Endogenous Interferences in Clinical Laboratory Tests

Preface vii 1 Accuracy Goals for Laboratory Tests 1 1.1 Accuracy and Precision 1 3 The Nature of Icteric Interference 21 3.1 Source Information on Bilirubin Interference 21 3.2 Allen C

Martin H Kroll, Christopher R McCudden

Endogenous Interferences in Clinical Laboratory Tests

Edited by

Oswald Sonntag and Mario Plebani

Volume 5

Martin H Kroll, Christopher R McCudden

Endogenous

Interferences in Clinical Laboratory Tests

Icteric, Lipemic and Turbid Samples

ISBN 978-3-11-026620-7

e-ISBN 978-3-11-026622-1

Library of Congress Cataloging-in-Publication Data

A CIP catalog record for this book has been applied for at the Library of Congress.

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publicatioin in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de.

© 2013 Walter de Gruyter GmbH, Berlin/Boston The publisher, together with the authors and editors, has taken great pains to ensure that all information presented in this work (programs, ap- plications, amounts, dosages, etc.) reflects the standard of knowledge and the time of publication Despite careful manuscript preparation and proof correction, errors can nevertheless occur Authors, editors and publisher disclaim all respondibility and for any errors or omissions or liability for the results obtained from use of the information, or parts thereof, contained in this work.

The citation of registered names, trade names, trade marks, etc in this work dies not imoly, even

in the absence of a specific statement, that such names are exempt from laws and regulations preotecting trade marks etc and therefore free for general use.

Typesetting: PTP-Berlin Protago-TEX-Production GmbH, Berlin

Printing and Binding: Hubert & Co GmbH & Co KG, Göttingen

Cover image: Comstock/Getty Images

♾ Printed on acid-free paper

Canada E-mail: cmccudde@uottawa.ca

The book has 28 figures and 19 tables.

To my wife Ellen and children Allison, Lauren and Jonathan.

Martin H Kroll

To my wife Liesje and children Katie and Sam

Christopher R McCudden

Medicine has evolved to a new level, where not only is it expected that physicians diagnose and treat patients efficaciously, but also that all patients are protected from harm Protecting patients from harm is part of patient safety and implies that the processes used in taking care of patients are free from error Medical care depends on obtaining useful information from laboratory tests Biochemical tests provide a great deal of information at relatively low cost and usually with rapid turnaround times The achievement of low cost and rapid turnaround times depends, to a large extent,

on the use of automation The dependence on automation subsequently results in a diminution of individualized attention to each individual sample To protect patient safety, laboratories need to establish detection systems to identify situations that could lead to biased results and rules to correct for the biased problems

A bias occurs when the result obtained during an assay deviates from the true value of the analyte in question A systematic bias occurs when there is an inher-ent problem in the measurement technique, as can occur with calibration errors and reagent deterioration All samples are affected by a systematic bias Interferences cause a non-systematic bias Here, the bias occurs only for the individual sample It

is important to identify common features that occur for interferences, and to identify ways of not only identifying the interferences, but also of quantifying their impact.For biochemical tests, especially those using serum or plasma as a matrix, a high concentration of bilirubin and turbidity can affect biochemical tests The most common cause of turbidity is lipemia The intent of this book is to provide a founda-tion for those running laboratories to identify, quantify and correct for the presence

of hyperbilirubinemia and lipemia (turbidity) Because most laboratories will need to perform these processes in an automated fashion, the people working in the labora-tory will need to design the appropriate procedures and to manage them

To establish the necessary foundation to effectively design processes to manage the interferences caused by bilirubin and lipemia (turbidity), this book contains several different perspectives The early chapters of the book provide information

on the physical and chemical mechanisms involved in interferences There is erable emphasis on the interaction of bilirubin and lipemic particles with light, the most common form of energy used to detect clinical biochemical species Additional chapters provide an emphasis on the clinical conditions where one might expect to encounter high concentrations of bilirubin or lipemia The latter half of the book dis-cusses means of detecting bilirubin or lipemia, as well as means to quantify their presence, to allow for appropriate reporting of results Finally, the last chapter dis-cusses means of characterizing and quantifying interferences in complex reactions,

consid-as frequently occurs with bilirubin, where the analyte may interact with the analyte or species directly related to the concentration of the analyte The intent of the book is to provide the laboratorian with sufficient background to deal with these interferences and protect patient safety.

Preface   vii

1 Accuracy Goals for Laboratory Tests   1

1.1 Accuracy and Precision   1

3 The Nature of Icteric Interference   21

3.1 Source Information on Bilirubin Interference   21

3.2 Allen Correction as a Source of Bilirubin Interference   21

3.3 Bilirubin Interference with Oximetry   22

3.3.1 Co-oximetry Interference   24

3.3.2 Pulse Oximetry   25

3.3.3 Cerebral Oximetry   26

3.3.4 Interference with Methemoglobin   27

3.4 Chemical Reactions as a Cause of Bilirubin Interference   28

3.4.1 Bilirubin Reaction with Creatinine Methods   29

3.4.2 Bilirubin Reactions with Peroxidase Methods   31

4 The Nature of Lipemic and Turbidity Interferences   35

4.1 Types of Interferences   35

4.2 Lipemia Causes Turbidity   36

4.3 Lipemia Interference Mechanisms   37

4.3.1 Light Scattering   37

4.3.2 Lipoprotein Particles   40

4.3.3 Intralipid® and Lipemia Simulation   42

4.3.4 Empirical Studies in Lipemia Turbidity   43

4.4 Lipoprotein Particles and Lipemia   44

5 Measurement of Interference   47

5.1 A Typical Commercial Study   47

5.2 Guidelines for Interference Studies   48

6 Origin of Icteric Samples   63

6.1 The Origin of Bilirubin   63

6.2 Bilirubin Toxicity   65

6.3 Transport of Bilirubin in the Blood   65

6.4 Uptake of Bilirubin by the Liver   66

6.5 Clinical Aspects of Bilirubin   66

7.2 Estimated Impacts Based on Interference Studies   75

7.3 Differential Interference with Different Bilirubin Isoforms   77

7.4 Non-spectrophotometric Icterus Interference   79

7.5 Resolving Icterus Interference   80

8.2.1 Frederickson Classification of Dyslipidemias   85

8.2.2 Obesity, Metabolic Syndrome and Diabetes   87

9.2 Estimated Impacts Based on Interference Studies   95

9.2.1 Interference by Light Scattering   95

9.2.2 Interference by Volume Displacement   96

9.2.3 Interference by Lipid Partitioning   99

10.2.2 Data Collection and Deconvolution of Non-Target Interferences   103

10.2.2.1 Subtraction Using Selected Wavelengths   104

10.2.2.2 Index Calculation Using Derivative Spectrometry   105

10.2.3 Establishing Indices and Defining Ranges   107

11.3 Reporting of Results in Icteric and Turbid Samples   115

11.4 Autoverification and Reporting Algorithms   116

11.5 Practical Issues: Education and Implementation   117

11.6 References   118

12 Analyte-dependent Interference   119

12.1 Complex Interferences   119

12.1.1 Model for Analyte-dependent Interference   120

12.1.2 Examples of Analyte-Dependent Interference   121

12.2 Statistical Testing for Significance   129

12.3 Failure to Design the Interference Study   133

12.4 Advantages of Using Multiple Regression Analysis   133

12.5 Concluding Remarks   135

12.6 References   137

Index   139

1 Accuracy Goals for Laboratory Tests

It is often said that laboratory tests account for 70 % of the objective information used

to diagnose and monitor patients Even though it is true that a good history and cal examination provide a significant amount of information, physicians and clini-cians, as well as nurses and other healthcare professionals, depend on laboratory test results to provide a final diagnosis, determine the degree of illness (the disease spectrum) and to monitor patients

physi-1.1 Accuracy and Precision

1.1.1 Definition

Accuracy of laboratory tests plays a vital role in health care, stipulating the quality and assuring patient safety [1] Typical process steps that infringe on the quality of laboratory results and thus patient safety include patient misidentification, failure of reagents, mismanagement, and failure to communicate [2] The accuracy of labora-tory tests is critically important for achieving and maintaining quality in delivering good medical care When the accuracy of laboratory tests is breached, the patient’s safety is put at risk Therefore, safe medical practice places a significant responsibil-ity on the laboratory to maintain a high accuracy of test results High accuracy of test results depends on good laboratory practice and includes such processes as Quality Control and Quality Assurance

