Diabetes Mellitus in pregnancy " Gestational diabetes mellitus''
Problems in the assessment of glycaemic control in diabetes mellitus
Transcript of Problems in the assessment of glycaemic control in diabetes mellitus
REVIEW
Problems in the Assessment ofGlycaemic Control in Diabetes MellitusE.S. Kilpatrick*
Department of Chemical Pathology, Withington Hospital,Manchester, UK
The measurement of glycated haemoglobin and serum fructosamine to assess the recentglycaemic control of diabetic patients has become well established. Likewise, the monitoringof blood glucose using glucose test strips and meters has become popular in both thecommunity and in the hospital inpatient environment. However, despite improvements inthe methods of analysis, clinically inaccurate assessments of glycaemia can still occur.Specific problems such as the lack of standardization in assays are in the process of beingresolved, but inherent difficulties associated with these measures remain. Clinicians shouldbe aware that these tests still need to be interpreted in conjunction with clinical prudence. 1997 by John Wiley & Sons, Ltd.
Diabet. Med. 14: 819–831 (1997)
No of Figures: 1. No of Tables: 3. No of Refs: 156
KEY WORDS Glycaemic control Glycated haemoglobin Fructosamine blood glucose
Received 14 January 1997; accepted 28 April 1997
Introduction Glycated Haemoglobin
The last 20 years have brought about a major advance Clinical Utilityin the assessment of glycaemic control in diabeticpatients. Two developments, in particular, have been Following the discovery that diabetic patients exhibited
an increase in some of the minor fractions of haemo-instrumental in this. The first has been the ability, sincethe 1970s, of patients to assess their own glucose control globin,3,4 it was established that this ‘diabetic haemo-
globin’ was structurally identical to the component ofthrough the self-monitoring of blood glucose (SMBG)using glucose test strips and portable meters. The second haemoglobin known as HbA1c.5 Haemoglobin A1c had
previously been described in non-diabetic adults,6 butis the subsequent development of glycated haemoglobinand fructosamine assays which have allowed clinicians in comparison to diabetic patients it was present in
much smaller proportions. Nearly a decade passed beforean independent means of judging a patient’s prior glucosecontrol. Although other indicators of glycaemia (such as a flurry of clinical studies showed that the increased
proportions of HbA1c in diabetic patients could be usedglycated albumin1 and 1,5-anhydroglucitol2) have beenproposed, they are not routinely used in clinical practice. as a reliable index of glycaemic control over the preceding
6–8 weeks. The studies demonstrated correlation ofThis ability to accurately assess present and priorglucose control has undoubtedly benefited the treatment glycated haemoglobin with other known indicators
of diabetic control such as 24-hour urinary glucoseof patients with diabetes. Clinicians can now set goalsof glycaemic control for their patients by using glycated excretion,7 plasma ‘glucose brackets’,8 daily mean plasma
glucose9 and the area under the curve of the glucosehaemoglobin or fructosamine values while patients canempower themselves to meet these aims through the tolerance test.10 Following these studies, and despite
considerable methodological difficulties in measurement,self-monitoring of blood glucose. However, errors in anyof these measurements may either lead to the setting of acceptance of the use of glycated haemoglobin by
diabetologists and other healthcare workers was rapidunattainable goals or conversely the acceptance ofspuriously good glucose control. because for the first time they had an apparently simple
tool which could give a completely objective assessmentAs these measurements become further established itis easy to assume that any faults which may have existed of a patient’s glucose control.11 Thus treatment regimes
could now be changed without relying solely on amust now have been removed. This article reviews theclinical utility, the analytical methods and the problems patient’s description of symptoms or home urine/blood
test results.associated with using glycated haemoglobin, serumfructosamine, and glucose test strips. As well as its use in monitoring diabetic patients, it
was hoped that measurement of glycated haemoglobincould also be used as an alternative to the laborious* Correspondence to: Dr E.S. Kilpatrick, Department of Chemical
Pathology, Withington Hospital, Manchester M20 2LR, UK and inconvenient glucose tolerance test (GTT) in diag-
819CCC 0742–3071/97/100819–13$17.50 1997 by John Wiley & Sons, Ltd. DIABETIC MEDICINE, 1997; 14: 819–831
REVIEWnosis. Even following the widespread recognition of ive haemoglobins by charge-dependent techniques. In
contrast, glycation at sites other than the b-chain aminoWHO criteria for the GTT, and the undertaking ofnumerous studies, there continues to be some debate as terminus results in a compound with a charge not
dissimilar to non-glycated haemoglobin and so is indis-to the usefulness of glycated haemoglobin in diagnosisand screening. A recent review of 25 studies confirmed tinguishable by these methods. However, this portion,
which accounts for about half of all haemoglobinthat in a subject with a mildly elevated glycatedhaemoglobin level the test was still not specific or glycation, can be detected if glycation-specific ‘total
glycated haemoglobin’ methods such as affinity chroma-sensitive enough to distinguish reliably between nor-mality, impaired glucose tolerance or diabetes.12 tography are used (see below).
The charge-separated haemoglobins of normal adultFollowing the first clinical studies it was realized thatglycated haemoglobin measurement could prove to be HbA0 are jointly known as haemoglobin A1 (HbA1).
Improved techniques have allowed further separation ofuseful in answering one of the fundamental questionsin diabetes, namely, whether long-term microvascular HbA1 into its constituent parts, HbA1a1, HbA1a2, HbA1b,
and HbA1c. Glucose is the carbohydrate in the majorcomplications of the disease related to the degree ofcontrol of hyperglycaemia.13 Although early studies had fraction, HbA1c, while other carbohydrates, some of
which have still to be established with certainty, constitutesuggested such a link,14,15 the technical advances ofglycated haemoglobin, capillary blood glucose measure- the other fractions (Table 1).23.
Glycation of haemoglobin occurs continuouslyment, and intensified insulin regimes now made itpossible to conduct more meaningful trials.16 Indeed, throughout the 120-day lifetime of the red cell,24 so the
eldest cells will be most glycated and the youngestonce available, glycated haemoglobin measurementbecame the cornerstone of treatment evaluation in all least.25 However, all ages of cells will have been exposed
to recent levels of glycaemia while only the eldest cellssuch studies.17 In particular, the Diabetes Control andComplications Trial (DCCT) showed conclusively that will also have been exposed to glucose levels from 4
months previously. Therefore, the more recent the periodglycaemic control as assessed by glycated haemoglobincan predict the risk of developing microvascular diabetic of glycaemia, the larger its influence will be on the
glycated haemoglobin value. Consequently, it has beencomplications in Type I patients.18 As a consequence,the usefulness of the assay was vindicated and its suggested that half of a HbA1c value is attributable to
changes in glycaemia over the preceding month, a furtherincreased use recommended.16,17,19,20 Moreover, it wassuggested that the cost of using the test in routine clinical quarter is due to the month prior to that, with the
remaining quarter a reflection of months 3 and 4.26practice was likely to be offset by much larger savingsdue to the decreased need for procedures such aslaser photocoagulation therapy, renal dialysis, and renal Analytical Methodstransplantation.16 The continued use of glycated haemo-globin measurement in diabetes assessment would thus Methods in routine use for the measurement of glycated
haemoglobin separate the molecule on the basis of eitherseem assured.its charge, or its structure or its antigenic properties.Table 2 gives the methods and instruments that are inGlycated Haemoglobin: Structure androutine use as described by the Wales External QualityFormationAssessment Scheme (WEQAS).
The most popular means of measurement rely onHuman haemoglobin demonstrates marked heterogen-eity, mainly as a consequence of post-translational the increased negative charge found in the glycated
haemoglobin molecule to distinguish it from its non-changes due to the non-enzymatic binding of variouscarbohydrates in a process known as glycation. Glycation glycated form. These assays include electrophoresis and
ion-exchange chromatography.27 The latter method isoccurs via a carbohydrate such as glucose reacting inits free aldehyde form with a haemoglobin molecule to becoming increasingly popular with the development of
rapid (as short as 2 min between samples) dedicated highform the Schiff-base compound, aldimine.21 The formedaldimine can then meet one of two further fates: it caneither dissociate back to its component carbohydrate Table 1. Carbohydrates involved in the formation of charge-and haemoglobin or, in a process that is 60-fold less separated haemoglobins with their approximate abundance in
non-diabetic individualsrapid, it can undergo an intermolecular transformationknown as the Amadori rearrangement to form a stable
Haemoglobin Modification Abundance (%)glycated ketoamine product.Haemoglobin glycation occurs at several sites, namely,
Ao 95the amino terminal of both its a and b-chains, as wellHbA1a1 fructose-1,6-diphosphate 0.2
as certain e-amino groups.22 However, it is only the HbA1a2 glucose-6-phosphate 0.2modification at the N-terminal valine amino acid of the HbA1b carbohydrate (?) 0.5HbA1 5HbA1c glucose 4b-chain which imparts enough negative charge to thehaemoglobin molecule to allow separation of the respect-
820 E.S. KILPATRICK
Diabet. Med. 14: 819–831 (1997) 1997 by John Wiley & Sons, Ltd.
