive control of 0.5

EU/ml, which is denoted in the following by PM1. In doing so, the true

classification of the samples changed, as now only the samples spiked with 1

EU/ml were considered as pyrogenic and the other samples as non-pyrogenic.

Second, with a modified prediction model, denoted as PM0, classifying a

sample as pyrogenic when the response was significantly larger than the

negative control response (significance level 1 %), we compared all spikes

against this control. Again, the true classification of the samples needed to be

adjusted considering the contaminated samples (0.25, 0.5, 1.0 EU/ml) as

pyrogenic and the unspiked samples as non-pyrogenic. The resulting

sensitivities and specificities are summarized together with the results from the

original PM for the four test systems in Fig. 3. All tests performed best for PM0,

where the sum of these two parameters was at least 1.90, while WBT-NI even

resulted in the maximum sum of 2.

0.95 0.96 0.97 0.980.92

0.87

1.000.89

0.69

0.94

0.73 0.72

0.58

0.900.98 0.97

0.95 0.98

1.00

0.97

0.99

0.96

0.93 0.97

0.88

1.00

0

1

2

PM0 PM PM1 PM0 PM PM1 PM0 PM PM1 PM0 PM PM1

MM6-IL6 PBMC-IL6 WBT-IL6 WBT-IL1 rabbit

su

m o

f sen

sit

ivit

y a

nd

sp

ecif

icit

y

specificity

sensitivity

Fig. 3. Sum of sensitivity and specificity resulting from three prediction

models for four in vitro assays in the validation study



50

The validation data of four tests were analysed with three prediction model

employing different controls for comparison and thus defining the true

classification of the samples (non-pyrogenic vs. pyrogenic) accordingly. The

test accuracy is described for each test and prediction model by the sum of

specificity and sensitivity allowing also for individual parameter assessment.

For comparison, the rabbit test performance according to the pre-defined

prediction model is added. The DLs also tested lipoteichoic acid (LTA) from

Bacillus subtilis, a BET-negative Gram-positive compound that activates

cytokine release from human monocytes (26, 57) prepared according to Morath

et al. (57), which was clearly detectable by the novel tests.

4.5. Discussion

Previous work (49, 65, 67, 70, 136, 143, 147, 150, 151) had established that

different sources of human monocytoid cells are valuable tools for mimicing the

human fever reaction in vitro. Not only can these cells detect the important

pyrogen LPS from E. coli and other Gram-negative bacteria but also a number

of compounds involved in the immune response to Gram-positive bacteria such

as LTA (58), exotoxins (67, 148), cell wall components like muramyl dipeptide

(148) or peptidoglycan (149), S. aureus Cowan (SAC) (147) or DNA (67) as

well as poly (I:C) (147), a synthetic double-stranded RNA used as a virus model

compound in fever research. It was also established that these novel test

systems overcome limitations of the BET and yield results comparable to the

rabbit pyrogen test (64, 67, 147, 148,). For the first time, six of these

monocytoid-cell based in vitro pyrogen tests were formally validated in the

present study. For this purpose, a harmonized analysis procedure was

established that allowed the direct comparison of the different tests and

incorporated various safety aspects. A conservative statistical approach

showed that four test systems met the criteria for safe detection of pyrogens.

The two test systems based on the use of THP-1 cells posed problems in

performance. These were related to insufficient transfer to one naive laboratory

(THP-Neo) and to use of an ELISA batch for the one-plate assay format (THP-



51

TNF) that, although qualifying for the detection of TNFα did not qualify for the

use with cells and caused their prestimulation. Both problems became obvious

only during validation and could not be overcome within the tight schedule of

validation. Thus, for these two systems additional validation processes would

be required. However, the data obtained for the other four test systems clearly

suggest that these have reached a stage of development that makes them

suitable for use in pyrogen testing as replacements for the rabbit pyrogen test.

For the purpose of this study, a threshold value of 0.5 EU/ml was chosen on the

basis of historical data from rabbit tests carried out in a national control

authority. This approach was conservative as only 50% of animals of the very

sensitive strain used showed a febrile reaction at this concentration.

Additionally, in order to be classified negative, the samples had to be,

according to the PM, significantly lower than 0.5 EU/ml. On the one hand, the

enormous challenge to the models by placing two samples at the threshold of

0.5 EU/ml, which had to be classified positive, resulted in reduced sensitivities.

On the other hand including a sample with 0.25 EU/ml, which had to be

identified as negative, was the reason for almost all false-positive classifications

resulting in the reduced specificities without representing any safety concern.

When tested against the negative control (PM0), i.e. when samples which are

not significantly different from the negative control were considered as pyrogen-

free, the tests performed even better, i.e. with increased sensitivity. However,

this approach increases consumers’ safety on the cost of rejecting drugs,

whose minor pyrogenic contamination would not induce adverse health effects

in humans. At the same time, this reflects the fact that the study design put

main emphasis on the threshold of 0.5 EU/ml. Similarly, decreased sensitivity

when given the task of identifying 1 EU/ml as threshold value shows that the

tests were especially designed for the threshold of 0.5 EU/ml. Since the test

performance when changing the threshold is still acceptable or even better, the

robustness of the alternative tests is underlined.

In summary, this study provides thus evidence of the validity of these tests and

should facilitate the regulatory acceptance of these novel tests and lead to their

introduction into Pharmacopoeias.



52

Conflict of interest

S. Poole is named as an inventor in Patent Number US 6,696,261 B2 , Feb 24,

2004: 'Pyrogenicity test for use with automated immunoassay systems'.

T. Hartung and A. Wendel are named as inventors in Patent Number US

5,891,728 , Apr 6, 1999: 'Test for determining pyrogenic effect of a material'.

4.6. Acknowledgements

We thank U. Lüderitz-Püchel from the Paul Ehrlich Institute, Langen,

Germany, for providing rabbit pyrogen test data.

This work was supported by the European Union [QLRT-1999-00811].

CRYOPRESERVATION OF HUMAN WHOLE BLOOD FOR PYROGENICITY

TESTING

53

5 Cryopreservation of human whole blood for pyrogenici-

ty testing

Stefanie Schindler, Silvia Asmus, Sonja von Aulock, Albrecht Wendel,

Thomas Hartung*, Stefan Fennrich

Biochemical Pharmacology, University of Konstanz, D-78457 Konstanz,

Germany

Corresponding address:

Thomas Hartung, PhD, MD

University of Konstanz

D-78457 Konstanz

Tel: +49-7531-88-4116

Fax: +49-7531-88-4117


Abbreviations:

DMSO, dimethylsulfoxide; ELC, endotoxin limit concentration; EU, endotoxin

equivalent units; IL-1β, interleukin-1β; LPS, lipopolysaccharide, endotoxin;

LTA, lipoteichoic acid; MVD, maximal valid dilution; RT, room temperature;

WHO, World Health Organization

5.1. Abstract

Human whole blood assays are increasingly employed to test immune functions

or detect pyrogenic contaminations, since they offer advantages such as easy

performance, few preparation artifacts and physiological cell environment. The

approach, however, is often limited by availability of freshly drawn blood,

putative safety concerns in case of infected donors and interindividual donor

differences. To overcome these limitations, a method was developed and

optimized to produce batches of cryopreserved blood that can be used directly

after thawing without any washing steps. Mononuclear cells remained intact as


TESTING

54

shown by FACS analysis. Cytokine release could be induced by a variety of

immunological stimuli. The cell preparation released higher amounts of IL-1β

and IL-6 compared to fresh blood, which could be attributed to the presence of

the cryoprotectant DMSO. Large batches of cryopreserved blood could be

produced by mixing blood donations of up to ten donors, independent of

differing blood groups. The detection limit for the WHO LPS reference

preparation (EC-6) with regard to induction of IL-1β release was at least 0.5

EU/ml. Endotoxin spikes at the limit concentrations prescribed by European

Pharmacopoeia could be detected in a series of drugs, showing that the In vitro

Pyrogen Test (IPT) can also be run with cryopreserved blood. Further possible

applications include high-throughput screening for immunomodulators or toxins

as well as preservation of patient samples for later analysis of cell functions.

Key words: Cryopreservation, Blood, Endotoxin, Interleukin-1β, In vitro

Pyrogen Test (IPT)

5.2. Introduction

Cryopreservation of cells represents a standard procedure in cell culture.

Human primary leukocytes are cryopreserved on a routine basis, for example to

store human bone marrow cells (151). Further cryopreservation protocols have

been established for various isolated blood cell populations including

lymphocytes and mononuclear cells and the retention of various cell functions

after thawing has been investigated (72, 152-155).

Although it is popular to isolate the respective immune cells from blood, it is

evident that such isolated cells do not reflect the in vivo situation: the cells are

often stimulated during the isolation procedure as indicated by basal mediator

release or adherence of the cells, interaction between different cell types

cannot take place and plasma components that often play an important role in

immune recognition are no longer present.

Methods employing whole blood have been developed to detect pyrogenic

(fever-inducing) contaminations, e.g. of batches of injectable drugs (49). This


TESTING

55

application has been successfully validated in a collaborative European study

and awaits incorporation into the pharmacopoeias (71). Further, we have

suggested the study of cytokine, histamine or eicosanoid release with this

method to allow the characterization of putative drugs or immunotoxins (156,

157). These methods can also be used ex vivo on treated volunteers or

patients to monitor the course and effects of treatment (158-160).

Many of these procedures could be simplified or optimized by the availability of

cryopreserved whole blood. The blood could be supplied in the form of a

standardized test reagent which could be stored until needed and be certified

free of infectious agents. A method to preserve and store cells from treated

patients might allow performance of the often laborious cell assays on a series

of collected samples in parallel or at a distant laboratory, thus reducing

variability and logistical problems.

We sought to develop a protocol which would allow the use of the thawed

whole blood samples directly without any washing steps to remove the

cryoprotectant, as such a step would eliminate essential advantages of the

human whole blood assay, i.e. the ease of performance which allows a high

degree of standardization as shown for various applications (161). Furthermore,

beside stress and handling artifacts, the cells would lose their autologous

plasma that enables a number of physiological responses, e.g. the sensitive

response to lipopolysaccharides (endotoxin, LPS) via lipopolysaccharide

binding protein (LBP) (162, 163).

In this report we describe the development of a protocol to freeze human whole

blood and demonstrate retention of sensitivity and functionality regarding

stimulation of cytokine release in response to inflammatory agents.

5.3. Materials and methods

Freezing procedure

Blood was drawn from healthy volunteers into tubes containing 15 IU/ml Li-

Heparin (Sarstedt, Nürnbrecht, Germany) and differential blood cell counts

were performed on each sample to rule out active infections (Pentra 60, ABX


TESTING

56

Diagnostics, Montpellier, France). In order to rule out pyrogenic contaminations

of any component used in the incubations, negative saline controls were

included in each experiment The heparinized blood was pre-cooled in ice water

for 15 minutes. Clinical grade dimethylsulfoxide (DMSO, Waco Chemicals,

Dessau-Thornau, Germany) was added to the blood in 50 ml centrifugation

tubes (Greiner bio-one, Frickenhausen, Germany) in small amounts to a final

concentration of 10% (vol/vol ratio) under constant gentle agitation to avoid cell

damage. Pooling was performed in 50 ml centrifugation tubes after the addition

of DMSO to the blood of the individual donors. Blood was pipetted as 1, 3 or 4

ml aliquots into pre-cooled cryotubes (1.8, 3.6 or 4.5 ml, Nunc, Wiesbaden,

Germany) and put into the rack of a programmable freezer with a TP type

nitrogen container (Nicool Plus PC, Air Liquide, Marne-la-Vallée Cedex 3,

France), pre-cooled to 4°C. A temperature probe was inserted into an extra

aliquot containing the same volume of blood to follow the freezing process. The

freezing program was started 5 min after closing the freezer. The blood was

cooled down to – 5°C at a rate of 1°C/minute. In order to compensate the latent

fusion heat generated by the blood when changing from the liquid to the solid

state, the temperature Tx in the freezing chamber was set to – 30°C. The

crystallization temperature was –12°C. When this temperature was reached,

the blood was cooled down to – 40°C at a rate of 2°C/min. The blood was given

120 seconds to stabilize before being cooled down to –120°C at a rate of

10°C/minute. After freezing, the tubes were removed from the freezer and put

immediately into the vapor phase of liquid nitrogen (nitrogen tank, Air Liquide,

Kryotechnik, Düsseldorf, Germany).

Thawing procedure

The closed tubes were left in an incubator at 37°C until completely thawed.

The aliquots of single donors were either pooled or the blood was pipetted

individually from each aliquot. Pooling of the blood of different donors could be

performed after thawing as an alternative to the procedure described above.

The whole blood incubation was started not more than 30 minutes after

complete thawing.


TESTING

57

Whole blood incubation

Human whole blood incubations were performed according to the protocol of

the in vitro Pyrogen Test (IPT; 49, 71). Briefly, 100 µl of fresh or cryopreserved

human blood were added to 1 ml physiological, clinical grade, pyrogen-free

saline in polypropylene reaction tubes (Eppendorf, Hamburg, Germany). After

the addition of stimuli, the tubes were closed, shaken gently and incubated

overnight (16-24 hours) at 37°C. The cells were resuspended and assayed

immediately or frozen until measurement. Within each experiment performed,

all samples were incubated and measured in parallel. When all samples of an

experiment were measured on the same ELISA plate, absorbance (A 450) was

given as the unit of measurement according to the IPT protocol. When the

samples could not be measured on the same ELISA plate, a recombinant

standard curve was run on each ELISA plate to allow interpolate comparison.

Endotoxin stimuli were LPS from Escherichia coli O113 (WHO standard

material), kindly provided by Dr. Stephen Poole, NIBSC, Hertfordshire, GB, or

LPS from E. coli O111 (IPT Kit, Charles River Endosafe, Charleston, SC, USA)

calibrated to the WHO standard material. One important criterion for the In Vitro

Pyrogen Test (IPT) is the ability to reproducibly detect the presence of 0.5 EU

(endotoxin equivalent units) per ml, equivalent to 50 pg/ml of the WHO

reference endotoxin standard or to 100 pg/ml of the LPS from E. coli O111,

respectively, in a sample solution, this being the fever threshold of the most

sensitive rabbit strain if applied at a dose of 10 ml/kg. Therefore, this LPS

concentration was included in every assay.

Non-endotoxin stimuli were lipoteichoic acid (LTA) from Bacillus subtilis (IPT

Kit, Charles River Endosafe) (49), glucan standard (Charles River Endosafe),

glucan from barley (Sigma, Munich, Germany), lectin from Phaseolus vulgaris

(PHA-L and PHA-E, Sigma), curdlan (Sigma) and zymosan A (Fluka, Buchs,

Switzerland).

Substances tested at MVD were furosemid (Lasix®), ampicillin (Binotal®),

Articain/Epinephrin (Ultracain®) (Aventis, Germany), Theophyllin

(Bronchoparat®), (Fujisawa, Munich, Germany), dimethindenmaleat

(Fenistil®) (Novartis, Munich, Germany), ranitidin (Sostril®) (Glaxo Smith


TESTING

58

Kline, Munich, Germany) and metoprolotartrat (Beloc®) (Astra Zeneca, Wedel,

Germany).

Cytokine ELISAs were based on commercially available antibody pairs against

IL-1β or TNFα (Endogen, Biozol, Eching, Germany), and IL-6 (R&D,

Wiesbaden, Germany). Binding of biotinylated antibody was quantified using

streptavidin-peroxidase (Biosource, Camarillo, CA, USA) and the substrate

TMB (3,3’,5,5’-tetramethylbenzidine, Sigma). Recombinant cytokines serving as

standards were gifts from Dr. S. Poole, NIBSC.

FACS Analysis

25 µl of fresh or cryopreserved blood was stained with 5 µl each of anti-CD45-

APC and anti-CD14-FITC antibodies (BD Biosciences, Heidelberg, Germany)

for 30 min at room temperature in the dark. 1 ml Cell Wash and propidium

iodide in a final concentration of 500ng/ml were added directly, immediately

before measurement in a FACSCalibur (all BD Biosciences). A live gate was

set on CD45-positive cells and 3000 leukocytes were counted. Whole blood

counts were determined by Türks staining and counting in a Neubauer

chamber.

Statistics

Statistics were performed with GraphPad InStat 3.0 (GraphPad Software, San

Diego, USA). Significance was tested by one-way ANOVA and Dunnetts post-

test/Dunn´s multiple comparison and with t-test, followed by Mann-Whitney

post-test.

5.4. Results

Freezing procedure

Different concentrations of the cryoprotectant DMSO were tested to determine

a concentration that would protect the cells and leave them functional after

thawing but which would be sufficiently low to have no toxic effects in the


TESTING

59

incubation (Fig. 1). While 1% DMSO was insufficient to protect the cells during

freezing and 20% DMSO had toxic effects in the subsequent incubation period,

cell preparations frozen with either 5 or 10% DMSO responded to stimulation

with LPS in a concentration-dependent manner after thawing.

0.0

0.5

1.0

1.5con0.5 EU/ml1.0 EU/ml

1% 5% 10% 20%DMSO (v/v)

ns

*

OD

[450n

m]

±± ±± S

D

Fig. 1. Optimization of the final DMSO concentration in cryopreserved

blood.

Blood was frozen with different concentrations of DMSO as shown and

stimulated with LPS from E. coli O113 after thawing (representative experiment

of 2). Blood from one donor in 4 replicates is shown, *, p<0.05 (one-way-

ANOVA, post test: Dunn´s multiple comparison). Incubation supernatants were

measured by ELISA technique and the endpoint IL-1β was

given as OD.

We compared whether the reactivity of the cryopreserved blood measured as

IL-1β response to endotoxin stimulation was affected by the blood temperature

(room temperature or 4°C) at which the DMSO was added and whether DMSO

should be added as a bolus or in several aliquots. The addition of DMSO at

room temperature seemed to cause an increase in reactivity rather than a

decline and addition of DMSO in several aliquots was preferable to the bolus.


TESTING

60

The mean OD of the aliquots with DMSO added at room temperature and

stimulated with 0.5 EU/ml O113 was 0.596 OD (SD 0.049) when the DMSO

was added as a bolus and 0.728 OD (SD 0.051) when the DMSO was added in

three aliquots. The cv was 8.17 and 7.02%, respectively. When the DMSO was

added at 4°C, the response was 0.404 OD (SD 0.056) when the DMSO was

added as a bolus and 0.547 (SD 0.034) when added in three aliquots (cv 13.82

and 6.29%).

Next, we determined how long the blood could be kept after addition of DMSO

before freezing and whether room temperature or 4°C is preferable. For this

purpose, DMSO was added to the blood and an aliquot was frozen immediately

while other aliquots were stored at room temperature or at 4°C for up to 200

minutes and then frozen and tested in parallel (Fig. 2). These data suggest that

storage at room temperature for up to 2 hours is tolerable and that storage at

4°C is beneficial when the blood is stored for longer.

0

2500

5000

7500

1.0 EU, RT

frozen immediately (25min)

110min 200min

storage time of blood + DMSO before freezing

0.5 EU, RT 0.5 EU, 4°C

1.0 EU, 4°C

IL1

ββ ββ [

pg

/ml]

±± ±± S

D

Fig. 2: Comparison of different storage temperatures and durations

before freezing.

Blood with 10% DMSO was frozen immediately or stored as indicated before

freezing, then stimulated with LPS from E. coli O111 after thawing

(representative experiment of two). Control values were < 6 pg/ml IL-1β for


TESTING

61

each condition. Four replicates of blood from one donor are shown (0.5 EU/ml:

n=5).