Accuracy is a generalized term In the vernacular it may refer to how good the quality of the test result is from an analytical perspective Theoretically, one judges the quality of the result arising from the laboratory by comparing it to a perfect method, i.e., a method without defect, for which one has obtained a perfect specimen and the reproducibility is perfect The term Reproducibility is an ISO term [3] and refers to the closeness of the agreement between the results of measurements of the same meas-urand (analyte ) carried out under controlled conditions of measurement Essentially, the term Reproducibility refers to the precision of the measurement made for a par-ticular analyte The laboratory easily determines the precision of an analyte by deter-mining values for control materials On a day to day basis, the results obtained for any particular analyte for any control material will tend to a mean or average value The typical scatter around this value will demonstrate a normal (Gaussian) distribution, and thus have a definable standard deviation (SD) Because results for any particular analyte may take on any value across the reportable range of the analyte, a standard deviation determined at a particular value for the given quality control material may not be directly applicable To extend the precision measurements over the report-able range, one can use the ratio of the standard deviation to the mean of the quality control value and express it as a percentage This ratio is called the coefficient of

variation and for most tests in Chemistry it ranges between 1 % to 10 %, depending

on the analyte being measured and the magnitude of the value in the quality control material The coefficient of variation provides a measure of the precision Ideally, the clinician would like the precision, as measured by the coefficient of variation, to be

as low as possible

1.1.2 Imprecision as a Form of Error

Another way to think of precision is that it represents the closeness of agreement between independent measurements to each other Of course, in the laboratory, in order to put structure into the analytical process, the laboratory develops rules to stip-ulate the conditions for performing the assay Clinicians assume that all the values for laboratory tests that they receive have an extremely high precision They presume that if they took a specimen and had the laboratory run that sample today, then if they gave the laboratory the same specimen tomorrow, they would receive exactly the same result

The laboratory has to conduct itself with the knowledge that most clinicians are not expecting that there are going to be errors in results For this reason, laboratories, and the people who manage them, spend a lot of time and effort in controlling the processes to minimize the errors generated by running laboratory tests Precision, or

in actuality, imprecision represents a non-systematic error A non-systematic error is not part of the designed process of deriving a value from the collection and analysis

of the specimen Even though imprecision can be measured for the process, random error causes the deviations from the central value (central tendency) Random errors, though characteristic of the process, occur independently of one another Even though the measure of a random error allows one to predict how the population of specimens will behave, one cannot predict for each individual specimen exactly what will happen In order to be able to predict exactly what will happen to each individual specimen, one needs to examine the systematic errors

1.2 Types of Error

1.2.1 Bias

Systematic errors are inherent in the process Systematic errors are part of the process

of measurement, that is, they are the result of the way the sample and reagent are mixed, the amplification of the detection system, and most importantly, how values are assigned to the readings generated in the sensing process How values are assigned to the readings generated by the sensing process relates to the calibration

of the method The calibration of the method can be biased if the standards used for

1.2 Types of error       3

calibrating the method are not properly assigned Most methods in the clinical ratory use calibrators instead of standards Standards contain purified analyte dis-solved in pure water or solvent of determined composition Calibrators contain puri-fied analyte or measured analyte dissolved in the matrix of the naturally occurring constituents comprising the environment of the samples used for testing The matrix often is serum, plasma, or urine Any of these matrices contains all sorts of unidenti-fied and unspecified materials, typically protein, lipids, and organics Typically the laboratories making the calibrators will control the concentration of the electrolytes and some of the organics What makes a matrix material different from a standard

labo-is the analyte of interest plus other analytes are bound or complexed with naturally occurring constituents The naturally occurring constituents may alter the way the analytical method interacts with the analyte of interest, altering the signal from the sensor

Testing and assigning values in the laboratory are separated into three phases: the pre-analytic , analytic and post-analytic phase The pre-analytic phase includes preparing the patient to obtain the specimen, collecting the specimen into an appro-priate container (often with an anti-coagulant for blood), labeling and transporting the specimen to the laboratory and processing of the specimen to present it to the analyzer The post-analytical phase includes communicating the value for the test result to the clinician The analytical phase includes physically introducing the speci-men into a reaction vessel, chemically or biologically reacting the specimen with other materials, physical interaction with some form of energy to produce a signal , and translation of that signal into a number or value that can be communicated to the clinician

In the analytical phase, calibrators do not always translate the signal into exactly the same set of values that a purified standard would The mistranslation results in

a systematic error Systematic errors can be separated into two types of error, based

on how they relate to the underlying true concentration If the error, for example for creatinine, were high or low and did not depend on the value for creatinine over the

entire range of results, then the error is constant To illustrate the constant error, take

a value of 115 μmol/L of creatinine If there is a constant error or bias of 27 μmol/L, then the reported value would be 88  μmol/L instead of 115  μmol/L Further, if the true value of creatinine were 71 μmol/L, then the reported value would be 44 μmol/L; and if the true value of creatinine were 398 μmol/L, then the reported value would be

371 μmol/L The deviation from the true value would always be the same What differs

in the error for each of these examples is the percentage of error that occurs For the

115 μmol/L the percentage error is a negative 23 %, for the 71 μmol/L, the percentage error is a negative 37 % and for the 398 μmol/L of creatinine, the percentage error is

a negative 7 % The impact of a constant bias decreases with an increasing true value

of the analyte More important is the effect that the error has on the interpretation of the laboratory result If the bias is negative and the true value falls within the refer-ence interval and values below the reference interval have no clinical impact, then

the negative bias itself has no clinical impact For a true value that exceeds the upper limit of the reference interval, if the negative bias causes the reported value to fall within the reference interval, then the interpretation would indicate that the patient does not have the condition implied by abnormal values Thus, if the upper limit for

creatinine in the reference interval were 106 μmol/L and the true value of the analyte was 115 μmol/L, a constant bias of −27 μmol/L would cause the reported value to be

88 μmol/L, which falls within the reference interval The reported result would

indi-cate that there is not a condition of renal dysfunction or impairment, which is

classi-fied as a false negative At a creatinine concentration of 398 μmol/L, the clinician is

already aware that the patient has renal dysfunction If the physician receives a result

of 371 μmol/L instead of 398 μmol/L, it would not change the assessment by the

phy-sician, because the interpretation of the test is that the patient has renal dysfunction and the interpretation of the test is unchanged by the creatinine result These exam-ples are typical of those used for the purpose of making a diagnosis

1.2.2 Impact of Bias

In addition to making a diagnosis, clinicians use laboratory tests to monitor the disease or condition that the patient is experiencing Here the situation is different, because the clinician has already made a diagnosis for the patient’s disease or condi-tion The clinician is interested in whether the patient is getting better or worse, how well the therapy is working or predicting the course of the disease and giving a prog-nosis The clinician may be observing the patient to follow the natural course of the disease, waiting until the patient crosses a particular threshold of disease severity or demonstrates enough change in their condition to indicate a time to institute therapy

