DISTR LIMITEEWORLD HEALTH ORGANIZATION

ORGANISATION MONDIALE DE LA SANTE

BS/95.1796

English Only

EXPERT COMMITTEE ON BIOLOGICAL STANDARDIZATIONGeneva, 17 to 24 October 1995

INTERNATIONAL MULTI-CENTRE STUDY FOR THECALIBRATION OF A PROPOSED REFERENCE PREPARATION

FOR THROMBOPLASTIN, RABBIT, PLAIN

A.M.H.P van den Besselaar

Thrombosis and Haemostasis Research UnitDepartment of Haematology

Leiden University Hospital, Leiden, the Netherlands

CONTENTS OF REPORT

Page

International multi-centre study for the calibration ofa proposed reference preparation for thromboplastin,rabbit, plain 1

Accelerated degradation of the proposed internationalreference preparation for thromboplastin, rabbit,plain 19

Precision of PT measurements with the proposed secondIRP for thromboplastin, rabbit, plain 20

Long-term stability monitoring of the InternationalReference Preparation for thromboplastin 22

Ce document n'est pas destine a etre distribue au grand public et to usles droits y afferents sont reserves par l'Organisation mondiale de laSante (OMS). lIne peut etre commente, resume, cite. reproduit outraduit. partiellement ou en totalite, sans une autorisation prealableecrite de I'OMS. Aucune partie ne doit etre chargee dans un systeme derecherche documentaire ou diffusee sous QuelQue forme ou par QuelQuemoyen que ce soit -electronique, mecanique, ou autre. sans une

autorisation prealable ecrite de I'OMS.

This document is not issued to the general public. and all rightsare reserved by the VJorld Health Organization (WH 0). Thedocument may not be reviewed. abstracted. Quoted. reproduced ortranslated. in part or in whole. without the prior written permissionof WHO. No oart of this document may be stored in a retrievalsystem or transmitted in any form or by any means. electronic.mechanical or other. without the prior written permission ofWHO.

BS/95.1796Page 2

SUMMARY.

in

a multi-centre study.

comparison of thromboplastins. The mean ISI of reagent 1020 calibrated with

RBT/79 was 1.0 (rounded off value)

(

BS/95.1796Page 3

INTRODUCTION.

The World Health Organization (WHO) scheme for standardization of the pro-

thrombin time in oral anticoagulant control depends on the provision of

international reference preparations (IRP) for thromboplastin (1). Three

IRP are held by the WHO:

IRP bovine, combined (coded 08T/79)

IRP rabbit, plain (coded RBT/79)

IRP human, plain (coded BCT/253)

Each of these IRP were established by means of multi-center calibration

studies against the primary IRP of Thromboplastin, human combined, coded

67/40 (2, 3). IRP 67/40 was discontinued by the WHO in 1983.

The WHO Expert Committee on Biological standardization stated that the ca-

libration of a thromboplastin is, in general, more precise when comparisons

are made between similar preparations from the same species (1).

It was suggested that laboratories and manufacturers use the IRP of the

same precies for the calibration of their working reference preparations,

and the WHO requirements have been formulated, therefore, taking into con-

sideration only the calibration of similar preparations (human with human,

rabbit wit rabbit, and bovine with bovine) (1) .

Each of the three WHO IRP is characterized by a value of the International

Sensitivity Index (ISI), being the slope of relationship with the primary

IRP 67/40. In theory, calibration of a given thromboplastin with any IRP

should yield the same ~ value of the ISI, taking into account the ISI of

the IRP used.

As stated above, the Qrecis;on of the ISI may depend on the choice of IRP,

however.

BS/95.1796Page 4

The stocks of the IRP for thromboplastin rabbit, plain (RBT/79) are now

practically depleted. A proposed replacement IRP was prepared in ~lass-sea-

led ampoules and calibrated in an international multi-centre study In this

study, all three lRP were included to assess their relative ISI values and

to compare these with the original values (2,3).