REVIEWTable 2. Glycated haemoglobin methods in routine laboratory fully in diabetic clinics.34 Its assay is based on latexuse as measured by involvement in the Welsh External immuno-agglutination inhibition methodology and usesQuality Assessment Scheme
a disposable cartridge to measure each sample individu-ally. In a situation similar to the affinity chromatographyHbA1c HbA1assays, all these immunoassay methods are calibratedwith material to which HbA1c values have been assignedIon exchange HPLC Ion exchange HPLC
Biomen HA 8121 Biomen HA 8121 using ion-exchange HPLC.Biorad Variant Corning GlycomatBiorad Diamat
Electrophoresis Factors Influencing Glycated HaemoglobinCorning Glycomat 745/765Helena RepShimadzu Measurement
Affinity chromatography Lack of StandardizationAbbott IMx
With such a wide range of instruments and methodsGamidor Primusavailable for glycated haemoglobin measurement it is
Affinity electrophoresis perhaps not surprising that instruments which measureBeckman Diatrac
different species (such as either HbA1, HbA1c or totalImmunochemical glycated haemoglobin) produce results which are not
Bayer DC 2000 directly comparable. However, due to a lack of stan-Boehringer Mannheim Tinaquant
dardization, even methods which purport to measureDako Novaclonethe same analyte can have widely differing referenceRoche Unimateintervals and give varying results with patient samples.Indeed, returns to UK external quality control schemeshave shown poor between-laboratory agreement forglycated haemoglobin, with an overall coefficient ofperformance liquid chromatography (HPLC) instruments
which are much less labour intensive than electro- variation of 20 %.35 As such, this lack of standardizationis regarded as one of the major drawbacks that limitsphoresis. These new analysers have also made feasible
the measurement of glycated haemoglobin ‘on-site’ at the clinical utility of glycated haemoglobin measure-ments,36 and both British Diabetic Association37 and thediabetic outpatient clinics,28 allowing ‘near patient’ test-
ing. National Diabetes Data Group38 see solving the problemas a major priority.Boronate affinity chromatography separates glycated
haemoglobin on the basis of its structure rather than A number of different approaches have been used toaddress the issue. European guidelines (established beforecharge. In this assay, separation occurs by the carbo-
hydrate moieties present on glycated haemoglobin bind- the DCCT report) have suggested that glycaemic controlbe classified according to how many standard deviationsing by condensation to the affinity reagent, di-hydrox-
yboronate.29 This method is specific for all glycated (SDs) a patient’s HbA1 or HbA1c lies from the non-diabetic mean value for the particular assay.39 Withinhaemoglobins irrespective of molecular charge or the
site of glycation on the haemoglobin molecule. In 3SD of the non-diabetic mean is defined as good control,between 3 and 5SD is borderline, and outwith 5SD isaddition, it is also able to detect the glycated portion of
haemoglobin in patients with haemoglobin variants such poor. However, these guidelines have a number ofdeficiencies, not least the fact that HbA1 assays classifyas HbS, HbC or HbF. Thus, the term ‘total glycated
haemoglobin’ has been used in describing this type of patients markedly different from those of HbA1c.40
Other groups have proposed alternatives such asassay. Nevertheless, despite some reservations,30 theseglycated haemoglobin values are usually expressed as comparing local laboratory values with those obtained
when the same samples are measured by the DCCT‘HbA1c equivalents’ by comparison with those obtainedusing an ion-exchange HPLC instrument. Some affinity central HbA1c laboratory.41 This pragmatic approach of
using a ‘designated comparison method’ is being activelychromatography methods have now been automated31
and made faster by the development of affinity HPLC sys- pursued by Goldstein and colleagues in the UnitedStates.42 The use of lyophilized calibrators to calibratetems.
Immunoassays have also been developed for the local instruments to produce similar values has also beenused with a degree of success in the Netherlands.43,44measurement of HbA1c. The first, the Dako Novoclone,
was based on an enzyme immunoassay performed on However, it has not been without problems. For example,lyophilized specimens were found to be incompatiblemicrotitre plates employing a monoclonal antibody
raised against HbA1c.32 More recently, high throughput with total glycated haemoglobin assays and delay in thetransport of samples between laboratories producedimmunoturbimetric methods have been developed which
can be adapted for use on a number of clinical chemistry variations in results. In addition, all these proposedschemes, with the exception of the European guidelines,analysers.33 However, such methods still require labora-
tory analysis, but a portable instrument, the DCA 2000 have the disadvantage of being time-consuming andexpensive to maintain.analyser from Bayer Diagnostics, has been used success-
821ASSESSMENT OF GLYCAEMIC CONTROL
1997 by John Wiley & Sons, Ltd. Diabet. Med. 14: 819–831 (1997)
REVIEWIt seems likely that all these suggestions will be globinopathies will fail to detect any glycated haemo-
globin and will necessitate measurements of derivativessuperseded by work being performed by the WorkingGroup on Standardization of HbA1c on behalf of the such as HbS1c or HbC1c. In heterozygous haemo-
globinopathies (e.g. HbAS, HbAC) or haemoglobin syn-International Federation of Clinical Chemistry (IFCC).45
Their approach is fundamental rather than pragmatic. thesis variants (e.g. b-thalassaemia), the same specificassays are likely to underestimate the degree of haemo-Thus, instead of comparing or calibrating equipment to
an arbitarily chosen instrument, for the first time a globin glycation by an amount dependent on the ratioof normal:abnormal haemoglobin. Examples of thesescientifically based reference system is to be established.
The reference material is to consist of highly purified specific methods include the electrophoretic, HPLC, andthe previously mentioned immunoassay methods. InHbA1c and HbA0 while the reference method consists of
measuring glycated and non-glycated peptides by a some of these systems the problem is easily identifiedbecause the abnormal haemoglobins separate to formhighly specific HPLC-mass spectrometry technique. This
system will then be used to produce calibrators for additional peaks, but in other HPLC and immunoassaytechniques these abnormal haemoglobins may be indis-routine methods. Values obtained by this system are
likely to differ from methods used in major clinical trials tinguishable from HbA0.48 The only assays able to givemeaningful measurements in these conditions are thosesuch as the DCCT, so ‘translation’ of these results will
be an important task for the IFCC group. based on affinity chromatography because of their abilityto measure glycation at any site of any haemoglobinThe full standardization of glycated haemoglobin
measurements as described is not expected until at least molecule. However, even measurement of this ‘totalglycated haemoglobin’ appears to be problematic sincethe end of the decade.45 In the meantime, proposals
such as the European guidelines are likely to continue it would seem that haemoglobins S and C glycate fasterthan native HbA0, thus leading to possible overestimationto be the only means of comparing results obtained from
different laboratories using differing methods. of the glycated haemoglobin value in the heterozygous(e.g. HbAC) state.49
HaemoglobinopathiesNormal adult haemoglobin is assembled from a-, b-, Fetal Haemoglobin
Fetal haemoglobin (HbF) has relevance to the measure-g-, and d-chains to form HbA0 (a2b2), fetal haemoglobinHbF (a2g2) and HbA2 (a2d2).24 As described above, ment of glycated haemoglobin beyond that of abnormal
haemoglobins. It is present physiologically as 75–100 %glycation of normal adult haemoglobin (HbA0) leads tothe formation of HbA1 and HbA1c. However, point of total haemoglobin in newborn children then declines
during childhood to less than 1 % in most adults.50mutations in the primary structure of the b-chain cangive rise to structural variants of haemoglobin, such as However, for reasons unknown, this decline can be
variable and incomplete, even in patients known not tothose found in HbS (b6 glu→val), HbC (b6 glu→lys), and HbE(b26 glu→lys). Also, impaired synthesis of b-chains leads have hereditary persistent HbF. The finding would be of
no consequence was it not for the fact that HbF, whetherto conditions such as b-thalassaemia and hereditarypersistent HbF. Patients with any of these haemo- glycated or not, co-migrates or co-elutes with the glycated
haemoglobin peak of many currently used electrophoreticglobinopathies (Table 3) are likely to form glycatedproducts such as HbS1c, HbC1c or HbE1c either in addition and HPLC methods.51,52 Thus, any detectable percentage
of HbF in a patient sample is likely to be included into or instead of HbA1c.46
The influence of abnormal haemoglobins on glycated the HbA1 or HbA1c result of these methods leading tofalsely elevated results.51,53,54 Moreover, compared tohaemoglobin measurement is variable depending on
the method of analysis used.47 In general, specific non-diabetic controls, increased concentrations of HbFhave been found in adult insulin-dependent and insulin-measurements of HbA1 or HbA1c in homozygous haemo-treated subjects thereby exacerbating the problem in thesepatients.55–57 Together, these findings have promptedTable 3. Factors potentially influencing glycated haemoglobin
measurements independent calls for the withdrawal of all glycatedhaemoglobin methods where fetal haemoglobin is a
Underestimation of Haemoglobinopathiesa, e.g. HbS potential interferent in measurement58,59 and, it is hoped,HbA1 or HbA1c Impaired haemoglobin synthesis e.g. have alerted clinicians and equipment manufacturers to
b-thalassaemiathe unacceptability of their continued use.Reduced red cell survival
Haemoglobin DerivativesOverestimation of HbF
Haemoglobin derivatives arise from post-translationalHbA1 or HbA1c Uraemiamodification of haemoglobin. Examples are thoseHigh-dose aspirin
Increasing age resulting from reactions with glucose (to form glycatedhaemoglobin), urea-derived isocyanate (to form carbamy-
Consult your local laboratory for details as errors are method and lated haemoglobin) and acetylsalicylic acid (to forminstrument dependent.