Comparison of different volumes

Different volumes of blood (1 ml, 3 ml and 4 ml aliquots) were frozen and

stimulated with endotoxin after thawing. The reactivity of the blood did not

depend on the volume of the frozen aliquots. The mean OD of the 1, 3, and 4

ml aliquots when stimulated with 1.0 EU/ml of O113 was 2.824 (SD 0.066),

2.463 (0.058) and 2.6 (0.087) OD, respectively, with a coefficient of variation of

2.32 , 2.35 and 3.35%.

Thawing procedure

A thawing protocol was developed in order to optimize the handling of the blood

aiming at maximum reactivity and viability. Aliquots of blood from the same

donors were thawed under different conditions, i.e. on ice, at room temperature

(20°C) and in an incubator (37°C) until completely thawed before stimulation

with endotoxin (Fig. 3). Quick thawing at 37°C resulted in the best response.

0.0

0.1

0.2

0.3

0.4

0.5

4°C RT (20°C) 37°C

***

***

OD

[450n

m]

±± ±± S

D

Fig. 3: Determination of a suitable thawing temperature.


TESTING

62

Cryopreserved blood with 10% DMSO was thawed at different temperatures

(4°C, room temperature, 37°C), then stimulated with 0.5 EU/ml (hatched bars)

and 1.0 EU/ml (black bars) LPS from E. coli O113 or saline (white bars)

(representative experiment of 4). Blood from one donor in 8 replicates is shown.

***, p<0.001 against the respective stimulation after thawing at 4°C or 20°C

(one-way ANOVA with Dunnett’s post-test). Incubation supernatants were

measured by ELISA technique and the endpoint IL-1β was given as OD.

An important issue was the potential cytotoxicity of the remaining

cryoprotectant DMSO after thawing and before dilution with saline. Therefore,

we tested how long the thawed blood samples could be kept at 37°C before

dilution and stimulation (Fig. 4). The reactivity of the blood towards the

endotoxin stimulation decreased after 45 minutes of thawing time. Therefore,

the blood was used within 30 minutes after thawing at 37°C in all subsequent

experiments.

Fig. 4: Effect of time between thawing of blood and incubation.

Blood was thawed and stored at 37°C for the times indicated, then stimulated

with 0.5 EU/ml LPS from E. coli O113 (representative experiment of 4). Blood

from one donor in five replicates is shown. ** = p<0.01 vs. the values at 15 min

15 30 45 600.0

0.1

0.2

0.3

0.4

0.5

control

0.5 EU/ml

****

ns

time between thawing and incubation [min]

OD

[450n

m]

±± ±± S

D


TESTING

63

(one-way ANOVA with Dunnetts post test). Incubation supernatants were

measured by ELISA technique and the endpoint IL-1β was given as OD.

FACS Analysis

Differential blood cell counts were done in parallel samples of fresh and

cryopreserved blood from 5 donors. Although the whole blood cell counts of the

cryopreserved blood did not differ from those of the fresh blood samples,

the differential blood cell count revealed that the neutrophilic granulocytes had

lost their surface markers and could no longer be identified as live, CD45

positive cells. The ratio of monocytes to lymphocytes in the differential blood

cell count was the same in the fresh and the cryopreserved blood (1 : 6.7 ± 0.9

versus 1 : 8.1 ± 1.6, n.s.) with a viability of these two populations of 99 vs. 90%

as shown by propidium iodide exclusion.

Establishment of a pooling protocol

Blood samples from five different donors were compared with each other and

with pools of the blood from the same donors combined either directly after

addition of the DMSO or after thawing of frozen blood. Establishing a pooling

protocol with blood from different donors with different blood groups proved

easier than anticipated (Fig. 5). There was no difference in the reactivity of the

blood pools, whether they were made before or after freezing. Also, the

reaction of the pooled blood was equal to the mean of the reaction of the

individual donors.


TESTING

64

0.5 EU

1.0 EU

0.0

0.5

1.0

1.5

0.5 EU

1.0 EU

1 2 3 4 5 pool 1 pool 2

OD

[450n

m]

±± ±± S

D

Fig. 5: Comparison of the reactivity of frozen blood from 5 donors and

their pooled blood.

Blood from five separate donors as well as a pool of their blood was frozen and

stimulated as shown (representative experiment of 4). The horizontal lines

indicate the calculated mean of the blood from the five donors to 0.5 or 1.0

EU/ml LPS from E. coli O113. Three replicates of all samples were measured

(0.5 EU/ml: 4 replicates). Pool 1, the blood of the single donors was pooled

after addition of DMSO; pool 2, the blood was pooled after thawing. Incubation

supernatants were measured by ELISA technique and

the endpoint IL-1β was given as OD.

Interlot variability

5 different pools of cryopreserved blood, each containing the blood of 5 donors,

were compared (Fig. 6). The interlot variability was very low, indicating that the

use of 5 donors in the pooling protocol is sufficient for producing highly similar

batches of blood.


TESTING

65

0.0

0.1

0.2

0.3

0.4

0.5

0.6

batch 142 batch 156 batch 161 batch 162

controlO-113; 0,25 EU/mlO-113; 0,5 EU/mlO-113; 1,0 EU/mlO-113; 2,0 EU/mln.s.

n.s.n.s.

OD

[450n

m]

±± ±± S

D

Fig. 6: Interlot variability of five different pools

Cryopreserved pools each consisting of five different donors and frozen over

a period of 23 weeks were thawed and stimulated on the same day (0.25

EU/ml, 2 EU/ml: 2 replicates each; control, 0.5 EU/ml, 1 EU/ml: 4 replicates

each) with LPS from E. coli O113. p > 0.05 of the 1.0 EU/ml value of batch 142-

161 against the 1.0 EU/ml value batch 162 (one-way ANOVA, Dunnett´s post-

test). Incubation supernatants were measured by ELISA technique and the

endpoint IL-1β was given as OD.

Stability

Numerous aliquots of a pool of blood from 5 donors were frozen and their

reactivity tested on different days over a period of 4 months. The IL-1β

response to 0.5 EU/ml endotoxin was significantly different from the saline

controls at each of the time points tested, indicating that the cryopreserved

blood remained stable over this time period and did not lose sensitivity

(Table I).


TESTING

66

Day after

freezing

Mean OD

saline control

Mean OD

0.5 EU/ml

Reactivity

(% saline control)

0 0.045 ± 0 0.311 ± 0.06 691

40 0.068 ± 0.01 0.298 ± 0.02 439

118 0.077 ± 0.01 0.755 ± 0.05 980

Table I: Stability of pooled cryopreserved blood from 5 donors over a

period of four months.

Blood was stimulated with 0.5 EU/ml LPS from E. coli O113, n=4. Incubation

supernatants were measured by ELISA technique and the endpoint IL-1β was

given as OD.

To determine inter-aliquot variability of aliquots from the same blood donor,

eight replicates each from 3 aliquots of 1 ml thawed cryopreserved blood

were stimulated with 0.5 EU/ml endotoxin and eight replicates each from three

aliquots were left unstimulated (Table II). The stimulated samples had mean

ODs of 0.27 – 0.49 and the coefficient of variation (cv) was 12.3 – 26.1%, while

the unstimulated samples had mean values ranging from 0.047 to 0.054 OD

and a cv of 5.4 – 42.2%.

The inter-aliquot variability of the same experiment was 0.051 ± 0.004 (cv

7.3%) for unstimulated versus 0.37 ± 0.11 (cv 31.3%) for blood stimulated with

0.5 EU/ml LPS.

aliquot 1

saline

aliquot 2

saline

aliquot 3

saline

aliquot 4

0.5 EU/ml

aliquot 5

0.5 EU/ml

aliquot 6

0.5 EU/ml

Minimum 0.040 0.043 0.048 0.242 0.216 0.384

Median 0.046 0.047 0.053 0.259 0.317 0.495

Maximum 0.107 0.051 0.066 0.339 0.490 0.582

Mean 0.053 0.047 0.054 0.274 0.327 0.493

SD 0.022 0.003 0.006 0.036 0.085 0.061


TESTING

67

SEM 0.008 0.001 0.002 0.012 0.030 0.021

cv (%) 42.17 5.4 10.66 12.83 26.08 12.26

Table II: Intra-aliquot variability of cryopreserved blood from one donor.

8 replicates from each aliquot of blood of the same donor were stimulated with

LPS from E. coli O113. Incubation supernatants were measured by ELISA

technique and the endpoint IL-1β was given as OD.

Comparison of the reactivity of cryopreserved with fresh whole blood

The reactivity of the cryopreserved blood to endotoxin stimulation was

compared to that of fresh blood of the same individual donors. As can be seen

in Figure 7, 0.5 EU/ml LPS induced significant IL-1β release both in the fresh

and the cryopreserved blood of every donor. This is the sensitivity limit

of the most sensitive rabbit strain for testing according to the European

Pharmacopoeia for injectable drugs.

0.0

0.1

0.2

0.3

0.4

0.5

0.6fresh blood, saline

fresh blood, 0.5 EU/ml

cryopreserved blood, saline

cryopreserved blood, 0.5 EU/ml

donor 1 donor 2 donor 3 donor 4 donor 5

*

***

*

***

****

* ***

*

***

OD

[450n

m]

±± ±± S

D

Fig. 7: Comparison of the reactivity of fresh and frozen blood of 5

separate donors.


TESTING

68

Fresh blood (4 replicates each) and thawed cryopreserved blood (6 replicates

each, saline control n=8) from the same five donors was stimulated with LPS

from E. coli O113. Cryopreserved blood was thawed immediately after

complete freezing and was incubated in parallel with the fresh blood of the

same donors (representative experiment of 3).

*, p<0.05, *** p<0.001, against the respective saline control (t-test and Mann-

Whitney post-test). Incubation supernatants were measured by ELISA

technique and the endpoint IL-1β was given as OD.

When the response to endotoxin stimulation of cryopreserved and fresh blood

from the same donors was compared in a kinetic study, a noticeable difference

in the kinetics of the LPS-inducible IL-1β release was observed (Figure 8, upper

panel). Measurable IL-1β release occurred with several hours delay in

cryopreserved compared to fresh blood. This could be attributed to the

presence of the cryoprotectant, since fresh blood containing 10% DMSO

showed the same delay. Furthermore, in both cases, the presence of DMSO

increased the maximum amount of IL-1β released 7-fold (fresh blood plus

DMSO) and 5-fold (cryopreserved blood). The same held true for IL-6 (Fig. 8,

center panel), though here the amount of IL-6 was increased nearly 20-fold.

TNFα release was no longer detectable after addition of DMSO (Fig. 8, lower

panel), both in fresh and cryopreserved blood.


TESTING

69

0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 40

1 0 0 0

2 0 0 0

f r e s h b lo o d

f r e s h b lo o d + D M S O

c r y o p r e s e r ve d b lo o d

2 5 0 0

5 0 0 0

7 5 0 0

1 0 0 0 0

in c u b a t io n t im e [ h ]

IL-1

ββ ββ [

pg

/ml]

±± ±± S

D

0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 40

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

5 0 0 0

1 0 0 0 02 0 0 0 03 0 0 0 04 0 0 0 05 0 0 0 06 0 0 0 0


IL-6

[p

g/m

l]±± ±±

SD

0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 40

5 0 0

1 0 0 0

1 5 0 0


TN

F [

pg

/ml]

±± ±± S

D

Fig. 8: Kinetics of cytokine response of fresh blood, fresh blood after

addition of 10% DMSO and cryopreserved blood.

Three replicates of blood samples pooled from five donors were challenged

with 1.0 EU/ml LPS from E. coli O111 for the times indicated.

Next, the reactivity of cryopreserved blood to a variety of immune stimuli was

tested in comparison to fresh blood. Different pyrogenic (fever-inducing) stimuli

including LPS, LTA, and phytohaemagglutinin-L (PHA-L) induced IL-1β release

in cryopreserved blood, but not the non-pyrogenic substances PHA-E, glucans

and monophosphoryl lipid A (data not shown). Differences were seen for

curdlan and zymosan A and, in terms of a higher sensitivity of the

cryopreserved blood, for endotoxin from Pseudomonas aeruginosa. Taken


TESTING

70

together, the cryopreservation procedure did not alter the spectrum of pyrogens

or immune stimuli detected and did not cause the cryopreserved blood to react

to substances which fresh blood does not react to either.

To test whether cryopreservation might interfere with the detection of

contaminations in pharmaceuticals, e.g. as a result of hemolysis, the

freezing/thawing procedure or the DMSO, a series of drugs was tested with

fresh and cryopreserved blood as to their interference with a given LPS spike.

Results are summarized in Table III.

Trade

name

Drug ELC

(EU/ml)

MVD MID

cryoblood

MID

fresh blood

Lasix furosemid-sodium 15 30 1 :10 1 :30

Ultracain articain/epinephrine 75 150 1:150 Not test.

Binotal ampicillin-sodium 75 150 1:50 1:100

Broncho

-parat

theophylline 37.5 75 1:25 1:75

Fenistil dimethindenmaleate 93.75 187,5 1:150 1:180

Sostril ranitidine 75 150 1:20 1:100

Beloc metoprolotartrate 75 150 1:40 1 :50

Table III: Interference testing of clinical-grade parenterals in fresh and

cryopreserved blood.

Drug samples were added to fresh blood of 2 donors (n=4) or cryopreserved

blood of one donor (n=4) and spiked with 0.5 EU/ml LPS from E. coli O113.

For positive spike retrieval, 50-200% of the response to the LPS spike in

saline had to be found in the spiked drug sample. ELC, endotoxin limit

concentration according to European Pharmacopoeia; MVD, maximum valid

dilution (ELC/0.5 EU/ml); MID minimum interference dilution. Incubation

supernatants were measured by ELISA technique and the endpoint IL-1β was

given as OD.


TESTING

71

The interferences differed, though surprisingly, the cryopreserved blood

proved to be less prone to interference than the fresh blood: the minimal

interference dilutions were lower for the cryopreserved blood and always at or

below the maximal valid dilution (MVD). These data show that a broad variety

of drugs can be controlled by a pyrogen test based on cryopreserved blood,

maintaining the endotoxin limit concentration (ELC) according to the

pharmacopoeias established for the Limulus amoebocyte lysate assay

without that test’s restriction to Gram-negative LPS. However, interference

testing must be performed in every case for both fresh and cryopreserved

blood for any given drug.

5.5. Discussion

The utility of human whole blood assays has been demonstrated in a broad

variety of applications. All applications so far have to be carried out within a

few hours after blood withdrawal, which makes parallel processing difficult,

leading to a high variability in clinical samples, complicating donor pre-testing

and posing problems with regard to continuous availability of fresh blood

samples.

The procedure described here offers a continuous supply with a highly

homogenous batch of blood. A regular blood donation (500 ml) would suffice

for 5000 tests (tube format) or 25.000 tests (microtiter plate format), which

can be increased even further by pooling blood from several donors before

freezing. Additionally, blood batches can be pre-tested with regard to

sensitivity, and infections such as HIV or hepatitis can be excluded. The

latter, however, can also be achieved by prescreening donors following the

standard guidelines for blood donation.

The established procedure has been optimized with regard to the retention of

sensitive cytokine response to pyrogenic contamination. Both the freezing

and the thawing protocol were optimized and the reactivity of the

cryopreserved blood was compared to that of fresh blood. The inter- and

intra-aliquot variabilities were tested, as was the reaction of cryopreserved


TESTING

72

blood to different pyrogenic and non-pyrogenic substances in comparison to

the reaction of fresh blood.

The differential blood cell counts of fresh versus cryopreserved blood showed

that monocytes and lymphocytes are still viable after the cryopreservation

procedure, thus implying that functional assays of these cells can still be

performed efficiently. Also the ratio of monocytes to lymphocytes remained

unchanged. However, the surface properties of neutrophilic granulocytes

were affected by the cryopreservation procedure, suggesting that neutrophil

function may be lost. The established pooling procedure allows the

preparation of large batches of cryopreserved blood and also reduces the risk

of possible abnormal individual reactions.

Apart from other application possibilities, the results presented indicate that

cryopreserved blood can be used as an alternative to fresh blood in the In

vitro Pyrogen Test to detect contaminations in batches of different drugs.

These first data show that a broad variety of drugs could be controlled using

the cryopreserved blood, maintaining the endotoxin limit concentration (ELC)

established for the Limulus amebocyte lysate assay without that test’s

restriction to Gram-negative LPS.

However, the hemolysis, dead PMN as well as the DMSO might result in

interferences with some drugs, e.g. drugs that bind to hemoglobin. It is also

possible that the synergy of LPS and hemoglobin influences the results.

Therefore, the suitability of cryopreserved blood for pyrogen testing of a given

product will have to be demonstrated by separate interference testing.

This approach may also find application in high-throughput screening. It is

tempting to base screening assays on homogenous preparations of human

primary cells, requiring no prior culture or isolation procedures. Given the

broad variety of immunomodulators in clinical use, several application

opportunities can be imagined. Since the whole blood model also allows the

determination of eicosanoid release (157), this might extend to modulators of

eicosanoid formation such as NSAID (non-steroidal anti-inflammatory drugs).

However, the feasibility and relevance of this approach will have to be

established.


TESTING

73

Taken together, a variety of immunological models and tests might benefit

from the availability of functional, cryopreserved blood. Efforts to make this

available on a large scale are ongoing. The advantages and respective

adaptations for different uses will have to be established. The approach

promises, however, to overcome problems of availability and standardization

of human primary blood leukocytes and to provide standardized blood as an

immunological reagent for a broad spectrum of applications.

5. 6. Acknowledgements

This work was supported by the Bundesministerium für Bildung und Forschung

(BMBF 11425A) and the Stiftung zur Förderung der Erforschung von Ersatz-

und Ergänzungsmethoden zur Einschränkung von Tierversuchen (set). The

procedure has been granted a European patent (97 122 072.8).

INTERNATIONAL VALIDATION OF PYROGEN TESTS BASED ON

CRYOPRESERVED HUMAN PRIMARY BLOOD CELLS

74

6 International validation of pyrogen tests based on

cryopreserved human primary blood cells

Authors: Stefanie Schindler a, Ingo Spreitzer b Bettina Löschner b, Sebastian

Hoffmann c, Kilian Hennes d, Marlies Halder c, Peter Brügger e, Esther Frey e,

Thomas Hartung c , Thomas Montag-Lessing b

a = Biochemical Pharmacology, University of Konstanz, Universitätsstr.10, D-

78457 Konstanz, Germany

b = Paul-Ehrlich Institute, Paul-Ehrlich-Strasse 51-59, D-63225 Langen,

Germany

c = European Centre for the Validation of Alternative Methods, Institute for

Health and Consumer Protection, European Commision, Joint Research

Centre, Via Fermi 1, I-21020 Ispra, Italy

d = Qualis Laboratories, Blarerstr. 56, D-78462 Konstanz (www.qualis-

laboratorium.com)

e = Biological Analytics, Novartis Pharma AG, CH-4002 Basel, Switzerland

Abbreviations: CV, coefficient of variation; DL, developing laboratory; DMSO,

dimethyl sulfoxide; ECVAM, European Center for the Validation of Alternative

Methods; ELC, endotoxin limit concentration; ELISA, enzyme-linked

immunosorbent assay; IU, international unit; GLP, good laboratory practice; h,

hours; IL, interleukin; IPT, in vitro pyrogen test; LoD, Limit of Detection; LPS,

lipopolysaccharide; LTA, lipoteichoic acid; MVD, maximum valid dilution; PEI,

Paul Ehrlich Institute; PM, prediction model; PPC, positive product control; NL,

naive laboratory; NPC, negative product control; OD, optical density; SOP,

standard operating procedure; WHO, World Health Organisation



75

Corresponding author:


European Commission

Joint Research Centre

Institute for Health & Consumer Protection

ECVAM, I-21020 Ispra (VA)


Tel: +39-0332-786256/ Fax: +39-0332-786297

6.1. Abstract

Pyrogens as fever-inducing agents can be a major health hazard in parenterally

applied drugs. For the control of these contaminants, pyrogen testing for batch

release is required by Pharmacopoeias. This has been done either by the in

vivo rabbit pyrogen test (since 1942) or the limulus amoebocyte lysate test

(LAL), since 1976. A new approach are cell-based assays employing in vitro

cultivation of human immune cells which respond e.g. with cytokine production

(IL-1β; IL-6) upon contact to pyrogens. Six variants of these assays have

recently been validated in a collaborative international study. Recently, the

development of successful cryopreservation methods promises to make

standardized immunoreactive primary human blood cells available for

widespread use. Furthermore, the pretesting of donors for infectious agents

such as HIV or hepatitis has made it possible to develop a safe and

standardised reagent for pyrogen testing. Using altogether 13 drugs, we have

validated here the pyrogen test based on fresh and cryopreserved human whole

blood in four laboratories. The test reached > 90% sensitivity and specificity. In

contrast to the LAL, the test is capable of detecting non-endotoxin pyrogens

derived from Gram-positive bacteria or fungi.

