David holme, hazel peck analytical biochemistry prentice hall (1998)

1 General principles ofanalytical biochemistry • The selection of a valid method of analysis • The quality of data • The production of results Analytical biochemistry involves the use of

Analytical Biochemistry Third Edition

David J Holme and Hazel Peck

•An imprint ofPearson Education

Harlow, England London New York· Reading, Massachusetts San Francisco

Toronto· Don Mills, Ontario Sydney· Tokyo· Singapore· Hong Kong' Seoul

Taipei Cape Town· Madrid· Mexico City· Amsterdam Munich· Paris· Milan

Edinburgh Gate

Harlow

Essex CM20 2JE

England

and Associated Companies throughout the world

Visit us on the World Wide Web at:

http://www.pearsoneduc.com

© Addison Wesley Longman Limited 1998

The rights of David J Holme and Hazel Peck to be identified

as authors of this Work have been asserted by

them in accordance with the Copyright,

Designs and Patents Act 1988.

All rights reserved; no part of this publication may be

reproduced, stored in a retrieval system, or transmitted

in any form or by any means, electronic, mechanical,

photocopying, recording, or otherwise without either

the prior written permission of the Publishers or a

licence permitting restricted copying in the United

Kingdom issued by the Copyright Licensing Agency

Ltd., 90 Tottenham Court Road, London WIP OLP.

First published 1983

Second edition 1993

This edition 1998

ISBN 0 582 29438-X

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

Typeset by 55 in10/12Times

Produced by Pearson Education Asia (Pte) Ltd.

Printed in Singapore (COS)

10 9 8 7 6 5 4 3

05 04 03 02 01

1 General principles of analytical biochemistry 1

5 Radioisotopes 196

Analytical techniques - precipitation reactions 238

Separation and identification of carbohydrate mixtures 335

Nucleic acid composition and structure

Isolation and purification of nucleic acids

Methods of nucleic acid analysis

Vectors

DNA sequencing

Appendix: Numbering and classification of enzymes

Self test questions: answers

Index

421 424 425 429

443

444 449 456 465 468

474479

481

TO THE THIRD EDITION

The technology associated with biochemical analysis is rapidly changing andnew laboratory instruments are constantly being introduced However, with afew exceptions, the innovations are not based on new principles of analysis,but offer analytical benefits often through a 'mix and match' approach Forexample, modem HPLC instruments may use columns that combine variousfeatures to effect separation and offer a range of detector options; the bound-aries between electrophoresis and chromatography become blurred in suchtechniques as capillary chromatography; and mass spectrometry, previouslyonly associated with GLC, is now linked to a much wider variety of chro-matographic techniques These 'state of the art' instruments are normallymicroprocessor controlled, offer some degree of automation and are attrac-tively designed for ease of use

Against this background it is perhaps tempting for analysts to mate the importance of understanding the principles of the techniques they areusing Unless this is the case they will be unlikely to be able to select, optimizeand develop new methods, troubles hoot existing ones and be confident in thequality of their results With increasing importance being attributed to qualityassurance and laboratory accreditation, in addition to the fact that employersrequire their staff to work efficiently, an appreciation of fundamental princi-ples of analysis is vital We therefore make no apologies for again concentrat-ing on these in this third edition

underesti-We have deleted some sections that contained detailed accounts of niques that are rarely encountered in modem laboratories, while retaining ref-erence to the important classical methods that do provide the basis of currentmethodology New material has been added to bring some topics up to dateand these include increased coverage of laboratory quality, safety and accred-itation, use of kits, mass spectrometry, and capillary electrophoresis Manyother changes have been made, not least of which is a completely new layout

tech-of the typescript with boxed areas for emphasis We hope this will aid standing and make the book more 'user friendly' Two types of self test ques-tion are also included, which are designed to be simple indicators of an under-standing of the basic concepts of the section and not a comprehensive test ofknowledge of the topics We have decided not to include photographs of par-ticular instruments, as they are often not particularly informative and thedesigns change so rapidly We would like to thank Dr Susan Laird and DrRobert Smith for revising their chapters on nucleic acids and immunological

under-methods respectively We are also indebted to the many colleagues who haveshared their knowledge and expertise with us over the years and whose advicehas been invaluable.

Analytical biochemistry is an extensive subject and both the actual tent and the balance of coverage in such a book as this is open to debate Withthis in mind, our aim in each edition has been to give a clear account of theprinciples of the subject that will aid the understanding of a wide range of sci-entists who are either studying for a qualification or who are working in a lab-oratory, or perhaps both The reading lists at the end of each chapter suggestadditional texts for readers who require more details of specific topics.David1.Holme

con-Hazel Peck

April 1997

TO THE SECOND EDITION

Since the publication of the first edition in 1983, several specialist bookswhich cover a range of specific techniques in detail have been published.However, the ability to select an appropriate technique for a particular analyt-ical problem still remains fundamental and the first edition of this book evi-dently proved useful in this respect Thus the principal objective for this sec-ond edition remains unchanged

Much of the information has been updated for the second edition toreflect substantial changes in the subject The edition of a chapter on nucleicacids was considered essential and complements the original chapters on thechemical nature and methods of analysis of other important biological mole-cules We are indebted to Dr Susan Laird for compiling this chapter and also

to Mr Robert Smith for the major update on immunoassays in the logical methods chapter

immuno-We have maintained the same balance of information in the new chapterand therefore details of specific applications of techniques are not discussed,for example, DNA fingerprinting Where appropriate, we have included titles

of books which have an emphasis on applications in the further reading list atthe end of each chapter These lists are not intended to be fully comprehensive,nor are the chapters referenced as we consider this to be inappropriate for thelevel of potential readership