If the clinician is waiting for the values reported from the laboratory to indicate that the patient has crossed into a more severe degree of their disease, then a constant bias may disturb the proper conclusion If the constant bias is negative, and the clini-cian is waiting for the laboratory values to exceed a reference interval limit, then the patient’s condition will exceed the limit before the reported laboratory values do In such a case, the clinician may not institute therapy soon enough and may inadvert-ently postpone therapy If the constant bias is positive, and the clinician is waiting for the laboratory values to exceed the reference interval limit, then the reported laboratory values will exceed the limit before the patient’s condition truly does, and the clinician may institute therapy too early, potentially exposing the patient to risk from the therapy If the institution of therapy is not warranted, because it is a false positive, then in addition to exposing the patient to the risk of therapy, the clinician may cause valuable resources to be expended when they are not needed In a cost-conscious world, expending resources when they are not required results in a waste

of resources, which potentially can risk the safety of the entire patient population, because abuse of resources may prevent the use of resources for another patient

1.2 Types of error       5

Clinicians often monitor patients observing changes in results A constant error or bias

may have minimal impact here, because if the reported value was initially 115 μmol/L with a true value of 141 μmol/L, when the next value is reported as 97 μmol/L with a true value of 124 μmol/L, the net change in value from before to after is 18 μmol/L for

both the reported and true values Thus, in observing the absolute change over time,

a constant bias has no effect

The situation is different for proportional bias With a proportional bias, the degree of bias depends on the true concentration For a positive proportional bias, the degree of bias increases with increasing concentration of the analyte , while for a neg-ative proportional bias, the degree of bias decreases with increasing concentration

For a proportional bias of 10 % and creatinine, at 71 μmol/L true value, the reported value would be 78 μmol/L At a creatinine concentration of 106 μmol/L, the reported value would be 117 μmol/L; while at a creatinine concentration of 398 μmol/L, the reported value would be 438 μmol/L, and so on The proportional bias demonstrates

a constant percentage of error over all the values of the reportable range The centage bias can be positive or negative Typically the proportional bias is reported

per-as a slope A positive biper-as of 10 % would have a slope of 1.1, while a negative biper-as of

10 % would have a slope 0.9 The proportional bias can cause the same problems with diagnosis as does the constant bias: false negative results and false positive results.Proportional bias shows a greater impact with monitoring of patients than the constant bias does Monitoring of patients entails comparing laboratory results from one time to the next If there is no change in the patient’s condition, then one would not expect the laboratory values to change and there would be no problem If there

is a change in the patient’s condition, one would expect the laboratory results to change In monitoring a patient for renal function, as their renal function worsens, one would expect their creatinine and urea values to increase If there was a nega-tive proportional bias, their values for creatinine and urea would not rise as quickly

as their condition If there was a positive proportional bias, their values for nine and urea would rise quicker than the actual condition For example, if pharmacy needed to adjust the dosage of a drug based on the patient’s renal clearance of that drug, then if the reported creatinine value was 20 % higher than the true concentra-tion, the calculated dosage would be too low and the patient would not receive a sufficient amount of drug; likewise, if the reported creatinine value was 20 % lower than the true value, the patient would be overdosed on the drug and run the risk of becoming drug toxic Acyclovir, amikacin, ceftazidime, ciprofloxacin, digoxin, gen-tamicin, lithium, ofloxacin, piperacillin, tobramycin, and vancomycin are just some

creati-of the medications that require adjustment creati-of dosage based on the creatinine and creatinine clearance values [4]

Even though there may be pure cases of constant bias by itself, or proportional bias by itself, most biases are mixed In mixed bias, both constant and proportional biases have an effect on the reported results Frequently, the constant and propor-tional biases run in opposite directions, i.e., if the constant bias is positive, the pro-

portional bias will be negative, or if the constant bias is negative, then the tional bias will be positive The net effect of the constant and proportional biases running in opposite directions is that there will be some central value where the net bias is zero If the biases balance out one another in this manner, then it means that the true value of the analyte dictates whether there is a positive or a negative bias For example, if there is a positive constant bias balanced by a negative proportional bias, then when the true value of the analyte is less than the central value with zero bias, the net bias is positive, but when the true value of the analyte is greater than the central value with zero bias, the net bias is negative Usually in calibration issues these biases are small Fortunately, one can characterize these biases quite well The laboratory expends considerable energy and resources, such as proficiency testing surveys and comparisons with other laboratories, in characterizing and minimizing the systematic error or biases.

propor-Constant and proportional biases represent systematic errors in analysis, while imprecision represents non-systematic error In both cases the error can be well-characterized and is predictable Often, the degree of error can be controlled Both types of errors depend on the mechanics of the instrumentation, the selection of the reagents, and the quality of the calibration Interferences in Laboratory Medicine rep-resent another type of error The error due to interference is not systematic, in that it does not depend on the mechanics of the analyzer nor on the quality of calibration

It does not apply to all specimens, but is specimen-specific It does depend on the choice of sensor and reaction Interference error, even though it is non-systematic, is not measurable in the same way that imprecision error is Most of the time, it is epi-sodic Because it is episodic, laboratories need to develop ways to detect its presence and report results in a suitable manner and reduce its impact on laboratory results [5]

1.3 Interference as a Type of Bias

Interferences depend on the specimen, the method and the type of reaction involved Owen and Keevil examined the interference cause by bilirubin with the Jaffe (alka-line picrate) and enzymatic methods for creatinine [6] They varied the concentration

of bilirubin and compared the measurements for both reaction methods with liquid chromatography tandem mass spectrometry (LC-MS/MS) as the reference method Samples with low concentrations of creatinine showed more than a 10  % reduc-

tion in creatinine at bilirubin concentrations greater than 220 μmol/L as measured

by the Jaffe method For samples measured by the enzymatic method, more than a

10 % reduction in creatinine was observed for bilirubin concentrations greater than

200  μmol/L The greatest percentage of reduction occurred for specimens with the

lowest concentration of creatinine

A 20  % reduction in the reported value for creatinine can affect the quality of care and patient safety The glomerular filtration rate is estimated by two different

1.3 Interference as a Type of Bias       7

equations, the Cockroft-Gault equation and the Modification of Diet in Renal Disease (MDRD ) study [7] The Cockroft-Gault equation is used to modify drug dosage, while the MDRD equation is used to assess risk in patients and identify patients for further workup for diminishing renal function

Examination of the use of a test in varying disease states provides a way to assess the full impact of an interference Clinicians are interested in assessing glomerular function Many diseases of the kidney or injuries to the nephron result in decreased glomerular function In physiologic studies the glomerular filtration rate provides

an assessment of the number of sufficiency of the glomerular and nephrons in the kidney In physiologic studies, the glomerular filtration rate is assessed by injecting and measuring inulin, which provides the inulin clearance

The inulin clearance is not a practical way to assess renal function Instead, cians use the creatinine clearance Creatinine is an endogenous compound, the break-down product of creatine produced by skeletal muscle Its main advantage is that there is no requirement to infuse it into the patient Creatinine is freely filtered by the glomerulus, but the renal tubules also secrete a small quantity of creatinine into the forming urine [8] In normal kidneys, the tubular secretion accounts for 10 % of the total secretion, but as renal disease progresses, the percentage of the total creatinine excretion contributed by the tubules increases Though imperfect as a marker, the use

of creatinine to assess renal function is a standard practice In the past, many cians used the creatinine clearance to assess renal function Calculating a creatinine clearance requires the collection of a 24-hour urine sample to measure the creatinine

clini-in it The collection must be done at the same time that the serum or plasma sample

is obtained for creatinine The clearance is calculated by dividing the urine creatinine production rate by the serum or plasma creatinine concentration [8] Because there is

a discrepancy in the orders of magnitude of the creatinine concentration in urine and serum (plasma) and there is some degree of error in the collection of the 24-hour urine specimen, there is considerable error in calculating a creatinine clearance Today, most clinicians depend on the Cockroft-Gault equation and the Modification of Diet

in Renal Disease (MDRD) formula to assess renal function

Renal disease can be acute or chronic Any patient who develops renal disease , whether acute or chronic, requires a further workup to establish the cause, especially

if they are young Other tests, more expensive than creatinine or with some degree of discomfort and risk are required, such as renal biopsy Invasive tests, such as renal biopsy, put the patient at risk, and are extremely costly

