Endotoxin Detection in Pharmaceuticals and Medical Devices ... · Quantitative Chromogenic Limulus Amebocyte Lysate Assay ... In den letzten 15lahren hat der Limulus Amo- ... (EU)

d":"\ KURZMITTEILUNG---~--------------------------------------------

Endotoxin Detection in Pharmaceuticals andMedical Devices with Kinetic-QCL, a Kinetic-Quantitative Chromogenic Limulus AmebocyteLysate AssayRonald N. BerzofskyBioWhittaker, Inc., USA-Walkersville

ZusammenJassung

Die Beobachtung, daj3 Endotoxin in Extrakten vonLimulus Amobozyten ein Gel erzeugt, wurde zu einem invitro Test ausgebaut, einem kinetischen, quantitativenChromogen-LAL-Assay (Kinetic-QCL), mit dem inwdssrigen Losungen Endotoxine bestimmt werdenkonnen. In den letzten 15 lahren hat der Limulus Amo-bozyten Lysat Test (LAL) zur Bestimmung von Endotoxi-nen in Pharmaka und medizinischen Hilfsmitteln breiteinternationale Anerkennung gefunden. Sowohl diePharmakopoen der USA als auch Europas enthaltenBeschreibungen und Vorschriften rum Einsat; des LAL-Tests fur die Bestimmung bakterieller Endotoxine. Inbeiden Pharmakopoen wurde damit begonnen, dieerforderlichen Kaninchen-Pyrogen- Tests in Arzneimit-tel-Monographien durch Endotoxin-Grenzwerte zuersetzen, die mit dem LAL-Test bestimmt werden konnen.Der Einsat: des LAL-Tests zur Endotoxin-Kontrolle vonEndprodukten hat sich als unschdtzbar erwiesen. Eben-sogut kann auch der Endotoxin-Gehalt in Ausgangsstof-fen und Verpackungsmaterialien bestimmt werden. Beider Kontrolle wdhrend kritischer Produktionsabldufekiinnen die Ursachen von Endotoxin-Kontaminationengefunden und Entpyrogenisierungsprozesse durch dieQuantifizierung des Endotoxin-Abbaus validiert werden.Schnelligkeit, Reproduzierbarkeit, Empfindlichkeit undWirtschaftlichkeit, verbunden mit der entsprechendenAusriistung und Software, machen aus dem kinetischenQCL-Assay eine sowohl dem in vivo Kaninchen-Pyro-gentest als auch dem traditionellen Limulus-Geliertestuberlegene Methode, wenn es um die Kontrolle vonEndotoxinen in Pharmazeutika und medizinischenHilfsmitteln geht.

Summary

The observation that endotoxin caused gelation inextracts of Limulus amebocytes has been expanded to thedevelopment of an in vitro kinetic, quantitative chromo-genic LAL assay (Kinetic-QCL) for the detection ofendotoxin in aqueous fluids. Within the last 15years, theuse of Limulus amebocyte lysate to detect and controlthe presence of pyrogenic substances in pharmaceuticalsand medical devices has gained wide internationalacceptance. Both the United States and EuropeanPharmacopoeias contain descriptions of and require-ments for the LAL Bacterial Endotoxin Test. Bothpharmacopoeias have begun to remove the rabbitpyrogen test requirement in a majority of drug mono-graphs and have substituted endotoxin limits to bedetermined by LAL. The use of LAL has proved invalua-ble in controlling the level of endotoxin in finishedproduct. The endotoxin contribution of raw materialsand packaging material can be monitored as well. In-process testing at critical production steps can identifyadditional sources of endotoxin contamination, anddepyrogenation processes can be validated by quantita-ting the degradation of endotoxin challenges. The speed,reproducibility, sensitivity, and economics of the Kinetic-QCL assay, in conjunction with the appropriate equip-ment and software, over both the in vivo rabbit pyrogentest and the more traditional LAL gel-clot assay allow amore in-depth approach to the control of endotoxin inpharmaceuticals and medical devices.