acetylated haemoglobin).60 As is the case for abnormalaSome haemoglobinopathies may lead to an overestimation when usingtotal glycated haemoglobin methods. haemoglobins, glycated haemoglobin methods are affec-
822 E.S. KILPATRICK
Diabet. Med. 14: 819–831 (1997) 1997 by John Wiley & Sons, Ltd.
REVIEWted differently by the latter two derivatives. Both electro- to use an alternative method of glycaemic control
assessment such as the measurement of serum fructosami-phoretic and HPLC methods show increases in glycatedhaemoglobin values with increasing degrees of uraemia ne.75
as a consequence of rising concentrations of carbamylatedhaemoglobin.61–63 For approximately every 15 mmol l−1
Serum Fructosamineof urea, glycated haemoglobin measurements increaseby 1 %.63 In contrast, affinity chromatography and Clinical Utilityenzyme immunoassay methods seem free from inter-ference.63,64 Acetylated haemoglobin affects the same
By contrast with glycated haemoglobin, serum fructosam-methods as the carbamylated form, although in patientsine measurement was quickly estabished as a usefulon chronic low doses of aspirin (200–300 mg day−1), theindicator of a patient’s glycaemic control over theeffect would appear negligible.63 However, in rheumatoidprevious 1–3 weeks. Johnson, Metcalf and Baker frompatients receiving 4 g day−1, rises in glycated haemoglobinNew Zealand were the first investigators to propose theof 1.5 % have been found when using HPLC methods.65
use of a simple colorimetric assay for the assessmentAge-related changes ‘serum glycosylprotein’ concentrations in 1982.76 They
Glycated haemoglobin values in non-diabetic individ- based their assay on the ability of serum proteinuals appear to increase with subject age,66,67 while those ketoamine linkages to glucose being able to act asof serum fructosamine and fasting plasma glucose do reducing agents in alkaline solution. These proteinnot.67 Although the cause of this remains speculative, ketoamines were generically termed ‘fructosamines’, notthe finding nevertheless has a number of important because of the involvement of fructose but because theimplications. Firstly, it means that HbA1c reference resulting compound had structural similarities to thisintervals are likely to be influenced by the age of the sugar.77 Since all serum proteins can form ketoaminessubjects chosen. This in turn adds a further complication through glycation, fructosamine was found to give ato methods (such as the European guidelines) which similar indication of glycaemic control to that foundattempt to assess glycaemic control by comparing diabetic when measuring total glycated protein and glycatedHbA1c values with those from non-diabetic subjects. albumin.78–81 However, in contrast to assaying the latter,Secondly, if, compared to younger individuals, the fructosamine could be measured by spectrophotometerglycated haemoglobin values from elderly subjects are instruments found in most clinical biochemistry labora-closer to those of diabetic patients then it may also tories. Therefore, the assay initially gave the promise ofaccount for the finding that HbA1c is a better screening being a less expensive and quicker alternative to thetest for diabetes in middle aged rather than elderly majority of glycated haemoglobin analyses methods usedpeople.68 Lastly, since it would seem appropriate that at the time.82,83 Indeed, early cross-sectional studiesdiabetic patients should have their glycated haemoglobin seemed to show good correlations between fructosaminevalues compared to those of their non-diabetic chrono- and glycated haemogobin,82–86 but subsequent studieslogical peers, then age related HbA1c reference intervals have since shown that marked discrepancies can existmay be required to facilitate more accurate glycaemic between the two measures,87–92 not least because theycontrol targets and for the better auditing of a clinic per- measure glycaemia over different time-scales. Neverthe-formance.67
less, the shorter time period measured by fructosaminehas especially lent itself to use in pregnant diabeticLabile Haemoglobinpatients93 where frequent objective monitoring is essentialMany original electrophoretic and liquid chromatogra-for the health of both mother and fetus.94,95phy methods for measuring glycated haemoglobin
Fructosamine, like glycated haemoglobin, has alsoshowed marked fluctuations in results with acute changesbeen evaluated as a tool for the diagnosis of diabetes.in blood glucose.69,70 This was due to these methodsAfter initial optimism96,97 studies have since found it tobeing unable to distinguish between HbA1/HbA1c andlack the sensitivity and specificity required of a screeningthe intermediate aldimine or Schiff base.71 This lattertest.98–102 Moreover, as a substitute for the GTT, it seemscompound, also known as the labile haemoglobin fractionto be inferior to measuring glycated haemoglobin.101,102or pre-HbA1c, has for over a decade been removed
chemically72 and should now only be of historical interest.
Fructosamine MethodsDecreased Red Cell SurvivalProcesses which lessen mean red cell life to less than
120 days will reduce the availability of haemoglobin for The fructosamine assay relies on the ability of serumAmadori compound to directly reduce nitroblue tetrazol-glycation. As a consequence, lower glycated haemoglobin
values have been recorded in patients with chronic ium (NBT) to the tetrazinolyl radical NBT+, whichdisassociates to yield a highly coloured formazan dye inrenal failure73, immune haemolytic anaemia,74 and
homozygous haemoglobinopathies.75 This phenomenon an alkaline environment.103 The reaction rate is followedby measuring the increase in dye absorbance at 530 nmis independent of the method of analysis used and
cannot be easily corrected, so often the only recourse is wavelength. While NBT is reduced, the oxidation
823ASSESSMENT OF GLYCAEMIC CONTROL
1997 by John Wiley & Sons, Ltd. Diabet. Med. 14: 819–831 (1997)
REVIEWdegeneration product of the Amadori compound would Clinical evidence would also support the fact that,
even with improvements in the assay, pathologicalappear to be d-glucosone.104
In 1991 the commercial assays from Boehringer changes in serum proteins can affect fructosamine.112
For example, patients with cirrhosis of the liver, nephroticMannheim and Roche underwent modifications byincluding a detergent and uricase in the reagent, and by syndrome113,114 and hyperthyroidism (associated with
increased protein turn-over)115 all demonstrate reducedincreasing the concentrations of both the carbonatebuffer and the dye.105 This helped reduce the effect of values. Even normoalbuminaemic Type 1 diabetic
patients have shown diurnal variation in fructosaminelipaemia and hyperuricaemia in samples as well asdiminishing protein matrix effects.106 At the same time, levels closely related with changes in their serum protein
concentration.116 However, there has been consensusthe detection wavelength of absorption changed from530 to 550 nm and glycated polylysine replaced deoxy- agreement that in patients with a serum albumin concen-
tration greater than 30 g l−1, the effect will be negligible.112morpholinofructose as the assay calibrant.107 Toaccompany these wholesale changes, the units for It is generally assumed that albumin makes the largest
contribution to total glycated protein concentration, butfructosamine measurement changed from mmol l−1 tommol l−1 with the results for the two assays not being it has also been suggested that certain proteins such as
IgA may have a greater influence on fructosamine values,directly comparable.108
presumably because they glycate at a different rate. Thisdisproportionate contribution of IgA, either in healthyFactors Influencing Serum Fructosaminesubjects117 or in patients with paraproteinaemias,118 mayMeasurement thus partly explain why the influence of serum proteinsin fructosamine measurement remains controversial.Lack of Standardization
Calibration of the fructosamine assay has proved to Interfering Substancesbe a major problem because of the lack of a trustworthy The original fructosamine assay was subject to inter-standard or calibrator. The obvious candidate, glycated ference from reducing substances other than ketoamineserum albumin, has proved to be extremely difficult to linkages, especially uric acid and ascorbic acid. Commonmanufacture to a known and reproducible amount.109 photometric interferents such as bilirubin, haemolysis,Even if feasible, storage in a matrix suitably similar to and lipaemia also interfered. Bilirubin and lipaemiahuman serum would pose further difficulties. As an caused increased fructosamine values while haemolysedalternative, the original fructosamine assay used the samples produced falsely low ones.77 The introductionless desirable deoxymorpholinofructose (DMF). Although of new fructosamine kits not only changed the calibrators,similarities between the reaction of NBT with DMF and but also addressed hyperuricaemic samples by includingwith human serum existed, the reaction kinetics and a uricase, and lipaemic samples by including surfactantabsorption spectra were markedly different. The situation detergents. Increasing the concentration of NBT in thehas since improved with the adoption of synthetically reaction mixture also reduced the error due to interferingcreated glycated polylysine as a calibrator in new reducing substances other than uric acid.109 Thus thefructosamine kits. Reaction of this compound with NBT new assay is significantly more robust than previously.produces an absorption spectrum more akin to human
Body Mass Indexserum than that found with DMF, thereby, it is hoped,An unusual but consistent finding has been observedresolving most fructosamine calibration problems.77
whereby obese diabetic patients and non-diabetic individ-uals have been found to have lower fructosamine valuesSerum Proteins
Debate continues as to whether fructosamine concen- than lean ones.119–121 Reductions of up to 25 % havebeen noted in subjects with a body mass indextration should be corrected for serum protein concen-
tration. Empirically, it would not seem surprising that a (BMI) . 30 kg m−2 despite having similar HbA1c andfasting glucose values.119,121 Serum from these patientsmeasurement of glycated proteins is apt to be influenced
by serum protein concentrations. Opponents of this view has also been found to glycate serum proteins at a slowerrate in vitro than serum from less obese individuals.121argue that, theoretically, glucose concentration is the
rate-limiting step in the glycation reaction because However, the precise mechanism for this phenomenonremains unknown.available protein lysine residues will always be vastly
in excess of reactive open-chain (carbonyl) glucosemolecules.110 However, much evidence has accumulated Monitoring of Blood Glucoseto suggest this is unlikely to be the case. An in vitrostudy has actually found that the amount of fructosamine Clinical Utilityformed is in first-order relation to albumin concentrationand not glucose.111 The authors of the study postulated Until the early 1980s, self-monitoring of glycaemic
control for most diabetic patients was limited to thethat as carbonyl glucoses were removed by glycation,more glucose molecules isomerized to the open-chain measurement of glucose concentrations in urine samples.