Primary glomerular diseases include glomerulosclerosis, acute postinfectious glomerulonephritis, membranoproliferative glomerulonephritis, IgA nephropathy, and chronic glomerulonephritis [9] Systemic diseases also affect the kidneys and include lupus nephritis, diabetic nephropathy, amyloidosis, Goodpasture syndrome, Wegener’s granulomatosis, Henoch-Schönlein purpura, bacterial endocarditis, and thrombotic micrangiopathy [9] Diseases that affect the tubules ultimately decrease glomerular filtration and increase creatinine concentrations, and include such dis-

eases as acute pyelonephritis, chronic pyelonephritis, and acute tubular necrosis [9] Some of these diseases represent medical emergencies, such as any acute glomerulo-nephritis, acute pyelonephritis, and acute tubular necrosis These medical emergen-cies require quick action, and a reported creatinine result that falls below the speci-fied limit slows the response for appropriate treatment and puts the patient at risk The blood vessels may be involved in renal disease and include stenosis or occlusion

of the renal artery secondary to atherosclerosis, malignant hypertension with ing nephrosclerosis and benign nephrosclerosis [9] Essential hypertension causes hyperplastic arteriolosclerosis in the kidneys Essential hypertension is one of the major causes of decreasing renal function The other major cause of renal dysfunc-tion is diabetes mellitus It causes diabetic macrovascular disease and hyaline arte-riolosclerosis and is related to the duration of the disease and the level of the blood pressure [9]

result-Dehydration and hypovolemia cause both blood urea nitrogen and creatinine to rise, usually urea rises faster than creatinine Dehydration and hypovolemia occur

in numerous diseases such as diabetes mellitus, acid base disorders, shock, rhage, infection, trauma and surgical emergencies In dehydration and hypovolemia, blood urea nitrogen rises fast than creatinine Normally, the urea to creatinine ratio

hemor-is around 10 (urea and creatinine in units of mg/dL (conversation ratio mg/dL into

μmol/L = 40)) During dehydration and hypovolemia the urea/creatinine ratio rises to

above 20 [10] If the value reported for creatinine is falsely depressed because of ference, then this ratio will rise even higher Clinicians use it in emergency situations and acute care to decide whether to give more fluids A falsely elevated value for this ratio would prompt clinicians to give fluid inappropriately, and, perhaps, not search for other causes for the azotemia and increased creatinine values

inter-Interferences represent a non-systematic type of error that is episodic and ficult to predict Interferences impact patient safety by creating false positive or false negative results when used for diagnosis, or by falsely increasing or decreasing the difference between two results separated by time used in monitoring Because it is difficult to predict the exact errors caused by interferences, especially for bilirubin, lipemia and turbidity , one needs to be aware of the nature of interferences, ways to measure them, the clinical situations that give rise to elevated bilirubin, lipemia and turbidity and ways to detect their presence

[3] VIM93 ISO, International Vocabulary of Basic and General Terms in Metrology Geneva,

Switzerland International Organization for Standardization, 1993.

2 Nature of Interferences

2.1 Definition

There are two commonly used definitions of interference The first definition claims that an analytic interference occurs when there is a component in the sample that causes an error in measurement of the analyte in the analyzer but by itself does not produce a signal [1] This definition presents a problem because it fails to include any effect that the interferent may have in the absence of the analyte It is better to define interference as any agent present in the sample, which causes the result of a measure-ment process to deviate (demonstrate a bias) from the true value or value that would have been obtained in the absence of the agent [2] Defining the term interference

in this broad fashion would allow the definition to cover such agents as bilirubin , lipemia and turbidity The definition for interference can include the effect of an agent

in the detection or determination of concentration or activity of an analyte occurring

at any step along the path to providing a result [3] For a more detailed discussion of the concepts of interference in analytical systems, see the review by Büttner [4] For the purposes of this book, we define interference as the effect of a substance present

in the sample that alters the correct value of the determination of the result, tively or qualitatively, for the analyte under consideration [2]

quantita-The interfering agent is a substance that alters the determination of the correct value That substance may be exogenous or endogenous An exogenous substance is one that is not a naturally occurring chemical or agent, e.g., an antibiotic An endog-enous substance is one that occurs naturally, but is in a higher concentration than normally encountered Hemolysis, lipemia, bilirubinemia, and paraproteinemia are examples of endogenous substances

2.2 Nature of Interferences

Bilirubin and lipemia generate interferences by three basic mechanisms Bilirubin and lipemia may alter the way light is absorbed or scattered Also, bilirubin may react with reagents or with the analyte itself Hyperlipidemia may alter the balance between aqueous and non-aqueous phases Melvin Glick et al., in their book, Interferographs , presented many examples of interferences caused by bilirubin and lipemia [5] This book is arranged by analyzers and demonstrates the effect of interference for bili-rubin, and for that matter, lipemia as well, over increasing concentrations on many different types of analyzers that were in use at that time Today, it remains a valuable resource, because the book clearly demonstrates the ubiquitiousness of interference over many different tests and varying methodologies

Awareness of bilirubin interference was heightened in the 1990s and manufacturers responded by improving their methods Improvement of methods occurred through better selection of wavelengths, addition of reagents that inhibited the interference effects, and changes in the way that analysis occurred in the analyzer Despite these changes for improvement, for which the manufacturers can be thanked, knowledge

of the methods affected by bilirubin is still helpful today, because the mechanisms behind the interferences have not been elucidated in all cases and the evolution of newer analyzers and the search for cheaper reagents may return one to the method-ologies used in the past Observation of the different methods and the noted inter-ferences serves as a resource to evaluate methods used in the present and in the future

con-The basic design of a spectrophotometer consists of a source of radiant energy,

a dispersive device, a vessel to hold the analyte in a solvent and for the reaction to occur, a photodetector and finally, a readout device [6] The most common sources of radiant energy consist of a lamp Typically these lamps emit electromagnetic radia-tion from 150–10,000  nm The spectrophotometer contains a dispersive device that separates the different wavelengths of light into discrete bandwidths of radiation [6] Examples of dispersive devices are absorption filters, interference filters, prisms and diffraction gratings The dispersive device may be placed either before or behind the sample Finally, the incident radiation (light) that passes through the sample will hit

a detector Commonly used detectors are photodiodes or photomultiplier tubes.The radiation or light hitting the sample may be absorbed by molecules with that ability The light absorbed by each of these molecules depends on the chemical structure of that particular molecule and the wavelength of the radiation In the light spectrum, on the left-hand side are the higher energies, shorter wavelengths associ-ated with ultraviolet light , while on the right-hand side are the lower energies, longer wavelengths associated with red light When electromagnetic radiation encounters

a molecule, it may be absorbed, scattered or completely pass by the molecule In

a spectrophotometer, when the wave of light passes by the molecule, that process

is known as transmission Spectrophotometers are designed to measure the light passing through a cuvette that contains the analyte of interest as well as a cuvette

in which the analyte of interest is omitted The cuvette without the analyte is known

as the blank The photometer provides a readout of the amount of light transmitted

2.3 Instrumentation       13

through the blank cuvette and the test cuvette, the one containing the analyte, for which the amount of light from both of these sources can be compared The amount

of light transmitted through the cuvette, i.e., the light that bypasses the molecules in

the sample, is known as the transmittance [6].