Keywords: Limulus amebocyte lysat test, kinetic quanti-tative chromogenic LAL assay, endotoxins, pharma-copeias, pharmaceuticals, medical devices

The association of pyrogenic reac-tions with the intravenous injectionof fluids is over 200 years old. In1874, Panum postulated the exist-

ence of a heat-stable pyrogenic mate-rial present in the injected fluid asthe causative agent. Subsequently,several investigators, Billroth(1862), Burdon-Sanderson (1876),Jona and Centanni (1916), studying

the nature of this fever-inducingsubstance suggested that it was ofbacterial origin and exclusively asso-ciated with Gram-negative bacteria.Pfeiffer is reported to have proposedthe term "endotoxin" to describe the

1 Introduction

ALTEX 12,2/95 93

BERZOFSKY wr.:'\---------------------------h~·--

~v~~..membrane-associated toxin of Vibrio(cited in Pearson, 1985).The concern over the presence of

endotoxin in pharmaceuticals wasraised by the studies of Wechsel-mann (1911), Muller (1911), Samel-son (1913), and Bendix and Berg-man (1913). Hort and Penfold(1912a, 1912b, 1912c) expanded thefindings that a Gram-negative bacte-rial substance was responsible forpyrogenicity. These workers arecredited also with designing the firstrabbit pyrogen assay.Subsequently, confirmatory obser-

vations by Seibert (1923, 1925),Seibert and Mendel (1923), and Ra-demaker (1930, 1932) reinforced theconcept that bacterial contaminationwas present in all pyrogenic injecta-bles, and that through careful tech-nique to avoid bacterial contaminati-on, non-pyrogenic injectables couldbe produced. Their work lead to acollaborative study, which standardi-zed the rabbit pyrogen test, andallowed pyrogen testing to obtainpharmacopeial recognition in 1942.The rabbit pyrogen test remained

the exclusive official pyrogen assay,referenced in the U.S. Pharamcopeia(USP) for over 25 years, and is stillreferenced exclusively in many ofthe USP drug monographs. Howe-ver, events occurring in the late1950s and early 1960s were to leadto a significant in vitro alternative tothe rabbit pyrogen test for determi-ning pyrogenicity in pharmaceuticalsand medical devices.

2 Limulus Amebocyte LysateAssay

The basis for using Limulus amebo-cyte lysate in the detection of endo-toxin lies entirely in the coagulationreaction inherent in Limulus blood.In 1956, Frederic Bang, reported thatinfecting the horseshoe crab, Limu-Ius polyphemus with Gram-negativebacteria resulted in fatal intravascu-lar coagulation. By 1964, Levin andBang had demonstrated that extractof the circulating amebocytes wouldgel in the presence of Gram-negative

94

endotoxin. Endotoxin catalyzes theactivation of a proenzyme containedwithin the LAL (Young et aI., 1972).The initial rate of activation is deter-mined by the concentration of endo-toxin present. The activated enzymecleaves the clotting protein, alsopresent in the LAL, resulting in theformation of an insoluble clot.

2.1 Gel-Clot Assay

The most commonly employed LALmethod is the gel-clot assay. Thisassay utilizes the entire endotoxin-mediated cascade in addition to theclotting protein to product a gelati-nous clot after incubation with endo-toxin. Basically equal volumes ofsample in LAL (typically 0.1 mleach) are combined in a 10 x 75 mmglass tube. After an incubation peri-od of 60 minutes at 37°C, the tubesare inverted 180°. A positive result isindicated by a clot, which withstandsthe inversion. By titrating the lysatewith an endotoxin of known potency,the minimum concentration of endo-toxin required to yield a positive clotcan be determined. This minimumendotoxin concentration, or end-point, is referred to as the lysatesensitivity.The gel-clot assay can be used as a

purely qualitative limits test to ranksamples as either positive or negati-ve, i.e. greater than or less than thelysate sensitivity. However, by titra-ting positive samples one can obtaina semi-quantitative measure of theendotoxin concentration in un-knowns by multiplying the last posi-tive sample dilution by the lysatesensitivity. The gel-clot assay suffersfrom the disadvantage that they arebest semi-quantitative and requirepreparing multiple dilutions of sam-ples to determine a positive/negativeend-point.