Many patients found this method to be insensitive,form to maintain equilibrium.
824 E.S. KILPATRICK
Diabet. Med. 14: 819–831 (1997) 1997 by John Wiley & Sons, Ltd.
REVIEWTable 4. Possible causes of inaccurate homeparticularly due to the problems associated with the renalblood glucose measurementthreshold for glucose. Indeed, at the time, adjustment of
insulin dosages according to urine tests was likened toUser errors Sample size
‘driving a car on roads with a 30 mph limit with a Sample placementspeedometer that only begins to work at greater than Test timing
Test strip storage70 mph’.122 The measurement by patients of blood ratherGlucose contamination bythan urine glucose was perceived as a means ofunwashed handssurmounting these problems.Falsification of results
Although for many years the accurate and precisemeasurement of blood glucose had been achievable in Analytical Sample haematocrit
errors Interfering substances:the laboratory setting,123 the technology for portableascorbateinstrumentation did not become available until the earlyparacetamol1970s. The first advance was in the development of ‘dry’salicylate
chemistry techniques which obviated the need for liquid uratereagents in glucose measurement. With this approach, creatinine
Sample oxygen tensionthe blood sample itself was the solvent in which theSample haemolysischemical reaction took place. The earliest demonstration
of such technology for glucose measurement occurredMagnitude of analytical errors vary depending onover 30 years ago using glucose oxidase ‘Dextrostix’ the meter used.
strips from Ames which were similar in principle tothose in use today.124,125 Reflectance meters were thenintroduced which gave the promise of more accurate and that for diagnosis, move away from laboratories into the
hands of diabetic patients and healthcare workers. Theconsistent measurement compared with visual reading.126
However, these first meters were bulky, expensive and use of meters has now become commonplace not onlyamong diabetic patients in the community but also fornot easy to use. For example, the Ames ‘Eyetone’
meter cost around £200, weighed 1.7 kg, measured the monitoring of acutely ill hospital patients. The latteris illustrated by the fact that in the author’s hospital trust,18 × 11 × 5 cm, required mains operation, a 30 minute
warm-up period, calibration before each sample, and over three ward glucose measurements are performedfor every sample sent to the biochemistry laboratory.employed a needle scale for reading.127
Despite these disadvantages, in 1978 a number of The next major advance in blood glucose monitoringis likely to be in the form of non-invasive measurementinvestigators demonstrated the potential usefulness of the
devices when used at home by diabetic patients.122,128,129 using near infra-red techniques132 but as yet no commer-cial system has been successfully developed.It was hoped that the ability to measure near physiological
blood glucose concentrations would allow more diabeticpatients to achieve this degree of control.122 It also meant Factors Influencing Blood Glucose Testthat glucose concentrations could be assessed accurately Strip Measurementwithout recourse to frequent hospital admission,especially in pregnant diabetic patients.128,130 Lack of Standardization
While the use of whole blood in a clinical setting hasNevertheless, it was recognized that the potentialmarket for meters would only be realized if they became the advantage of being convenient, whole blood glucose
measurement, unlike plasma glucose, lacks a definedless expensive and genuinely portable. Thus, therefollowed a period of steady improvement in the products standard against which other methods can be cali-
brated.133 However, the Yellow Springs Instrument (YSI)of all manufacturers in terms of ease of use and cost. In1987, the ExacTech meter from Medisense was particulary glucose analyser has become a de facto standard used
by most glucose meter manufacturers to calibrate theirgroundbreaking. It used a pen sized meter which, insteadof detecting a colour change in a reagent strip, employed instruments. The choice of the YSI owes more to it being
one of the first widely available whole blood glucosea disposable biosensor for the electrochemical detectionof blood glucose.131 It was also in the vanguard of instruments than to any analytical superiority. Indeed,
because the YSI measures blood glucose indirectly (i.e.introducing products which did not require the user towipe blood from the test strip during measurement. after dilution of the sample), results are liable to be
affected by deviations in plasma protein concentrations,Most new glucose meters are now compact and ‘non-wipe’ and several also include features such as automatic high lipid levels, and the haematocrit of the blood
specimens.133 Most reflectance meters in fact measuretiming of the test, time/date stamping of results, and therecording of these results in the meter memory. Figure 1 plasma glucose because the red cells are retarded from
entering the reagent layer of the test strip, but becauseillustrates a variety of new technology instrumentscurrently available in the UK. the meter is calibrated against the YSI it gives a whole
blood equivalent result. Boehringer Mannheim haveThe ensuing success of glucose test strips and metershas seen the measurement of blood glucose, other than broken with tradition by calibrating their Advantage
825ASSESSMENT OF GLYCAEMIC CONTROL
1997 by John Wiley & Sons, Ltd. Diabet. Med. 14: 819–831 (1997)
REVIEW
Figure 1. A selection of ‘new technology’ blood glucose meters. Top (L to R): Boehringer Mannheim Advantage; Bayer Glucometer4; Lifescan One Touch II. Bottom: Medisense meter.