Keywords: Pyrogen testing; validation study; IL-1β; cryopreservation



76

6.2. Introduction

Testing for pyrogens has employed animals, either the rabbit (since 1942) or

the horseshoe crab, Limulus polyphemus (since 1976). Pyrogens, especially

lipopolysaccharide (LPS) as part of the cell walls of Gram-negative bacteria, are

an ubiquitous threat to human health due to their stability. Other pyrogens are

LTA (Morath et al., 2001), exotoxins (149, 18, 164), muramyldipeptide (165) or

peptidoglycan (22, 23,149). Alternative pyrogen tests, altogether 6 cell-based in

vitro pyrogen tests using the production of inflammatory mediators, (i.e. IL-1β,

TNF-α or IL-6 as well as neopterin) (detailed by 49, 65, 67, 142, 143, 150) of

human blood cells have been developed and validated in an international

collaborative study (48).

Since then, two variants, one using freshly drawn human whole blood and

measuring the IL-1β response (48, 53) and another using freshly drawn isolated

PBMCs and measuring IL-6 (150), have undergone further development and

refinement by developing a standardized freezing procedure using DMSO as a

cryoprotective agent. This allows both methods to become more standardized

and more widely available. For the human whole blood test (IPT), even two

variations of freezing were developed and optimized. The newer one was

optimized for storage at -80°C, the other one making use of storage in liquid

nitrogen has been described previously (73).

In a second study identical to the one previously performed, the refined human

whole blood assay was validated. This study focused again on LPS as a

pyrogen, since it is the most frequent contamination, and the capability of the

cell-based assays of detecting non-endotoxin pyrogens had been demonstrated

in the previous study.

6.3. Materials and Methods

The test system was validated in the developing lab and in three different naïve

labs (Table 1) after a detailed SOP had been compiled by the developing lab

(DL) and made available for the naïve labs (NL) by ECVAM. The successful



77

technology transfer had been assessed in a prevalidation phase.

Since the developing lab of the IPT (endpoint IL-1β) cannot work under GLP, the

assay was performed by three naïve labs (PEI, NL 1, Qualis, NL 2 and Novartis,

NL 3). The IPT was performed in three variants, one involving fresh blood and

two involving frozen whole blood stored at -80°C and liquid nitrogen,

respectively. (* The results of this lab did not enter the formal evaluation since it

is not operating under good laboratory practice (GLP)).

Test

system

Developing

lab (DL)

1st lab

(NL 1)

2nd lab

(NL 2)

3rd lab

(NL 3)

IPT

IL-1β

University of

Konstanz *

Paul-Ehrlich

Institute

Qualis

laboratories

Novartis

Pharma

Table 1: Laboratories performing the assays

Endotoxin stimulus

The second international WHO standard for endotoxin 94/580 from E. coli O113:

H10, which was used in the previous validation study, served again as the

standard endotoxin (Poole et al., 1997). 100 pg/ml of this endotoxin are

equivalent to 1 IU (International Unit)/ml.

Fresh and cryopreserved human whole blood test IL-1β

Blood collection

Blood from healthy donors was collected into heparinized tubes (Li-Heparin, 15

IU/ml) (Sarstedt-monovette, 7.5ml, Nümbrecht, Germany) using a multifly

needle set and used within 4h. Additionally, for the preparation of

cryopreserved blood, a sample from each of the five donors was drawn into

Serum- and EDTA tubes.



78

(Sarstedt) for differential blood counts and infection serology. For the fresh

blood assay, one single donor was used and his blood was subjected to a cell

count (Pentra 60, ABX Diagnostics, Montpellier, France), in order to exclude

infections.

Testing for infectious agents in cryopreserved blood

The additionally drawn blood of each donor was tested by a qualified laboratory

(Dr. U. Brunner, Konstanz, Germany) for Hepatitis A, B, C and HIV according to

the standards for blood donations for transfusion purposes in Germany. In the

meantime, the freezing of the blood took place.

Freezing for storage at -80°C (Method A)

Endotoxin-free Soerensen Buffer (Acila GMNmbh, Mörfelden-Walldorf,

Germany) was mixed with 20% v/v endotoxin-free DMSO (Wak-Chemie

Medical GmbH, Steinbach, Germany). 1.8ml cryotubes (Nalge Nunc

International, Denmark) were screwed open under a laminar flow bench and

0.6ml of the pooled blood of five donors were pipetted into each one. Using a

multipette with a 5ml combitip, the cryoprotective solution was added in three

aliquots of 200µl each, gently swirling the blood in between. The tubes were

closed and placed in a storage box (Nalge Nunc), leaving a space of about 1

cm between each vial in order to ensure a homogenous freezing process. The

boxes were then placed in a freezer at –80°C and left to freeze. They were

stored at –80°C.

Freezing for storage in liquid nitrogen (Method B)

An alternative freezing procedure involved a controlled freezing process using

the vapour phase of liquid nitrogen and has been described earlier (73). For this,

the DMSO was added directly to the blood of the individual donors at 10% final

concentration (v/v). The blood was then pooled and aliquoted at 1.2 ml/vial and

frozen in a computer-controlled freezer (Nicool Plus PC, AirLiquide, Marne-la-

Vallée Cedex 3, France) until at -120°C. The aliquots were then taken out of the

machine and placed in the vapour phase of liquid nitrogen.



79

Shipment

The required number of aliquots of frozen blood was shipped to the other

laboratories using a voyager containing liquid nitrogen (Air Liquide Kryotechnik

GmbH, Düsseldorf, Germany). The temperature was monitored using a

computer-controlled temperature probe (Thermory Mobile, Air Liquide, France;

software Logiciel Recwin, Marne la vallée Cedex, France). After arrival, the

aliquots were kept either in the transportation vehicle itself or transferred to the

vapour phase of liquid nitrogen, if available. Alternatively, the aliquots that had

been frozen at -80°C (Method A) could be retransferred to the freezer at -80°C.

Thawing procedure

The vials were taken out of the voyager/ the nitrogen tank and placed

immediately in an incubator at 37°C. After 15 min, the blood was pooled in a

centrifuge tube if more than one aliquot was used, and gently swirled in order to

ensure complete mixing.

Pretesting of the aliquots

After all donors were clearly negative regarding the infectious agents in

question, the cryopreserved blood was pretested before sending it out to the

other laboratories by carrying out a dose-response curve using the WHO

standard or an endotoxin calibrated to it. The criteria were an absorbance for

the saline control of 0.1 OD or lower, and a response to the 0.5 IU/ml of at least

1.6 times the OD of the saline control.

Incubation procedure

Cryopreserved blood (Method A and B)

Method A: 180µl of RPMI (Charles River Endosafe), 20µl sample/control and

40µl of thawed blood were added to a pyrogen-free microtiter plate (Falcon).

After adding the blood, the contents of the wells were mixed by gently



80

aspiring/dispensing 5 times using a multichannel pipette and sterile, pyrogen-

free tips, changing the tips in-between the rows in order to avoid cross

contaminations. The plates were then covered with a lid and placed in an

incubator at 37°C and 5% CO2 for 10-24 h.

Method B: The alternative version using nitrogen-stored blood was handled the

same, except that the incubation involved 200µl of RPMI, 20µl of

samples/controls and 20µl of thawed blood, since the blood had not been

prediluted in the freezing process.

Fresh blood (Method C)

In order to allow direct comparisons, the method validated in 2005 (Hoffmann et

al., 2005a) was adapted from 1ml vial incubation to microtiter plates as used for

the cryopreserved blood. This variant was also included in the validation to

exclude an effect of this changed format.

200µl of saline (Charles River Endosafe, Charleston, South Carolina, USA),

20µl sample/control and 20µl of blood were added to a pyrogen-free microtiter

plate (Falcon 96well flatbottom tissue culture plate, Becton Dickinson Labware,

Meylan Cedex, France). After adding the blood, the contents of the wells were

mixed by gently aspiring/dispensing 5 times using a multichannel pipette and

sterile, pyrogen-free tips, changing the tips in between the rows in order to

avoid cross- contaminations. The plates were then covered with a lid and

placed in an incubator at 37°C for 10-24 h.

ELISA procedure

The IPT Kit was used (Charles River Endosafe). Aliquots of 100µl of each well

of the incubation plates were added to the wells of the ELISA plate. When

transferring the supernatants, they were mixed by aspiring and dispensing them

2-3 times using a multichannel pipette. The ELISA was done according to the

manufacturer´s instructions.



81

Data analysis

Data analysis was the same as in the previous validation. The quality criterion

for acceptable variability, i.e. allowing a maximum coefficient of variation (CV)

of 0.45 has been empirically established in the previous study for both assays

in order to ensure the interpretability of test results and was adopted here. The

data of those samples and control exceeding this CV-value were tested for

outliers by the Grubbs-test. If an outlying replicate caused the excessive

variation it was excluded and further analysis was performed with the remaining

three replicates. The samples and controls, whose large variation was not

caused by an outlier were excluded from further analysis. In the cases when

the positive product control (PPC) CV exceeded 45%, the corresponding 0.5

IU/ml in saline, which was part of the dose-response curve using the WHO

standard endotoxin, were used instead. If the CV of this standard also

exceeded 45%, the whole set of data was not considered for analysis.

Furthermore, the response of the 0.5 IU/ml had to be significantly higher, i.e. a

p-value below 0.01, than the respective response of the negative saline control.

Accepted data were analysed by a so-called prediction model (PM): the data of

a blinded sample were compared with the PPC data or, if the PPC did not fulfil

the qualitiy criterion, the 0.5 IU/ml control using a one-sided t-test with log-

transformed data and a local significance level of 1%.

Blinding procedure

All test items are registered medicinal products and were obtained from a

pharmaceutical supplier. For the validation, test items and endotoxin spiking

samples were prepared by the University of Konstanz and blinded/coded under

GLP by personnel (G. Bowe and J. de Lange) from ECVAM, Italy. These were

then shipped by the University of Konstanz to each of the appropriate test

facilities participating in the study.

Prevalidation

The drugs used were the same as in the previous validation study (Hoffmann et



82

al., 2005a), that is Gelafundin, a volume-replacement therapy for transfusion

with high protein content (B. Braun Melsungen AG, Melsungen, Germany),

Jonosteril, an electrolyte infusion (Fresenius AG, Bad Homburg, Germany) and

Haemate, a factor VIII preparation (Aventis Behring GmbH, Marburg,

Germany). Additionally, a negative and a positive control (0.5 IU/ml) were

included in each run.

This set was tested in the developing laboratory (DL) Konstanz as well as in

two naive laboratories PEI (NL 1) and Qualis (NL 2) with the three approaches

of the IPT (Method A-C) in order to prove successful transfer.

Prior to preparing the spikes, an interference test was performed with all three

substances by the DL. Although this had been done in the previous validation,

a shift in the interference due to the DMSO/freezing process could not be

excluded. Interferences differed indeed for the IPT (data not shown), and the

spikes were calculated according to the required dilution.

Sample preparation and blinding was done at the University of Konstanz using

pyrogen-free clinical grade saline and the WHO reference standard endotoxin.

Validation

For the validation phase, in order to maximise comparability with the previous

validation study, the same ten drugs were employed. The concentrations were

based on a recent in-depth analysis of the fever response of a sensitive rabbit

strain (Hoffmann et al., 2005b): Five blinded spikes, two of them defined as

non-pyrogenic, that is below 0.5 IU/ml (0 and 0.25 IU/ml), and three as

pyrogenic (2 x 0.5 and 1.0 IU/ml) were tested in the different laboratories. All

drugs were tested at their MVD (maximum valid dilution), thus adopting the

rationale of the pharmacopoeial LAL reference (limit) test. The MVD is

calculated from the endotoxin limit concentration (ELC) in IU/ml, defined for a

drug by the European Pharmacopoeia (146), divided by the threshold of

pyrogenicity as the limit of detection (LoD), in this case 0.5 IU/ml. Drugs,

sources, ELCs and MVDs are summarized in Table 2.



83

Drug code Source Agent Indication MVD

Glucose

5 % (w/v)

GL Eifel Glucose nutrition 70

Ethanol

13 % (w/w)

ET B. Braun Ethanol diluent 35

MCP ME Hexal Metoclopramid antiemetic 350

Syntocinon SY Aventis Oxytocin initiation

of delivery

700

Binotal BI Aventis Ampicillin antibiotic 140

Fenistil FE Novartis Dimetinden-

maleat

antiallergic 175

Sostril SO GlaxoSmith

Kline

Ranitidine antiacidic 140

Beloc BE Astra Zeneca Metoprolol

tartrate

heart

dysfunction

140

Drug A LO - 0.9 % NaCl - 35

Drug B MO - 0.9 % NaCl - 70

Table 2: Drugs employed in the validation

All ten clinical-grade drugs, that had been used in the previous validation using

freshly drawn cells, were used again. All drugs were used at their respective

MVD (maximum valid dilution).

6.4. Results

Pretesting of the cryopreserved blood

The blood was tested employing an E. coli O113: H10 dose-response curve

(Fig. 1). The blood was considered suitable since the OD of the saline control

was below 0.1 OD and the mean OD of the 0.5 IU/ml was 1.6 times the mean

OD of the saline or higher.



84

0

1

2

3

version - 80°C(Method A)

nitrogen version(Method B)

controlO-113; 0.25 EU/mlO-113; 0.5 EU/mlO-113; 1.0 EU/mlO-113; 2.0 EU/ml

OD

[450n

m]

Fig. 1: Pretesting of both versions of the cryopreserved blood prior to

shipping

Both versions of the cryopreserved blood (Method A and B) were pretested

measuring the release of IL-1β in response to a pyrogenic stimulus prior to

shipping them to the participating labs. For this purpose, a dose-response

curve (0.25-2 IU/ml) using the international WHO standard was done (n=4).

Prevalidation

Method A – Cryoblood, -80°C version

The data produced with the method based on cryoblood frozen at -80°C are

summarized in Figure 2 as an example for all three methods. It has to be noted

that for the NL 1 the level approximated the maximum response level of 4 OD.

This might cause problems for discriminating pyrogenic spikes, especially when

the positive control of 0.5 IU/ml produces such high-level responses. The

background OD-levels were small in the NL 1 and in the DL. NL 2 provided the

data with the background level subtracted. The three standard curves,

consisting of C-, (0 IU/ml), 0.25, 0.5 and 1 IU, indicate a typical monotone

increase in OD-response with increasing concentration.



85

J-0

J-0

J-0.5

J-1

G-0

G-0

G-0

.5G-1

H-0

H-0

H-0

.5H-1 C

-

0.25

EU

0.5 E

U (C+)

1 EU

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9 Konstanz (DL)

OD

J-0

J-0

J-0.5

J-1

G-0

G-0

G-0

.5G-1

H-0

H-0

H-0

.5H-1 C

-

0.25

EU

0.5 E

U (C+)

1 EU

0

1

2

3

4PEI (NL1)

OD

J-0

J-0

J-0.5

J-1

G-0

G-0

G-0

.5G-1

H-0

H-0

H-0

.5H-1 C

-

0.25

EU

0.5 E

U (C+)

1 EU

0.0

0.5

1.0

1.5

2.0

2.5

3.0Qualis (NL2)

OD

Fig. 2: Prevalidation data for Method A of the three involved laboratories.

The treatments and controls are abbreviated (J = Jonosteril: G = Gelafundin;

H = Haemate; C- = saline; C+ = positive control) indicating the endotoxin

contamination in IU (0, 0.5 and 1 IU/ml).



86

As Figure 2 only gives an indication about variability of replicates, the CVs were

calculated for all samples and controls for all laboratories. While the major part

of the total of 48 CV-values was smaller than 30%, four samples, all of which

were unspiked, showed a CV larger 45%. In all of these, one of the four

replicate values was much larger than the others, and thus caused the high

variability, and was excluded by outlier analysis.

Application of the prediction model (PM) to these data revealed that eleven out

of the twelve spikes were classified in the same way in all laboratories.

Comparing the laboratories pairwise showed that 34 of the total of 36 single

comparison, i.e. 94.4%, resulted in the same classification.

Assessing in the final step preliminarily the predictive capacity, revealed that all

negative samples were classified correctly and that one 0.5 IU spike (NL 1: H-

0.5) at the rabbit classification threshold was classified false-negative. In terms

of performance parameters, this resulted in a specificity of 18/18 = 100% and a

sensitivity of 17/18 = 94.4%.

Method B – Cryopreserved blood (liquid nitrogen version)

Background OD-levels were small in the DL and NL 1. NL 2 provided the data

with the background level subtracted. The three endotoxin standard curves

indicate a higher limit of detection as the 0.25 IU standards, and for NL 2 also

the 0.5 IU standard, gave low OD-responses (data not shown).

The CVs were calculated for each treatment or control for all laboratories. While

the major part of the CVs was smaller than 40%, six samples (mainly from DL)

and one standard showed a CV larger than 45%. Nine out of the twelve spikes

were classified in the same way in all laboratories. Comparing the laboratories

pairwise showed that 30 of the total of 36 single comparison, i.e. 83.3%,

resulted in the same classification.

Assessing in the final step preliminarily the predictive capacity, revealed that

one negative samples was classified wrongly (NL 2: J-0) due to one outlying

value, and that two times a Haemate 0.5 IU/ml sample (DL and NL 1) at the



87

rabbit classification threshold was classified false-negative. In terms of

performance parameters, this resulted in a specificity of 17/18 = 94.4% and a

sensitivity of 16/18 = 88.9%.

Method C – fresh whole blood

Background OD-levels were small in the DL and NL 1. NL 2 provided the data

with the background level subtracted. The three standard curves, consisting of

the negative control C- (0 IU/ml), 0.25, 0.5 and 1 IU/ml, showed a typical

monotonous increase in OD-response with increasing concentration.

The CVs were calculated for each treatment or control for all laboratories. In

general, the CVs were smaller than 30%. Only two samples resulted in a CV

larger than 45%. These two samples were an H-0 tested at NL 2, which was

caused by an aberrant value, and a G-0 tested in DL with a CV of 48.8%.