We have received many pleasing reports of the usefulness of the first tion in a range of analytical laboratories, in areas such as pharmaceuticals,biotechnology, agrochemicals, clinical biochemistry, molecular biology, etc.Our own experience and comments from colleagues in other universities havereinforced our initial purpose of writing a book for students on a range ofcourses that include the analytical aspects of biochemistry We are thereforedelighted that this softback edition is now available which will encouragewider access for student use

edi-Hazel Peck

David Holme

Sheffield Hallam University

July 1992

TO THE FIRST EDITION

The initial stimulus for writing this book arose out of difficulties experienced

in recommending a single suitable textbook for students on courses in whichthe analytical aspects of biochemistry were a major component Althoughthere are many books on analytical chemistry in general and clinical chemistry

in particular, many omit the biochemical aspects of analysis such as ogy and immunology while others do not cover the basic science of the sub-ject The objective was to bring together in one book those topics which weconsider to be essential to the subject of analytical biochemistry

enzymol-In the introductory section to each chapter, there is a brief explanation ofthe scientific basis of the topic and this is followed by a discussion of the ana-lytical methods which are relevant While it is not intended that it should be abook of 'recipes', technical details for many of the methods described aregiven This will help those readers with no practical experience to appreciatethe steps involved in the analysis while at the same time giving sufficient detailfor the method to be developed in practice It is intended that the book willprovide enough information to enable a student to select a technique or series

of techniques which would be appropriate for a particular analytical problemand to be able to develop a valid and reliable analytical method

The topics covered in this book fall into three main groups Analyticaltechniques such as spectroscopy, chromatography, etc are particularly impor-tant in analytical biochemistry as well as in analytical chemistry generally Theprinciples of each technique are explained and the scope and applications arediscussed There are chapters on enzymes, antibodies and radioisotopes, sub-stances which it may be necessary to detect and measure but which also can

be very useful in a variety of analytical methods Here again, the basic theory

is explained before discussing their applications in analytical biochemistry.Finally, there are four chapters which explain the chemical nature and meth-ods of analysis of the major groups of biologically important compounds,namely, carbohydrates, amino acids, proteins and lipids While it is appreciat-

ed that the range of compounds in this final section could be considerablyextended it has been deliberately restricted to those groups which we consid-

er to be of particular biochemical importance

At the end of each chapter, several books are listed for further reading

on the subject but it is suggested that the following books would be suitablefor further reading on the topic of biochemistry of amino acids, carbohydrates,proteins and lipids,

I.W Suttie, Introduction to biochemistry Holt, Rinehart and Winston, New

York, USA

H.R Mahler and E.H Cordes, Biological chemistry Harper and Row, New

York, USA

A.White, P Handler andE.L Smith, Principles ofbiochemistry McGraw-Hill

Book Co., New York, USA

We would like to thank Dr Rodney Pollitt for reading the draft text and

for his invaluable comments In addition, we would like to thank those

col-leagues who have helped in various ways and Mrs P Holme for typing themanuscript

DavidI. HolmeHazel Peck

Sheffield City Polytechnic

February 1982

About the boxes and

self test questions

The book contains margin notes and two types of boxes, which are designed

to enable the reader to identify certain types of information easily

The margin boxes highlight important points in the text with a shortstatement or definition to give some background to the topic or they refer toother sections in the book which give additional information on the topic.The procedure boxes give technical details of some procedure either toillustrate a technique or to provide technical details for readers who wish touse it

The self test questions are at the end of most sections in a box The fourquestions are designed to test the reader's understanding of the basic princi-ples of the topic without going into the details of the subject There are twotypes of questions:

1 Multiple choice question - any number or none of the alternatives may becorrect

2 Relationship analysis - consists of two statements joined by the wordBECAUSE Each statement should be considered separately and identified

as being either TRUE or FALSE.Ifboth statements are true, then the wholesentence should be considered to decide whether, overall, it is correct(YES) or not (NO), i.e whether the second statement provides a correctexplanation for the first statement Itshould be appreciated that one shortstatement will not provide a complete explanation but the overall sentencecan still be true

1 General principles of

analytical biochemistry

• The selection of a valid method of analysis

• The quality of data

• The production of results

Analytical biochemistry involves the use of laboratory methods to determinethe composition of biological samples and it has applications in many widelydiffering areas of biological science The information gained from an analysis

is usually presented as a laboratory report, which may simply say what stances are present (a qualitative report) or may specify the precise amount of

sub-a substsub-ance in the ssub-ample (sub-a qusub-antitsub-ative report)

Aqualitative reportwill often indicate whether a particular substance

or group of substances is present without commenting on the complete position of the sample In many cases the report will also specify the individ-ual members of that group of substances.Itmight, for instance, name only thedifferent carbohydrates present although the sample contained other sub-stances, e.g lipids and proteins Itis possible, when using some qualitativemethods, to compare the amount of substance in the sample with the amount

com-in a reference sample and to report the presence of either com-increased ordecreased quantities Such a report is said to be semi-quantitative.Chromatographic and electrophoretic methods often give results which can beinterpreted in this way

Aquantitative report will state the amount of a particular substancepresent in the sample and it is important that the units of measurement aremeaningful and appropriate in order to prevent subsequent misunderstandings.When reporting quantitative results it is desirable to indicate their reliability, afeature which can often be assessed statistically In practice it may not be nec-essary to present this information with each report but it should be readilyavailable for reference