The transmittance represents a comparison between the amount of light mitted through the test cuvette and the amount of light transmitted through the blank cuvette A comparison is required because light may be absorbed or scattered by the material composing the cuvette itself, even if it is glass, quartz or plastic, and by the solvent within the cuvette In the clinical laboratory, the most commonly used solvent consists of water, but may include other chemicals as well, typically such chemicals as salts, protein, surfactants, etc By matching the cuvettes, the amount

trans-of light absorbed by the cuvette and solvent by both the blank and the sample can

be essentially handled and accounted for The amount of light transmitted through

the blank is referred to as I o, while the amount of light transmitted through the test

sample is I [6] The transmittance for the test sample is defined as T = I/I o [6] The spectrophotometer will translate the amount of light transmitted, i.e., the amount of light impinging on the detectors, into a voltage difference Ultimately this voltage dif-ference is sent to a readout, which can quantify the amount of voltage difference for the test sample and the blank For our purposes here, the quantity can be considered

to be arbitrary If the amount of transmitted light had a value of 500 for the blank and

400 for the test sample, then the transmittance would be 0.9

The amount of light transmitted through the test sample, where our analyte is located, depends on the concentration of the analyte; however, it does so in an expo-nential manner Therefore, the relationship between the concentration of analyte and transmittance is not linear To make it easier to understand, the logarithm of the reciprocal of the transmittance is used to define absorbance , A [6]

According to Beer’s law the absorbance, A, is related to the capacity of a molecule to

absorb light, a, the pathlength of the cuvette, b, and the concentration of the analyte,

C, in a linear fashion.

(2.2)

One knows the pathlength of the cuvette Also, one can determine the capacity of the molecule to absorb light, known as the specific absorptivity or specific absorptivity coefficient Then, one can directly relate the absorbance to the concentration of the molecule being identified A transmittance of 1.0 yields an absorbance of 0, that of 0.5 yields an absorbance of 0.3, that of 0.1 an absorbance of 1.0, that of 0.01 an absorb-ance of 2.0, and that of 0.001 an absorbance of 3.0 Absorbance provides an easy way

to measure the molecule of interest Spectrophotometers are limited in the range of concentrations that they can measure because of error of detection There is consider-able error of detection when the concentration of the molecule of interest is very low

or very high For most practical purposes, the relative error is considered high when the absorbance, A, falls below 0.1 or above 2.4 [6]

Methods using spectrophotometers to measure analytes are designed to relate a set of standards or calibrators that cover absorbances from 0.1–2.0 The standards or calibrators represent known entities The reaction is run and the absorbance for each concentration is determined, yielding, in most cases, a linear relationship, known

as a standard curve The slope of the line determined by this standard curve directly relates the absorbance to a concentration When the unknown test sample is ana-lyzed, the absorbance is compared to this standard curve, from which one can read the concentration for the unknown In non-automated methods, one typically reads the concentration from a graph In automated analyzers, one can use the inverse of the standard curve If the standard curve is represented as

2.4 The Chemistry of the Absorbance of Light       15

absorbs a sufficient quantity of light within a spectrum of wavelengths that can be considered to be fairly specific for that molecule If the molecule of interest does not absorb a sufficient quantity of light within the range of the spectrophotometer, or the specificity is not sufficient, then the molecule can be reacted with reagents that will produce another molecule that can meet these criteria The challenge in Clini-cal Chemistry and much of Laboratory Medicine is to design methods that allow the determination of a molecule of interest in a sea of other molecules that have similar physical and chemical characteristics

2.4 The Chemistry of the Absorbance of Light

The spectral range that one may utilize to determine molecules of interest is what limited The infrared region of the spectrum begins at 700 nm and extends to 5,000  nm Molecules absorb light in the infrared region as the result of vibrational motions of atoms as well as bending, rotating and twisting [6] Infrared spectroscopy

some-is useful in identifying fairly pure solutions of a molecule, because the absorbance some-is specific for functional groups making up the molecule One needs to examine absorb-ance versus a fairly wide spectrum and relate the absorbance peaks to known func-tional groups Sometimes the absorbance spectrum may give way to a ‘fingerprint’ pattern compared to known compounds

At the other end of the spectrum, with wavelengths below 400  nm, begins the ultraviolet region The human eye cannot detect ultraviolet light There are many molecules that absorb in this region; however, the use of the ultraviolet region is limited because of absorbance by materials used to hold the sample, such as glass, and by solvents Most organic solvents absorb significant amounts of light between

100 and 300  nm [6] For example, acetone absorbs light at 340  nm and ethanol at

210 nm Water, the most common solvent for Laboratory Medicine purposes, absorbs

at 191  nm  [6] For these reasons, most wavelengths usable in determinations fall between 340 and 700 nm

Most molecules of interest do not absorb enough light between the ranges of 340–700 nm without some chemical reaction Usually, as in the case of creatinine , the molecule of interest will shift the absorption band of a reagent, allowing for quanti-fication of the analyte Bilirubin is unique in that it absorbs light without the need of reaction Lipemia is unique in that it is the major source of turbidity , a form of light scattering

As mentioned before, light, a form of electromagnetic radiation can interact with matter If the light is scattered by the matter, no change in the matter occurs If the matter absorbs the light, then there is an alteration of the molecule in some fashion Radiation with short wavelengths, such as X-rays, when they interact with matter can break chemical bonds Often, this interaction is destructive At long wavelengths,

2.4 The Chemistry of the Absorbance of Light       17

such as infrared light, nearly all the chemical bonds can absorb the light and the energy is given off in some form of heat, because it represents vibrational energy

In regards to analytical methodology and spectrophotometry , absorption of visible and near ultraviolet radiation is the most interesting and useful form of absorption Molecules absorbing light in this region undergo electron excitation [7] If the surface

of a bulky material is very smooth, it may reflect light, for example, a mirror If the surface is rough, it will reflect light diffusely The latter effect is the one we experience

in everyday life Diffuse reflection demonstrates the color of the object [7] Scattered light , as occurs with a powder, does not demonstrate a color, because there is no interaction with the matter itself [7]

For absorption in analytical chemistry, the most important sources of color for objects are transition metals and transitions between molecular orbitals [7] The absorption by transition metals is an important facet for the chemistry of minerals and inorganic chemistry and has been used to detect and quantify transition metals

in clinical chemistry They are notable for the absorption of a specific wavelength of light, with the transition of an electron in a d-orbital to a specific higher energy level, which is known as a ligand field effect [7] The resulting absorption band often is very sharp

Far more important for the purposes of clinical chemistry are absorption and color in organic molecules Absorption by organic molecules can be described by the molecular orbital theory [7] Much of the knowledge concerning color in organic molecules is derived from the study of organic dyes In dyes, it has been noted that conjugated double bonds , as occur in carotene , give rise to a particular color (Fig. 2.2 and Fig. 2.3)

Dyes possess several resonance structures [7] In molecular orbital theory, there are several types of orbitals involved, including bonding or antibonding orbitals of sigma (σ or σ*) or pi (π or π*) types, and in addition, n-type nonbonding orbitals [7]