meter against a whole blood hexokinase reference fabricated results, in 8 % results had been omitted andin 26 % there was disagreement between the meter resultinstrument rather than the YSI. However, it means results
from this meter tend to be more comparable to plasma and that written in the diary.140 The authors concludedthat about a third of patients could not be relied onvalues (i.e. higher) than if calibrated to the Yellow
Springs Instrument. Thus, the lack of an agreed standard, to keep unambiguous diaries, a figure confirmed insubsequent studies.141and the limitations of the adopted YSI instrument means
that whole blood glucose measurements cannot be asAnalytical Errors
universally concordant as that of plasma.Analytical compromises are unavoidable when using
a glucose meter instrument costing £30–50 and test stripsUser ErrorsAlthough newer glucose meters have helped address priced at £0.25 each. Indeed, it is perhaps surprising
how accurate and precise these meters can be consideringspecific problems, many errors in test strip glucosemeasurement are still attributable to the user. These such cost restraints. Nevertheless, some of the compro-
mises present can lead to clinically important errors ininclude errors due to the size and placement of theblood sample, the timing of the test and wiping of blood glucose measurement, especially under the wide range
of physiological conditions commonly found in hospitalfrom the test strip.134 Any other conceivable mistake inmeasurement such as the use of glucose-contaminated patients. Errors include those listed below and summar-
ized in Table 4.hands or inappropriate storage of test strips is sure tobecome apparent when any group of diabetic patients Sample Haematocrit. Variations in blood samples haema-
tocrit have been shown to be a source of inaccuracy inor health care staff are allowed use of the analysis.As reviewed previously in Diabetic Medicine,135 visual many blood glucose systems.142–145 In general, blood
samples with low haematocrits can give spuriously highreading of test strips adds another source of error whichcan considerably reduce the accuracy of measurements glucose results and vice versa. This effect can be well
in excess of that found when using instruments such asin general use.134 Patients tend to cluster their resultsaround the standard colours on the container, despite the YSI analyser. For example, for every 10 % change
in sample haematocrit the glucose concentration meas-instructions to interpolate.136 Colour blindness andimpaired colour discrimination, either inherited or ured by Bayer Glucostix can vary by approximately
20 %.142 Newer instruments such as the Bayer Gluco-acquired as a result of diabetes,137 can also lead todifficulty in strip interpretation.138,139 meter 4 and Boehringer Mannheim Accutrend have
effectively addressed this problem, but other contempor-Another important source of user ‘error’ is the falsifi-cation by patients of results obtained by glucose self- ary instruments, such as the Boehringer Mannheim
Advantage, the Medisense meter, and (to a lesser extent)monitoring. In the first study to highlight the problem,19 diabetic patients were given meters which had the Lifescan One Touch II still remain affected.146–150
Interfering Substances. When the glucose oxidase(unknown to them) memory chips built in. It was foundthat in 30 % of glucose record diaries the patient had enzyme is used for glucose measurement, reducing
826 E.S. KILPATRICK
Diabet. Med. 14: 819–831 (1997) 1997 by John Wiley & Sons, Ltd.
REVIEWagents found in the blood sample can interfere by a and test strip blood glucose measurement, many clinically
relevant problems remain when assessing glycaemiccompetitive effect on the redox detection system.151
Hence, exogenously given substances such as ascorbic control by these routinely used means. Thus, althoughtechnological advances have led to improvements,acid, paracetamol and salicylate, as well as large amounts
of endogenous uric acid, urea and creatinine can lead interpretation of these indices cannot yet be used as asole substitute for clinical judgement.to falsely low results with many systems.151 Recent
products have approached this problem by using teststrips with alternative enzymes to the glucose oxidase
Referencessystems. For example, the Glucometer 4 meter fromBayer uses hexokinase while the Advantage meter from
1. Winocour PH, Bhatnager D, Kalsi P, Hillier VF, AndersonBoehringer Mannheim uses glucose dehydrogenase. Both DC. Relative clinical usefulness of glycosylated serumthese enzymes show little influence from chemical albumin and fructosamine during short-term changes ininterference. An alternative solution used in the Medi- glycemic control in IDDM. Diabetes Care 1989; 12:
665–672.sense meter has been to continue to use glucose oxidase,2. Frattali AL, Wolf BA. 1,5-anhydroglucitol: a novel serumbut to correct for interference by using an additional
marker for screening and monitoring diabetes mellitus?‘blanking’ electrode. Clin Chem 1994; 40: 1991–1993.Sample Oxygen Tension. In test strips which use glucose 3. Huisman THJ, Dozy AM. Studies on the heterogeneityoxidase, the substrates of the enzyme are glucose, of hemoglobin. V. Binding of hemoglobin with oxidised
glutathione. J Lab Clin Med 1962: 60: 302–319.molecular oxygen, and water. Thus, it is not particularly4. Rahbar S. An abnormal hemoglobin in red cells ofsurprising that changes in sample and atmospheric
diabetics. Clin Chim Acta 1968; 22: 296–298.oxygen tension have been found to affect some of 5. Rahbar S, Blumenfeld O, Ranney HM. Studies of anthese glucose meter systems. The ExacTech meter from unusual hemoglobin in patients with diabetes mellitus.Medisense is especially influenced such that it may give Biochem Biophys Res Comm 1969; 36: 838–843.
6. Schnek AG, Schroeder WA. The relation betweeninappropriately elevated results in hypoxic patients whilethe minor components of whole normal human adultrecording reduced values in patients on oxygen treat-hemoglobin as isolated by chromatography and starchment.152 Other manufacturers have circumvented the block electrophoresis. J Am Chem Soc 1961; 83:
problem by using detection enzymes which do not 1472–1478.require molecular oxygen, while Medisense itself seems 7. Gabbay KH, Hasty K, Breslow JL, Ellison RC, Bunn HF,
Gallop PM. Glycosylated hemoglobins and long-termto have effectively resolved the issue in its newer meterblood glucose control in diabetes mellitus. J Clinwith their use of a 3 electrode test strip.153
Endocrinol Metab 1977; 44: 859–864.Sample Haemolysis. Extra-laboratory measurement of 8. Koenig RJ, Peterson CM, Jones RL, Saudek C, Lehrmananalytes such as blood glucose has led to an increase M, Cerami A. Correlation of glucose regulation andin the use of whole blood rather than serum or haemoglobin A1 in diabetes mellitus. N Engl J Med
1976; 295: 417–420.plasma as a specimen for biochemical measurement.9. Gonen B, Rubenstein AH, Rochman H, Tanega SP,Unfortunately, unlike plasma or serum, it is not possible
Horwitz DL. Haemoglobin A1: an indicator of theto tell if the sample being used is in any way haemolysed. metabolic control of diabetic patients. Lancet 1977; ii:A recent study has shown that most blood glucose meters 734–737.can give spurious results when extremely haemolysed 10. Koenig RJ, Peterson CM, Kilo C, Cerami A, Williamson
JR. Haemoglobin A1c as an indicator of the degree ofsamples are used.154 However, only the Boehringerglucose intolerance in diabetes. Diabetes 1976; 25:Mannheim Accutrend has been found to give inaccurate230–232.results at more modest degrees of haemolysis. For every 11. Goldstein D, Little R, Wiedmeyer H, England J, McKenzie
7 % of red cells lysed, Accutrend results increased E. Glycated haemoglobin: methodologies and clinicalby 15 %. applications. Clin Chem 1986; 32 (suppl): B64–B70.
12. Davidson MB, Schriger DL, Peters AL. An alternativeVisual Confirmation of Blood Glucose. Meter measure-approach to the diagnosis of diabetes with a review ofment of glucose test strips has been replacing visualthe literature. Diabetes Care 1995; 18: 1065–1071.reading mainly because of its ease of use and potential 13. Bunn HF. Evaluation of glycosylated hemoglobin in
for more accurate results. However, case reports have diabetic patients. Diabetes 1981; 30: 613–617.documented examples where extremely high glucose 14. Johnsson S. Retinopathy and nephropathy in diabetes
mellitus: comparison of the effects of two forms ofvalues have given considerably lower results on thetreatment. Diabetes 1960; 9: 1–8.Boehringer Mannheim Reflolux S meter while manual
15. Pirart T. Diabetes mellitus and its degenerative compli-reading of the strip has confirmed the gross hypergly- cations: a prospective study of 4400 patients observedcaemia.155,156 With more meters being introduced which between 1947 and 1973. Diabetes Care 1978; 1:do not have any form of visual backup, there may be 168–188.
16. Goldstein DE, Little RR, Wiedmeyer HM, England JD,more scope for such errors to pass unnoticed.Rohlfing CL, Wilke AL. Is glycohemoglobin testing usefulin diabetes mellitus? Lessons from the Diabetes ControlConclusionsand Complications Trial. Clin Chem 1994; 40: 1637–
This article has shown that despite the undoubted 1640.17. Dahl-Jørgensen K, Brinchmann-Hansen O, Bangstad HJ,advantages of glycated haemoglobin, serum fructosamine
827ASSESSMENT OF GLYCAEMIC CONTROL
1997 by John Wiley & Sons, Ltd. Diabet. Med. 14: 819–831 (1997)
REVIEWHanssen KF. Blood glucose control and microvascular values: standardisation is essential (Letter). Br Med J
1994; 309: 1511.complications—what do we do now? Diabetologia 1994;37: 1172–1177. 38. Baynes J, Bunn H, Goldstein D. National Diabetes Data
Group: report of the expert committee on glycosylated18. The Diabetes Control and Complications Trial ResearchGroup. The effect of intensive treatment of diabetes hemoglobin. Diabetes Care 1984; 7: 605–615.
39. European IDDM Policy Group. Consensus guidelines foron the development and progression of long-termcomplications in insulin-dependent diabetes mellitus. N the management of insulin-dependent (Type 1) diabetes.
Diabetic Med 1993; 10: 990–1005.Engl J Med 1993; 329: 977–986.19. Santiago JV. Lessons from the Diabetes Control and 40. Kilpatrick ES, Rumley AG, Dominiczak MH, Small
M. Glycated haemoglobin values: problems in theComplications Trial. Diabetes 1993; 42: 1549–1554.20. American Diabetes Association. Implications of the assessment of blood glucose control in diabetes mellitus.