In order to be able to choose a suitable analytical method it is essential to knowsomething about the chemical and physical properties of the test substance(Table 1.1) Because the relationship between the property and the amount ofsubstance is not always a simple one, some methods are only suitable for thedetection of the substance (qualitative) while others may be quantitative Forany method it is important to appreciate the nature of the relationship betweenTable 1.1 Physical basis of analytical methods

Physical properties thatcan be measured with somedegree of precision

Examples of properties used inthe quantitation of

ExtensiveMassVolumeMechanicalSpecific gravityViscositySurface tensionSpectral

AbsorptionEmissionFluorescenceTurbidityRotationElectricalConductivityCurrent/voltageHalf-cell potentialNuclear

Radioactivity

+

+ + + +

ic wavelengths have all been used quantitatively

The lead content of biological samples is usually very small, rendering gravimetricmethods impracticable, and methods have often relied upon the formation of colouredcomplexes with a variety of dyes More recently, the development of absorption spec-troscopy using vaporized samples has provided a sensitive quantitative method.Oxygen measurements using specific electrodes offer a level of sensitivity which isunobtainable using volumetric gas analysis

the measurement obtained and the amount of substance in the sample.

Most analytical methods involve several preparative steps before the final surement can be made and it is possible to produce a flow diagram represent-ing a generalized method of analysis (Table 1.2) Not all the steps may be nec-essary in any particular method and it may be possible to combine two or more

mea-by careful choice of instrumentation.Itis important when selecting a particularmethod to consider not only its analytical validity but also the cost of the analy-sis in terms of the instrumentation and reagents required and the time taken.Table 1.2 Generalized method of analysis

The major manipulative

Ifa compound does not show an easily detectable characteristic it may

be possible to modify it chemically to produce a compound which can be sured more easily In the early part of this century, this approach to analysis

mea-led to the development of many complex reagents designed to react cally with particular test substances Generally these reagents resulted in theformation of a colour which could be measured using visual comparators.Most of these reagents have been superseded by improved instrumental meth-ods but some very reliable ones still remain in use They were often namedafter the workers associated with their development, e.g Folin andCiocalteu's reagent, originally described in 1920 for the detection of pheno-lic compounds.

specifi-Interference occurs when other substances, as well as the test compound,are also detected, resulting in erroneously increased values Occasionallyinterference effects can result in suppression of the test reaction For anymethod it is important to be aware of substances that may cause interferenceand to know if any are likely to be present in the sample

Ifinterference is a major problem the sample must be partially purifiedbefore analysis This breaks the analysis into preparatory and quantitativestages In order to reduce the technical difficulties resulting from such two-stagemethods much work has gone into the development of analytical techniquessuch as gas and liquid chromatography in which separation and quantitationare effected sequentially

While it may be possible to devise quantitative methods of analysis for manybiochemical compounds, the only practical method of measurement for others

is through their physiological effects A bioassay involves the measurement of

a response of an organism or a target organ to the test compound and may be

conducted in vivo using live animals or in vitro using isolated organ or tissue

preparations Many bioassays are quantitative but those that give only a tive or negative result are said to be quantal in nature

posi-A satisfactory bioassay demands that the response of the animal to thesubstance can be measured in some fairly precise manner but it must beremembered that different animals respond in different ways to the same stim-ulus Bioassays must therefore be designed to take account of such variationsand replicate measurements using different animals must be made In allassays it is important that the external factors that may influence the responseare standardized as much as possible The age and weight of an animal mayaffect its response as may also the environmental conditions, route of injectionand many other factors

In the absence of absolute chemical identification it is often necessary toestablish that different samples contain the same physiologically active sub-stance This may be achieved by comparing the dose-response relationship forboth samples This involves measuring the response to varying amounts ofeach sample and demonstrating that the slope of the resulting relationship isthe same in both cases In such graphical or statistical methods it may be nec-essary to use the logarithm of the amount in order to produce a straight linerather than a curve.Itis often necessary to use such a technique to confirm thevalidity of using synthetic or purified preparations as standards in quantitativeassays

The use of cells from specific tissues grown in cultures, rather thanfreshly isolated, provides a technique of bioassay which reduces the need forthe use of animals with all the implications of costly resources and ethical con-flict Alternatively there is a wide range of cell lines available of different tis-sue origin (Table 1.3).

Table 1.3 Bioassays using cell lines

Parameter measuredCell growth

Cell growthInhibition of interleukin 5(IL5) stimulated growth

The measurement of the catalytic activity of an enzyme is also a say despite the fact that chemical methods may be used to measure the amount

bioas-of substrate bioas-of product Although the use bioas-of radioimmunoassays may enablethe determination of the molar concentration of an enzyme, the problem of therelationship between molar concentration and physiological effects stillremains

In recent years many methods for a wide variety of analytes have been oped by reagent or instrument manufacturers and are marketed as 'kits' Theserange from relatively simple colorimetric assays, which only require the addi-tion of the chemical solutions provided to the test sample, to more sophisti-cated procedures involving complex reagents such as labelled antibodies ornucleic acids The kits include all the necessary standards and assay compo-nents They may be designed to be used in a manual procedure or, more com-monly, on a particular automated instrument Full assay protocols are given,together with details of the composition of all the reagents, any associated haz-ard data and specified storage conditions

devel-The increased availability of kits has greatly reduced the necessity forindividual laboratories to develop their own methods Itis a requirement thatall kits are validated before they can be sold and that details of the expectedanalytical performance are included with the product Nevertheless, each lab-oratory is responsible for its own results and staff should ensure that the man-ufacturer's instructions are followed and that they are satisfied with all aspects

of any kit that they use This may necessitate checks on the quoted analyticalperformance being made