In the lowest energy state of the molecule, all of the bonding and nonbonding als are fully occupied, while the antibonding orbitals are empty [7] When the organic molecule absorbs light, the energy of the light now becomes part of the molecule, often described as a transition of orbitals from the highest occupied molecular orbital (HOMO) to the lowest unoccupied molecular orbital (LUMO), i.e., an electron transfers from a nonbonding or bonding orbital to an antibonding orbital, usually n → π* and

orbit-π → π* [7] In formaldehyde the orbit-π → π* produces a strong absorption band at 185 nm and the n → π* produces a strong absorption band at 290 nm [7] Neither of these two bands produce a color in the visible region, but addition of another double bonded carbon to formaldehyde produces acrolein (Fig. 2.4), and shifts the n → π* absorption from 290–330 nm, much closer to the visible region in the spectrum [7] Addition of another double bonded carbon to acrolein (Fig. 2.4) produces a molecule with yellow absorbance

Continued addition of double bonds in a polyene fashion produces dyes, such as carotene (Fig. 2.2 and Fig. 2.3) Absorbance of this molecule is predominated by the

π → π* transition [7] In organic molecules, the p-orbitals are unshielded from static interactions; the electrostatic interactions from other electronic orbitals have their effect on the absorption bands, broadening them out, with the result that the absorption band for organic molecules are typically very broad Also, in organic mol-ecules, the change in energy levels between the LUMO and the HOMO states depend

electro-on the length of the π system in the molecule and the belectro-ond length; molecules like bilirubin are rather plastic and there can be small variations in the length of the π system, causing a dispersed population of molecules absorbing light [8]

As the number of double bonds increases in a dye, so does the wavelength of light absorbed by the dye [8] If the wavelength of light absorbed is 420  nm, then the color of the light absorbed is blue, and the solution appears yellow; if the wave-length is 540 nm, the color of the light absorbed is green and the color of the solu-tion is red; if the wavelength of light absorbed is 640 nm, then the color of the light absorbed is orange and the color of the solution is blue; and if the wavelength of the light absorbed is 740 nm, then the color of the light absorbed is red and the color of the solution is bluish green [8] As the number of π electrons increases in a polyene,

so does the wavelength, with 16–30 p-electrons placing its absorbance in the visible region between 400 and 550 nm [8] Beta-carotene is a typical example, with a strong orange absorption (Fig. 2.2 and Fig. 2.3) Bilirubin also has a polyene structure with nitrogens in the place of some of the carbons (Fig. 2.3) Carotene has 11 double bonds while bilirubin has ten In both carotene and bilirubin, the double bonds are laid out in adjacent fashion without ring structure formation Benzene shows aromatic resonance because it has three double bonds arranged in hexagonal ring structure Neither carotene nor bilirubin demonstrate aromatic resonance Because bilirubin does not demonstrate aromatic resonance as in benzene, it acts as a strong chromo-phore in the visible region of light (Fig. 2.2) The absorbance spectra for carotene and bilirubin are almost identical in the visible region Bilirubin is not quite as strong as carotene in its molar absorptivity, but both molecules represent important natural dyes

Polyenes do not need to exist only in a straight chain, but may also occur in cyclic form, and they can represent strong absorbers of light if they do not present with benzenoid conjugation and aromatic resonance [7] The porphyrins are strong rep-resentatives of this class and include chlorophyll and heme Of course, bilirubin is the breakdown product of heme Bilirubin has a strong potential for causing interfer-ence because its absorption band covers a broad area of the spectrum In the past, if

O

C

O C H

Fig 2.4: Structure of formaldehyde and acrolein.

2.4 The Chemistry of the Absorbance of Light       19

one directly measured absorbance without a chemical reaction, bilirubin has a strong potential for causing interference, because its absorbance might be measured as well This interfering potential is an important problem for co-oximetry, where hemoglobin

in its various forms is measured without chemical alteration

One of the most common ways to minimize interferences with methods is using

a sample blank In a manual spectrophotometer, that process is rather easy, because one would use a sample diluted with buffer as the blank in the analyzer Automated equipment used to present a challenge, because automated analyzers used a single blank channel for multiple samples More modern automated analyzers solved this problem by taking multiple readings Thus, an initial reading is taken after the addi-tion of the sample and the buffer or other diluents, prior to adding the reagents neces-sary to start the reaction This blank reading for the sample is held in the analyzer’s computer and subtracted from the final readout answer The multiple readings approach works well, but may be confounded if the blank reading is so high that the final total absorbance is very high If the final total absorbance reading is very high,

it may exceed the photometric error allowed for the assay and result in a flag or code, indicating that there is a problem with the sample [9]

An alternate approach to taking a blank reading is to subtract out a baseline of the interferent This approach is commonly used to deal with the effects of hemo-globin present in the sample and is mentioned here because its use has the potential

to cause problems with bilirubin This method is commonly referred to as the Allen correction In the classic Allen correction, in addition to the wavelength chosen for the primary chromogen (usually chosen close to its maximum value of the absorption curve after subtracting out the absorbance for the reagent), two other wavelengths are chosen, one to the left of the chromogen wavelength and the other to the right of the chromogen wavelength (Fig. 2.2) [9] This method works if the spectrum of the inter-ferent differs from that of the chromogen in the reaction It assumes that the spectrum

of the interferent is approximately the same on both the left-hand side (lhs) and the right-hand side (rhs), and that the chromogen has minimal absorbance at these two wavelengths The formula for calculating the Allen correction is

(2.5)

In many modern analyzers, a modified version of the Allen correction is used, where only one wavelength is used to adjust for the presence of hemoglobin Use of only one wavelength may incur effects from other potential interferents, such as bilirubin

Corrected A chromogen = A chromogen − (A lhs +2A rhs )

[3] chemistry; proposed guidelines NCCLS Document EP7-P Villanova, PA, USA, National

Committee for Clinical Laboratory Standards, 1986.

[4] Büttner J Unspecificity and interference in analytical systems: concepts and theoretical aspects DG Klin Chem Mitteilungen Germany 1991,22,3–11.

[5] Glick MR, Ryder KW, Glick SJ Interferographs User’s Guide to Interferences in Clinical Chemistry Instruments 2nd ed Indianapolis, IN, USA, Science Enterprises, Inc, 1991.

[6] Willard HH Merritt Jr LL, Dean JA, Settle Jr FA Instrumental Methods of Analysis 6th ed Belmont, CA, USA, Wadsworth Publishing Company, 1981.

[7] Nassau K The Physics and Chemistry of Color: The Fifteen Causes of Color New York, NY, USA, John Wiley & Sons, 1983.

[8] Kuhn H, Forsterling H-D Principles of Physical Chemistry: Understanding Molecules, Molecular Assemblies, Supramolecular Machines Chichester, UK, John Wiley & Sons, Ltd, 2000.

[9] Kaplan LA, Pesce AJ Clinical Chemistry:Theory Analysis Correlation 5th ed St. Louis, MO, USA, Mosby Elsevier, 2010.

