!~----------------------------~~

Comparative Study of RabbitPyrogen Test and Human whole Blood Assayon Human Serum AlbuminIngo Spreitzer, Matthias Fischer, Katja Hartrsch. Ursel Liideritz-Piichel and Thomas MontagPaul-Ehrlich-Institut, D-Langen

SummaryA comparative study of rabbit pyrogen test and human wholeblood assay was performed on released preparations of humanserum albumin. In addition, the samples were spiked with 5IVlml (in whole blood 0.5 IVlml too) and 10 IU endotoxin/ml.The unspiked samples were negative in both assays..The human whole blood test resulted in the same level of secu-rity for the products as the rabbit pyrogen test did. Both, theborderline 5 Ill/kg and the 10 Ill/kg-Spike partially lead toresults of the rabbit test which would cause further testing withadditional animals. In contrast, the human whole blood assayresulted in a 100% detection for the 5 IVlml and 10 IVlml-Spike. We designed a study protocol for a minimised number oftest animals and were able to show the general usefulness of thehuman whole blood assay.

Zusammenfassung: Vergleichsstudie von Kaninchen-Pyrogen-test und humanem in vitro Pyrogentest am Beispiel von Albu-minpraparatenIn einer vergleichsstudie wurden Albuminpraparate verschieden-er Hersteller im Kaninchen-Pyrogentest und im humanen Voll-bluttest miteinander verglichen. Zusdtrlicb wurden die Produktemit 5 IU (Vollblut tusatzlicli 0.5 IU) brw. 10 IU LPSlml dotiert.Die undotierten Chargen waren samtlicli negativ.Der Vollbluttest ergab fur die getesteten Albuminpraparate diegleiche Sicherheit wie der Kaninchen-Pyrogentest. Sowohl beider grenzwertigen 5 Ill/leg als auch bei der 10 Ill/kg-Dotierungkam es im Kaninchen-Pyrogentest teilweise zu Ergebnissen,welche eine Wiederholungsprufung mit weiteren. Versuchstierennotwendig gemacht hdite. 1m Gegensatz zum Kaninchen-Pyrogentest lieferte der Vollbluttest eindeutige Ergebnisse for5 IU sowie 10 IUlm!' Mit dem auf einen miiglichst geringenTierverbrauch ausgelegten Versuch konnte die grundsiitzlicheEignung des Vollbluttests als sichere Alternative zumKaninchen-Pyrogentest bestiitigt werden.

Keywords: comparative study, human whole blood assay, rabbit pyrogen test, albumin

1 Introduction

The screening for pyrogenic contamina-tions (pyrogens) is a crucial step in thequality control of pharmaceuticals. Sinceits introduction in 1942 (United StatesPharmacopoeia), the rabbit pyrogen testbecame the golden standard of pyrogentesting and has established a high level ofsecurity in pharmaceutical products. Thesensitivity of man and rabbit for pyro-gens is comparable.After intravenous injection of the test

substance into a group of three rabbits,the individual changes in body tempera-tures are recorded over a defined periodof time. In general, the decision whethera substance passes, fails or may be re-tested is based on the total of temperaturerises in a group of animals. The rabbitpyrogen test detects any pyrogen which

ALTEX 19, SuppL 1102

causes a temperature rise (in rabbits).Rare pyrogenic reactions of patientsunder treatment with rabbit-tested phar-maceuticals and experimental data (e.g.with human immunoglobulin prepara-tions) indicate species-specific differ-ences between man and rabbit.Endotoxin of Gram-negative bacteria

is the most famous pyrogen. The Limu-Ius-Amebocyte-Lysate test (LAL) wasdeveloped and established as an in vitroalternative to the rabbit pyrogen test.The LAL is highly specific for endo-

toxin, but fails to detect non-endotoxinpyrogens. The specific detection of endo-toxin by LAL may be performed in aqualitative or quantitative manner, de-pending on the LAL-method chosen. Thedevelopment of quantitative LAL-assaysmade it possible to define and to controlendotoxin threshold-limits for pharma-