2.2 Kinetic-QCL Assay

A truly quantitative assay has beendeveloped by substituting a syntheticchromogenic substrate for the endo-

toxin activated clotting enzyme inplace of the natural clotting protein.As with the gel-clot assay, theamount of active clotting enzymeproduced is proportional to theamount of endotoxin in the sample.However, instead of resulting in asimple yes/no, qualitative, assay; theuse of the chromogenic substrateresults in an assay which allows forthe endotoxin concentration Inunknowns to be quantitated.The combination of LAL and the

chromogenic substrate (Kinetic-QCLReagent) provides a better under-standing of the endotoxin mediatedreaction (Figure 1). Solutions contai-ning known amounts of endotoxinranging in concentration from 50 to0.005 EU/ml were combined with anequal volume of the Kinetic-QCLReagent, and incubated at 37°C whi-le the absorbance of the reaction at405 nm monitored. The graph illu-strates the time, on the X-axis, beforethe creation of the active enzyme asseen by the increase in absorbance,on the Y-axis, at different endotoxinconcentrations. The concentration ofendotoxin is expressed in EndotoxinUnits (EU) based on the currentFDA/USP Reference EndotoxinStandard, EC-S. With high concen-trations of endotoxin, i.e. 50 EU/ml,there is a short lag time followed bya rapid increase in absorbance and aneventual plateau when all the sub-strate has been consumed. As theendotoxin concentrations decrease,the corresponding lag time signifi-cantly increase and the rates of colorgeneration become slower.This family of curves is different

from what one normally sees whenperforming a kinetic enzyme analy-sis, where the reaction begins imme-diately and different concentrationsof analyte only cause different reac-tion rates. It is important to keep inmind, that unlike "classical" kinetics,in the kinetic-quantitative chromoge-nic LAL (Kinetic-QCL) assay theenzyme does not exist but is createdduring the reaction by the activationof endotoxin.In order to estimate the endotoxin

concentration in unknown samples,

ALTEX 12,2/95

d'~"\ BERZOFSKY--~~~-----------------------------~v~

it is necessary to develop a standardcurve which correlates a unique reac-tion parameter with a correspondingendotoxin concentration. It was deci-ded to focus on the time required forthe creation of the active enzyme.This time, designated the ReactionTime, was arbitrarily defined as thetime when the absorbance of thereaction mixture increased by 200mOD. The horizontal line in Figure 1represents this 200 mOD threshold.This value was chosen to allowextremely low concentrations of en-dotoxin to be detected as soon aspossible and still provide adequateseparation between this lowest endo-toxin concentration and the blank.There is an inverse relationship bet-ween Reaction Time and endotoxinconcentration. The larger concentra-tions of endotoxin have the shorterReaction Times and the smaller con-centrations of endotoxin have thelonger Reaction Times.A standard curve for Kinetic-QCL

assay can be constructed from thedata obtained from the kinetic analy-sis (Figure 2). The log endotoxinconcentration, on the X-axis, is plot-ted against the log Reaction Time, onthe Yvaxis. Prepared in this manner,the curve is linear and can quantitateendotoxin in unknowns over a con-centration range of 50 to 0.005 EUIml without the need to prepare multi-ple dilutions of the sample.Because of the unique method

chosen to perform the Kinetic-QCLassay and analyze the data, it wasnecessary to develop a Kinetic-QCl,specific software package combinedwith the appropriate instrumentation,the Kinetic-QCL Reader. The Kine-tic-QCL Reader is an incubatingmicroplate spectrophotometer capa-ble of repeatedly reading a dispos-able microplate over time whilemaintaining the reaction at 37°C.The Kinetic-QCL Reader is suppliedwith a version of the Kinetic-QCLsoftware on-board, or the Kinetic-QCL Reader can be control by anexternal PC and the Kinetic-Qf'LPC-based software. The Kinetic-QCL PC based software simplifiesdata entry and provides for greater