2 Which of the following could be said to be a quantitative result?(a) The test is positive.

(b) The sample contains more than 5 g of glucose

(c) The sample contains 0.3 g of glucose

(d) The sample contains both glucose and lactose

3 Weighing a sample is a qualitative methodBECAUSE

weighing a substance will not give the identity of the substance

4 Many hormones lend themselves to bioassaysBECAUSE

bioassays involve measuring the effect of the test substance on livingcells

All data, particularly numerical, are subject to error for a variety of reasons butbecause decisions will be made on the basis of analytical data, it is importantthat this error is quantified in some way

1.2.1 Variability in analytical data

The results of replicate analyses of the same sample will usually show somevariation about a mean value and if only one measurement is made, it will be

an approximation of the true value

Random errorVariation between replicate measurements may be due to a variety of causes,the most predictable being random error which occurs as a cumulative result

of a series of simple, indeterminate variations These are often due to ment design and use, e.g the frictional effects on a balance, variable volumesdelivered by auto-pipettes owing to wear, and operator decision when readingfluctuating signals Such error gives rise to results which, unless their meanvalue approaches zero, will show a normal distribution about the mean.Although random error cannot be avoided, it can be reduced by careful tech-nique and the use of good quality instruments

instru-The plotting of a histogram is a convenient way of representing the ation in such a set of replicate measurements All the values obtained are ini-tially divided into a convenient number of uniform groups, the range of eachgroup being known as the class interval In Figure 1.1 there are 11 suchgroups and the class interval is 1.0 unit The number of measurements fallingwithin a particular class interval is known as the frequency if) and is plotted

vari-as a rectangle in which the bvari-ase represents the particular clvari-ass interval and theheight represents the frequency of measurements falling within that interval.The class interval with the greatest frequency is known as the modal class andthe measurement occurring with the greatest frequency is known as the mode.The average of all measurements is known as the mean and, in theory, to

15 Class interval

Frequency

5 18 32 55 86 95 83533515 6

determine this value(11-), many replicates are required In practice, when thenumber of replicates is limited, the calculated mean (x) is an acceptableapproximation of the true value

The most acceptable way of expressing the variation that occurs betweenreplicate measurements is by calculating the standard deviation (s) of thedata:

s= J!,(x:x?

where x is an individual measurement and n is the number of individual

mea-surements An alternative formula which is more convenient for use with culators is:

cal-s= J!'x2

-~

n - IThe calculation of standard deviation requires a large number of repli-

cates For any number of replicates less than 30 the value for s is only anapproximate value and the function(n - I) is used in the equation rather than

(n).This function(n - I)is also known as the degrees of freedom(4J)ated with the mean and is important when tests of significance are used.Knowledge of the standard deviation permits a precise statement to bemade regarding the distribution of the replicate measurements about the meanvalue Table 1.4 lists the relationship between a standard deviation and the pro-

associ-Table 1.4 Normal distribution about a mean

Defined limits about the mean interms of the standard deviation(s)

68.2786.6495.4598.7699.7399.96 99.99

portion of measurements lying within defined ranges Two convenient limitsoften used are ± Is and ± 2s of the mean value Out of 100 replicate mea-surements, for instance, approximately 95 will fall within the range of ± 2softhe mean value This allows the probability of a single measurement lyingwithin specified limits of the mean value to be predicted For example, theprobability of a single measurement lying within a range of approximately ±

2s of the mean value would be 0.95 (95%).Ifa limited number of replicateswere done instead of a single measurement, a greater degree of confidencecould be placed in the resulting mean value This confidence can be expressed

as the standard error of the mean (SEM), in which the standard deviation isreduced by a factor of the square root of the number of replicates taken:

s

SEM=-,jn

There is therefore a considerable advantage in making a limited number

of replicate analyses rather than a single analysis but, in practice, it is sary to balance the improved confidence that can be placed in the data againstthe increased time and effort involved

neces-Systematic errorSystematic errors are peculiar to each particular method or system They areconstant in character and although they can be controlled to some extent, theycannot be assessed statistically A major effect of the introduction of system-atic error into an analytical method may be to shift the position of the mean

of a set of readings relative to the original mean.Itmay not obviously affect

the distribution of readings about the new mean and so the data would show

similar values for the standard deviation Such a method is said to show bias

towards either the positive (an increase in the mean) or the negative (a decrease

in the mean) depending upon the direction of displacement

Instrumental factors

Instability in instruments contributes to random error described earlier butsometimes features of their design or the fatigue or failure of componentsmay result in readings being consistently lower or greater than they should

be This may sometimes be seen as a gradual drift over a period of time.Such variations are said to be systematic in origin and will result in biasedresults Itis essential in order to minimize the danger of such systematicerror that great care is exercised in the choice and use of analytical instru-ments

Errors of method

The chemical or biological basis of an analytical method may not permit asimple, direct relationship between the reading and the concentration ofthe analyte and failure to appreciate the limitations and constraints of amethod can lead to significant systematic error Carbohydrates may bequantified, for instance, using a method based on their reducing propertiesbut results will tend to be higher than they should be if non-carbohydratereducing substances are also present in the samples

It is also possible for a perfectly valid analytical method to becomeless valid when used under different conditions For example, the poten-tiometric measurement of pH is temperature-dependent and the use of ref-erence and test solutions at different temperatures without any compensa-tion will result in values being consistently higher or lower than theyshould be