Br Med J 1994; 309: 983–986.Diabetes Control and Complications Trial. Diabetes1993; 42: 1555–1558. 41. Marshall SM, Parnham AJ, Gibb I, Lord C, Bucksa J,
Steffes MW. Is it possible to standardize glycated21. Koenig RJ, Blobstein SH, Cerami A. Structure of carbo-hydrate of hemoglobin A1c. J Biol Chem 1977; 252: haemoglobin assays using standard deviation scores?
Diabetic Med 1994; 11(suppl 1): S40.2992–2997.22. Bunn HF, Shapiro R, McManus M, Garrick L, McDonald 42. Goldstein DE, Little RR, Lorenz RA, Malone JI, Nathan
D, Peterson CM. Tests of glycemia in diabetes. DiabetesMJ, Gallop PM, et al. Structural heterogeneity of humanhemoglobin A due to nonenzymatic glycosylation. J Biol Care 1995; 18: 896–909.
43. Weykamp CW, Penders TJ, Muskiet FAJ, van der SlikChem 1979; 254: 3892–3898.23. McDonald MJ, Shapiro R, Bleichman M, Solway J, Bunn W. Effect of calibration on dispersion of glycohemoglobin
values determined by 111 laboratories using 21 methods.HF. Glycosylated minor components of human adulthemoglobin. J Biol Chem 1978; 253: 2327–2332. Clin Chem 1994; 40: 138–144.
44. Weykamp CW, Penders TJ, Miedema K, Muskiet FAJ,24. Bunn HF, Haney DN, Kamin S, Gabbay KH, GallopPM. The biosynthesis of human hemoglobin A1c. Slow van der Slik W. Standardization of glycohemoglobin
results and reference values in whole blood studied inglycosylation of hemoglobin in vivo. J Clin Invest 1976;57: 1652–1659. 103 laboratories using 20 methods. Clin Chem 1995;
41: 82–86.25. Fitzgibbons JF, Koler RD, Jones RT. Red cell age-relatedchanges in hemoglobin A1a+1b and A1c in normal and 45. Hoelzel W, Miedema K. Development of a reference
system for the international standardization ofdiabetic subjects. J Clin Invest 1976; 58: 820–824.26. Tahara Y, Shima K. The response of GHb to stepwise HbA1c/glycohemoglobin determinations. J Int Fed Clin
Chem 1996; 9: 62–67.plasma glucose change over time in diabetic patients(Letter). Diabetes Care 1993; 16: 1313–1314. 46. Aleyassine H. Glycosylation of hemoglobin S and
hemoglobin C. Clin Chem 1980; 26: 526–527.27. Menard L, Dempsey ME, Blankstein LA, Aleyassihe H,Wacks M, Soeldner JS. Quantitiative determination of 47. Weykamp CW, Penders TJ, Muskiet FAJ, van der Slik
W. Influence of hemoglobin variants and derivatives onglycosylated haemoglobin A1 by agar gel electrophoresis.Clin Chem 1980; 26: 1598–1602. glycohemoglobin determinations, as investigated by 102
laboratories using 16 methods. Clin Chem 1993; 39:28. Rumley AG, Carlton G, Small M. Within-clinic glycosyl-ated haemoglobin measurement. Diabetic Med 1990; 7: 1717–1723.
48. Kilpatrick ES. Assessing glucose control in diabetes:838–840.29. Mallia AK, Hermanson GT, Krohn RI, Fujimoto EK, proceed with caution. Spectrum International 1995; 35:
37–39.Smith PK. Preparation and use of a boronic acid affinitysupport for separation and quantitation of glycosylated 49. Goujon R, Thiivolet C. Glycation of hemoglobin C in
the heterozygous state in diabetic patients (Letter).hemoglobins. Anal Lett 1981; 14: 649–661.30. Turpeinen U, Karjalainen U, Stenman U. Three assays Diabetes Care 1994; 17: 247.
50. Rutland PC, Pembury ME, Davies T. The estimation offor glycohemoglobin compared. Clin Chem 1995; 41:191–195. fetal haemoglobin in healthy adults by radioimmunoas-
say. Br J Haematol 1983; 53: 673–682.31. Wilson DH, Bogacz JP, Forsythe CM, Turk PJ, Lane TL,Gates RC, et al. Fully automated assay of glycohemoglo- 51. Yatscoff RW, Tevaarwerk JM, Clarson CL, Warnock LM.
Interference of fetal haemoglobin with the measurementsbin with the Abbott IMx analyser: novel approachesfor separation and detection. Clin Chem 1993; 39: of glycosylated hemoglobin. Clin Chem 1983; 29:
543–545.2090–2097.32. John WG, Bullock DG, MacKenzie F. Methods for 52. Thivolet C, Goujon R, Vassy V, Revol A, Tourniaire J.
Interference of elevated fetal hemoglobin on HbA1cthe analysis of glycated haemoglobins: what is beingmeasured? Diabetic Med 1992; 9: 15–19. measurements in adult type I diabetic patients. Diabetes
Care 1991; 14: 1108.33. John WG. Hemoglobin A1c measurement: new preciseimmunoassay method involving latex particle aggluti- 53. Krause JR, Stolc V, Campbell E. The effect of hemoglobin
F upon glycosylated hemoglobin determinations. Am Jnation. Clin Chem 1996; 42: 1874–1875.34. Rumley AG, Kilpatrick ES, Dominiczak MH, Small M. Clin Path 1982; 78: 767–769.
54. Paisey RB, Read R, Palmer R, Hartog M. Persistent fetalEvaluation of glycaemic control limits using the AmesDCA 2000 HbA1c analyser. Diabetic Med 1993; 10: haemoglobin and falsely high glycosylated haemoglobin
levels. Br Med J 1984; 289: 279–280.976–979.35. John WG. Performance of glycated hemoglobin analysis: 55. Kilpatrick ES, Rumley AG, Small M, Dominiczak MH.
Increased fetal haemoglobin in insulin-treated diabetesdata from the UK national external quality assessmentscheme (Abstract). Clin Chem 1990; 36: 1002. mellitus contributes to the imprecision of glycohaemoglo-
bin measurements. Clin Chem 1993; 39: 833–835.36. Bruns DE. Standardisation, calibration, and the care ofdiabetic patients. Clin Chem 1992; 38: 2363–2364. 56. Diem P, Mullis P, Hirt A, Schuler JJ, Burgi W, Zuppinger
KA, Teuscher A. Fetal hemoglobin levels in adult type I37. Alexander WD, Paterson KR. Glycated haemoglobin
828 E.S. KILPATRICK
Diabet. Med. 14: 819–831 (1997) 1997 by John Wiley & Sons, Ltd.
REVIEW(insulin-dependent) diabetic patients. Diabetologia 1993; approach to the estimation of serum glycosylprotein. An36: 129–132. index of diabetic control. Clin Chem Acta 1982; 127:
57. Koskinen LK, Lahtela JT, Koivula TA. Fetal hemoglobin 87–95.in diabetic patients Diabetes Care 1994; 17: 828–831. 77. Koch DD. Fructosamine: how useful is it? Lab Med
58. Goldstein DE, Little RR. More than you ever wanted to 1990; 21: 497–503.know (but need to know) about glycohemoglobin testing. 78. Zoppi F, Mosca A, Granata S, Montalbetti N. GlycatedDiabetes Care 1994; 17: 938–939. proteins in serum: effect of their relative proportions on
59. Bulusu S. Assessing blood glucose in diabetes mellitus: their alkaline reducing activity in the fructosamine test.variant haemoglobin may affect measurements (Letter). Clin Chem 1987; 33: 1895–1897.Br Med J 1995; 310: 740. 79. Seng LY, Staley MJ. Plasma fructosamine is a measure
60. Ladenson JH, Chan KM, Kilzer P. Glycsoylated hemoglo- of all glycated proteins. Clin Chem 1986; 32: 560.bin and diabetes: a case and an overview of the subject. 80. Mosca A, Carenini A, Zoppi F. Plasma protein glycationClin Chem 1985; 31: 1060–1067. as measured by fructosamine assay. Clin Chem 1987;
61. Paisey R, Banks R, Holton R, Young K, Hopton M, 33: 1141–1146.White D, Hartog M. Glycosylated haemoglobin and 81. Johnson RN, Metcalf PA, Baker JR. Relationship betweenuraemia. Diabetic Med 1986; 3: 445–448. albumin and fructosamine in diabetic and non-diabetic
62. Lamb E, Dawnay A. Glycated haemoglobin measurement sera. Clin Chem Acta 1987; 164: 151–162.in uraemic patients. Ann Clin Biochem 1992; 29: 82. Hindle EJ, Rostron GM, Gatt JA. The estimation of serum118–120. fructosamine: an alternative measurement to glycated
63. Weykamp CW, Penders TJ, Siebelder CWM, Muskiet haemoglobin. Ann Clin Biochem 1985; 22: 84–89.FAJ, van der Slik W. Interference of carbamylated and 83. Pandya HC, Livingstone S, Colgan ME, Percy-Robb IW,acetylated hemoglobins in assays of glycohemoglobin Frier BM. Serum fructosamine as an index of glycaemia:by HPLC, electrophoresis, affinity chromatography, and comparison with glycated haemoglobin in diabetic andenzyme immunoassay. Clin Chem 1993; 39: 138–142. non-diabetic individuals. Practical Diabetes 1987; 4:
64. Engbaek F, Christensen SE, Jespersen B. Enzyme immu- 126–128.noassay of hemoglobin A1c: analytical characteristics 84. Cefalu W, Parker TB, Johnson CR. Validity of serumand clinical performance for patients with diabetes fructosamine as an index of short-term glycemic controlmellitus, with and without uremia. Clin Chem 1989; 35: in diabetic outpatients. Diabetes Care 1988; 11: 662–664.93–97. 85. Lloyd DR, Nott M, Marples J. Comparison of serum
65. Nathan DM, Francis TB, Palmer JL. Effect of aspirin on frcutosamine with glycosylated serum proteindeterminations of glycosylated hemoglobin. Clin Chem (determined by affinity chromatography) for the assess-1983; 29: 466–469. ment of diabetic control. Diabetic Med 1985; 2: 474–478.