ALTEX 12.2/95

2000••••1a • A A A • A • t• • • • * * *

6' • • * *1500 • *0 • • **.s • • *Q) *o 1000 • *c • 0ItS • * 0.0 0•.. • * 00 .01. 0(/) * 0.0 500 • .01. *<l:

• .01. *200 mOD * •••0 ••••0 500 1000 1500 2000 2500 3000 3500

TIME (seconds)

I.Blank .0,005 EU 00,05 EU * 0,5 EU .01. 5 EU • 50 EU IFigure 1: Kinetic-QCL assay, absorbance vs timeSolutions containing endotoxin ranging in concentration from 50 to 0.005 EU/ml werecombined Kinetic-QCL reagent, and incubated at 3rC while the absorbance of thereaction was monitored at 405 nm.

storage capacity of assay protocols,sample information and archiveddata.Because the endotoxin concentrati-

on in an unknown is calculated bycomparing its corresponding Reac-tion Time to that of a series ofendotoxin standards, it is essentialthat the Reaction Times be reprodu-cible across the entire microplate. Ofconcern is that different portions ofthe microplate may heat-up to incu-bating temperature at different rates.Since the assay is monitored kineti-cally these differences in heatingmay result in differences in theobserved Reaction Time and causevariability in quantitating endotoxin.In using the Kinetic-QCL Reader,

the microplate, filled with just thestandards and unknowns, is preincu-bated in the Reader to 37°C prior tothe addition of the Kinetic-QCL Re-agent. In this way, independent ofthe heat-up rate, all positions of themicroplate are at the appropriateincubation temperature before theassay is initiated. In addition, thedesign of the Kinetic-QCL Readerallows for the addition of the Kine-tic-QCL Reagent while the micropla-te remains in the Reader, and ismaintained at 37°C. The data presen-ted in Figure I was generated by

assaying each endotoxin solution intriplicate. The coefficients of variati-on for the individual Reaction Timesobtained for the 50 EU/ml, 5 Eu/ml,0.5 EU/ml, 0.05 EU/ml, and 0.005EU/ml solutions were 0.88%, 1.23%,0.89%, 0.31%, and 1.22%, respec-tively. To assess the overall reprodu-cibility of the Kinetic-Qf'L assayperformed in this way, 100 ~l of asolution containing 0.5 EU/ml ofendotoxin was dispensed across theentire microplate, preincubated inthe Kinetic-QCL Reader for 10 mi-nutes prior to the addition of 100 IIIof the Kinetic-Qf.L Reagent, and theReaction Times of each of the 96wells across the microplate determi-ned. The overall coefficient of varia-tion for the 96 Reaction Times acrossthe entire plate was 0.82%. Compa-rable CV's were observed bothacross individual rows or down indi-vidual columns.

3 Regulatory Aspects

The United States Pharmacopeia(USP), the United States Food andDrug Administration (FDA), and theEuropean Pharmacopoeia (EP) havepublished guidelines on the use ofLAL for detection of endotoxin in

95

BERZOFSKY ~r.:"\-------------------------------~~~--~~"'~.

10000

Slope = -0.204(jJ Y int. = 3.008D Corr. = -0.996c0oQ)2-ill 1000EFc0 •f)cuillIT:

1000,005 0,05 0,5 5 50

Endotoxin Concentration

Figure 2: Kinetic-QCL assay, standard curveStandard curve for Kinetic-QCL assay. The log endotoxin concentration is plottedagainst the log reaction time. The curve can quantitate endotoxin in unknowns over aconcentration range of 50 to 0.005 EU/ml

pharmaceuticals and medical de-vices. Collectively, these documentsoutline the acceptable procedures for1) determining endotoxin limits,2) validating the LAL tests and3) developing a routine testing pro-

tocol.