Analytical methods should be precise, accurate, sensitive and specific but,because of the reasons outlined earlier, all methods fail to meet these crite-ria fully.Itis important to assess every method for these qualities and theremust be consistency in the definition and use of these words

Precision

The precision, or reproducibility, of a method is the extent to which a ber of replicate measurements of a sample agree with one another and isaffected by the random error of the method It is measured as imprecision,which is expressed numerically in terms of the standard deviation of a largenumber of replicate determinations (i.e greater than 30), although for sim-plicity in the calculation shown in Procedure 1.1 only a limited number ofreplicates are used The value quoted for s is a measure of the scatter ofreplicate measurements about their mean value and must always be quotedrelative to that mean value

num-The significance of any value quoted for standard deviation is notimmediately apparent without reference to the mean value to which it relates

The coefficient of variation or relative standard deviation expresses thestandard deviation as a percentage of the mean value and provides a valuewhich gives an easier appreciation of the precision:

Coefficient of variation(V) = _s_ X 100%

mean

Itis difficult to appreciate what the statement that a standard deviation

of 1.43 g1- Iimplies but quoting it as a coefficient of variation of 10.3% nifies that the majority of the replicate results (68%) are scattered within arange of :::': 10.3% of the mean value Compared with quoting the values for

sig-standard deviation, the implication of coefficient of variation values for twoanalytical methods, for instance, of 5% and 10% are immediately obvious.While the value for coefficient of variation is a general statement aboutthe imprecision of a method, only the value for standard deviation can be used

in any statistical comparison of two methods The use of coefficient of tion assumes a constant relationship between standard deviation and the meanvalue and this is not always true (Table 1.5)

varia-Table 1.5 An example of standard deviation and

coefficient of variation for different mean values

The use of coefficient of variation(V) reveals that

maxi-mum precision in the example given is achieved at the

mid-range concentration values, a fact that would not be

so obvious if only the standard deviation(s) were quoted

Itmay be possible to demonstrate a high degree of precision in a set ofreplicate analyses done at the same time and in such a situation the withinbatch imprecision would be said to be good However, comparison of repli-cate samples analysed on different days or in different batches may showgreater variation and the between batch imprecision would be said to be poor

In practice this may more closely reflect the validity of the analytical data thanwould the within batch imprecision

A comparison of the imprecision of two methods may assist in thechoice of one for routine use Statistical comparison of values for the standarddeviation using the 'F'test (Procedure 1.2) may be used to compare not onlydifferent methods but also the results from different analysts or laboratories.Some caution has to be exercised in the interpretation of statistical data andparticularly in such tests of significance Although some statistical tests areoutlined in this book, anyone intending to use them is strongly recommended

to read an appropriate text on the subject

From a knowledge of the imprecision of a method it is possible to assess

the number of significant figures to quote in any numerical result There isalways the temptation to imply a high degree of precision by quoting numeri-cal data to too many decimal places On the other hand, the error due to'rounding off' decimals must not be allowed to impair the precision which isinherent in the method A zero at the end of a series of decimals is often omit-ted but it may be significant It is a convenient rule of thumb not to report anydata to significant figures less than a quarter of the standard deviation, e.g.

1.43 gl-l0.35 gl-l

by comparing the means of replicate analyses by the two methods using the't'

test.Anexample of such a comparison is given in Procedure 1.3 with the ment that only a limited number of replicates are used solely to simplify thecalculation

com-Some authors use the word 'trueness' instead of 'accuracy' to describethe closeness of the mean of many replicate analyses to the true value Thisallows the word 'accuracy' to carry a more general meaning which relates tothe accuracy or difference of a single result from the true value, as a conse-

quence of both the imprecision (random error) and any bias (systematic error)

of a method Throughout this book the word is used as initially defined above.The method under study is usually compared with an accepted referencemethod or, if one is not available, with a method which relies on an entirelydifferent principle In the latter case the two methods are unlikely to showexactly the same degree of bias and if they give very similar results it can beassumed that neither shows any significant bias

The formula for the 't' test described in Procedure 1.3 compares themean of replicate analyses of only one sample but it may be preferable to com-pare the accuracy over the analytical range of the method To do this a paired

't' test may be used in which samples with different concentrations areanalysed using both methods and the difference between each pair of results iscompared A simplified example is given in Procedure lA.

One criticism of such a paired '1'test is that for a wide range of trations the difference between the pairs is accorded equal significance regard-less of the size of the numerical value The difference between values of 8 and

concen-10, for instance, is more significant than, although numerically equal to, thedifference between 98 and 100

An additional approach to handling paired data is to assess the degree ofcorrelation between the pairs The data can be presented as a graph in whichone axis is used for the results obtained by one method and the other axis forthe results of the same samples obtained by the other method.Ifeach sampleanalysed gave an identical result by both methods then a characteristic graphwould result (Figure 1.2(a» The closeness of the fit between all the points and

the straight line can be assessed by the correlation coefficient(r).Perfect relation as illustrated in Figure 1.2(a) gives a correlation coefficient of I Acoefficient of 0 indicates no correlation between the data; values between 0and I indicate varying degrees of correlation In general a correlation coeffi-cient greater than 0.9 indicates fair to good correlation and together with anacceptable result for the paired 't' test would provide strong evidence for acommon degree of accuracy between the two methods The method of calcu-lating the correlation coefficient is illustrated in Procedure 1.5.

cor-Itis possible that two methods that differ significantly in their accuracymay give a good correlation of data but would fail by the 't'test Plotting thedata as a regression plot would show a divergence from the pattern of Figure'1.2(a) Many complex variations might occur but two fairly simple ones wouldshow characteristic features.Ifone method gave results that were consistentlyhigh or low by a fixed amount (owing to a lack of specificity for instance), agraph similar to Figure 1.2(b) would result, in which the intercept was not zerobut corresponded to the fixed error involved Similarly if one method showed

Figure 1.2 Regression plot Graph (a) represents the data obtained by analysing samples by two methods which show perfect correlation.