Furthermore, a tendency for larger CV of endotoxin-free samples/treatments

was observed, as the background OD-level was lower compared to comparable

assays, e.g. in the main validation study. Ten out of the twelve spikes were

classified in the same way in all laboratories. Comparing the laboratories

pairwise, showed that 32 of the total of 36 single comparison, i.e. 89.9%,

resulted in the same classification.

Assessing in the final step preliminarily the predictive capacity, revealed that all

negative samples were classified correctly and that two 0.5 IU/ml spikes (DL: J-

0.5; NL 1: H-0.5), which are at the rabbit classification threshold, were classified

false-negative.

In terms of performance parameters, this resulted in a specificity of 18/18 =

100% and a sensitivity of 16/18 = 88.9%.

Validation

Inter-laboratory reproducibility

As within-laboratory reproducibility was generally successfully shown in

prevalidation, only inter-laboratory reproducibility, based on three laboratories

per test method, was assessed in the validation. The similarity of laboratories

was based on the classification resulting from the PM (Table 3) to compare the



88

laboratories with each other with respect to concordance, i.e. without taking the

true classifications of the samples into account. The results are presented in

Table 4. The overall inter-laboratory reproducibilities of IPT Method A and C are

consistently high. Regarding IPT Method B, only about 65% of the samples

were classified the same way due to NL 3 where four drugs caused problems.

IPT Method A IPT Method B IPT Method C

drug

spike

(IU/ml) truth NL 1 NL 2 NL 3 NL 1 NL 2 NL 3 NL 1 NL 2 NL 3

0.0 0 0 0 CV 0 0 0 0 0 0

0.25 0 0 1 CV 0 0 CV 1 0 0

0.5 1 1 1 1 0 1 1 1 1 1

0.5 1 1 1 1 1 1 1 1 1 1

Beloc

1.0 1 1 1 1 1 1 1 1 1 1

0.0 0 0 0 0 0 0 0 0 0 0

0.25 0 0 1 1 0 1 CV 0 1 0

0.5 1 0 1 1 1 1 1 1 1 1

0.5 1 1 1 1 1 1 1 1 1 1

Binotal

1.0 1 1 1 1 1 1 1 1 1 1

0.0 0 0 0 nq 0 0 0 0 0 CV

0.25 0 CV 1 nq 0 0 CV 0 1 0

0.5 1 1 1 nq 1 1 0 1 1 1

0.5 1 1 1 nq 1 1 0 1 1 1

Ethanol

1.0 1 1 1 nq 1 1 1 1 1 1



89

0.0 0 0 0 0 0 0 0 0 0 0

0.25 0 0 1 1 0 CV 1 CV 1 1

0.5 1 1 1 CV CV 1 CV CV 1 1

0.5 1 1 1 1 1 1 1 CV 1 1

Fenistil

1.0 1 1 1 1 1 1 1 1 1 1

0.0 0 0 0 0 0 0 0 0 0 0

0.25 0 CV 0 0 0 1 0 0 1 1

0.5 1 1 1 1 0 1 CV 1 1 1

0.5 1 1 1 1 0 1 1 1 1 1

Glucose

1.0 1 1 1 1 0 1 1 1 1 1

0.0 0 0 0 0 0 0 0 0 0 0

0.25 0 CV 1 CV 0 1 0 0 0 1

0.5 1 1 1 1 0 1 CV 0 1 1

0.5 1 1 1 1 0 1 1 1 1 1

MCP

1.0 1 1 1 1 1 1 1 CV 1 1

0.0 0 0 0 nq 0 0 nq 0 0 0

0.25 0 0 1 nq 0 1 nq CV 0 1

0.5 1 0 1 nq 0 1 nq 1 1 CV

0.5 1 1 1 nq 1 1 nq 1 1 1

Sostril

1.0 1 1 1 nq 1 1 nq 1 1 1

0.0 0 0 0 0 0 0 nq 0 0 0

0.25 0 0 CV CV CV CV nq 0 0 0

0.5 1 1 1 1 1 1 nq 1 1 1

0.5 1 1 1 1 1 1 nq 1 1 1

Synto-

cinon

1.0 1 1 1 1 1 1 nq 1 1 1



90

0.0 0 CV 0 nq 0 0 nq CV 0 0

0.25 0 0 0 nq 0 0 nq 0 0 0

0.5 1 1 1 nq 0 1 nq CV 1 1

0.5 1 1 1 nq 0 1 nq 1 1 1

A

(saline)

1.0 1 1 1 nq 1 1 nq 1 1 1

0.0 0 0 0 nq 0 0 nq 0 0 0

0.25 0 0 0 nq 0 0 nq 0 0 CV

0.5 1 1 1 nq 0 1 nq CV 1 1

0.5 1 1 1 nq 0 1 nq 1 1 1

B

(saline)

1.0 1 1 1 nq 1 1 nq 1 1 1

Sample size n 46 49 25 48 48 24 42 50 47

Specificity 100 68.4 75 100 77.8 88.9 94.1 80.0 76.5

Sensitivity 93.3 100 100 62.1 100 86.7 96.0 100 100

Table 3: Classifications of samples by all methods and all laboratories in

validation

Grey shading indicates that for these drugs the PPCs did not qualify so that the

PC was used in the PM.

CV = sample showed a variability resulting in exclusion, i.e. CV > 45 % and no

significant outlier present.

nq = not qualified according to quality criteria, i.e. failure of PPCs and PCs

0 = considered/classified negative/ 1 = considered/classified positive

False classifications are in bold type

Predictive capacities

Table 4 summarises the sensitivity and specificity for each method together with

the respective sample sizes. For IPT Method A, eight samples at the NL 1 and

three at the NL 3 were excluded due to their high variability, i.e. CVs > 45%. For

IPT Method B and Method C, the sample sizes were reduced for both methods in



91

addition to ten samples with high variability by four drugs tested at the NL 3 that

did fail the quality criteria for both the PPC and the PC.

Test

Inter-laboratory

reproducibility (%)

Sample size:

sensitivity

Sensitivity

(%)

Sample size:

specificity

Specificity

(%)

IPT

Method A

DL-NL 1: 86.7

DL-NL 2: 87.5

NL 1-NL 2: 100

77

97.4

45

82.2

IPT

Method B

DL-NL 1: 66.0

DL-NL 2: 63.3

NL 1-NL 2: 83.3

74

82.4

46

89.1

IPT

Method C

DL-NL 1: 88.1

DL-NL 2: 89.7

NL 1-NL 2: 91.5

84

98.8

55

83.6

Table 4: Inter-laboratory reproducibility and sensitivity/specificity with the

respective sample sizes in validation

Inter-laboratory reproducibility was calculated by the proportion of samples

classified identically for each pair-wise laboratory comparison.

The overall performances of the IPT Methods A and C were very good: High

sensitivities over 90% could be achieved, while specificities around 80% were

established, reflecting the safety approach in the PM emphasizing sensitivity. In

contrast to these methods, the IPT Method B performed differently with a higher

specificity of 89% on cost of a decreased sensitivity (82%). Misclassifications

occurred with one exception only for samples with contaminations close to the

pyrogenicity threshold, i. e. 0.25 and 0.5 IU/ml. The 95%-confidence intervals

for the parameters were calculated, assuming binomial distribution, and are



92

presented in Figure 3. The according parameters for the rabbit test, calculated

with the model of Hoffmann et al. 2005 (16), were also included. While the new

test except IPT Method B had a slightly lower specificity than the rabbit test, the

sensitivity was substantially increased by 20% up to 40%.

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1,0

0,0 0,1 0,2 0,3 0,4 0,5

1-specificity

sensitiv

ity

IPT Method A

IPT Method B

IPT Method C

Rabbit

Fig. 3: Sensitivity and specificity of validation with 95% confidence

intervals including the rabbit pyrogen test

The sensitivity and specificity of the in vitro methods was assessed after the

validation phase and compared to the rabbit pyrogen test assuming binomial

distribution. The y-axis shows the sensitivity in percent, the x-axis the specificity

for the rabbit pyrogen test and all four in vitro assays.

6.5. Discussion

This follow-up study had the aim to refine and improve the already established,

validated systems for pyrogen testing using freshly drawn human blood and

make them available as a safe and standardized reagent in cryopreserved

form. In comparison to the fresh blood assays, the cryopreserved cells showed

a higher response with regard to IL-1β, but the variance also tended to be

higher.



93

Therefore, some sets of data of all of the IPT methods had to be excluded due

to variance-controlling quality criteria. In the former validation (48), the whole

blood test, at this time performed with fresh blood in reaction tubes, achieved a

sensitivity of 72.7% (n= 88) and a specificity of 93.2% (n= 59). The inter-lab

reproducibility had been 72.9 and 81.6% between the developing lab and the

two naïve labs. In this study, the sensitivity of the fresh blood assay could be

improved by transferring it to the microtiter plate (98.8%) with a minor reduction

of specificity (83.6%). This approach proved to be more easily transferable to

the naïve labs (inter-lab reproducibility 88-92%). The frozen cells performed

well, although the -80°C method (A) was better transferable than the nitrogen

method (86-100% vs. 63-83% for the inter-laboratory reproducibility). However,

one of the laboratories (NL 3) had particular problems with both cryopreserving

methods as the data of four drugs failed quality criteria and thus had to be

excluded from analysis.

We assume that the IPT assay at the NL 3 posed problems in performance

because of insufficient transfer of the method. This problem could not be

overcome because of the very tight time schedule of the validation.

Nevertheless, IPT-data generated in parallel at Konstanz (Table 5) yielded

good results, although they did not enter the formal evaluation according to the

study protocol which foresaw only the participation of three GLP-concordant

laboratories. The importance of successful assay transfer is stressed

considering the fact that, had the results of the developing lab instead of those

of NL 3 entered the evaluation, the sample size for method A qualifying for

evaluation had increased from 120 to 143.

IPT A IPT B IPT C

sample size n 48 37 47

Specificity 100 100 94.4

Sensitivity 96.7 90.5 86.2

Table 5: Performance of the IPT in the developing laboratory



94

The developing laboratory in Konstanz performed all three methods of the IPT

in parallel to the naïve labs. This table shows the sample size that qualified for

evaluation and the achieved specificity and sensitivity for these samples.

The specificity of this method would have increased slightly to 89.5% and the

sensitivity would hardly have changed at all, indicating that the method of

evaluation used in this study was adequate. In general, the –80°C version

produced higher ODs and was more sensitive than the nitrogen version of the

cryoblood. It was noticeable, though, that the linear range in the dose response

curve from 0.25 IU/ml to 1 IU/ml is much smaller in the –80°C than in the

nitrogen version. All in all, this validation has shown that the novel, recently

validated pyrogen test based on human blood can be performed with

cryopreserved cell preparations. The IPT could be improved with regard to

performance making it at the same time easier to handle by transferring it to the

microtiter plate. The optimization employing cryopreserved cells allows the

assay to become more standardized and cells as test reagents more widely

available. The fact that safety standards of blood transfusions can be

implemented as shown for the IPT stresses that concerns of possibly infected

donors can be ruled out.

PYROGEN TESTING OF LIPIDIC PARENTERALS WITH A NOVEL IN

VITRO TEST

95

7 Pyrogen testing of lipidic parenterals with a novel in

vitro test

Application of the InVitro Pyrogen Test (IPT) based on cryopreserved human

whole blood

Stefanie Schindler+, Ute Rosenberg*, Detlef Schlote*, Katja Panse*, Andreas

Kempe*, Stefan Fennrich+ and Thomas Hartung+#

* Schering AG, Biological Quality Control, Müllerstrasse 178, D-13342 Berlin

+ Biochemical Pharmacology, University of Konstanz, Universitätsstr.10, D-

78457 Konstanz, Germany

# European Commission, Joint Research Centre, Institute for Health &

Consumer Protection, ECVAM, I-21020 Ispra (VA)

Abbreviations: CV, coefficient of variation; DMSO, dimethylsulfoxide;

ECVAM, European Center for the Validation of Alternative Methods; ELC,

Endotoxin Limit Concentration; ELISA, Enzyme-linked immunosorbent assay;

EU, Endotoxin Unit; IL, Interleukin; LAL, Limulus Amoebocyte Lysate; LPS,

lipopolysaccharide; MVD, Maximum Valid Dilution; PPC, positive product

control; NPC, negative product control; IPT, In Vitro Pyrogen Test; WHO,

World Health Organisation

Corresponding author


European Commission, Joint Research Centre

Institute for Health & Consumer Protection

ECVAM/ I-21020 Ispra (VA)


Tel: +39-0332-785939/ Fax: +39-0332-786297


VITRO TEST

96

7.1. Abstract

The European Pharmacopoeia has made the testing of small volume

parenterals (< 15 ml) obligatory since 2004. This concerns many

formulations, e.g. vitamins, steroids and hormones, many of which are

applied intramuscularly using a lipidic carrier. Lipopolysaccharides, the best

established endotoxins from Gram-negative bacteria, bind strongly to

lipophilic substances which mask them in Limulus amebocyte lysate assays.

End-product testing, therefore, can only be carried out in rabbit pyrogen tests.

This will of course lead to a pronounced increase in animal experiments if no

alternative procedures become available. We have described a novel in vitro

pyrogen test (IPT) based on human whole blood, which has recently been

validated in a collaborative study including the European Pharmacopoeia.

Here, the utility of the IPT for lipophilic substances and lipid-containing end-

products was assessed.

For a variety of lipids commonly added to formulations of injectable

endproducts, namely peanut oil, sesame oil, miglyol and paraffin, a protocol

which allows interference-free testing was established applying the

pharmacopoeial criterium of 50 to 200% retrieval of an LPS spike.

Furthermore, end-product testing for three formulations was possible. In all

cases a method could be established which allows to determine given or

calculated ELC (endotoxin limit concentrations) according to Pharmacopoeia.

It is concluded that the monocytes react to lipid-bound LPS showing that

immune responses to contaminated endproducts must be anticipated and

that the IPT is suitable for endproduct control of these formulations.

Keywords: Pyrogen testing; alternatives to animals; lipids; IL-1β


VITRO TEST

97

7.2. Introduction

The Limulus amebocyte lysate assay (LAL) has replaced about 80% of the in

vivo tests in rabbits. However, complex preparations challenge this assay

because of interferences. This holds true especially for biologicals such as

blood products through binding of LPS to lipoproteins (166), but also for

lipophilic preparations (167) or liposomes (168). Similar problems are seen

for vaccines (169), which usually contain adjuvants such as aluminium

hydroxide which binds and masks endotoxins, making them undetectable in

the LAL (170).

If such endproducts are controlled (some are not controllable), dilutions in

pyrogen-free distilled water are established to allow endotoxin spike retrieval in

the LAL. Preparing dilutions of lipids in water is extremely difficult, since the

phases will only remain mixed for a few seconds after vortexing and a

homogenous and representative distribution of a contamination in the

hydrophilic diluent cannot be determined. A considerable proportion of LPS

could remain in the lipophilic phase and remain unaccessible for testing and,

additionally, be lost by binding to plastic pipette tips and vials. Very viscuous

samples, such as castor oil, are almost impossible to pipet. Therefore, the true

pyrogen content of the original sample is extremely difficult to judge.

The obvious alternative, the rabbit pyrogen test, is costly, laborious, and

especially in the case of lipid substances, ethically problematic. The procedure

of injecting lipid samples intravenously bears the risk of seriously harming the

animal through the formation of miniscule drops which can obstruct small

vessels e.g. in the kidney. The rabbit assay itself has some shortcomings which

are often overlooked, namely the lack of standardization with regard to animal

strain and age (27), the lack of positive and negative controls, and the small

number of animals used per sample.

Shortly after the human fever reaction was elucidated by identifying

proinflammatory cytokines of leukocytic origin (37), Duff and Atkins (134) as

well as Dinarello et al. (133) suggested using this in


VITRO TEST

98

vitro for future pyrogen testing. Problems in standardizing such cellular assays

initially hampered their routine application. But, recently a collaborative study,

including the European Pharmacopoeia, compared and validated such cell-

based pyrogen tests for inclusion into Pharmacopoeia (48). We contributed a

simplified and therefore highly standardized pyrogen test employing human

whole blood (IPT) in which the release of IL-1.beta. is the endpoint (49) to this

validation exercise. The advantage of testing the pyrogenic reactions of

monocytes in their natural environment comprises the presence of the

endotoxin-presenting plasma protein Lipopolysaccharide-Binding Protein (LBP)

which is capable to transfer LPS to the cellular binding site (171). Further

advantages in using primary cells instead of cell lines lie in their availability, not

requiring laborious maintenance and culture steps, and the fact that, unlike cell

lines, they do not harbour the risk of undergoing dedifferentiation or oncogenes,

therefore ensuring that they express all receptors and signalling proteins

necessary for their response to pyrogens.

Other cell-based approaches (reviewed by Poole and Gaines Das, 2001) (172)

measured IL-6 (63, 142) or TNF-α (65) as well as neopterin (66, 67). The whole

blood assay has since then been further developed and optimized for use with

cryopreserved blood (73) and is available in a standardized format.

Here, we addressed the pressing question whether lipid formulations of drugs

can be adequately controlled by IPT.

7.3. Materials and Methods

Lipids

Sesame oil and castor oil (both Schering AG, Berlin, Germany) as well as

peanut oil, miglyol and paraffin (L+S, Bad Bocklet, Germany) were used.

Another hydrophobic additive was benzyl benzoate (Schering). The

endproducts tested were Testoviron®, Gynodian® and Noristerat® provided by

Schering. Calibration and end-product testing was performed using


VITRO TEST

99

commercially available sesame oil (Sigma-Aldrich, Steinheim, Germany). All oily

substances were kept at room temperature or at 4°C and placed in an incubator

or in a water bath at 37°C for 30 minutes prior to use. Dilution of the castor oil

was done using endotoxin-free DMSO (Cryo-Sure DMSO, Wak-Chemie,

Steinbach, Germany)

Standards

The 2nd International WHO Standard for endotoxin (from E. coli O113: H10:K(-)

(94/580), which is identical to FDA/USP standard EC6/Lot G was used as the

standard endotoxin (56) for calibration. 100 pg of this standard correspond to

the pyrogenic activity of 1 EU (Endotoxin Unit).

The endotoxin used for the testing of the lipids was a calibrated lyophilisate of

E. coli O111: B4 provided in the IPT Kit (Charles River Endosafe, Charleston,

South Carolina, USA).

The Gram-positive standard used for the testing was a calibrated lyophilisate of

B. subtilis (57) provided in the IPT Kit (Charles River Endosafe).

Cryopreserved human whole blood

The procedure of freezing human whole blood has been described previously

(73). Briefly, the freshly drawn heparinized blood of five healthy donors is mixed

with pyrogen-free DMSO at a final concentration of 10% (v/v ratio), and pooled.

The blood is frozen in 1.5 or 4ml aliquots in a computer-controlled freezer

(Nicool Plus PC, Air Liquide, Marne-la-Vallée Cedex 3, France) and stored in

the vapor phase of liquid nitrogen. Before use the cryopreserved blood was

thawed at 37°C for 15 minutes.

Incubation

Basic procedure

A) Tube method

The procedure of the whole blood incubation has been described previously

(Hoffmann et al., 2005a). The incubation was carried out in pyrogen-free


VITRO TEST

100

polypropylene 1.5 ml reaction tubes (Eppendorf, Hamburg, Germany). Briefly,

1000 µl of pyrogen-free physiological saline (Charles River Endosafe or Berlin-

Chemie, Berlin, Germany) were pipetted into each tube, and 100 µl of

standard/samples and 100 µl of cryopreserved blood were added. The tubes

were closed, inverted twice to achieve complete mixing of the contents and

placed overnight in an incubator at 37°C. The next day, the tubes were taken

out and inverted again. The resuspended incubations were then transferred to

the wells of the ELISA plate.