ceuticals (depending on the applicationroute (e.g. subcutaneous, intravenous,intrathecal.j). Although the LAL hasreplaced the rabbit pyrogen test in mostcases, it has to be pointed out that theLAL is not a pyrogen test. The LAL is anendotoxin test. The national and interna-tional Pharmacopoeias clearly distin-guish between pyrogen tests and endo-toxin tests. In comparison to the rabbitpyrogen test, the LAL-readout reveals noinformation on the biological impact of apositively tested substance.Many biologicals like blood products

interfere with the LAL, and still have tobe tested with the rabbit pyrogen test.Because most of these biologicals areimmunogenic, test animals can only beused once (in general, re-use of testrabbits after a defined safety period isin agreement with different pharma-

73

_S_PR_E_IT_Z_E_R_E_T_A_L_. i~-- __~~<

copeias). Therefore users of the rabbitpyrogen test are in a dilemma: due to thehigh quality in pharmaceutical productionpyrogenic contaminations are rare,nevertheless the obligatory rabbit pyro-gen test causes a high consumption ofanimals.The human whole blood assay was

developed as a real in vitro alternative tothe rabbit pyrogen test. The basic ideawas to mimic the fever reaction. Ingeneral, the detection of exogenous pyro-gens (e.g. endotoxin) by blood cellscauses them to release endogenous pyro-gens like IL-l/3, IL-6 and TNFa. Thesecytokines affect the thermal regulationcentre in the brain and increase the bodytemperature by changing its set point.The test principle is based on the detec-tion of IL-lB (ELISA), which was re-leased by human blood cells after contactwith pyrogens. Like the rabbit pyrogentest the human whole blood assay revealsthe biological impact of pyrogenic con-taminations. The predictability of thehuman whole blood assay is high. It is aphysiological test system withoutspecies-specific differences to man.Comparative studies between animal

experiments and new test methods arenecessary for the replacement of animalexperiments. The general usefulness ofthe human whole blood assay was shownin a first project supported by the BMBF(German ministry of education and re-search: "Evaluation and pre-validation ofa whole blood model for the replacementof the rabbit pyrogen test", 1.7.97 -30.6.00). The study on albumin presen-ted here was part of this first project. Ear-lier data on different materials werealready published inALTEX 15 SuppI. 98(Fischer, M. et aI., 1998; Hartung, T. etal., 1998). Meanwhile, validation of thehuman whole blood assay is performedwithin a follow-up BMBF-project ("Re-placement of the rabbit pyrogen test by awhole blood assay", 1.10.00 - 30.9.03),by testing product groups like factorVIII. In parallel, an EU-project, ("Com-parison and validation of nivel pyrogentests based on the human fever reaction",1.1.00 - 31.12.02) was started in whichthe human whole blood assay as well asother alternative in vitro pyrogen tests arecompared in a Europe wide double-blinded study.