Method 2 in graph (b) shows a constant positive bias compared with Method 1.

In graph (c), Method 2 shows a negative bias which is proportional to the tion of the analyte.

concentra-a systemconcentra-atic error which wconcentra-as proportionconcentra-al to the sconcentra-ample concentrconcentra-ation in someway, the slope would be different (Figure 1.2(c» These values can be calcu-lated using the method of linear regression analysis (Procedure 1.6) Using thisstatistical method, the equation for the straight line is determined and the val-

ues for the slope and intercept calculated Ifthese differ from 1.0 and zerorespectively, the graph differs from the characteristic one of Figure 1.2(a) andthe two methods differ in their accuracy.

Sensitivity

The sensitivity of a method is defined as its ability to detect small amounts ofthe test substance Some confusion may arise from the ways in which sensi-tivity is measured Itcan be assessed by quoting the smallest amount of sub-stance that can be detected; for example, the smallest reading after zero thatcan consistently be detected and measured The slope of the calibration graph

is a conventional way of expressing sensitivity and is particularly useful whencomparing two methods.Itis essential that for such a comparison, the units ofboth axes are the same for each method While there may be a significant dif-ference between the mean values of replicate determinations of two sampleswith slightly different concentrations, there may not be a significant differencebetween single or even duplicate analyses of these two samples In such casesthe lack of precision is more significant than the sensitivity of the method,which cannot be better than the precision if only single or duplicate analysesare undertaken

1.2.3 Quality assurance in analytical biochemistry

In order to produce reliable results, all analytical methods should be carefullydesigned and their precision and accuracy must be determined The stability ofsamples should be investigated and their subsequent handling controlled in an

appropriate manner The attitude of the staff involved is of vital importance:they must be motivated to produce valid data and to take a pride in the quali-

ty of the final product All of these factors together with the scheme for itoring performance will be scrutinized if a laboratory seeks formal accredita-tion

mon-Quality control

Quality control refers to an internal scheme that will give a warning whenunforeseen factors cause a reduction in the analytical performance of themethod This allows an immediate decision to be made on whether the testresults are acceptable or must be rejected This is usually done by the analysis

of a control sample with each batch of tests

Control samples

A control sample is a sample for which the concentrations of the test analyte

is known and which is treated in an identical manner to the test samples It

should ideally be of a similar overall composition to the test samples in order

to show similar physical and analytical features For instance, if serum ples are being analysed for their glucose content, the control sample shouldalso be serum with a known concentration of glucose A control sample will

sam-be one of many aliquots of a larger sample, stored under suitable conditionsand for which the between batch mean and standard deviation of manyreplicates have been determined It may be prepared within the laboratory

or purchased from an external supplier Although values are often stated forcommercially available control samples, it is essential that the mean andstandard deviation are determined from replicate analyses within each par-ticular laboratory

Control samples should be analysed along with the test samples but theanalyst should not know which are the control samples Knowing the meanvalue for the control sample and the precision expected from the method, it ispossible to forecast limits within which a single control result should normal-

ly fall The basis of a quality control programme is the assumption that if a gle control result falls within these defined limits, the method is under controland the test results produced at the same time are valid Ifthe control valuefalls outside the defined limits it is likely that the test results are in error andmust be rejected

sin-Control charts

Itis often helpful to record the results of control samples in a visible mannernot only because of the greater impact of a visual display but also for the rel-ative ease with which it is possible to forecast trends A variety of styles ofquality control charts have been suggested but the most commonly used arethose known as Levey-Jennings or Shewart charts, which indicate the scatter

of the individual control results about the designated mean value (Procedure1.7)

Incorporated in the chart are control limits set at ± 2s and± 3s, which

approximate to the 95% and 99% confidence limits respectively Ifa controlresult falls outside the 95% limit there is only a maximum probability of 0.05(5%) that the result lies in a normal distribution about the accepted mean The

most obvious implication is that the result could be wrong Similar deductions,with an increased degree of confidence, can be made about the 99% controllimits The precise position of the control limits can be set according to thedemands of the situation and in some instances it may be necessary to set morestringent control levels::!: Is and ::!: 2s.

The narrower limits are usually known as the warning limits Failure to

meet these limits implies that the method must be investigated and any knownweakness, such as unstable reagents, temperature control, etc., should be rec-tified However, results obtained at the same time as the control result can still

be accepted Probably the first step in a case like this is to repeat the controlanalysis.Ifthe original result was a valid random point about the mean thenthe repeat result should be nearer to the mean value Ifthe repeat analysisshows no improvement or the original control result lay outside the wider con-

trollimits (known as action limits) then it must be assumed that all the results

are wrong The method must be investigated, the fault rectified and the sis of samples and controls repeated

analy-The chart can give additional information about any change in the racy of the method Itwould normally be expected that a series of controlresults would show scatter about the mean value.Ifthe points showed a ten-dency to lie to one side of the mean but still within the accepted range, thiswould be an indication that the method was showing a bias in one particulardirection

accu-An alternative visual method of presenting quality control results is

known as the Cusum plot (cumulative sum) This is particularly useful when

control samples are not available (possibly owing to their lack of stability) andalso when large numbers of analyses which give a constant mean value areundertaken The mean value for each large batch of analyses will vary in a nor-mal distribution about the 'mean of means' calculated from a large number ofbatches over a period of time The difference between the mean of a batch andthe 'mean of means' should therefore be zero and the cumulative sum of thesedifferences should also be zero A graph is plotted of the cumulative sum ofthe differences against the date of the analysis or the batch number (Procedure1.8) For quality control samples the cumulative sum of the individual resultcompared to the previously calculated mean value is plotted