B) Microtiter plate method

During the testing of the lipid substances, the procedure was adapted to a 96-

well sterile, pyrogen-free microtiter plate (96well flatbottom tissue culture plate,

Becton Dickinson Labware, Meylan, Cedex, France). 200 µl of RPMI (Charles

River Endosafe) were pipetted into each well and 20 µl of sample/control and

20 µl of blood were added. Mixing was achieved by aspiring and dispensing the

incubations 4-6 times with a multichannel pipette using sterile, pyrogen-free tips

and changing the tips between the rows in order to avoid cross-contaminations.

The plate was then covered with a lid and placed in an incubator at 37°C and

5% CO2. The next day, the incubations were resuspended using a multichannel

pipette with disposable tips and transferred to the wells of the ELISA plate.

C) ELISA procedure

Cytokine ELISAs were based on commercially available antibody pairs against

IL-1.beta. (R&D, Wiesbaden, Germany). Binding of biotinylated antibody was

quantified using streptavidin-peroxidase (Biosource, Camarillo, California, USA)

and the substrate TMB (3,3’,5,5’-tetramethylbenzidine) (Sigma-Aldrich). The

substrate is transformed from colorless to blue, and the intensity is measured in

terms of optical density (OD), in our case at a wavelength of 450 nm. For final

endproduct testing, the IPT Kit provided by Charles River Endosafe was used

according to the manufacturer’s instructions.

.


VITRO TEST

101

Adaptation of the procedure to the lipid substances

A) Peanut oil, sesame oil, miglyol, paraffin and benzyl benzoate

The endotoxin E. coli O111: B4, serving as Gram-negative control, was diluted

either in saline (Charles River Endosafe) or in the respective oil as indicated.

Dilutions were performed either with saline or oil down to a concentration of 0.5

EU/ml. The saline dilutions served as reference for judging the interferences

caused by the oil. All dilutions were vortexed before use for about 10 seconds at

maximum speed.

B) Testing of the endproducts Testoviron®, Gynodian® and Noristerat®

The Gram-negative control was resuspended in the respective endproduct,

yielding an artificial contamination of 20 EU/ml. After a suitable diluent had been

found for pipetting dose-response curves, a protocol for a PPC (positive product

control) and an NPC (negative product control) was developed for the Gram-

negative and the Gram-positive control, respectively. Furthermore, the

procedure was optimized for the microtiter plate. Since the resuspended

lyophilisate was very viscuous, 1 ml syringes (Kendall, Wollerau, Switzerland)

for used for the dilutions and a multipette with pyrogen-free combitips was used

to pipet the required volumes of the lipid samples onto the microtiter plate. All

dilutions were vortexed before use for about 30 seconds at maximum speed.

7.4. Results

A) Peanut oil, sesame oil, miglyol, paraffin and benzyl benzoate

The protocol established for the testing of hydrophilic parenterals had to be

adapted to the lipidic drugs. In order to avoid problems caused by redistribution

of the LPS, the lyophilized LPS standards in the IPT kit were resuspended and

diluted in the oils. An additional reference curve in saline was performed to

judge the amount of pyrogen retrieved by the monocytes under these

conditions. In all following experiments, a pool of cryopreserved blood from five

individual donors was used. It turned out that the oils differed in their

characteristics with regard to interference in spike retrieval (Table 1).


VITRO TEST

102

With peanut oil and sesame oil, the dilutions of the reconstituted LPS in the oil

resulted in a dose-response curve with a positive signal down to 0.5 EU/ml. A

comparison to the standard curve in saline showed no interference, except for

the 0.5 EU/ml in sesame oil, which gave a higher response than the saline

reference.

Oil control

oil

0.5

EU/ml

oil

1.0

EU/ml

oil

control

saline

0.5

EU/ml

saline

1.0

EU/ml

saline

mean OD 0.1027 0.3413* 0.6507 0.08900 0.1460 0.6840 sesame

oil SD 0.009866 0.04167 0.1134 0.007550 0.03208 0.06077

mean OD 0.0850 0.9547 1.864 0.09567 0.7103 1.610 peanut

oil SD 0.004000 0.02542 0.08154 0.03326 0.1760 0.1664

meanOD 0.07167 0.3063 0.4360 0.05933 0.2773 0.3763 miglyol

SD 0.02060 0.02290 0.09032 0.00208 0.02084 0.05160

mean OD 0.07867 0.2597 0.4520 0.05933 0.2773 0.3763 paraffin

SD 0.02554 0.05952 0.1410 0.00208 0.02084 0.05160

mean OD 0.09167 0.08533 0.1240 0.09567 0.3363 1.392 castor oil

SD 0.007572 0.006506 0.06022 0.009018 0.05659 0.06558

mean OD 0.2493 0.2073 0.2047 0.3083 0.9740 2.000 benzyl

benzoate SD 0.02574 0.02434 0.01617 0.08093 0.1888 0.2435

Table 1: Interference test of six different oily preparations towards a

reference E. coli O111: B4 endotoxin curve

bold letters: negative interference: the mean OD of the sample is below 50%

the mean OD of the saline reference

positive interference: the mean OD of the sample is above 200% the mean

OD of the saline reference

The saline reference curve is the same for miglyol and paraffin, since the oils

were incubated in the same experiment. They tested interference-free at all

dilutions and were easy to pipet. This did not apply to the other substances


VITRO TEST

103

though, and castor oil and benzyl benzoate showed a strong interference when

compared to saline.

Part B) Endproduct testing

Since the endproducts consist mainly of castor oil and benzyl benzoate, a

strong interference occurred for all three (Fig. 1). The reference curve was

performed with saline and with sesame oil, since the latter had proven to be

advantageous with regard to high spike retrieval and very little interference

(Table 1).

0 2.5 50 .0

0 .5

1 .0

1 .5

T e s to vi r o n

G y n o d ia n

N o r is te r a t

s e s a m e o i ls a lin e

O 1 1 1 ( E U / m l)

OD

450

±± ±± S

D

Fig. 1: Dose-response curve of E. coli O111: B4 in the endproducts, in

sesame oil and saline

After modification of the testing protocol for castor oil, this oil could be tested in

the IPT. In this experiment, the castor oil was diluted with 15% endotoxin-free

DMSO (v/v ratio) and LPS dilutions were performed in the oil with the added

DMSO. The reference in saline was done with and without 15% DMSO. An

approach using the Gram-positive stimulus (LTA, lipoteichoic acid from B.

subtilis) was done in parallel.

The modified protocol brought an improvement for the castor oil, but did not

resolve the interference problem with the endproducts (data not shown),

probably due to remaining interference of the benzylbenzoate or the


VITRO TEST

104

pharmaceutically active substances themselves. This approach was therefore

not pursued further.

Using the approach that allowed interference-free testing with some lipids, it

was attempted to dilute the interfering drugs to concentrations at which the

interference would be diminished or entirely gone was developed. For this

purpose, the drug itself was used to reconstitute the Gram-negative standard to

a concentration of 20 EU/ml. This stock solution was then diluted with sesame

oil. Sesame oil was chosen since it is commercially available and showed

excellent pipetting characteristics in the previous experiments. At the same

time, the procedure was transferred from the tubes to the microtiter plate

incubation. The CV (coefficient of variation) was determined in order to see

whether lipids could be pipetted reproducibly at a volume of 20 µl. The CV

equals the standard deviation divided by the mean (expressed as a

percentage). This proved to be successful for all three products (Table 2).

Dose-response curve 1

Mean OD CV (%)

LPS (EU/ml) 0 1 2 4 0 1 2 4

Testoviron 0.032 1.187 1.209 1.483 6.82 8.46 9.77 6.38

Gynodian 0.033 0.804 1.236 1.400 8.65 23.84 6.65 5.39

Noristerat 0.035 0.664 1.072 1.226 35.55 34.74 4.55 13.27

Table 2a: E. coli O111: B4 was reconstituted in the endproduct and

diluted with sesame oil (dose-response curve 1).


VITRO TEST

105

Dose-response curve 2

Mean OD CV (%)

LPS (EU/ml) 0 1 2 4 0 1 2 4

Testoviron 0.042 1.379 1.547 1.674 66.43 5.26 5.56 4.48

Gynodian 0.029 1.290 1.557 1.664 8.16 7.59 4.18 3.26*

Noristerat 0.031 1.438 1.636 1.742 29.26 6.2 4.12 1.65

Table 2b: E. coli O111: B4 was reconstituted and diluted in saline (dose-

response curve 2).

Next, the Gram-positive standard, lipoteichoic acid (LTA), was tested with a

similar protocol. The lyophilisate was reconstituted in saline and in sesame oil,

respectively. Both the reactivity and the variance were satisfying (Fig. 3),

although the retrieval in the oil was lower.

Establishment of a positive product control (PPC) is necessary in order to prove

that a contamination would have been retrieved at a certain product dilution. For

this purpose, the diluted “clean” product is spiked with a sample from the

sesame oil reference dose-response curve and must prove interference-free,

i.e. have an OD between 50 and 200% of the signal of the same spike

concentration in the absence of the product. A PPC of 2 EU/ml at a product

dilution in sesame oil of 1:10 and one of 4 EU/ml at a product dilution of 1:5 was

done for Testoviron® and Noristerat® (Fig. 4 and 5). A corresponding negative

product control (NPC) was done accordingly, spiking with the negative control

from the sesame oil dose-response curve. The mean OD (solid line) was

calculated for the PPC 2 EU/ml as well as the 50 and 200% cut-off (dotted

lines).


VITRO TEST

106

0 LTA 0 LTA0.0

0.1

0.2

0.3

0.4

0.5

saline sesame oil

5.49

19.93one outlierexcluded

5.1one outlierexcluded

20.67

OD

450

Fig. 3: replicates and reactivity of the Gram-positive standard

(Lipoteichoic acid, LTA) in saline and sesame oil.

Both the PPC with 4 and with 2 EU/ml were interference-free when compared to

the dose-response curve. The PPC 2 EU/ml was chosen for end-product

testing, since it lies in the linear range of the reference curve. Additionally, in the

same experiment as in Fig. 4 and 5, a PPC for the Gram-positive control was

established at a product dilution of 1:10 for both drugs. The 1:10 dilution was

chosen since it corresponds to the product dilution of the Gram-negative PPC 2

EU/ml.


VITRO TEST

107

0.0

1.0

2.0

4.0

TV 1 E

U/ml

TV 2 E

U/ml

TV 4 E

U/ml

NPC

PPC 2 E

U/ml

PPC 4 E

U/ml

0.0

0.5

1.0

1.5

O111 Dose-Response insesame oil

Testoviron + O111(20 EU/ml), dilutedin sesame oil

NPC/PPCO111

200%

PPC

50%

OD

450

±± ±± S

D

LTAreferencein sesame oil

LTA in 1:10product dilution

Gram-negative PPC Gram-positive PPC

Fig. 4: Establishment of a PPC (positive product control) for the Gram-

negative and the Gram-positive stimulus: Testoviron®

0.0

1.0

2.0

4.0

1 EU/m

l

2 EU/m

l

4 EU/m

lNPC

PPC 2 E

U/ml

PPC 4 E

U/ml

0

1

2

O111 Dose-Response insesame oil

Testoviron + O111(20 EU/ml), dilutedin sesame oil

NPC/PPCO111

200%

PPC

50%

LTAreferencein sesame oil

LTA in 1:10product dilution

Gram-negative PPC Gram-positive PPC

OD

450

±± ±± S

D

Fig. 5: Establishment of a PPC (positive product control) for the Gram-

negative and the Gram-positive stimulus: Noristerat®


VITRO TEST

108

A commercially available sesame oil preparation (Sigma-Aldrich) was tested to

determine whether it would be a suitable diluent for the preparation of an LPS

dose-response curve. For this purpose, the reaction was compared to that of an

E. coli O113: H10 WHO standard dose-response curve in saline. An O111

dose-response curve in saline was set up in parallel. With this sesame oil

preparation, negative interference occurred, but the reactivity of O111 in

sesame oil and O113 in saline was very similar (Fig. 6).

0 1 2 40.0

0.5

1.0

1.5O111 in salineO111 in sesame oilO113 in saline

O111 (EU/ml)

OD

450

Fig. 6: Comparison of the activity of LPS from E. coli O111: B4 in saline

and in sesame oil with the E. coli O113: H10 WHO standard in saline


VITRO TEST

109

With these results, it was concluded that the O111 saline curve would not be

suitable for judging lipidic endproducts, since the pyrogenic contaminations

would be underestimated. The similarity of the O111 in sesame oil to the WHO

standard allows the conclusion that the legend “EU/ml” is adequate for

endproduct testing. Since it is useful to set up the reference in the same diluent

used for the sample, the O111 dose-response curve was from then on

performed in sesame oil in order to serve as reference for product testing.

Additionally, the following minimum assay suitability requirements for the testing

of the endproducts were determined:

A) The 2.0 EU/ml concentration of the Gram-negative standard in sesame oil

must test positive (more than 1.6 times the mean OD of the negative sesame oil

control). In the In vitro Pyrogen Test, a value higher than 1.6 times the mean of

the corresponding negative control is considered significantly positive. The

factor 1.6 derives from the average cv (below 20%), equal to 0.2 in the

calculation. This 0.2 value is multiplied by 3 in order to create a broad safety

margin, should the cv be higher than average.

B) The Gram-positive (LTA) control must test positive (more than 1.6 times the

mean of the negative sesame oil control).

C) The 2.0 EU/ml PPC of the Gram-negative control must be in the

interference-free (50-200%) range of the 2.0 EU/ml concentration of the dose-

response curve.

D) The Gram-positive PPC must be in the interference-free range of the Gram-

positive control in sesame oil.

E) The OD of the negative sesame oil control as well as the NPC must not

exceed 0.1.

Applying these criteria, all three end products were tested three times

independently and with three different lots of cryopreserved blood in the IPT.

Results are summarized in Table 3.


VITRO TEST

110

Quality criteria (A-E) met Experiment

Number Testoviron Gynodian Noristerat

1 Yes Yes Yes

2 Yes Yes Yes

3 No* Yes Yes

* Criterion B not met (Gram-positive control did not test positive), therefore

criterion D could not be met as well

Table 3: Results of the final testing of all three substances in the IPT

7.5. Discussion

In this study, the strongly varying characteristics of different oily preparations

concerning their interference with the human whole blood test based on pooled

cryopreserved blood were assessed with altogether five oils and three

endproducts. While most oils tested interference-free when compared to an

LPS dose-response curve in saline, others caused strong negative interference.

Interference testing is one of the most critical aspects of pyrogen testing. First of

all, it requires the availability of a pyrogen-free reference material, which is often

difficult to obtain or to judge, especially in the case of negative interference in

the respective pyrogen test. The lipid A portion as the active part of the LPS

molecule is integrated into lipid layers, e.g. into the membranes of liposomes

and reduces IL-1.beta. production (173). Therefore, a strong interference with

the pyrogen test was expected, but the oils as biological products showed

different characteristics that appeared to be independent of integration and

inactivation of LPS. No solution could be found for the substance benzyl

benzoate. In principle, all oils that are interference-free in the test are suitable

as diluents for more complicated preparations that cannot be tested in their

undiluted form.


VITRO TEST

111

Taking into account the fever threshold of humans at 5 EU/kg (48) on which

ELC calculations are based, the maximum contamination that can be tolerated

was calculated. Since the endproducts are applied at a dose of 1 ml

intramuscularly and assuming an average body weight of 70 kg, a theoretical

contamination of 350 EU/ml would still be non-pyrogenic. It should be noted that

this criterion was set for intravenous drugs which might suggest an even higher

possible ELC. In general, these calculations suggest that most end-products

can be controlled at the respective maximum valid dilution. With a protocol

allowing 20 EU/ml to be detected, a broad safety margin could be achieved.

The calibration of the O111 LPS in sesame oil against the international WHO

endotoxin standard from E. coli O113 (56) showed that “endotoxin unit”

(EU/ml) is appropriate in the case of dose-response curves in lipids. At the

same time, this experiment showed that the pyrogenic activity of the O111 in

saline would have been overestimated. This led to the establishment of the

reference dose-response curve in sesame oil. In contrast to the rabbit test, a

positive product control as well as a negative control could be established;

the successful retrieval of the former serving as proof that a contamination of

the “clean” drug would have been detected at the respective dilution. A Gram-

positive control and a corresponding PPC could be established as well.

Furthermore, the distribution of LPS plays a major role. A homogenous

redistribution of LPS in lipidic products could be demonstrated by calculating the

coefficient of variation, thus proving that the lipid samples can be pipetted in a

satisfying way even at very small volumes (20 µl in the microtiter plate). For all

experiments, cryopreserved human whole blood was used (73). An interesting

result was that the reactivity of the three lots of cryopreserved blood differed in

their response to the Gram-positive stimulus. In endproduct testing, one lot of

cryopreserved blood failed to detect the Gram-positive control in the testing of

the Testoviron®; while when was used to test the other two endproducts, the

signals were very low, though positive. The other two lots were highly sensitive

towards this stimulus. The reactivity towards the Gram-negative controls was far

more homogenous in all experiments. This stresses the importance of prior

testing before releasing the cryopreserved blood by the manufacturer and


VITRO TEST

112

indicates at the same time that the individual sensitivities towards Gram-positive

contaminations are much higher than those for LPS.

Lipidic substances in general remain a challenge for the IPT. The established

protocol has been optimized for the given endproducts, but can be modified,

e.g. concerning the diluent and/or the necessary dilutions, in order to meet the

requirements of other endproducts. Large volume parenterals for total

parenteral nutrition (TPN), are particularly difficult to test, since, apart from

containing fats, they also require an extremely low detection limit for successful

pyrogen testing. Further interesting preparations to be examined in the IPT are

liposomes. These are used as vehicles for drugs, releasing the drug at or

nearby the desired site of action where they are degraded and could release

possible pyrogenic contaminations.

7.6. Conclusion

This study shows that contaminations of lipidic formulations of drugs cause a

strong and sensitive reaction of human monocytes, indicating that they have the

potential to be a health hazard to patients receiving these drugs.

Taking this into account, this study provides compelling evidence that the IPT

assay can overcome most of the problems associated with the control of lipidic

endproducts. It therefore offers a viable alternative to testing these preparations

in rabbits.

SUMMARIZING DISCUSSION

113

8 Summarizing discussion

The desire to make available an in vitro pyrogen test as an alternative method

to the rabbit pyrogen test is prompted by ethical considerations of animal

protection as well as to matters of patient safety. It is known long that the raise

in temperature in the rabbit represents an unreliable parameter due to stress-

induced hyperthermia, hypothermia (29), sensitivity differences of rabbit strains

(27), age, sex and even between individuals of the same breeds who have

been found to be high, low, and intermediate responders (174). Therefore, the

LAL was introduced in pyrogen testing in the 70s and is now widely used. The

major limitations of this test is, as its alternative name bacterial endotoxin test

(BET) implies, being limited to the pyrogens derived from Gram-negative

bacteria and missing known pyrogens as lipoteichoic acid and peptidoglycans

as well as exotoxins and other non-endotoxin pyrogens. Pyrogenic reactions in

healthy volunteers have been reported during a clinical study when a

preparation of human growth hormone was injected that had tested negative in

the LAL and in the rabbit (133). Adverse reactions after the application of

vaccine against early summer meningoencephalitis have been observed,

although the LAL had tested negative (59). An assay capable of detecting all

relevant pyrogens should mirror closely the physiological events which take

place in the mammalian organism during a pyrogenic reaction.