Twenty laboratories were invited to participate in the study (see Appendix

1 )

DESIGN OF THE COLLABORATIVE STUDY.

The proposed replacement IRP for rabbit thromboplastin (reagent coded 1020)

and the current IRPs RBT/79, OBT/79 and BCT/253 were tested on freshly pre-

pared plasma specimens taken from anticoagulated patients and from norma

individuals The laboratories were requested to select patient plasmas with

prothrombin time ratios corresponding to an interval of 1.5 -4.5 INR. In

addition to the fresh plasma specimens, two lyophilized plasmas were tested

on each day The laboratories were requested to test the four thromboplast-

ins by the manual technique

It was considered that results obtained by 20 different expert laboratories

were sufficient to assess the effect of interlaboratory variation. To in-

clude the effect of inter-day variation, PT determinations were performed

in each laboratory on 10 different days. On each daYt each laboratory had

to include specimens obtained from two normal individuals and from six pa

tients.

Two different orders of testing were used On odd-numbered days, the order

of the thromboplastins was: OBT/79, BCT/253, RBT/79, reagent 1020. On even

the thromboplastins were tested in reserved order. Each plasma wasdays,

tested with the four thromboplastins, before the next plasma was tested.

BS/95.1796Page 5

In this way the effect of plasma instability on the relationship between

the thromboplastins was minimized. Plasmas were tested on each day, accor-

ding to the following order: lyophilized plasma R 82 A, lyophilized plasma

AK 2 BOA, fr'esh' normal specimen 1, fresh patient specimens 1 through 6,

fresh normal specimen 2.

STATISTICAL ~IETHODS.

The statisti~:al methodology used in this exercise was that developed for

the calibration of thromboplastins which was described previously (2, 3).

same strategy was adopted in the present study using orthogonal regres-

sion to estimate the slope of the linear relationship between the loga-

rithms of the prothrombin times with different thromboplastin reagents

Outliers were rejected on the basis of their orthogonal distance from the

fitted calibration line. As in the previous studies (2, 3), an observation

considered to be outlying if its orthogonal distance was more than 3

times the standard deviation of regression (calculated with al points in-

cluded).

The overall calibration was obtained by taking an unweighted average of the

separate orthogonal regression lines calculated for the different centres.

ISI values were calculated as the product of the ISI of the reference

thromboplastin and the slope of the orthogonal regression line. To detect

outliers among ISI values determined by individual laboratoriest the follo-

wing algorithm was used. The ISI value at the largest distance from the

mean of n laboratories (Xn)was omitted for calculation of a new mean (Xn-l)

and standard deviation (SOn-I).

If the omitted value was outside the interval

xn-1 .t Kn-1 .SDn-1

it was rejected as an outlier. The values of K were chosen with a 1% change

BS/95.1796Page 6

of inadvertently identifying a result as an outlier.Typical values of K

are: 2.95 (n = :19),2.98 (n = 18).

ISIPT Datient-

geometric mean PT of normals equation 1\

RESULTS.

Results were obtained from the 20 aboratories. All laboratories managed to

complete the study, but one (no. 18) performed the tests with the four rea-

gents on only four days. Nevertheless, the results of Lab.no.IS were inclu-

ded in the analysis

The geometric means of the clotting times of the fresh normal specimens are

shown in Table 1. These values are very similar to the corresponding values

obtained in the original calibration studies performed in 1979 and 1983 (2,

3).

ISI values of all preparations were determined with each of the three IRP.

Table 2 shows the ISI values obtained with 08T/79 as reference preparation.

There were two ISI outliers: one with BCT/253, and one with reagent 1020

(see also Table ~I) The between-laboratory coefficient of variation ranged

from 3.8% to 5. o~; including the outliers) or from 3.8% to 4.1% (excluding

the outliers}.

Table 4 shows the ISI values obtained with BCT/253 as reference preparati-

on The between-laboratory coefficient of variation ranged from 4.8% to

including the outliers or from 3.7% to 7.2% (excluding the out

1 iers}.