66. Simon D, Senan C, Garnier P, Saint-Paul M, Papoz L. 86. Baker JR, Metcalf PA, Holdaway IM, Johnson RN. SerumEpidemiological features of glycated haemogobin A1c— fructosamine concentration as a measure of blooddistribution in a healthy population. Diabetologia 1989; glucose control in type I (insulin-dependent) diabetes32: 864–869. mellitus. Br Med J 1985; 290: 352–355.
67. Kilpatrick ES, Dominiczak MH, Small M. The effects of 87. Dominiczak MH, MacRury SM, Orrell JM, Paterson KR.ageing on glycation and the interpretation of glycaemic Long-term performance of the fructosamine assay. Anncontrol in Type 2 diabetes. Q J Med 1996; 89: 307–312. Clin Biochem 1988; 25: 627–633.
68. Tsukiu S, Kobayashi I. Effect of age and obesity on 88. Dominiczak MH, Smith LA, McNaught J, Patersonglycated haemoglobin and 1,5-anhydroglucitol in screen- KR. Assessment of past glycaemic control: measureing for Type 2 diabetes mellitus. Diabetic Med 1995; fructosamine, haemoglobin A1 or both? Diabetes Care12: 899–903. 1988; 11: 359–360.69. Goldstein DE, Peth SB, England JD, Hess RL, Da Costa 89. Fisken RA, Chan AW, Hanlon A, MacFarlane IA.J. Effects of acute changes in blood glucose on HbA1c. Longitudinal changes in serum fructosamine do notDiabetes 1980; 29: 623–628.
parallel those in glycated haemoglobin in young adults70. Svendsen PA, Christiansen JS, Søegaard U, Welinderwith insulin-dependent diabetes. Clin Chim Acta 1990;BS, Nerup J. Rapid changes in chromatographically191: 79–86.determined haemoglobin A1c induced by short-term
90. Watts GF, Macleod AF, Benn JJ, Slavin BM, Morris RW,changes in glucose concentration. Diabetologia 1980;Williams CD, et al. Comparison of the real-time use of19: 130–136.glycosylated haemoglobin and plasma fructosamine in71. Nathan DM. Labile glycosylated hemoglobin contributesthe diabetic clinic. Diabetic Med 1991; 8: 573–579.to haemoglobin A1 as measured by liquid-chromatogra-
91. MacRury SM, Kilpatrick ES, Paterson KR, Dominiczakphy or electrophoresis. Clin Chem 1981; 27: 1261–1263.MH. Serum fructosamine/HbA1 ratio predicts the future72. Nathan DM, Avezzano E, Palmer JL. Rapid method forchanges in haemoglobin A1 in type II (non-insulineliminating labile glycosylated hemoglobin from thedependent) diabetic patients. Clinica Chimica Acta 1991;assay for hemogobin A1. Clin Chem 1982; 28: 512–515.199: 43–51.73. Dandona PD, Freedman D, Moorheah JF. Glycosylated
92. Shield JPH, Poyser I, Hunt L, Pennock CA. Fructosaminehaemoglobin in chronic renal failure. Br Med J 1979; i:and glycated haemoglobin in the assessment of long1183–1184.term glycaemic control in diabetes. Arch Dis Child74. Panzer S, Kronik G, Lechner K, Bettelheim P, Neumann1994; 71: 443–445.E, Dudczak R. Glycosylated hemoglobins (GHb): an
93. Frandsen EK, Sabagh T, Bacchus RA. Serum fructosmaineindex of red cell survival. Blood 1982; 59: 1348–1350.in diabetic pregnancy. Clin Chem 1988; 34: 316–319.75. Martina WV, Martijn EG, van der Molen M, Schermer
94. Steel JM, Johnstone FD, Hepburn DA, Smith AF. CanJG, Muskiet FAJ. b-N-Terminal glycohemoglobins inprepregnancy care of diabetic women reduce the risksubjects with common hemoglobinopathies: relationof abnormal babies? Br Med J 1990; 301: 1070–1074.with fructosamine and mean erythrocyte age. Clin Chem
95. Kitzmiller JL, Gavin LA, Gin GD, Jovanovic-Peterson L,1993; 39: 2259–2265.76. Johnson RN, Metcalf PA, Baker JR. Fructosamine: a new Main EK, Zigrang WD. Preconception care of diabetes:
829ASSESSMENT OF GLYCAEMIC CONTROL
1997 by John Wiley & Sons, Ltd. Diabet. Med. 14: 819–831 (1997)
REVIEWglycemic control prevents congenital anomalies. J Am minaemia and hypergammaglobulinaemia. Ann Clin
Biochem 1992; 29: 437–442.Med Assoc 1991; 265: 731–736.96. Baker JR, O’Connor JP, Metcalf PA, Lawson MR, 115. Ford HC, Lim WC, Crooke MJ. Hemoglobin A1 and
serum fructosamine levels in hyperthyroidism. Clin ChimJohnson RN. Clinical usefulness of estimation of serumfructosamine concentration as a screening test for Acta 1987; 166: 317–321.
116. Fluckiger R, Woodtli T, Berger W. Evaluation of thediabetes mellitus. Br Med J 1983; 287: 863–867.97. Salmans THB, van Dieijen-Visser MP, Brombacker PJ. fructosamine test for the measurement of plasma protein
glycation. Diabetologia 1987; 30: 648–652.The value of HbA1 and fructosamine in predictingimpaired glucose tolerance—an alternative to OGTT to 117. Rodriguez-Segade S, Lojo S, Camina MF, Paz JM, Del
Rio R. Effects of various serum proteins on quantificationdetect diabetes mellitus or gestational diabetes. Ann ClinBiochem 1987; 24: 447–452. of fructosamine. Clin Chem 1989; 35: 134–138.
118. Handley G, Johnson PG, Peel D. Raised serum fructos-98. Swai ABM, Harrison K, Chuwa LM, Makene W, McLartyD, Alberti KGMM. Screening for diabetes: does measure- maine concentration caused by IgA paraproteinaemia.
Ann Clin Biochem 1993; 30: 106–107.ment of serum fructosamine help? Diabetic Med 1988;5: 648–652. 119. Skrha J, Svacina S. Serum fructosmaine and obesity
(Letter). Clin Chem 1991; 37: 2020–2021.99. Begley JP, Stafford KA, Clark JDA. Fructosamine as ascreening test for diabetes mellitus. Practical Diabetes 120. Woo J, Cockram C, Lau E, Chan A, Swaminathan R.
Influence of obesity on plasma fructosamine concen-1992; 9: 104–105.100. Guillausseau PJ, Charles MA, Paolaggi F, Timsit J, tration. Clin Chem 1992; 38: 2190–2192.
121. Ardawi MSM, Nasrat HAN, Bahnassy AA. FructosamineChanso P, Peynet J, et al. Comparison of HbA1 andfructosamine in diagnosis of glucose-tolerance abnor- in obese normal subjects and type 2 diabetes. Diabetic
Med 1994; 11: 50–56.malities. Diabetes Care 1990; 13: 898–900.101. Shima K, Abe F, Chikakiyo H, Ito N. The relative value 122. Sonksen PH, Judd SL, Lowy C. Home monitoring of
blood glucose—method for improving diabetic control.of glycated albumin, hemoglobin A1c and fructosaminewhen screening for diabetes mellitus. Diab Res Clin Lancet 1978; i: 729–732.