3 The Nature of Icteric Interference

3.1 Source Information on Bilirubin Interference

The most comprehensive and informative source of information on bilirubin ference comes from the book, Interferographs , by Melvin Glick, Kenneth Ryder and Starla Glick [1] As might be expected, the degree of interference varies by manufac-turer and by method Some analyzers demonstrate next to no interference for their methods, while other analyzers demonstrate many methods with an interference due

inter-to bilirubin Some of the more common methods affected by a bilirubin interference are creatinine, uric acid, and alkaline phosphatase Often the creatinine, cholesterol and uric acid interferences are negative, that is, yield lower results, while the bicarbo-nate results are positive, that is yielding higher results than expected [1]

The College of American Pathologists (CAP), through its Instrumentation Resource Committee, offers a survey that provides materials for testing the impact

of interfering substances with common chemistry tests The survey materials include samples with specified amounts of either hemoglobin or bilirubin The survey is designed for verifying manufacturers’ interference specifications and investigating discrepant results caused by interfering substances The survey is designed to cover more than 20 common analytes

The survey provides information for the individual laboratory, showing the effects of hemoglobin and bilirubin interference for their own analyzers In addition, the survey provides information at the instrument or group level The survey allows

a laboratory to evaluate the quality of their detection system for hyperbilirubinemia The CAP survey has demonstrated bilirubin can cause interference with many common chemistry tests, including ALT, albumin, alkaline phosphatase, calcium, creatinine, sodium, glucose, lipase, total protein, creatine kinase, magnesium, phos-phorus, urea and uric acid Because bilirubin interference varies by peer group, plat-form, test methodology and reagent formulation, each laboratory should evaluate the methods used in their laboratory The College of American Pathologists Interfering Substance Survey provides a convenient process to obtain samples containing biliru-bin and analysis of the results

3.2 Allen Correction as a Source of Bilirubin Interference

Bilirubin might also cause an interference if the Allen correction is used to handle hemoglobin absorbance interference This approach is called the bichromatic correc-tion and uses a primary wavelength for following the analyte in the reaction and a secondary wavelength to subtract contributions for other substances, usually hemo-globin Use of the bichromatic correction works well in most cases, but it can present

a problem if the choice of the secondary wavelength shows absorbance with bin As an example, consider the problem with using two wavelengths for measure-ment, 400 nm for the primary wavelength and 450 for the secondary wavelength Bilirubin’s extinction coefficient is 22,910 at 400 nm and 55,000 at 450 nm If the reaction is measured at 400 nm and the absorbance at 450 nm is subtracted from the total absorbance, there will be an overestimation of the amount of absorbance due to bilirubin, in this case because the ratio of bilirubin’s absorbance at 450 nm compared with 400 nm is 2.4–1 The effect is small for very low concentrations of bilirubin, but

biliru-as the concentration of bilirubin increbiliru-ases it may have a much greater effect Such

an occurrence can explain why sometimes the interference from bilirubin may be negative

3.3 Bilirubin Interference with Oximetry

Bilirubin may cause an interference in the measurement of the different forms of hemoglobin by absorbing light at the same wavelengths as that of hemoglobin The active portion of the hemoglobin molecule is the heme unit, which is a porphyrin ring It contains one atom of iron, which may be in the ferrous (Fe2+) or ferric (Fe3+) oxidation state In order to bind oxygen (O2), the iron must be in the ferrous state On binding oxygen, there is a tendency for the iron to become oxidized from the ferrous

to the ferric state, which is known as methemoglobin Normal individuals reduce the ferric form of iron back to the ferrous form by means of the enzyme NADH-cyto-chrome-b5 reductase [3] The blood of normal persons contains approximately 1.5 % methemoglobin

Hemoglobin provides a way for the blood to capture oxygen in the lungs and transport it back to the tissues Lung diseases, both acute and chronic, demonstrate decreased concentrations of oxyhemoglobin Determination of oxyhemoglobin and percent saturation of hemoglobin by oxygen represent key measurements in the care

of many acutely ill patients and that information is critical in reducing morbidity and mortality The percent saturation of hemoglobin by oxygen can be determined by both pulse oximetry and co-oximetry Pulse oximetry is an in vivo, non-invasive technique where the oxy and deoxy forms of hemoglobin are measured through the skin Co-oximetry measures oxy and deoxy forms of hemoglobin, but requires a whole blood sample

One can distinguish between the different forms of hemoglobin by using try There are two types of analyzers for oximetry, the pulse form and the co-oximeter Both of these types of devices work on the principle of using multiple wavelengths

oxime-of light to measure the different forms Both oxyhemoglobin and deoxyhemoglobin absorb light at the same wavelengths, but the amount of light absorbed varies depending on the wavelength, which is expressed as the molar absorptivity or molar extinction coefficient

3.3 Bilirubin Interference with Oximetry       23

One could measure the total amount of hemoglobin at 431 nm, which has an tion coefficient of 528,600 (extinction coefficients are in units of per Mole/L for a pathlength of 1 cm) The other forms of hemoglobin absorb light at wavelengths very close to 431 nm with nearly similar extinction coefficients (Fig 3.1) [3]

Fig 3.1: Absorbance spectra for oxyhemoglobin, deoxyhemoglobin and bilirubin.

The extinction coefficients are too high, near 431 nm, to be very useful; however, one can measure the hemoglobin molecule absorption at slightly above 500 nm to dis-criminate among the varying hemoglobin forms At a wavelength of 555 nm, deoxyhe-moglobin has an extinction coefficient of 54,520, oxyhemoglobin an extinction coeffi-cient of 36,815, carboxyhemoglobin and methemoglobin have extinction coefficients that are distinct from these other species The absorption maximum in the 500–700 nm region for oxyhemoglobin is 578 nm, while 621 is the absorption maximum for meth-emoglobin, with carboxyhemoglobin showing absorbance maxima at 541 nm and 577

nm [4] By measuring the peaks for deoxyhemoglobin, oxyhemoglobin, moglobin and methemoglobin at three other wavelengths, one can distinguish among the varying forms, because the peaks occur at different wavelengths and the extinc-tion coefficient are different

carboxyhe-3.3.1 Co-oximetry Interference

By measuring the absorbance at four carefully chosen wavelengths, one can discern among the four species of hemoglobin In some methods, one measures the total hemoglobin as cyanohemoglobin, so a fifth wavelength is added To find the concen-trations of the various species one solves the set of simultaneous equations given by

(3.2)

where C represents the concentration of each hemoglobin species, ε represents the extinction coefficient for that species at that particular wavelength, λ, and Aλ rep-resents the absorbance at that particular wavelength The equations represent five equations and five unknowns, and can be directly solved for the concentration of each species The problem is that bilirubin can also absorb light in the region of the selected wavelengths and thus masquerade as one of the hemoglobin species

An example of this type of interference occurs in oximetry where it has been noted that pulse oximetry is often not affected by bilirubin, but co-oximetry is [5] Beall and Moorthy reported a case that is interesting A patient was treated for nodular scleros-ing Hodgkin’s lymphoma with several rounds of chemotherapy and whole body irra-diation followed by autologous bone marrow transplantation He developed hepatic venous occlusive disease, which resulted in hepatic failure with bilirubin concentra-

tions ranging between 633–770  μmol/L and respiratory failure requiring intubation

and mechanical ventilator support The patient was monitored with continuous pulse oximetry and blood gas analysis The blood gas analysis included determination of hemoglobin oxygen saturation by co-oximetry

The staff were able to maintain the patient’s arterial pO2 in a range of 92–

133  mmHg, which should be sufficient for normal oxygenation of hemoglobin; however, the results from the co-oximeter of the blood gas unit were lower than expected, with hemoglobin saturations of 88–93 % (IL 282 Co-oximeter from Instru-mentation Laboratory) These results for the co-oximeter do not correlate well with the arterial pO2, obtained on the same specimen as used for the co-oximetry deter-mination of hemoglobin saturation Further evidence helped to discern this discrep-ancy The staff recorded values for the hemoglobin saturation from pulse oximetry using two different devices, the Nellcor N100c and the Ohmeda Biox 3700® [5] The results from pulse oximetry gave results of hemoglobin saturation of 98–99 %, which

is within the normal range and consistent with the arterial pO2 measurements.Additional information from the co-oximeter showed a slight increase in the frac-tion of hemoglobin represented as carboxyhemoglobin , in this case the values ranged

3.3 Bilirubin Interference with Oximetry       25

from 2.4–2.9 % and increased fractions measured as methemoglobin , in this case the values ranged from 3.2–11.9  % [5] Differences besides the specimen, transcutane-ous nonvasive for pulse oximetry and an arterial blood sample for co-oximetry are required to explain these differences