74

2 Animals and methods

29 batches of human albumin prepara-tions (5%, 20%, 25%; with one excep-tion three batches of each product; allavailable on the German market) of fivecompanies were tested in parallel in therabbit pyrogen test and in the humanwhole blood assay. The pure productswere tested in the PEI as routine batchrelease. The tests with spiked albuminpreparations were done as officiallyapproved animal experiment. Chinchillabastards (Charles River, Kisslegg, SPF-animals) were used for the rabbit pyro-gen test. The albumin preparations wereadditionally spiked with 5 IU and 10IU/ml endotoxin (Control Standardendotoxin E. coli 0113, Pyroquant,Walldorf; 1 IU = 100 pg). The spikedpreparations were applied with 1 ml/kgbody weight (resulting in doses of 5 IUand 10 IU/kg). The 5 IU/kg dose (lmlcontaining 5 IU per kg) is at the detectionlimit of the most sensitive rabbits (de-fined as dose at which 50% of animalsshow a defined rise in temperature).Depending on the preparation (and itsmonographs in the different pharma-copoeias) a maximal volume of 10 ml/kgbody weight can be applied. Taken intoaccount the detection limit of 5 IU/kg,under optimal test conditions (injectionof 10 ml test solution per kg) the mostsensitive rabbits can detect 0.5 IU endo-toxin/ml (0.5 IU per ml in 10 ml resultsin a dose of 5 IU/kg). The detection limitdepends on the endotoxin used and therabbit species. For the whole bloodassay, the 5 IU/ml-samples were addi-tionally pre-diluted 1:10 to achieve therabbit-test limit concentration of 0.5IU/mI. Therefore, 0 IU/kg (batchrelease), 5 IU/kg and 10 IU/kg were test-ed in the rabbit pyrogen test, whereas 0IU/ml, 0.5 IU/ml, 5 IU/ml and 10 IU/mlwere tested in the human whole bloodassay. The human whole blood assay wasperformed as described previously(Hartung, T. and Wendel, A., 1995).

100 III heparinised blood of twohealiliy donors (two exceptions: one ex-periment one donor, one experiment 5donors) was added to 1000 ul clinicalsaline and 100 III sample, and incubatedat 37°C for 20 h. After incubation, sam-ples were mixed and centrifuged, IL-IB

in the supernatant was detected byELISA. The test kit used was the Py-recheck-Kit (Milenia Biotech), which isnow distributed in a modified version byCharles River Endosafe (Kisslegg, Ger-many) as "in vitro pyrogen test (lPT)".The test result was only evaluable if theblood of the given donor was capable todetect the 50 pg/ml endotoxin-control(resembling the sensitivity of the mostsensitive rabbits). Samples were classi-fied "positive" if the resulting OD in theELISA exceeded the individual cut-off ofthe given donor.

3 Results

All pure albumin preparations were neg-ative in both assays (Tab. 1), resemblingthe high quality of pharmaceuticalproduction. The spike of 5 IU/kg wasclearly found by the rabbit pyrogen testin five batches, 23 batches resulted in atemperature rise which would haveallowed repetitive testing (each batchthree animals = 23 x 3), one batch wastested negative. The same samples(5 IU/ml; test volume 100 III resulting in0.5 IU per assay) were all tested positivein me human whole blood assay (29batches positive). For the 1:10 pre-dilu-tion of the 5 IU/ml spike (resulting in0.05 IU per assay, only performed in thewhole blood assay), 18 batches were pos-itive, II were negative. The dose of 10IU/kg was detected by the rabbit pyrogentest in 21 batches, 8 would have allowedrepetition (8 x 3 animals), no batch wastested negative. All batches spiked with10 IU/ml were found positive in thehuman whole blood assay.

4 Discussion

The rabbit pyrogen test has remarkablemerits in drug safety. Nevertheless, in mesense of animal protection it should bereplaced as far as possible by an appro-priate in vitro method which guaranteesthe same level of product safety. Newtherapeutic strategies like geneticallyengineered human cells enforce thedevelopment of new pyrogen tests, asthey can not be tested in the rabbit pyro-gen test or the LAL-test. The LAL-assay

ALTEX 19, Suppl. 1/02

i~ S_PR_E_I_TZ_E_R_E_T_A_L_.