123. Burrin JM, Price CP. Measurement of blood glucose.Prac 1989; 7: 243–250.102. Tsuji I, Nakamoto K, Hasegawa T, Hisashige A, Inawa- Ann Clin Biochem 1985; 22: 327–342.
124. Cohen SL, Legg S, Bird R. A bedside method of blood-shiro H, Fukao A, et al. Receiver operating characteristicanalysis on fasting plasma glucose, HbA1c and fructosam- glucose determination. Lancet 1964; ii: 883–884.
125. Rennie IDB, Keen H, Southon A. A rapid enzyme-stripine on diabetes screening. Diabetes Care 1991; 14;1075–1077. method for estimating blood-sugar. Lancet 1964; ii:
884–886.103. Baker JR, Zyzak DV, Thorpe SR, Baynes JW. Mechanismof the fructosamine assay: evidence against a role for 126. Jarrett RJ, Keen H, Hardwick C. Instant blood sugar
measurement using dextrostix and a reflectance meter.superoxide as an intermediate in reduction of nitrobluetetrazolium. Clin Chem 1993; 39: 2460–2465. Diabetes 1970; 19: 724–726.
127. Skyler JS. Self-monitoring of blood glucose. Med Clin N104. Baker JR, Zyzak DV, Thorpe SR, Baynes JW. Chemistryof the fructosamine assay: d-glucosone is the product Am 1982; 66: 1227–1250.
128. Walford S, Gale EA, Allison SP, Tattersall RB. Self-of oxidation of Amadori compounds. Clin Chem 1994;40: 1950–1955. monitoring of blood glucose—improvement of diabetic
control. Lancet 1978; i: 732–735.105. Vogt BW, Ziegenhorn J. Improved colorimetric fructosa-mine assay (Abstract). Clin Chem 1989; 35: 1110. 129. Ikeda Y, Tajima N, Minami N, Ide Y, Yokoyama J, Abe
M. Pilot study of self measurement of blood glucose106. Baker JR, Metcalf PA, Scragg P, Johnson RN. Fructosam-ine Test-Plus: a modified fructosamine assay evaluated. using the Dextrostix-Eyetone system for juvenile onset
diabetes. Diabetologia 1978; 15: 91–93.Clin Chem 1991; 37: 552–556.107. Schleicher ED, Vogt BW. Standardization of serum 130. Peacock I, Hunter JC, Walford S, Allison SP, Davison J,
Clarke P, et al. Self-monitoring of blood glucose infructosamine assays. Clin Chem 1990; 36: 136–139.108. Cefalu WT, Bell-Farrow AD, Petty M, Izlar C, Smith JA. diabetic pregnancy. Br Med J 1979; 280: 1333–1336.
131. Matthews DR, Holman RR, Brown E, Steemson J, WatsonClinical validation of a second-generation fructosamineassay. Clin Chem 1991; 37: 1252–1256. A, Hughes S, et al. Pen-sized digital 30-second blood
glucose meter. Lancet 1987; i: 778–779.109. Hegstrand LR, Miller RC, Koch DD. Fructosamine:modified NBT assay to improve clinical utility. Am J 132. Hall JW, Pollard A. Near-infrared spectrophotometry: a
new dimension in clinical chemistry. Clin Chem 1992;Clin Path 1990; 93: 446.110. Staley MJ. Fructosmine and protein concentrations in 38: 1623–1631.
133. Wiener K. Whole blood glucose: what are we actuallyserum. Clin Chem 1987; 33: 2326–2327.111. Lamb E, Mainwaring-Burton R, Dawnay A. Effect of measuring? Ann Clin Biochem 1995; 32: 1–8.
134. Consensus Statement. Self-monitoring of blood glucose.protein concentration on the formation of glycatedalbumin and fructosamine. Clin Chem 1991; 37: Diabetes Care 1993; 16(suppl 2): 60–65.
135. Page SR, Peacock I. Blood glucose monitoring: does2138–2139.112. Henrichs HR, Lehmann P, Vorberg E. An improved technology help? Diabetic Med 1993; 10: 793–801.
136. Wing RR, Lamparski D, Zaslow S, Betschart J, Simineriofructosamine assay for monitoring blood glucose control.Diabetic Med 1991; 8: 580–584. J, Becker D. Frequency and accuracy of self monitoring
of blood glucose in children: relationship to glycemic113. Dominiczak MH, Orrell JM, Finlay WEI. The effectof hypoalbuminaemia, hyperbilirubinaemia and renal control. Diabetes Care 1985; 8: 214–218.
137. Daley ML, Watzke RC, Riddle MC. Early loss of bluefailure on serum fructosamine concentration in nondia-betic individuals. Clin Chim Acta 1989; 182: 123–130. sensitive color vision in patients with type I diabetes.
Diabetes Care 1987; 10: 777–781.114. Constanti C, Simo JM, Joven J, Camps J. Serum fructosam-ine concentration in patients with nephrotic syndrome 138. Graham K, Kesson CM, Kennedy HB, Ireland JT.
Relevance of colour vision and diabetic retinopathy toand with cirrhosis of the liver: the influence of hypoalbu-
830 E.S. KILPATRICK
Diabet. Med. 14: 819–831 (1997) 1997 by John Wiley & Sons, Ltd.
REVIEWself monitoring of blood glucose. Br Med J 1980; 281: 148. Kilpatrick ES, Brear GS. The Advantage blood glucose971–973. meter in an intensive care environment (Abstract).
139. Allwood MC, Tyler R. Colour vision and blood glucose Diabetic Med 1996; 13(Suppl 7): S9.self-monitoring in diabetics. Practical Diabetes 1988; 5: 149. Barlow IM, Jones E. Evaluation of the Medisense110–112. Companion 2 glucose meter (Abstract). In Martin SM,
140. Mazze RS, Shamoon H, Pasmantier R, Lucido D, Murphy Halloran SP, Green AJE, eds. Proceedings of the ACBJ, Hartmann K, et al. Reliability of blood glucose National Meeting: 1994, 103. PRC Associates, Cam-monitoring by patients with diabetes mellitus. Am J Med bridge.1984; 77: 211–217. 150. Kilpatrick ES, Rumley AG, Small M. Evaluation of the
141. Gonder-Frederick LA, Julian DM, Cox DJ, Clarke WL, One Touch II blood glucose meter (Abstract). In: MartinCarter WR. Self-measurement of blood glucose. Accuracy SM, Halloran SP, Green AJE, eds. Proceedings of the ACBof self-reported data and adherence to recommended National Meeting. 1993: 89. PRC Associates, Cambridge.regimens. Diabetes Care 1988; 11: 579–585. 151. Sylvester ECJ, Price CP, Burrin JM. Investigation of the
142. Barreau PB, Buttery JE. Effect of hematocrit concentration potential for interference with whole blood glucoseon blood glucose values determined on Glucometer II. strips. Ann Clin Biochem 1994; 31: 94–96.Diabetes Care 1988; 11: 116–118. 152. Kilpatrick ES, Rumley AG, Smith EA. Variations in
143. Wiener K. The effect of haematocrit on reagent strip sample pH and pO2 affect Exactech meter glucosetests for glucose. Diabetic Med 1991; 8: 172–175. measurements. Diabetic Med 1994; 11: 506–509.144. Clark JDA, Goldberg L, Jones K, Hartog M. Are blood
153. Kilpatrick ES, Rumley AG. ExacTech blood glucoseglucose reagent strips reliable in renal failure? Diabeticmeter (Letter). Diabetic Med 1995; 12: 451.Med 1991; 8: 168–171.
154. Kilpatrick ES, Rumley AG, Rumley CN. The effect of145. Wiener K. An assessment of the effect of haematocrithaemolysis on blood glucose meter measurements.on the HemoCue Blood Glucose Analyser. Ann ClinDiabetic Med 1995; 12: 341–343.Biochem 1993; 30: 90–93.
155. Evans JR, McIntosh JP, Hartung H, Anderson C, Sawers146. Wiener K. The effect of haematocrit and sample tempera-JSA. Falsely low blood glucose readings by a bloodture on the Glucotide/Glucometer 4 blood glucose assayglucose meter system (Letter). Diabetic Med 1994; 11:system (Letter). Diabetic Med 1995; 12: 362–363.326–327.147. Scott R, Fitzpatrick L. Evaluation of the Boehringer
156. Jenkins RD, Bakhat A. Self-monitoring of blood glucoseMannheim Accutrend blood glucose meter. Clin Biochem(Letter). Br Med J 1993; 306: 332.Revs 1992; 13: 122.
831ASSESSMENT OF GLYCAEMIC CONTROL
1997 by John Wiley & Sons, Ltd. Diabet. Med. 14: 819–831 (1997)