3.1 Endotoxin Limits

In the United States, endotoxin pot-encies are expressed in EndotoxinUnits (EU) based on the currentFDAfUSP Reference EndotoxinStandard, EC-5. In Europe, endoto-xin potencies are expressed in Inter-national Units (IU) based on theWHO Endotoxin Standard, 84-650.For pharmaceuticals, endotoxin li-mits are based on the product'shuman dose. A typical parenteralsolution would have an endotoxinlimit of 5 units of endotoxin perkilogram body dose of the product.Alternative procedures exist for in-trathecals, radiopharmaceuticals andanti-neoplastics. Medical devices aretypically tested by extracting thedevice with 40 mls of an appropriateextraction solution. Each ml of theextraction fluid must contain lessthan 0.5 units of endotoxin.

96

3.2 Validation

Validation of the LAL assay is com-posed of two parts,1) qualifying the laboratory analyst

and2) qualifying the specific product.Analyst qualification involves con-firming the endotoxin sensitivity ofthe assay method using an endotoxinpreparation of known potency. Pro-duct qualification involves spikingthe product with known amounts ofendotoxin and subsequently detec-ting an acceptable percentage of thespike. This recovery experiment in-dicates that the product does notadversely interfere with the LALassay. Many pharmaceutical pro-ducts interfere with the performanceof LAL test. Guilfoyle and Munson(1982) reported that of 587 pharma-ceuticals tested at use concentration,78% interfered with the LAL assayin some manner. Interference may bedefined as the inability to recover anendotoxin spiked within acceptablelimits. Enhancement, on the otherhand, is characterized by the appa-rent recovery of more than theknown endotoxin spike. However,the authors demonstrated that all ofthese products could be pre-treated,most notably by dilution alone, and

successfully validated using the LALassay.The specific validation require-

ments for analyst qualification usingthe Kinetic-QCL assay involve con-firming the linearity of a standardcurve generated using an endotoxinpreparation of known potency. Thecoefficient of correlation, r, determi-ned by linear regression must fallbetween -0.980 and -1.000. Thesevalues are negative because of theinverse relationship between Reac-tion Time and endotoxin concentrati-on. The product validation require-ments using the Kinetic-QCL assayinvolve spiking a sample of theproduct with a known amount ofendotoxin, typically 0.5 EU/ml, and,after accounting for any backgroundendotoxin contained in the unspikedsample, recovering +/- 50% of thespike

ReferencesBang, F. (1956). A bacterial disease ofLimulus polyphemus. Bull. JohnsHopkins Hosp. 98, 325-35l.

Bendix, B. und Bergmann, 1. (1913).Uber das sogenannte Kochsalzfie-ber. Monatsschr. Kinderheilkd. 11,387-404.

Billroth, T. (1862). Beobachtungstudi-en uber das Wundfieber und acci-dentelle Wundkrankheiten. Arch.Klin. Chir. 2, 578 (cited in Pearson,1985).

Burdon-Sanderson, J. (1876). On theprocess of fever. Part III. Pyrexia.Practitioner, 417 (cited in Pearson,1985).

Centanni, E. (1921). La immunita sto-mogene (terza immunita). In Tratta-to di Immunologia, (149) Rome:Societa Endrice Libraria (cited inPearson, 1985).

Friberger, P., Knos, M. and Mellstam,L. (1982). A quantitative endotoxinassay utilizing LAL and a chromo-genic substrate. In S. W. Watson, J.Levin and T. 1. Novitsky (eds.),Endotoxins and their Detection withthe Limulus Amebocyte Lysate Test(195-206). New York: Alan R. Liss.

Guilfoyle, D. E. and Munson, T. E.(1982). Procedures for improvingdetection of endotoxin in products

ALTEX 12,2/95

wl-'\ BERZOFSKY--~~~-----------------------------~~

found incompatible for direct analy-sis with Limulus amebocyte lysate.In S. W. Watson, J. Levin and T. J.Novitsky (eds.), Endotoxins andtheir Detection with the LimulusLysate Test (79-90). New York:Alan R. Liss.

Hort, E. C. and Penfold, W. J. (l912a).The relation of salvarsan fever toother forms of infection fever. Proc.R. Soc. Med. (Part III) 5 (Path.Sect.),131-139.