Upon contact with so-called pathogen-associated molecular patterns (PAMPS)

(127), e.g. bacterial, fungal or viral components, the blood monocytes are

activated. The Cluster of Differentiation (CD) 14 receptor, a glycoprotein whose

significance was recognized in 1990, exists in a membrane-bound (mCD14, a

53 kDa protein) and a soluble form. The former is embedded in the plasma

membrane, the latter, sCD14, is capable of inhibiting LPS activity (175), but

additionally LPS/sCD14 complexes can induce biological responses in certain

CD14-negative cells such as endothelium (176). LPS binding takes place via

the Lipid A component and is of high affinity (177).


114

The innate immune system of mammals uses the family of toll-like receptors

(TLR) to engage microorganisms by recognizing PAMPS (178). For the

recognition of LPS, the TLR4 receptor is crucial (179). The role of TLR4 was

determined in TLR4 deficient mice which proved to be unresponsive to LPS

(180). Upon binding, the TLR4 receptor dimerizes and the intracellular signal

cascade is initiated.

In contrast, the pyrogen LTA, but also peptidoglycans and lipoproteins, bind to

the TLR2 (181, 182) and cause fever along the same lines as LPS. The

receptor engagement activates intracellular signaling cascades which lead to

the formation of proinflammatory cytokines such as Interleukin-1β (IL-1β), IL-6

and others by the blood monocytes. The IPT measures the endpoint IL-1β.

The IL-1β exists in the mononuclear cells as a preformed molecule, the pro-IL-

1β. Upon stimulation, a protease is activated, the IL-1 Converting Enzyme

(ICE), whose catalytic function is essential for the generation of mature,

extracellular IL-1β (183). It is an interesting phenomenon that in studies with

human subjects, no IL-1β could be found in the blood stream after LPS

injection or in naturally occurring sepsis, whereas IL-6 was always detectable

(184-186). Messengers this potent are probably quickly bound to their

corresponding receptors, e.g. to the soluble Interleukin-1 receptors (187), or

internalized, e.g. by binding to/crossing the blood-brain barrier (188, 189),

making them disappear fast from the blood serum. The individual roles of the

proinflammatory cytokines in the generation of fever are still controversially

discussed, but several lines of evidence suggest that IL-1β is indeed the most

important and potent inducer of fever. Several studies in humans (190-193)

have demonstrated the pyrogenic potency of this cytokine and studies in rabbits

have shown clearly that the doses of IL-1β required are 100fold lower than

those of IL-6 and around 10fold lower than those of TNF-α (46, 47).

Furthermore, the cytoplasmic domains of TLRs and the IL-1β receptor share

the same signaling areas (38). The necessary intracellular enzymatic cleavage

of pro-IL-1β might provide an additional control mechanism of a mediator which

is crucial in the genesis of pyrogenic reactions.

These events in the mammalian organism provided the idea for the principle of

the whole blood test. When in contact with a pyrogen, the monocyte, a


115

subfraction of the white blood cells, reacts with the production of

proinflammatory cytokines, one of which is IL-1β. By using the whole blood as a

reagent, the monocyte is left in its natural environmant with all components

necessary for a reaction that is as close to the in vivo situation as possible.

IL-1β was chosen as a readout since it is a very stable monomer, which

survives repeated freezing-thawing processes, something that might be

necessary during the establishment of the system and for research purposes,

and since it is highly regulated, which might prevent or reduce false-positive

reactions or basal levels in unstimulated samples. Furthermore, IL-1β has been

recognized as the most potent cytokine when injected in vivo (47, 190-193) and

appears to play a crucial, if not the crucial role in the pathogenesis of fever.

In order to make the new IPT available on a large scale and therefore a true

alternative to animal experiments, all reagents had to be standardized and

certified. A major issue was the certification and standardization of the human

blood. Freshly drawn human blood cannot be stored for more than 4 hours until

the IL-1 formation declines, and is difficult to measure. Individual differences if

not in the sensitivity, but the amount of cytokine produced can make

interpretation of the results difficult. Furthermore, a risk of infection for the user

with e.g. hepatitis cannot be excluded easily. In the first part of this work, a

protocol for pooling and freezing whole blood was developed which maintains

all the characteristics of the fresh blood. The cryopreservation of isolated blood

cells has been successfully performed since the 1960s for transfusion and

research purposes (73, 194-197). By freezing the entire blood without any

isolation procedures, the monocyte is maintained in its natural environment with

all the known and unknown factors that might influence the response, thus

maintaining their reactivity and specificity with regard to pyrogen testing. An

important plasma protein is for example the LPS binding protein (LBP) which in

the blood of healthy donors presents the LPS to the CD14 receptor (162, 163).

While red blood cells require very high freezing rates in order to remain intact,

the white blood cells are best frozen at rates of 1-2°C/min. This is due to the

process of exosmosis of intracellular water which depends on parameters like

cell surface/volume ratio and membrane permeability and which was described

by Mazur in 1963, 1965 and 1977 (198-200). Our protocol therefore had to


116

tolerate the complete destruction of the erythrocytes with the risk, that the

reaction of the monocytes would be influenced by the cell detritus. This was, at

least in our experiments, never the case. The procedure involved clinical,

endotoxin-free DMSO at a 10% final concentration in the blood (v/v ratio) and a

computer-controlled freezing process using liquid nitrogen. For higher

standardization and in order to level out individual differences in the response,

the blood of 5 donors was pooled and frozen together. The frozen blood proved

to be a highly sensitive and robust reagent with a very high inter-lot

comparability. The possibility of pretesting the donors in question for infectious

agents eliminates the abovementioned health hazard for the personnel.

In order to make the IPT available for routine application, the assay using

cryopreserved blood, allowing storage at -80°C as compared to liquid nitrogen

as well as an improved variation of the fresh blood was validated in an

international collaborative study which followed the procedure in which the

fresh blood had been validated previously in detail. The validation included

three laboratories working under good laboratory practice (GLP) with 10

substances and altogether 50 blinded endotoxin spikes at or around the

pyrogenic threshold of 0.5 EU/ml (16). The IPT achieved sensitivities around

90% and specificities around 80%. The cryopreserved blood described in part

one had a lower sensitivity of 80% with at the same time the best specificity.

Compared to the former study (48), the whole blood assay could be improved

regarding consumer-friendlyness as well as performance. Based on this

outcome, the inclusion of the IPT into the Pharmacoopoeias should be

possible.

In the last part of this work, the validated assay using the cryopreserved blood

was adapted to suit a special application. So far, only hydrophilic substances

mainly for intravenous administration had been used in the test. A change in

regulation in 2004 by European Pharmacopoeia made the testing of so-called

small volume parenterals (SVP) obligatory that had so far not been subjected to

pyrogen testing. Suddenly, several lipophilic parenterals, e.g. with fat-soluble

vitamins or hormones, for intramuscular and subcutaneous application were


117

concerned. Testing lipophilic substances in the rabbit by intravenous injection

into the ear vein which is the accepted procedure, is extremely dangerous since

the insoluble miniscule drops can clot capillaries in vital organs such as the

kidneys. Apart from the viewpoint of animal protection, the outcome of such an

experiment must be doubtful. The LAL, on the other hand, is impeded by the

fact that the pyrogenic portion, the lipid A, is masked by lipid substances, lipidic

parenterals (122), lipoproteins (121, 168), and liposomes (173) and therefore

no longer accessible to the components of the coagulation cascades, leading to

an underestimation of the pyrogenic contamination. It was therefore a pressing

question whether such substances can be controlled by the IPT. For this

purpose, the standard protocol was modified and different pure oils as well as

three endproducts were measured. It turned out that all products can be

controlled, although the pyrogenic threshold of 0.5 EU/ml for hydrophilic

substances could not be maintained in this product group. Still, since the drugs

are given at a very small volume, a higher contamination can be tolerated

which will predictably not cause any adverse reactions in the recipient.

Taken together, with the new assay based on human whole blood, a

standardized, reliable and highly sensitive method based on the human fever

reactions and measuring all relevant pyrogens is now available for widespread

use. We hope that the replacement of the rabbit pyrogen test by this assay is

therefore only a question of time, simultaneously maintaining and even

exceeding the already high level of safety for patients receiving parenterals.

ZUSAMMENFASSUNG

118

9 Summary

The detection of bacterial contaminations of all origins in parenterals has been

recognized as an important issue. In 1943, the in vivo rabbit pyrogen test was

introduced into the US Pharmacopoeia and has been used since then. In the

70s, another pyrogen test, the limulus amoebocyte lysate test (LAL) was

introduced which uses the lysates of blood cells of the horseshoe crab and

cannot measure a variety of common pyrogens nor distinguish between the

potencies of given pyrogens in the mammal. The immune system is able to

respond to pathogens with production and secretion of cytokines. In 1995, a

novel in vitro alternative based on human whole blood which uses this reaction

for the detection of all possible kinds of microbial contaminations has been

developed. Its reliability and wide spectrum of application possibilities make it a

promising candidate for entirely replacing the rabbit test.

• As a first step, the human whole blood test using freshly drawn human

whole blood was validated in an international collaborative study.

• The most critical reagent of the new test, the human whole blood, was made

available by developing a protocol for pooling and cryopreserving the fresh

blood, at the same time maintaining the characteristics of the fresh blood

regarding the reactivity towards all relevant pyrogens.

• This blood was then validated in an international study in order to be able to

include the assay into Pharmacopoeias.

• With this validated procedure, the applications of the test were extended

from hydrophilic to lipophilic parenterals meeting new requirements of

European Pharmacopoeia.

With all these measures, the new pyrogen assay based on whole blood which

measures all relevant pyrogens inducing a fever reaction is ready for routine

use. This work adds to the replacement of an important and widely used animal

test while at the same time maintaining, even superceding the existing safety

standards and extending the possibilitites of applications beyond those of the

rabbit test and even the LAL.

ZUSAMMENFASSUNG

119

10 Zusammenfassung

Die Detektion von bakteriellen Kontaminationen jeglicher Herkunft in

Parenteralia ist als wichtiges Problem erkannt. 1943 wurde der in vivo

Kaninchenpyrogentest in die US Pharmacopoe eingeführt und wird seither

verwendet. In den 1970er Jahren wurde ein weiterer Pyrogentest, der Limulus

Amoebozyten Lysattest (LAL) eingeführt, der das Lysat der Blutzellen der

Hufeisenkrabbe einsetzt und der weder alle relevanten Pyrogene messen noch

zwischen ihrer jeweiligen biologischen Potenz im Säugetier unterscheiden

kann. Das Immunsystem antwortet auf Pathogene mit der Produktion und

Sekretion von Zytokinen. 1995 wurde eine in vitro Alternative mit humanem

Vollblut entwickelt, die diese Reaktion zur Detektion aller möglichen

mikrobiellen Kontaminationen nutzt. Ihre Zuverlässigkeit und das weite

Spektrum an Anwendungsmöglichkeiten machen sie zu einem

vielversprechenden Kandidaten, den Kaninchentest vollständig zu ersetzen.

• Als ein erster Schritt wurde der Test, basierend auf frisch abgenommenem

humanem Vollblut, in einer internationalen Studie validiert.

• Der wohl kritischste Bestandteil des Tests, das menschliche Vollblut, wurde

verfügbar gemacht, indem ein Protokoll zum Mischen und Einfrieren von

frischem Blut entwickelt wurde. Gleichzeitig blieben die Eigenschaften des

frischen Blutes in Bezug auf die Reaktivität gegenüber allen relevanten

Pyrogenen erhalten.

• Parallel zu dem Frischblut wurde dieses Blut in einer internationalen Studie

validiert, um den Test in die Pharmakopoen einführen zu können.

• Mit dieser validierten Methode wurden die Anwendungen für den Test

erweitert, um den neuen Anforderungen der Europäischen Pharmacopoe

Genüge zu tun: von hydrophilen zu lipophilen Parenteralia.

Damit ist der neue Pyrogentest auf Vollblutbasis, der alle relevanten

fiebererzeugenden Pyrogene mißt, zum routinemäßigen Einsatz bereit. Die

vorliegende Arbeit trägt dazu bei, einen wichtigen und viel genutzten

Tierversuch zu ersetzen, gleichzeitig die bestehenden Sicherheitsstandards

beizubehalten bzw. zu übertreffen und die Anwendungsmöglichkeiten über die

des Kaninchentests und sogar des LAL hinaus auszuweiten.

REFERENCES

120

11 References

1. Semmelweis, I. P. (1861). Die Ätiologie der Begriff und die Prophylaxis des

Kindbettfiebers. C. A. Hartleben´s Verlags-Expedition: Pest

2. Godlee, R. J. (1917). Lord Lister, London

3. von Haller, A. (1757-66). Elements physiologiae corporis humanae Vol III

4. Panum, P. L. (1874). Das putride Gift, die bakterien, die putride infektion

oder intoxikation und die septikämie

Arch Pathol Anat Physiol Klein Med (Virchow´s Arch) 60: 301

5. Billbroth, T. (1862). Beobachtungsstudien über das wundfieber und

accidentelle wund-krankheiten. Arch Klin Chir 2: 578

6. Centanni, E. (1921). Trattato di Immunologia

Societa Editrice Libraria, Rome: 149

7. Hort, E., Penfold, W. (1912). Microorganisms and their relation to fever. J

Hyg 12: 361

8. Seibert, F. B. (1925). The cause of many febrile reactions following

intravenous injection. Am J Physiol 71: 621

9. Rademaker, L. (1933)

The cause and elimination of reactions after intravenous infusions

Surg Gynaecol Obstet 56: 956

10. Welsh, H., Calvery, H. O., McClosky, W. T., and Price, C. W. (1943)

Method of Preparation and Test for Bacterial Pyrogen

J Am Pharm Assoc 3: 65-69

REFERENCES

121

11. McClosky, W. T., Price, C. W., van Winkle, W. J., Welch, H., Calvery, H.

O. (1943). Results of the first USP collaborative study of pyrogens.

J Am Pharm Assoc 32: 69

12. Nowotny, A. (1983). Shedding of bacteria. Biomembranes 11: 1-20

13. Galanos, C., Rietschel, E. T., Luderitz, O., Westphal, O., Kim, Y. B.,

Watson, D. W. (1972). Biological activities of lipid A complexed with bovine-

serum albumin. Eur J Biochem 31: 230-3

14. Rietschel, E. T., Kim, Y. B., Watson, D. W., Galanos, C., Luderitz, O., and

Westphal, O. (1973). Pyrogenicity and immunogenicity of lipid A complexed

with bovine serum albumin or human serum albumin. Infect Immun 8: 173-7

15. U. S. Food and Drug Administration (1980)

Human and veterinary drugs; availability of draft guidelines for use of Limulus

amoebocyte lysate. Fed Reg 45: 3668

16. Hoffmann, S., Luederitz-Puechel, U., Montag-Lessing T., and Hartung, T.

(2005). Optimisation of pyrogen testing in parenterals according to different

pharmacopoeias by probabilistic modelling. J Endotoxin Res 11: 25-31

17. Bodel, P.T., and Atkins, E. (1965). Studies in staphylococcal fever.

V. Staphlococcal filtrate pyrogen. Yale J Biol Med 38(3): 282-98

18. Brunson, K.W., and Watson, D.W. (1974)

Pyrogenic specificity of streptococcal exotoxins, staphylococcal enterotoxin,

and gram-negative endotoxin. Infect Immun 10: 347-51

19. Watson, D. W. (1960). Host-parasite factors in group A streptococcal

infections: pyrogenic and other effects of immunologic distinct exotoxins

related to scarlet fever toxins . J Exp Med 111: 255-284

REFERENCES

122

20. Atkins, E., and Huang W. C. (1958). Studies on the pathogenesis of fever

with influenza virus. I. The appearance of an endogenous pyrogen in the

blood following intravenous injection of virus. J Exp Med 107: 383-401

21. Atkins, E., and Morse, S. I. (1967). Studies in staphylococcal fever. VI.

Responses induced by cell walls and various fractions of staphylococci and

their products. Yale J Biol Med 39: 297-311

22. Rotta, J. (1975). Endotoxin-like properties of the peptidoglycan.

Z Immunitätsforsch Exp Klin Immunol 149: 230-44

23. Schleifer, K. H. (1975). Chemical structure of the peptidoglycan, its

modifiability and relation to the biological activity.

Z Immunitätsforsch Exp Klin Immunol 149: 104-17

24. Kobayashi, G. S., and Friedman, L. (1964). Characterization of the

pyrogenicity of Candida albicans, Saccharomyces cervisiae, and

Cryptococcus neoformans. J Bacteriol 88: 660-6

25. Braude, A. I., McConnell, J., and Douglas, H. (1960). Fever from

pathogenic fungi. J Clin Invest 39: 1266-76

26. Morath, S., Geyer, A., and Hartung, T. (2001). Structure-function

relationship of cytokine induction by lipotechoic acid from Staphylococcus

aureus. J Exp Med 193: 393

27. van Dijck, P. and H. van de Voorde (1977). Factors affecting pyrogen

testing in rabbits. Dev Biol Stand 34: 57-63

28. Deeter, L. B., L. W. Martin, and Lipton, J. M. (1989).

Age- and sex-related differences in febrile response to peripheral pyrogens in

the rabbit. Gerontology 35: 297-304.

REFERENCES

123

29. Grant, R. (1950). Emotional hypothermia in rabbits. Am J Physiol 160:

285

30. Greisman, S. E. and Hornick, R. B. (1969). Comparative pyrogenic

reactivity of rabbit and man to bacterial endotoxin. Proc Soc Exp Biol Med

131: 1154

31. Levin, L. and Bang, F. B. (1964). A description of cellular coagulation in

the limulus. Bull Johns Hopkins Hosp 115: 337-45

32. Levin, J. and Bang, F. B. (1964). The role of endotoxin in the

extraacellular coagulation of limulus blood.

Bull Johns Hopkins Hosp 115: 265-74

33. Yin, E. T., Galanos, C., Kinsky, S., Bradshaw, R. A., Wessler, S., Luderitz,

O., and Sarmiento, M. E. (1972). Picogram-sensitive assay for endotoxin:

Gelation of Limulus polyphemus blood cell lysate induced by purified

lipopolysaccharides and lipid A from gram-negative bacteria

Biochem Biophys Acta 26: 284

34. Devleeschouwer, M. J., Cornil, M. F., and Dony, J. (1985). Studies on the

sensitivity and specificity of the Limulus amebocyte lysate test and rabbit

pyrogen assays. Appl Environ Microbiol 50: 1509-11

35. Wildfeuer, A., Heymer, B., Schleifer, K. H., and Haferkamp, O. (1974)

Investigations on the specificity of the Limulus test for the detection of

endotoxin. Appl Microbiol 28: 867-71

36. Beeson, P. B. (1948). Temperature-elevating effect of a substance

obtained from polymorphonuclear leucocytes (abstract)

J Clin Invest 27: 524

REFERENCES

124

37. Dinarello, C. A., Goldin, N P., and Wolff, S. M. (1974). Demonstration and

characterization of two distinct human leukocytic pyrogens. J Exp Med 139:

1369-81

38. Gay, N. J., and Keith, F. J. (1991). Drosophila Toll and IL-1 receptor.

Nature 351(6325): 355-6

39. Li, S., Wang, Y., Matsumura, K., Ballou, L R., Morham, S. G., and

Blatteis, C. M. (1999). The febrile response to lipopolysaccharide is blocked in

cyclooxygenase-2 (-/-), but not in cyclooxygenase-1(-/-) mice.