In many cases a Cusum plot will not show the expected horizontal line

but rather a line with a small but constant slope owing to the value attributed

to the 'mean of means' being incorrect The plot is still acceptable and in suchcases a change in the slope will indicate a change from the expected value andthe possibility of error Some of the difficulties with the Cusum plot is thatvariations are most obvious retrospectively, little information can be gainedfrom a single point and errors are only apparent from several consecutivepoints Thus it is debatable whether this type of plot can be classed as truequality control

Quality assessment

As well as operating quality control programmes, laboratories undertaking a

similar range of analyses can cooperate in a group programme In such a group

scheme, the same control samples are analysed by each laboratory and theresults compared within the group This form of assessment is usually a retro-spective process enabling overall quality to be maintained or improved Groupschemes do not necessarily demand that the control samples are of known con-centration because even using unknown samples, comparisons of single or repli-cate analyses and of mean values and standard deviation (precision) for the

group canbemade Results are frequently published in a coded manner so thateach member of the group can identify their result and where they lie in thegroup profile but cannot identify results from other membersinthe group.This type of external scheme is often administered by a commercialorganization, which distributes the samples and the results to all participants.

Accreditation formally recognizes that a laboratory is competent to carry outits analytical service and is increasingly becoming accepted as a necessaryrequirement.Itis an overall assessment of the performance of a laboratory andcovers the quality of management and associated organizational procedurestogether with the quality of testing Accreditation is an extensive processrequiring a quality audit and review associated with a series of visits from theexternal accrediting body, to whom a fee is payable

Various bodies have been established such as NAMAS (NationalMeasurement Accreditation Service) and CPA (Clinical PathologyAccreditation), which operate in the UK Quality standards appropriate to awide range of organizational activities, e.g BS 5750, ISO 9000 and EN 45000series in Europe, together with the internationally recognized GLP (good lab-oratory practice) may form part of the accreditation requirement In the USA,CLIA'88 (Clinical Laboratory Improvement Amendments, 1988) togetherwith the Federal Drug Administration and the US Environmental ProtectionAgency are involved in accreditation of laboratories

Each accrediting body produces detailed documents that outline its all requirements and assessment process They could include the followingaspects: general requirements, organization and management, quality systems,quality audit and review, staff, equipment, measurement traceability and cali-bration, methods and procedures for tests, laboratory accommodation and envi-ronment, handling of test items, records, test reports, handling of complaintsand anomalies, sub-contracting of tests, outside support services and supplies.Several specific aspects of laboratory management which are essential inthe process of accreditation are discussed in the following sections These top-ics are also very important for laboratories not seeking formal accreditationbut concerned about the quality of their work, their credibility and the safety

over-of their employees

Health andsafetyMany of the chemicals and much of the equipment used in laboratories arepotentially hazardous Itis essential that these hazards are clearly identifiedand appropriate working procedures defined together with adequate training ofstaff and readily available facilities to deal with the effects of any possibleaccident These aspects of laboratory work are covered in the UK by COSHH(Control of Substances Hazardous to Health) Regulations and by OSHA(Occupational Safety and Health Administration) in the USA

Every activity in the laboratory should be assessed for the potential ards involved and all the relevant information should be presented in a stan-dard format known variously as procedure hazard forms or material safety data

haz-sheets (MSDS) There are several steps in producing such a hazard document(Procedure 1.9) The document should then be approved by the laboratorysafety officer or equivalent, who should satisfy him- or herself that all the nec-essary equipment, resources and training are in place for the procedure to beundertaken.

Information on hazards is available from various sources Chemicalmanufacturers produce hazard data sheets for their products and some of themajor companies produce comprehensive databases Each data sheet containsinformation on the physical description of the compound, stability, hazards,first aid measures, storage, transport and disposal requirements

Chemical hazards are classified under five main headings

Explosive

Ingeneral laboratories there are some compounds that are potentially sive, e.g picric acid.Itis important that such substances are stored under suit-able conditions, e.g under water, and that they are regularly inspected in order

explo-to maintain these conditions The use of such substances should be carefullycontrolled and only small amounts used

Flammable

Flammable liquids and solids are subdivided according to their flash pointsand many oxidizing substances can cause fire when in contact with com-bustible materials Storage, handling and disposal are obviously major features

to be considered when using such substances

Toxic

Substances are graded in toxicity and by the route, e.g inhalation, swallowing

or contact, the latter being particularly important in the design of the workingenvironment In some instances it is possible to specify exposure limits and insuch cases monitoring of the environment may be essential

Corrosive and irritant

The result of skin contact with corrosive substances is usually obvious but the

effects of irritants are less obvious and hence are potentially likely to be

treat-ed more casually Protective clothing is essential when using toxic or irritantsubstances

Radioactive

These substances are subject to very strict control and laboratories must beapproved to handle the different categories of radioactivity

Standard operating procedures (SOPs)