Hort, E. C. and Penfold, W. J. (1912b).Microorganisms and their relation tofever. 1. Hyg. 12,361-390.

Hort, E. C. and Penfold, W. J. (1912c).A critical study of experimental fe-ver. Proc R. Soc. London, Ser. B 85,174-186.

Jona, 1. L. (1916). A contribution tothe experimental study of fever. 1.Hyg. 15, 169 (cited in Pearson,1985).

Levin, J. and Bang, F. B. (1964). Therole of endotoxin in the extracellularcoagulation of Limulus blood. Bull.Johns Hopkins Hosp. 115,265-274.

MUller, P. T. (1911). Uber den Bakte-riengehalt des in Apotheken erhaltli-chen destillierten Wassers. Munch.Med. Wochenschr. 58,2739-2740.

Panum, P. L. (1874). Das putride Gift,die Bakterien, die putride Infektionoder Intoxikation und die Septik-amie. Arch. Pathol. Anat. Pysiol.Klin. Med. 60, 301 (cited in Pearson,1985).

Pearson, F. C. III (1985). Pyrogens,Endotoxins, LAL Testing, and De-pyrogenation. In J. R. Robinson(Hrsg.), Advances in Parenteral Sci-ences 2. New York: Marcel Dekker.

Rademaker, L. (1930). The cause andelimination of reactions after intra-venous infusions. Ann. Surg. 92,195-201.

Rademaker, L. (1933). Reactions afterintravenous infusion. A further re-port on their elimination. Surg.Gynecol. Obstet. 56, 956-958.

Samelson, S. (1913). Uber das soge-nannte Kochsalzfieber. Monatsschr.Kinderheilkd. 11, 125-133.

Seibert, F. B. (1923). Fever-producingsubstance found in some distilldwaters, Am. 1. Physiol. 67, 90-104.

Seibert, F. B. (1925). The cause ofmany febrile reactions following in-travenous injections. Am. 1. Physiol.71,621-652.

ALTEX 12,2/95

Seibert, F. B. and Mendel, L. B.(1923). Protein fevers, with specialreference to casein. Am. 1. Physiol.67, 105-123.

Wechselmann, W. (1911). Neuere Er-fahrungen uber intravenose Salvar-san-Injektionen ohne Reaktionser-scheinungen. Munch. Med. Wochen-schr. 58, 1510-1511.

AddressRonald N. Berzofsky, Ph. D.BioWhittaker, Inc.8830 Biggs Ford RoadWalkersville, MD 21793-0127, USA

Richtigstellung:

Die Prasidenten der Schweizerischen Akademie der MedizinischenWissenschaften (SAMW; Prof. A. F. Muller) und der SchweizerischenAkademie der Naturwissenschaften (SANW, Prof. B. Hauck), weisenuns auf einen Irrtum bei der Besprechung der Ethischen Grundsatzeund Richtlinien (EGR) in ALTEX 1114, 1994, S. 220ff hin.

Da war zu lesen:

"Sie (die EGR) sind fi.ir Mitglieder der beiden Akademien freiwilligund unverbindlich. Die Industrie muf diese Empfehlungen nichtberucksichtigen. "

Richtig miisse es heiBen:

Die SAMW und die SANW haben die EGR bereits 1983 als Kodex furaIle in der Schweiz tatigen Wissenschaftler und Wissenschaftlerinnenund deren Mitarbeiter und Mitarbeiterinnen verbindlich erklart (1).Auch die chemische Industrie der Schweiz steht uneingeschrankthinter diesen Richtlinien und erachtet sie als Basis fur ihre Tatigkeit(2).

(1) Ethische Grundsatze und Richtlinien fur wissenschaftliche Tier-versuche (1994). Schweizerische Arstezeitung 75, Heft 33, 1255-1263.

(2) Tierversuche sind notwendig (1992). Eine Dokumentation desArbeitskreises Gesundheit und Forschung, Zi.irich, Seite 21.

97