Brain Res 825: 86-94

40. Li, S., Ballou, L. R., Morham, S. G., and Blatteis, C. M. (2001)

Cyclooxygenase-2 mediates the febrile response of mice to interleukin-1beta.

Brain Res: 163-73

41. Li, S., Goorha, S., Ballou, L. R., and Blatteis, C. M. (2003).

Intracerebroventricular interleukin-6, macrophage inflammatory protein-1 beta

and IL-18: pyrogenic and PGE(2)-mediated? Brain Res 992: 76-84

42. Chai, Z., Gatti, S., Toniatti, C., Poli, V., and Bartfai, T. (1996)

Interleukin (IL)-6 gene expression in the central nervous system is necessary

for fever response to lipopolysaccharide or IL-1 beta: a study on IL-6-deficient

mice. J Exp Med 183: 311-6

43. Ushikubi, F., Segi, E., Sugimoto, Y., Murata, T., Matsuoka, T., Kobayashi,

T., Hizaki, H., Tuboi, K., Katsuyama, M., Ichikawa, A., Tanaka, T., Yoshida,

N., and Narumiya, S. (1998). Impaired febrile response in mice lacking the

prostaglandin E receptor subtype EP3. Nature 395 (6699): 281-4

44. Dinarello, C. A., Gatt, S., and Bartfai, T. (1999)

Fever: links with an ancient receptor

Curr Biol 9: R147-50

REFERENCES

125

45. Laburn, H. P., Rosendorff, C., Willies, G., and Woolf, C. (1974)

Proceedings: A role for noradrenaline and cyclic AMP in prostaglandin E1

Fever. J Physiol 240: 49P-50P

46. Dinarello, C. A., Cannon, J. G., Wolff, S. M., Bernheim, H. A., Beutler, B.,

Cerami, A., Figari, I. S., Palladino, M. A., and O'Connor, J. V. (1986).

Tumor necrosis factor (cachectin) is an endogenous pyrogen and induces

production of interleukin 1. J Exp Med 163:1433-50

47. Dinarello, C. A., Cannon, J. G., Mancilla, J., Bishai, I., Lees, J., and

Coceani, F. (1991). Interleukin-6 as an endogenous pyrogen: induction of

prostaglandin E2 in brain but not in peripheral blood mononuclear cells.

Brain Res 562:199-206

48. Hoffmann, S., Peterbauer, A., Schindler, S., Fennrich, S., Poole, S.,

Mistry, Y., Montag-Lessing, T., Spreitzer, I., Löschner, B., van Aalderen, M.,

Bos, R., Gommer, M., Nibbeling, R., Werner-Felmayer, G., Loitzl, P., Jungi,

T., Brcic, M., Brugger, P., Frey, E., Bowe, G., Casado, J., Coecke, S., de

Lange, J., Mogster, B., Naess, L. M., Aaberge, I. S., Wendel, A., and Hartung,

T. (2005). International validation of novel pyrogen tests based on human

monocytoid cells. J Immunol Methods: 161-73

49. Hartung, T., and Wendel, A. (1995). Detection of Pyrogens using human

whole blood. ALTEX 12: 70-75

50. Rudloe A, Hernkind A. E. (1983). The effect of heavy bleeding on mortality

of the horseshoe crab, Limulus polyphemus, in the natural environment.

J Invertebr Pathol 42: 167–176

51. Thompson, M.(1998). Assessments of the population biology and critical

habitat for the horseshoe crab, Limulus polyphemus, in the South Atlantic Bight

Master’s thesis Medical University of South Carolina, University of Charleston.

REFERENCES

126

52. Walls E. A., Berkson, J. (2003). Effects of blood extraction on horseshoe

crabs (Limulus polyphemus). Fish Bull 101. 457-459

53. Fennrich, S., Fischer, M., Hartung, T., Lexa, P., Montag-Lessing, T.,

Sonntag, H. G., Weigandt, M., Wendel, A (1999). Detection of endotoxins and

other pyrogens using human whole blood. Dev Biol Stand 101: 131-139

54. Bennett, I. L., Beeson, P. B. (1953). Studies on the pathogenesis of fever.

II. Characterization of fever-producing substances from polymorphonuclear

leukocytes and from the fluid of sterile exsudates. J Exp Med 98: 493-508

55. Dinarello, C. A. (2004). Infection, fever, and exogenous and endogenous

pyrogens: some concepts have changed. J Endotoxin Res 10: 201-22

56. Poole, S., Dawson, P., Gaines Das, R. E. (1997). Second international

standard for endotoxin: calibration in an international collaborative study.

J Endotoxin Res 4: 221-231

57. Morath, S., Geyer, A., Spreitzer, I., Hermann, C., Hartung, T. (2002).

Structural decomposition and heterogeneity of commercial lipoteichoic acid

preparations. Infect Immun 70 (2): 938-44

58. Schindler, S., Bristow, A., Cartmell, T., Hartung, T., Fennrich, S. (2003).

Comparison of the reactivity of human and rabbit blood towards pyrogenic

stimuli. ALTEX 20: 59-63

59. Fischer, M., Keller-Stanislawski, B., Schober-Bendixen, S., Schosser, R.,

Hacke, K., Hartung, T., Montag, T. (2001). Effect of the preservative thiomersal

on the release of interleukin-1 beta from human peripheral blood cells.

ALTEX 18 (1): 47-9.

REFERENCES

127

60. Carlin, G., Viitanen, E. (2003). In vitro pyrogenicity of a multivalent vaccine:

Infanrix. Pharmeuropa 15: 418-23.

61. Carlin, G., Viitanen, E. (2005). In vitro pyrogenicity of the diphtheria, tetanus

and acellular pertussis components of a trivalent vaccine. Vaccine 23: 3709-15.

62. Steere, A. C., Rifaat, M. K., Seligmann, E. B., Hochstein, H. D., Friedland,

G., Dasse, P., Wustrack, K. O., Axnick, K. J., Barker, L. F. (1978). Pyrogen

reactions associated with the infusion of normal serum albumin (human).

Transfusion 18 (1): 102-7

63. Pool, E. J., Johaar, G., James, S., Petersen, I., Bouic, P. (1998). The

detection of pyrogens in blood products using an ex vivo whole blood culture

assay. J Immunoassay 19: 95-111

64. Spreitzer, I., Fischer, M., Hartzsch, K., Luderitz-Puchel, U., Montag, T.

(2002). Comparative study of rabbit pyrogen test and human whole blood assay

on human serum albumin. ALTEX 19 Suppl 1: 73-5

65. Eperon, S., Jungi, T. W. (1996). The use of human monocytoid lines as

indicators of endotoxin. J Immunol Methods 194: 121-129

66. Werner-Felmayer, G., Baier-Bitterlich, G., Fuchs, D., Hausen, A., Murr, C.,

Reibnegger, G., Werner, E. R., Wachter, H. (1995). Detection of Bacterial

Pyrogens on the Basis of their Effects on Gamma Interferon-Mediated

Formation of Neopterin or Nitrite in Cultured Monnocyte Cell Lines. Clin. Diagn.

Lab. Immunol. 2 (3): 307-313

67. Peterbauer, A., Werner-Felmayer, G. (1999). Further development of a cell

culture model for the detection of bacterial pyrogens. ALTEX 16: 3-8

REFERENCES

128

68. Ziegler-Heitbrock, H. W., Thiel, E., Futterer, A., Herzog, V., Wirtz, A.,

Riethmuller, G. (1988). Establishment of a human cell line (Mono Mac 6) with

characteristics of mature monocytes. Int J Cancer 41: 456-461

69. Hartung, T., Fennrich, S., Fische, M., Montag-Lessing, T., Wendel, A.(2000).

Prevalidation of an alternative to the rabbit pyrogen test based on human whole

blood. In: Progress in the reduction, refinement and replacement of animal

experimentation. Balls, M., van Zeller, A. M., Halder, M. E.

2000 Elsevier Science B. V. 2000, pp 991-99

70. Hartung, T., Aaberge, I., Berthold, S., Carlin, G., Charton, E., Coecke, S.,

Fennrich, S., Fischer, M., Gommer, M., Halder, M., Haslov, K., Jahnke, M.,

Montag-Lessing, T., Poole, S., Schechtman, L., Wendel, A., Werner-Felmayer,

G. (2001). Novel pyrogen tests based on the human fever reaction. The report

and recommendations of ECVAM workshop 43. European Centre for the

Validation of Alternative Methods. Altern Lab Anim 29 (2): 99-123

71. Hartung, T. (2002). Comparison and validation of novel pyrogen tests based

on the human fever reaction. Altern Lab Anim 30 Suppl 2: 49-51

72. De Boer, M., Reijneke, R., Van de Griend, R. J., Loos, J. A., Roos, D.

(1981). Large-scale purification and cryopreservation of human monocytes.

J Immunol Methods 43 (2): 225-39

73. Schindler, S., Asmus, S., von Aulock, S., Wendel, A., Hartung, T., Fennrich,

S. (2004). Cryopreservation of human whole blood for pyrogenicity testing.

J Immunol Methods 294 (1-2): 89-100

74. Schindler, S., Spreitzer, I., Löschner, B., Hoffmann, S., Hennes, K., Halder,

M., Brügger, P., Frey, E., Hartung, T., Montag-Lessing, T. (2006). International

validation of pyrogen tests based on cryopreserved human primary blood cells.

J Immunol Methods in print

REFERENCES

129

75. Association for the Advancement of Medical Instrumentation

AAMI Standards and Recommended Practices, vol 3

American National Standards Institute, Arlington 3: 1-332, 2001

76. Shmunes , Darby, T. (1984). Contact dermatitis due to endotoxin in

irradiated latex gloves. Contact Dermatitis 10: 240-4

77. Kure, R., Grendahl, H., Paulssen, J. (1982). Pyrogens from surgeons' sterile

latex gloves. Acta Pathol Microbiol Immunol Scand [B] 90: 85-8

78. Groetsch, W., Leimbach, R., Sonnenschein, B. (1992). On the safety of

medical products: The detection of endotoxin on sterile surgical gloves.

Hyg Med 17: 200-206

79. Mazzotti, F., Beuttler, J., Zeller, R., Fink, U., Schindler, S., Wendel, A.,

Hartung, T., von Aulock, S. (2006). In vitro Pyrogen Test - a new test method for

solid medical devices. J Biomed Mater Res Part A, in print

80. Anderson, J. M., Miller, K. M. (1984). Biomaterials biocompatibility and the

macrophage. Biomaterials 5: 5-10

81. Ziats, N. P., Miller, K. M., Anderson, J. M. (1988). In vitro and in vivo

interactions of cells with biomaterials. Biomaterials 1988; 9: 5-13

82. Rich, A., Harris, A. K. (1981). Anomalous preference of cultured

macrophages for hydrophobic and roughened substrata. J Cell Sci 50: 1-7

83. Soskolne, W. A., Cohen, S., Sennerby, L., Wennerberg, A., Shapira, L.

(2002). The effect of titanium surface roughness on the adhesion of monocytes

and their secretion of TNF-α and PGE2. Clin Oral Implants Res 13 (1): 86-93

REFERENCES

130

84. Refai, A. K., Textor, M., Brunette, D. M., Waterfield, J. D. (2004). Effect of

titanium surface topography on macrophage activation and secretion of

proinflammatory cytokines and chemokines. J Biomed Res 70A: 194-205

85. Kullmann, K. (2002). Adaption des In Vitro Pyrogen tests (IPT) für

prothetische Materialien. Master Thesis, Fachbereich Biologie der Universität

Konstanz

86. Rosenberg, S. A. (2001). Cellular therapy: an introduction.

Cancer J 7 (Suppl 2): 51-2

87. Blajchman, M. A., Beckers, E. A., Dickmeiss, E., Lin, L., Moore, G., Muylle,

L. (2005). Bacterial detection of platelets: Current problems and possible

resolutions. Transfus Med Rev 19 (4): 259-272

88. Perez, P., Salmi, L. R., Follea, G., Schmit, J. L., de Barbeyrac, B., Sudre, P.,

Salamon, R. (2001). Determinants of transfusion-associated bacterial

contamination: results of the French BACTHEM case-control study. Transfusion

41 (7): 862-871

89. Kuehnert, M. J., Roth, V. R., Haley, N. R., Gregory, K. R., Elder, K. V.,

Schreiber, G. B., Arduino, M. J., Holt, S. C., Carson, L. A., Banerjee, S. N.,

Jarvis, W. R. (2001). Transfusion-transmitted bacterial infection in the United

States, 1998 through 2000. Transfusion 41 (12): 1493-1499

90. McDonald, C. P., Hartley, S. , Orchard, K. , Hughes, G., Brett, M. M., Hewitt,

P. E., Barbara, J. A. (1998). Fatal Clostridium perfringens sepsis from a pooled

platelet transfusion. Transfus Med 8(1): 19-22

91. Wollowitz, S. (2001) Fundamentals of the psoralen-based Helinx

technology for inactivation of infectious pathogens and leukocytes in platelets

and plasma. Semin Hematol 38(4 Suppl 11): 4-11

REFERENCES

131

92. Raij, L., Shapiro, F. L., Michael, A. F. (1973). Endotoxemia in febrile

reactions during hemodialysis. Kidney Int 4: 57-60

93. Favero, M S., Petersen, N. J., Boyer, K. M., Carson, L. A., Bond, W. W.

(1974). Microbial contamination of renal dialysis systems and associated health

risks. Trans Am Soc Artif Intern Organs 20A: 175-83

94. Klein, E., Pass, T., Harding, G. B., Wright, R., Million, C. (1990). Microbial

and endotoxin contamination in water and dialysate in the central United States.

Artif Organs 14: 85-94

95. Pegues, D. A., Oettinger, C. W., Bland, L. A., Oliver, J. C., Arduino, M. J.,

Aguero, S. M., McAllister, S. K., Gordon, S. M., Favero, M. S., Jarvis, W. R.

(1992). A prospective study of pyrogenic reactions in hemodialysis patients

using bicarbonate dialysis fluids filtered to remove bacteria and endotoxin. J Am

Soc Nephrol 3 (4): 1002-7

96. Kulander, L., Nisbeth, U., Danielsson, B. G., Eriksson, O. (1993).

Occurrence of endotoxin in dialysis fluid from 39 dialysis units. J Hosp Infect 24:

29-37

97. Bambauer, .R, Schauer, M., Jung, W. K., Daum, V., Vienken, J. (1994).

Contamination of dialysis water and dialysate. A survey of 30 centers. ASAIO J

40(4): 1012-6

98. Phillips, G., Hudson, S., Stewart, W. K. (1994). Persistence of microflora in

biofilm within fluid pathways of contemporary haemodialysis monitors (Gambro

AK-10). J Hosp Infect 27: 117-25

99. Schouten, W. E., Grooteman, M. P., van Houte, A. J., Schoorl, M., van

Limbeek, J., Nube, M. J. (2000). Effects of dialyser and dialysate on the acute

phase reaction in clinical bicarbonate dialysis. Nephrol Dial Transplant 15 (3):

379-84

REFERENCES

132

100. Arvanitidou, M., Spaia, S., Katsinas, C., Pangidis, P., Constantinidis, T.,

Katsouyannopoulos, V., Vayonas, G. (1998). Microbiological quality of water

and dialysate in all haemodialysis centres of Greece.

Nephrol Dial Transplant 13(4): 949-54

101. Laurence, R. A., Lapierre, S. T. (1995). Quality of hemodialysis water: a 7-

year multicenter study. Am J Kidney Dis 25: 738-50

102. Marion-Ferey, K., Leid, J. G., Bouvier, G., Pasmore, M., Husson, G.,

Vilagines, R. (2005). Endotoxin level measurement in hemodialysis biofilm using

"the whole blood assay". Artif Organs 29 (6): 475-81

103. Evans, R. C., Holmes, C. J. (1991). In vitro study of the transfer of

cytokine-inducing substances across selected high-flux hemodialysis

membranes. Blood Purif. 9: 92-101

104. Lonnemann, G., Behme, T. C., Lenzner, B., Floege, J., Schulze, M.,

Colton, C. K., Koch, K. M., Shaldon, S. (1992). Permeability of dialyzer

membranes to TNF alpha-inducing substances derived from water bacteria.

Kidney Int 42 (1): 61-8

105. Urena, P., Herbelin, A., Zingraff, J., Lair, M., Man, N. K., Descamps-

Latscha, B., Drueke, T. (1992). Permeability of cellulosic and non-cellulosic

membranes to endotoxin subunits and cytokine production during in-vitro

haemodialysis. Nephrol Dial Transplant 7: 16-28

106. Tsuchida, K., Takemoto, Y., Yamagami, S., Edney, H., Niwa, M., Tsuchiya,

M., Kishimoto, T., Shaldon, S. (1997). Detection of peptidoglycan and endotoxin

in dialysate, using silkworm larvae plasma and limulus amebocyte lysate

methods. Nephron 75 (4): 438-43

107. Dinarello, C. A., Koch, K. M., Shaldon, S. (1988). Interleukin-1 and its

relevance in patients treated with hemodialysis. Kidney Int Suppl 24: 21-6

REFERENCES

133

108. Miyasaka, N., Sato, K., Kitano, Y., Higaki, M., Nishioka, K., Ohta, K.

(1992). Aberrant cytokine production from tenosynovium in dialysis associated

amyloidosis. Ann Rheum Dis 51 (6): 797-802

109. Baz, M., Durand, C., Ragon. A., Jaber, K., Andrieu, D., Merzouk, T.,

Purgus, R., Olmer, M., Reynier, J. P., Berland, Y. (1991). Using ultrapure water

in hemodialysis delays carpal tunnel syndrome. Int J Artif Organs 14 (11): 681-5

110. Schwalbe, S., Holzhauer, M., Schaeffer, J., Galanski, M., Koch, K. M.,

Floege, J. (1997). Beta 2-microglobulin associated amyloidosis: a vanishing

complication of long-term hemodialysis. Kidney Int 52 (4): 1077-83

111. Daneshian, M., Guenther, A., Wendel, A., Hartung, T., von Aulock, S.

(2006). In vitro pyrogen test for toxic or immunomodulatory drugs

in print

112. Prausnitz, C. (1936). Investigations on respiratory dust disease in

operatives in the cotton industry. Medical Research Council, Special report

series 1936; 212: 1-73. His Majesty´s Stationery Office, London

113. Neal, P. A., Schneiter, R., Caminita, B. H. (1942). Report on acute illness

among rural mattress makers using low grade, stained cotton. JAMA 119: 1074-

1082

114. Cinkotai, F. F., Lockwood, M. G., Rylander, R. (1977). Airborne micro-

organisms and prevalence of byssinotic symptoms in cotton mills.

Am Ind Hyg Assoc J 38: 554-9

115. Donham, K. J., Zavala, D. C., Merchant, J. A. (1984). Acute effects of the

work environment on pulmonary functions of swine confinement workers.