SOPs give written details of the protocol that must be followed for any ular procedure being undertaken They are not restricted to laboratory opera-tions but are applicable to a wide range of activities in an organization and assuch are linked with the overall quality programme Within a laboratory theyare a convenient way of documenting, in a standard, unambiguous format, theinformation that staff require to be able to carry out their duties in a safe andappropriate manner Evidence of up-to-date information of this nature is nec-essary for laboratories seeking accreditation

partic-SOPs include details of the procedures for collecting and handling thesamples; performing the analysis; analysing, storing and retrieving data; andpreparing reports They contain instructions on the use, maintenance, servic-ing, calibration and repair of equipment and instruments, including computers.The hazards associated with the overall protocol are stated and safe workingpractices are specified The quality assurance measures that must be compliedwith are included together with instructions on the steps to be taken if themethod does not perform to specification The SOP is given a serial numberand dated before being signed and stored for future reference A copy, in a suit-able protective cover, must be available at the work bench and the SOP serialnumber should be quoted on all records relating to its use

Much information on the mode of operation and verification of mance of laboratory instruments is often available from manufacturers or sup-pliers in a form that is suitable for incorporation into an SOP

perfor-Computerization

Computers were first used in laboratories to calculate results and generatereports, often from an individual instrument As automated analysers weredeveloped, so the level of computerization increased and computers now play

a major role in the modern laboratory They are associated with both the lytical and organizational aspects and the term Laboratory InformationManagement System (LIMS) is often used to describe this overall function.Such systems are available that link the various operations associated with theproduction of a validated test result, from the receipt of the sample to the elec-tronic transmission of the report to the initiator of the request, who may be at

ana-a site removed from the lana-aborana-atory Other uses include stock control, humana-anresource management and budgets

The successful introduction of a system that is appropriate for a ular laboratory is a lengthy process requiring widespread consultation Currentand future needs must be taken into account, as also must the requirements ofany external body which may specify the extent of computerization required

partic-for accreditation The introduction of any new computer facilities will sitate the training of staff and their complete familiarization with the system.

neces-As with all other laboratory instruments, SOPs must be written that includehazard assessments and procedures to be followed in the event of systems fail-ure Regular checks on performance must be carried out and full servicing andmaintenance records must be kept

Good laboratory practice

Good laboratory practice (GLP) is a set of procedures within which the all performance of a laboratory can be monitored Itis applicable to the orga-nization and functioning of any laboratory but it is particularly relevant to thepharmaceutical industry Compliance with GLP may be required for accredi-tation of a laboratory by an external regulating agency

over-GLP involves all aspects of the organization which is involved in ating an analytical result, from senior management to the bench workers Theessential features of GLP can be summarized as follows

gener-Staff. All staff must be adequately trained with designated ties and appropriate qualifications Full details of all staff must be keptfor ten years

responsibili-Equipment.This must be of an adequate standard and full records of allmaintenance and faults must be kept for ten years

Procedures.All methods and procedures must be in the form of an SOP

Data. After the analysis has been completed, all the details of themethod, equipment, SOP and the raw results must be stored for tenyears

The validity of a laboratory report is affected by additional factors as well asthose related to the analytical method used Of particular importance is the sam-ple itself and the manner in which it is collected and stored prior to analysis.The collection procedure should be designed to provide a representativesample of the system under investigation and not adversely affect the analyticalprocess The correct storage conditions for the sample are vital to preserve theintegrity of the biological components (Table 1.6); in some situations ensuringthat the cellular morphology is not significantly affected The optimal storageconditions depend on the nature of both the sample and the analysis The chem-ical and physical environment are both important and stability studies may indi-cate the need for the addition of chemicals such as anti-microbial agents, anti-coagulants, etc and storage at a specific temperature for a certain period oftime These conditions should be clearly stated in the laboratory protocol

In addition to these considerations, the report that is issued is also

affect-ed by the procaffect-edures that follow the actual analysis (post-analytical) includingcalculations and transcriptions The use of computers, particularly in associa-tion with bar-coded sample containers, has improved the procedures of ensur-ing correct sample identity, the production of work lists and the presentation

of results Instructions relating to all the non-analytical aspects should be ified in the SOP associated with the analytical method

spec-Table 1.6 Examples of storage conditions of biological samplesPossible changein the sample Examples of methods of preventionMicrobial degradation Addition of anti-microbial agent,

e.g sodium azide

Store at temperatures below -20 DC

Denaturation of enzymes Store in 50% glycerol at low temperatures

Store in liquid nitrogenLeakage of intracellular components Separate cells immediately

Store in isotonic mediumUsually do not freezeOxidation Add antioxidant, e.g 2-mercaptoethanol,

dithiothreitolStore in the darkStore under nitrogen or hydrogenEnzymic conversion of analyte Add enzyme inhibitor, e.g fluoride

Store at temperatures below -20°CCoagulation Add anticoagulant, e.g heparin,

ethylenediamine tetraacetic acid (EDTA)Gaseous loss Store under oil, e.g liquid paraffin

(a) Reject all the test results

(b) Accept all the test results

(c) Re-analyse the control

(d) Re-analyse the test samples

3 The imprecision of a method can be assessed by determining thestandard deviation of replicate analyses

BECAUSEprecision falls as random error increases

4 If a control sample in a quality control programme gives a value that isgreater than the mean value for all the control samples by more than 2

SD it suggests that errors have been introduced into the assayBECAUSE

in a normal distribution of replicate results, no more than approximately2.5% of the values should exceed the mean value by more than 2 SD