The basis for measurements of oxygen saturation depends on the differences in extinction coefficient s at various wavelengths for deoxyhemoglobin and oxyhemo-globin Oxyhemoglobin absorbs less light in the longer wave, red region, near 660

nm, than does deoxyhemoglobin [6] Because deoxy hemoglobin absorbs a able amount of light in the 660 nm range, associated with red color, its color does not appear as red as oxyhemoglobin The color of the substance is determined by the wavelengths that the light reflects If a substance, such as oxyhemoglobin, absorbs more light in the blue, yellow and green regions of the spectrum , it reflects more red light and therefore looks red Carboxyhemoglobin, the form of hemoglobin which has combined with carbon monoxide, CO, instead of oxygen, absorbs even less light in the red region, and thereby appears even brighter red than oxyhemoglobin

consider-3.3.2 Pulse Oximetry

A pulse oximeter works by synchronizing the absorbances with the arterial pulse [6]

In the skin, the capillaries dilate in response to the pulse and synchronizing the urements with the pulse gives the most reliable results In pulse oximetry, absorb-ances are measured at two absorbances, one at 660 nm, at which the absorbance of oxyhemoglobin is less than that of deoxyhemoglobin, and at a much longer wave-length The secondary measurement of the absorbance is typically done at a wave-length between 815 and 940 nm In this region of the spectrum, the absorbance of the oxyhemoglobin is slightly greater than that of deoxyhemoglobin [6] Pulse oximeters are normalized to account for variation in skin absorbance from person to person

meas-By examining a ratio of the absorption of light in the red region (R) and the red region (IR), pulse oximetry can account for variations from person to person The ratio is calculated as

infra-(3.3)

Manufacturers calibrate the pulse oximeters empirically by observing the ratio from a group of normal volunteers and comparing them with results obtained by a co-oxime-ter [6] Subjects are asked to breathe hypoxic gas mixtures so that a calibration curve may be generated Around an oxyhemoglobin saturation of 85 %, the absorbance at the red and infrared wavelengths measures are about the same, the ratio of these two absorbances would be 1.0 Using a standard or calibration curve provides for a robust relationship because there is a large difference in the values for the ratio; at 99  %

Ratio (660 : 815−940) = IR R

oxygenation, the ratio is around 0.4, while at 50 % it is around 2.0 [6] Bilirubin does not absorb much light at either of the two wavelengths chosen for measurement with pulse oximeters; therefore, bilirubin is fairly unlikely to cause an interference with this methodology.

Pulse oximeters do have limitations though Because pulse oximeters measure light at only two wavelengths, they can only distinguish two different species of hemoglobin At 660 nm, methemoglobin absorbs light in a fashion similar to oxy-hemoglobin [6] Carboxyhemoglobin absorbs similarly to oxyhemoglobin as well Caution must be used in depending on pulse oximetry alone, because it cannot provide information about these two important species of hemoglobin Frequently, acutely ill patients have their oxygen saturation determined by both methods.The IL 282 Co-oximeter (Instrumentation Laboratories) used in the case measured light at four wavelengths, 535, 585, 594 and 626 nm Even though bilirubin exhibits its absorption peak at 450 nm, the bilirubin absorption band extends its tail into the 535–585 nm range, adding its absorption to that measured for hemoglobin Based on the set of equations used to calculate deoxyhemoglobin, oxyhemoglobin, carboxyhe-moglobin and methemoglobin, the presence of bilirubin in the specimen will add to the apparent absorbance of carboxyhemoglobin and methemoglobin, which is why the carboxyhemoglobin and methemoglobin were overestimated in the co-oximeter

In the co-oximeter, the percent hemoglobin saturation by oxygen is determined by dividing the calculation fraction of oxyhemoglobin by the total hemoglobin Total hemoglobin is determined from the fractions for deoxyhemoglobin, carboxyhemo-globin and methemoglobin Because the calculations for the co-oximeter are per-formed differently than for the pulse oximeter, bilirubin may cause an interference with co-oximeters not seen with pulse oximeters Many co-oximeters today use an increased number of wavelengths, so that they can calculate the contribution to absorbance from bilirubin and subtract it from the total, thus avoiding bilirubin inter-ference

Pulse oximetry is utilized by transmitting light through skin, usually a finger New techniques in oximetry provide a noninvasive method assessing cerebral oxygen saturation using dual-wavelength near-infrared spectrophotometry (NIRS) [7] NIRS has been used for monitoring cerebral oxygenation during carotid endarterectomy, acute heart failure and orthotopic liver transplantation [7] NIRS works by measuring absorbance in the cerebral tissue at 733 and 809 nm [7] The wavelength at 733 nm provides a measure of deoxygenated hemoglobin, while the wavelength at 809 nm provides a sum of deoxygenated and oxyhemoglobin

3.3.3 Cerebral Oximetry

Cerebral oximetry employs infrared light which can penetrate the tissues The tion of the two wavelengths, one near 730 nm and the other near 810 nm allows for

selec-3.3 Bilirubin Interference with Oximetry       27

maximal tissue penetration, with the light being scattered back from the skin, the skull, and up to 15 mm of cerebral tissue [8] The operation of orthotopic liver trans-plantation is divided into four phases: dissection, anhepatic, reperfusion and end Patients requiring liver transplantation are likely to be jaundiced During the reperfu-sion phase a rise in the cerebral oxygen saturation is expected; however, in patients with elevated bilirubin this increase in the cerebral oxygen saturation is blunted [7].Examination of the hemoglobin oxygen saturation showed no effect of bilirubin

in arterial or venous samples; however, examination of the cerebral oxygen tion demonstrated a negative interference with increasing concentrations of blood bilirubin [7] It is suspected that rather than bilirubin deposited in the cutis, it is biliverdin, the oxidative product of bilirubin, deposited in the cutis that is interfer-ing with the cerebral oxygen saturation measurement [7] The absorption of light by biliverdin changes depending on the orientation that the molecule takes: in nonpolar solvents it takes on a ring form and absorbs light in the ultraviolet region, around 350

satura-nm, but in polar solvents or attached to proteins, it takes on a straight chain form with four connected pyrrole groups and shifts its light absorbance to the visible region [9]

3.3.4 Interference with Methemoglobin

Falsely elevated values for methemoglobin , caused by bilirubin when using metry, can cause erroneous diagnosis of methemoglobinemia Methemoglobinemia is defined when the hemoglobin in the red blood cells possess greater than 1 % hemo-globin, and even though a small amount of methemoglobin is normal and does not pose a health risk, elevated fractions of methemoglobin decrease the oxygen carry-ing and delivery capacity of the blood and may pose a risk, resulting in a functional anemia [10] Methemoglobin is formed when the iron in normal hemoglobin is oxi-dized from the ferrous to the ferric form When this occurs the skin become cyanotic

co-oxi-or blue in appearance; neurologic and cardiac symptoms begin when the fraction increases above 15 % because of hypoxia and death occurs for fractions of 70 % or above [10]

The normal method for reduction of ferric iron to ferrous iron involves the adenine dinucleotide (NADH)-dependent reduction, called the diaphorase pathway, the enzyme cytochrome b5 reductase playing the major role in the pathway [10] The diaphorase pathway reduces 95–99 % of the methemoglobin, while another enzyme system, the nicotinamide adenine dinucleotide phosphate (NADPH)-dependent methemoglobin reduction accounts for the rest; this enzyme uses glutathione and glucose-6-phosphate dehydrogenase (G6PD) to reduce methemoglobin to hemo-globin [10] Hereditary methemoglobinemia is a rare condition and is most commonly found among Native American tribes such as the Navajo, Athabascan Alaskans and the Yakutsk people of Siberia; it involves a deficiency in cytochrome b5 reductase [10] Most cases of methemoglobinemia are acquired through exposure to drugs or toxins,