~v'~~

Tab. 1: Comparison of rabbit pyrogen test and human whole blood assay

endotoxin rabbit pyrogen test human whole blood assay

rabbit human whole positive repetition negative positive negativeblood assay "> 1.15°C:>

IU per kg IU per assay "> 2.65°C 2.65°C :>1.15°C

0 0 0 0 29 0 29

0.05 not done not done not done 18 11

5 0.5 5 23 1 29 0

10 1 21 8 0 29 0

29 different batches of albumin were tested pure and spiked with 5 IU/ml and 10 IU/ml (the 5 IU/ml samples were additionally prediluted1:10 for the in vitro pryrogen test, to achieve a test solution at the sensitivity limit of the rabbit pyrogen test of 0.5 IU/ml).

is a highly specific endotoxin test, there-fore it is not a replacement for the rabbitpyrogen test. Within the last years it hasbecome obvious that there are severalnon-endotoxin pyrogens (Lipoteichoicacids, lipoproteins, bacterial DNA etc.),which further emphasise the need for anin vitro pyrogen test. Nevertheless, weemployed only LPS (Control StandardEndotoxin) in this study, as it is the onlyavailable and accepted pyrogenic stan-dard material. In the meantime, we haveaccess to purified pyrogens from Gram-positive bacteria and are evaluating thehuman whole blood assay with thesesubstances.In this study, the rabbit pyrogen test

and the human whole blood assay werecompared on albumin preparations. Theanimal experiments were inevitable, aswe had to compare the sensitivity of bothtest systems with defined samples. Awide comparison of marketed albuminpreparations like in this study wouldnot have been possible with the rarepyrogenic batches which are foundduring quality control. Albumin belongsto those substances which still are testedin the rabbit pyrogen test. Humanalbumin is immunogenic, repetitive testshave to be done with new animals, whichcauses further animal consumption.Comparing the two test systems, one

has to consider the volume to injectaccording to the regulations. The endo-

ALTEX 19, Supp!. 1/02

toxin concentrations used were 5 IU/mland 10 IUlml, respectively. The rabbitsreceived 1 ml per kg body weight, in theblood assay 100 ul each were applied.According to the European Pharma-copoeia, low concentrated therapeuticalalbumin (3.5 - 5%) has to be applied in avolume of 10 ml per kg body weight. Asa consequence, in these cases for exam-ple 5 IU endotoxin per ml (concentrationin the drug) in the rabbit pyrogen testhave to be compared with 0.5 IU endo-toxin per ml (5 IU in 10 ml = 0.5 IU perml) in the human whole blood assay.Testing albumin spiked with 0.5 IUendotoxin per ml in the human wholeblood assay (0.05 IU per assay) most ofour experiments were positive but wesaw uncertain and negative results too.Therefore, the sensitivity of the new testsystem is comparable to the rabbit'ssensitivity assuming even worst caseconditions. If necessary, the sensitivity ofthe test can be improved by increasingthe sample volume.On the other hand, most of the drugs to

be tested in the rabbit pyrogen test haveto be injected in smaller volumes. There-fore, in practice the new test wouldproduce more sensitive results in most ofthe cases. An interesting example repre-sents coagulation factor VIII, which weare currently testing in a similar study. Inorder to obtain a comparable sensitivityof both tests, one would have to inject 10

ml of factor VIII per kg body weight (ap-proximately 30 ml per rabbit, 90 ml pertest). Taking into account the marketprice of factor VIII (1000 IU per contain-er) the drug value would be around4,500 per batch. The same safety levelwould be achieved using 100 I-li in thehuman whole blood assay.

References

Fischer, M., Hartzsch, K., Hartung, T.and Montag-Lessing, T. (1998). FirstResults in the pre-evaluation of thehuman whole blood assay for pyrogensin biological pharmaceuticals. ALTEX15,10-13.

Hartung, T., Fennrich, S., Fischer, M. etal. (1998). Development and evalua-tion of a pyrogen test based on humanwhole blood. ALTEX 15,9-10.

Hartung, T. and Wendel, A. (1995).Detection of Pyrogens using humanwhole blood. ALTEX 12,70-75.

Correspondence toDr. Ingo SpreitzerPaul-Ehrlich- InstitutFG 1/3Paul-Ehrlich-Strasse 51-59D-63225 LangenTel.: + 49-6103-77 26 69E-mail: sprin@pei.de

75