Am J Ind Med 5: 367-75

REFERENCES

134

116. Donham, K.J., Zavala, D. C., Merchant, J. A. (1984). Respiratory

symptoms and lung function among workers in swine confinement buildings: a

cross-sectional epidemiological study. Arch Environ Health 39: 96-101

117. Thelin, A., Tegler, O., Rylander, R. (1984). Lung reactions during poultry

handling related to dust and bacterial endotoxin levels. Eur J Respir Dis. 65 266-

71

118. Zucker, B. A., Draz, A., Müller, W. (2000). Comparison of filtration and

impingement for sampling airborne endotoxin. J Aerosol Sci 31: 751-755

119. Kindinger, I., Daneshian, M., Baur, H., Gabrio, T., Hofmann, A., Fennrich,

S., von Aulock, S., Hartung, T. (2005). A new method to measure air-borne

pyrogens based on human whole blood cytokine response. J Immunol Methods

298 (1-2): 143-53

120. Rietschel, E. T., Kirikae, T., Schade, F. U., Ulmer, A. J., Holst, O., Brade,

H., Schmidt, G., Mamat, U., Grimmecke, H. D., Kusumoto, S. (1993). The

chemical structure of bacterial endotoxin in relation to bioactivity.

Immunobiology 187 (3-5):169-90

121. Emancipator, K., Csako, G., Elin, R. J. (1992). In vitro inactivation of

bacterial endotoxin by human lipoproteins and apolipoproteins. Infect Immun 60:

596-601

122. Paulsson, J., Michaelsen, P. (1984). The limulus amoebocyte lysate test

(LAL) assay for the detection of endotoxin in fat emulsions for total parenteral

nutrition (TPN). Acta Pathol Microbiol Immunol Scand 92: 177-179

123. Schindler, S., Rosenberg, U., Schlote, D., Panse, K., Kempe, A., Fennrich,

S., Hartung, T. (2006) Application of the InVitro Pyrogen Test (IPT) based on

cryopreserved human whole blood for lipidic parenterals. Pharmeuropa 2006, in

print

REFERENCES

135

124. Zimmermann, M., Busch, K., Kuhn, S., Zeppezauer, M. (1999). Endotoxin

adsorbent based on immobilized human serum albumin. Clin Chem Lab Med

37: 373-379

125. Russel, W. M. S. and Burch R. L. (1959). The principles of humane

experimental technique. Methuen, London

126. European Union (1986). EU-Directive 86/609/EEC. Official Journal of the

European Union L 358

127. Beutler, B. and Rietschel, E. T. (2003). Innate immune sensing and its

roots: the story of endotoxin. Nat Rev Immunol 3: 169-176

128. Mascoli, C. C. and Weary, M. E. (1979). Applications and advantages of

the Limulus amebocyte lysate (LAL) pyrogen test for parenteral injectable

products. Prog Clin Biol Res 29: 387-402

129. Mascoli, C. C. and Weary, M. E. (1979). Limulus amebocyte lysate (LAL)

test for detecting pyrogens in parenteral injectable products and medical

devices: advantages to manufacturers and regulatory officials. J Parenter

Drug Assoc 33: 81-95

130. Twohy, C. W., Duran, A. P., Munson, T. E. (1984). Endotoxin

contamination of parenteral drugs and radiopharmaceuticals as determined

by the limulus amebocyte lysate method. J Parenter Sci Technol 38: 190-201.

131. Roslansky, P. F. and Novitsky, T. J. (1991). Sensitivity of Limulus

amebocyte lysate (LAL) to LAL-reactive glucans. J Clin Microbiol 29: 2477-

2483

132. Dinarello, C. A. (1999). Cytokines as endogenous pyrogens.

J Infect Dis 179 Suppl 2: 294-304

REFERENCES

136

133. Dinarello, C. A., O'Connor, J. V., LoPreste, G., Swift, R. L. (1984)

Human leukocytic pyrogen test for detection of pyrogenic material in growth

hormone produced by recombinant Escherichia coli

J Clin Microbiol 20: 323-329

134. Duff, G. W. and Atkins, E. (1982). The detection of endotoxin by in vitro

production of endogenous pyrogen: comparison with limulus amebocyte

lysate gelation. J Immunol Methods 52: 323-331

135. Tsuchiya, S., Yamabe, M., Yamaguchi, Y., Kobayashi, Y., Konno, T.,

Tada, K. (1980). Establishment and characterization of a human acute

monocytic leukemia cell line (THP-1). Int J Cancer 26: 171-176.

136. Poole, S., Mistry, Y., Ball, C., Gaines Das, R. E., Opie, L. P., Tucker, G.,

Patel, M. (2003). A rapid 'one-plate' in vitro test for pyrogens.

J Immunol Methods 274: 209-220

137. Balls, M., Blaauboer, B., Brusick, D., Frazier, J., Lamb, D., Pemberton,

M., Reinhardt, C., Roberfroid, M., Rosenkranz, H., Schmid, B., Spielmann, H.,

Stammati, A. L., Walum, E. (1990). Report and recommendations of the

CAAT/ERGATT workshop on validation of toxicity test procedures.

Altern Lab Anim 18: 303-337

138. Balls, M., Blaauboer, B., Fentem, J. H., Bruner, L., Combes, R. D.,

Ekwall, B., Fielder, R. J., Guillouzo, A., Lewis, R. W., Lovell, D. P., Repetto,

G., Sladowski, D., Spielmann, H., Zucco, F. (1995). Practical aspects of the

validation of toxicity test procedures. The report and recommendations of

ECVAM workshop 5. Altern Lab Ani. 23: 129-147

139. Worth, A. P. and Balls M. (2002). The principles of validation and the

ECVAM validation process. Altern Lab Anim 30 Suppl 2: 15-21

REFERENCES

137

140. Poole, S. and Mussett, M. V. (1989). The International Standard for

Endotoxin: evaluation in an international collaborative study.

J Biol Stand 17: 161-171

141. Council of Europe (2001). Biological Tests, 2.6.8. Pyrogens. In:

European Pharmacopoeia. Council of Europe, Strasbourg: 131-132

142. Peterbauer, A., Eperon, S., Jungi, T. W., Werner, E. R., Werner-

Felmayer, G. (2000). Interferon-gamma-primed monocytoid cell lines:

optimizing their use for in vitro detection of bacterial pyrogens.


143. Taktak, Y. S., Selkirk, S., Bristow, A. F., Carpenter, A., Ball, C., Rafferty,

B., Poole, S. (1991). Assay of pyrogens by interleukin-6 release from

monocytic cell lines. J Pharm Pharmacol 43: 578-582

144. Schins, R. P., van Hartingsveldt, B., Borm, P. J. (1996). Ex vivo cytokine

release from whole blood. A routine method for health effect screening

Exp Toxicol Pathol 48: 494-496

145. Clopper, C. J., and Pearson, E. S. (1934). The use of confidence or

fiducial limits illustrated in the case of the binomial. Biometrika 26: 404-413

146. Council of Europe (2001). Biological Tests, 2.6.14. Bacterial endotoxins.

In: European Pharmacopoeia. Council of Europe, Strasbourg: 140-147

147. Nakagawa, Y., Maeda, H., Murai, T. (2002). Evaluation of the in vitro

pyrogen test based on proinflammatory cytokine release from human

monocytes: comparison with a human whole blood culture test system and

with the rabbit pyrogen test. Clin Diagn Lab Immunol 9: 588

REFERENCES

138

148. Eperon, S., De Groote, D., Werner-Felmayer, G., Jungi, T. W. (1997)

Human monocytoid cell lines as indicators of endotoxin: comparison with

rabbit pyrogen and Limulus amoebocyte lysate assay.


149. Harada, T., Misaki, A., Saito, H. (1968)

Curdlan: a bacterial gel-forming beta-1,3-glucan.

Arch Biochem Biophys 124: 292-298

150. Poole, S., Thorpe, R., Meager, A., Hubbard, A. R., Gearing, A. J. (1988)

Detection of pyrogen by cytokine release. Lancet 8577: 130

151. Goldman, J.M., Th'ng, K.H., Park, D.S., Spiers, A.S., Lowenthal, R.M.

and Ruutu, T. (1978). Collection, cryopreservation and subsequent viability of

haemopoietic stem cells intended for treatment of chronic granulocytic

leukaemia in blast-cell transformation. Br J Haematol 40: 185-95

152. Cavins, J. A., Djerassi, I., Roy, A. J. and Klein, E. (1965). Preservation of

viable human granulocytes at low temperature in dimethylsulfoxide.

Cryobiology 2: 129-33

153. Knorpp, C. T., Merchant, W. R., Gikas, P. W., Spencer, H. H. and

Thompson, N. W. (1967). Hydroxyethyl starch: extracellular cryophylactic

agent for erythrocytes. Science 157: 1312-1313

154. Handin, R. I. and Valeri, C. R. (1972). Improved viability of previously

frozen platelets. Blood 40: 509-513

155. Frim, J. and Mazur, P. (1983). Interactions of cooling rate, warming rate,

glycerol concentration, and dilution procedure on the viability of frozen-

thawed human granulocytes. Cryobiology 20: 657-76

REFERENCES

139

156. Hermann, C., von Aulock, S., Graf, K. and Hartung, T. (2003)

A model of human whole blood lymphokine release for in vitro and ex vivo

use. J Immunol Methods 275: 69-79

157. von Aulock, S., Hermann, C. and Hartung, T. (2003). Determination of

the eicosanoid response to inflammatory stimuli in whole blood and its

pharmacological modulation ex vivo. J Immunol Methods 277: 53-63

158. Hartung, T., Docke, W.D., Gantner, F., Krieger, G., Sauer, A., Stevens,

P., Volk, H.D. and Wendel, A. (1995). Effect of granulocyte colony-stimulating

factor treatment on ex vivo blood cytokine response in human volunteers.

Blood 85: 2482-9

159. Hartung, T., Doecke, W.D., Bundschuh, D., Foote, M.A., Gantner, F.,

Hermann, C., Lenz, A., Milwee, S., Rich, B., Simon, B., Volk, H.D., von

Aulock, S. and Wendel, A. (1999). Effect of filgrastim treatment on

inflammatory cytokines and lymphocyte functions.

Clin Pharmacol Ther 66: 415-24

160. von Aulock, S., Boneberg, E.M. and Hartung, T. (2000).

Intermittent G-CSF (filgrastim) treatment cannot induce lymphocytosis in

volunteers Clin Pharmacol Ther 68: 104

161. Fennrich, S., Wendel, A. and Hartung, T. (1999). New Applications of the

Human Whole Blood Pyrogen Assay (PyroCheck). Altex 16: 146-149

162. Schumann, R.R. (1992)

Function of lipopolysaccharide (LPS)-binding protein (LBP) and CD14, the

receptor for LPS/LBP complexes: a short review

Res Immunol 143: 11-5

163. Fenton, M.J. and Golenbock, D.T. (1998). LPS-binding proteins and

receptors. J Leukoc Biol 64: 25-32

REFERENCES

140

164. Blackman, M. A., and Woodland, D.L. (1995). In vivo effects of

superantigens. Life Sci 57: 1717-35

165. Nakatani, T., Tsuchida, K., Sugimura, K., Yoshimura, R., and Takemoto,

Y. (2002). Response of peripheral blood mononuclear cells in hemodialyzed

patients against endotoxin and muramyldipeptide

Int J Mol Med 10: 469-72

166. Emancipator, K., Csako, G., Elin, R. J. (1992). In vitro inactivation of

bacterial endotoxin by human lipoproteins and apolipoproteins. Infect Immun 60:

596-601

167. Paulsson, J., Michaelsen, P. (1984). The limulus amoebocyte lysate test

(LAL) assay for the detection of endotoxin in fat emulsions for total parenteral

nutrition (TPN). Acta Pathol Microbiol Immunol Scand 92: 177-179

168. Richardson, E. C., Banerji, B., Seid, R. C., Levin, J., and Alving, C. R.

(1983). Interactions of lipid A and liposome-associated lipid A with Limulus

polyphemus amoebocytes. Infect Immun 39: 1385-91

169. Kreeftenberg, J. G., Loggen, H. G., van Ramshorst, J. D., and Beuvery,

E. C. (1977). The limulus amoebocyte lysate test micromethod and

application in the control of sera and vaccines. Dev Biol Stand 34: 15-20

170. Ochiai, M., Yamamoto, A., Kataoka, M., Toyoizumi, H., and Horiuchi, Y.

(2001). Interfering effect of Diphteria-Tetanus-Acellular Pertussis Combined

(DtaP) Vaccines on the Bacterial Endotoxin Test. Biologicals 29: 55-58

171. Hailman, E., Lichenstein, H. S., Wurfel, M. M., et al. (1994).

Lipopolysaccharide (LPS)-binding protein accelerates the binding of LPS to

CD14. J Exp Med 179: 269-77

REFERENCES

141

172. Poole S., Gaines Das RE. Towards a humane pyrogen test. Eur J

Parenteral Sci 2001; 6 (2): 63-64.

173. Dijkstra, J., Mellors, J. W., Ryan, J. L., and Szoka, F. (1987)

Modulation of the biological activity of bacterial endotoxin by incorporation

into liposomes. J Immunol 138: 2663-2670

174. Stitt, J. T. (1985). A study of the variability in the febrile responses of

rabbits to endogenous pyrogen. J Appl Physiol 59: 1254-7

175. Schutt, C., Schilling, T., Grunwald, U., Schonfeld, W., and Kruger, C.

(1992). Endotoxin-neutralizing capacity of soluble CD14.

Res Immunol 143: 71-8

176. Frey, E. A., Miller, D. S., Jahr, T. G., Sundan, A., Bazil, V., Espevik, T.,

Finlay, B. B., and Wright, S. D. (1992). Soluble CD14 participates in the

response of cell to lipopolysaccharide. J Exp Med 176: 1665-1671

177. Kirkland, T. N., Finley, F., Leturcq, D., Moriarty, A., Lee, J.-D., Ulevitch,

R. D., and Tobias, P. S. (1993). Analysis of lipopolysaccharide binding by

CD14. J Biol Chem 268: 24818-24823

178. Beutler, B., K. Hoebe, X. Du, and R. J. Ulevitch. (2003). How we detect

microbes and respond to them: the Toll-like receptors and their transducers.

J Leukoc Biol 74:479

179. Hirschfeld, M., Weis J. H., Vogel, S. N., and Weis J. J. (2000). Cutting

Edge: repurification of lipopolysaccharide eliminates signalling through both

human and murine toll-like receptor 2. J Immunol 165: 618-622

REFERENCES

142

180. Poltorak, A., He, X., Smirnova, I., Liu, M. Y., Huffel, C. V., Du, X.,

Birdwell, D., Alejos, E., Silva, M., Galanos, C., Freudenberg, M., Ricciardi-

Castagnoli, P., Layton, B., and Beutler, B. (1998). Defective LPS-signaling in

C3H/HeJ and C57BL/10ScCr mice: mutations in the tlr4 gene.

Science 282: 2085-2088

181. Schwandner, R., R. Dziarski, H. Wesche, M. Rothe, and C. J.

Kirschning. (1999). Peptidoglycan- and lipoteichoic acid-induced cell

activation is mediated by toll-like receptor 2. J Biol Chem 274: 17406.

182. Takeuchi, O., Hoshino, K., Kawai, T., Sanjo, H., Takanda, H., Takeda,

K., and Akira, S. (1999). Differential roles of TLR2 and TLR4 in recognition of

gram-negative and gram- positive bacterial cell wall components.

Immunity 11: 443-51

183. Thornberry, N. A., Bull, H. G., Calaycay, J. R., Chapman, K. T., Howard,

A. D., Kostura, M. J., Miller, D. K., Molineaux, S. M., Weidner, J. R., et al.

(1992). A novel heterodimeric cysteine protease is required for interleukin-1β

processing in monocytes. Nature 356: 768-774

184. Cannon, J. G., Tompkins, R. G., Gelfand, J. A., Michie, H. R., Stanford,

G. G., van der Meer, H. W. M., Endres, S., Lonnemann, G., Corsetti, J.,

Chrenow, B., Wilmore, D. W., Wolff, S. M., Burke, J. F., and Dinarello, C. A.

(1990). Circulating interleukin-1 and tumor necrosis factor in septic shock and

experimental endotoxin fever. J Infect Dis 161: 79-84

185. Engel, A., Kern, W. V., Murdter, G., and Kern, P. (1994)

Kinetics and correlation with body temperature of circulating interleukin-6,

interleukin-8, tumor necrosis factor alpha and interleukin-1 beta in patients

with fever and neutropenia . Infection 22: 160-164

REFERENCES

143

186. van Zee, K. J., Coyle, S. M., Calvano, S. E., Oldenburg, H. S. A., Stiles,

D. M., Pribble, J., Catalano, M., Moldawer, L. L., and Lowry, S. F. (1995)

Influence of IL-1 receptor blockade on the human response to endotoxemia.

J Immunol 154: 1499-1507

187. Luheshi, G., Miller, A. J., Brouwer, S., Dascombe, M. J., Rothwell, N. J.,

and Hopkins, S. J. (1996). Interleukin-1 receptor antagonist inhibits endotoxin

fever and systemic interleukin-6 induction in the rat.

Am J Physiol 270: E91-5

188. Banks, W. A., Ortiz, L., Plotkin, S. R., and Kastin, A. J. (1991)

Human interleukin (IL) 1α, murine IL-1α and murine IL-1β are transported

from blood to brain in the mouse by a shared saturable mechanism.

J Pharmacol Exp Ther 259: 988-996

189. Banks, W. A., Kastin, A. J., and Gutierrez, E. G. (1994). Penetration of

interleukin-6 across the murine blood-brain barrier. Neurosci Lett 179: 53-56

190. Tewari, A., Buhles, W. C., Starnes, H. F. (1990). Preliminary report:

effects of interleukin-1 on platelet counts. Lancet 336(8717): 712-4

191. Ogilvie, A. C., Hack, C. E., Wagstaff, J. , et al.(1996). IL-1 beta does not

cause neutrophil degranulation but does lead to IL-6, IL-8, and nitrite/nitrate

release when used in patients with cancer. J Immunol 156: 389-94

192. Nemunaitis, J., Ross, M., Meisenberg, B., et al. (1994). Phase I study of

recombinant human interleukin-1 beta (rhIL-1 beta) in patients with bone

marrow failure. Bone Marrow Transplant 14: 583-8

193. Crown, J., Jakubowski, A., Gabrilove, J. (1993). Interleukin-1: biological

effects in human hematopoiesis. Leuk Lymphoma 9: 433-40

REFERENCES

144

194. Du Bois, M. J., Schellekens, P. T., De Wit, J. J., and Eijsvoogel, V. P.

(1976). In vitro reactivity of human lymphocytes after cryopreservation using a

programmed cooling device. Scand J Immunol Suppl 5: 17-22

195. Doebbler, G.F., and Rinfret, A. P. (1962)

The influence of protective compounds and cooling and warming conditions

on hemolysis of erythrocytes by freezing and thawing

Biochim Biophys Acta 58: 449-58

196. Lionetti, F. J., Hunt, S. M., Gore, J. M., et al. (1975)

Cryopreservation of human granulocytes. Cryobiology 12: 181-91

197. Lionetti, F. J., and Hunt, S. M. (1975). Cryopreservation of human red

cells in liquid nitrogen with hydroxyethyl starch. Cryobiology 12: 110-8

198. Mazur. P. (1963). Kinetics of water loss from cells at subzero

temperatures and the likelihood of intracellular freezing.

J Gen Physiol 47: 347-69

199. Mazur, P. (1965). Causes of injury in frozen and thawed cells.

Fed Proc 24:1 75-82

200. Mazur, P. (1977). The role of intracellular freezing in the death of cells

cooled at supraoptimal rates. Cryobiology 14: 251-72

Documents

Diss Schindler