gg clinchem

7/29/2019 Gg Clinchem

1/4

Estimation of the Glycation Gap inDiabetic Patients With StableGlycemic ControlSANTIAGO RODRGUEZ-SEGADE, PHD1,2

JAVIER RODRGUEZ, PHD1,2

JOS M. GARCA LOPEZ, MD3

FELIPE F. CASANUEVA, MD3,4FLIX CAMIA, PHD1

OBJECTIVEdThe glycation gap (the difference between measured A1C and the value pre-dicted by regression on fructosamine) is stable and is associated with microvascular complica-tions of diabetes but has not hitherto been estimated within a clinically useful time frame. Weinvestigated whether two determinations 30 days apartsuffice fora reasonably reliable estimate ifboth A1C and fructosamine exhibit stability.

RESEARCH DESIGN AND METHODSdWe studied 311 patients with type 1 or type 2diabetes for whom simultaneous measurements of A1C and serum fructosamine had been madeonat least two occasions separatedby 1 month (t0 and t1). Glycemia wasdeemedstableif A1C(t1)

A1C(t0) and fructosamine(t1)2

fructosamine(t0) were both less than their reference changevalues (RCVs). Instantaneous glycation gaps [gg(t0) and gg(t1)] and their mean (GG), were cal-culated using the data from all stable patients for the required regression.

RESULTSdStable glycemia was shown by 144 patients. In 90% of unstable case subjects, achange in medication was identified as the cause of instability. Among129 stable patients with anaverage of eight gg determinations prior to t0, GG correlated closely with the mean of these priordeterminations (r2 = 0.902, slope 1.025, intercept 20.038).

CONCLUSIONSdThe glycation gap can be calculated reliably from pairs of A1C and fruc-tosamine measurements taken 1 month apart if these measurements satisfy the RCV criteria forglycemic control.

Diabetes Care 35:24472450, 2012

It is now well established that amongpatients with type 1 diabetes (14), pa-tients with type 2 diabetes (5), and

nondiabetic patients (69), there are con-siderable interindividual differences in

A1C that are not accounted for by corre-sponding differences in glycemia butinstead seem to reflect the different gly-cabilities of hemoglobin in different indi-viduals. Yudkin et al. (6) referred topersons with A1C levels higher thanwas expected from their serum glucoselevels as high glycators and to thosewith lower-than-expected A1C as low

glycators.

McCarter et al. (1) quantifiedthese tendencies as the hemoglobin gly-cation index, the difference between mea-sured A1C and the value predicted byregression of A1C on mean blood glucosein the study sample, and found that thehemoglobin glycation index was a statis-tically significant predictor of retinopathyand nephropathy in the Diabetes Controland Complications Trial (DCCT).

More recently, two studies have usedfructosamine instead of mean blood glucoseas the predictor of A1C in quantifyingglycability. In both (2,5), A1C and

fructosamine were both measured at oneor more visits to the clinic; A1C was re-gressed on fructosamine using all validmeasurements from all patients in the sam-ple; an instantaneous glycation gap (gg)wascalculated foreach visit as thedifferencebetween the measured A1C and the valuepredicted from the fructosamine measure-ment using the regression equation; in oneof these studies (5), a characteristic glycationgap was calculated for each patient as themean of that patientsgg values (GG). Cohenet al. (2) reported that in type 1 diabetic

patients, gg partly explained the excessbetween-patient variance in A1C. In ourret-rospective analysis of 2,314 type 2 diabeticpatientsfollowed up for a mean of 6.5 years,GG predicted the progression of nephropa-thy independently of fructosamine even af-ter adjustment for A1C (5).

Authors finding fault with the notionof the glycation gap have asserted that it isitself nothing but the residual of theregression of A1C on fructosamine andthat therefore no difference betweenhigh-, medium-, and low-GG groups

would be expected after statistical adjust-ment for A1C and fructosamine (10).They have accordingly questioned thefindings of Rodrguez-Segade et al. (5).This argument ignores the distinction be-tween the regression residual, gg, and theintrinsic glycation gap of the individual(GG*) (estimated by Rodrguez-Segadeet al. [5] as an average over gg) and as-sumes the validity of a statistical modelin which measured A1C deviates ran-domly, both between and within individ-uals, from the value predicted fromfructosamine (A1C*). Yet, this statisticalmodel is precisely what is called intoquestion by the glycation gap hypothesis,which replaces it by a model in which,for a given patient, measured A1C is theresult of random deviation from A1C* +GG* and not A1C*. The fact that GG* isnot directly measurable is inconvenientbut should not cause confusion aboutthe model that is being investigated.

GG, calculated as an average over sev-eral years measurements (5), is evidentlynot usable clinically. A single gg also can-not be used with confidence because

c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c

From the1Departmentof Biochemistryand Molecular Biology, Universityof Santiago de Compostela,Santiagode Compostela, A Corua, Spain; the 2Department of Biochemistry, University Hospital Clinical Bio-chemistry Laboratory,Universityof Santiago de Compostela,Santiago de Compostela, A Corua, Spain;the3Division of Endocrinology, Department of Medicine, University of Santiago de Compostela, Santiago deCompostela, A Corua, Spain; and the 4Physiopathology of Obesity and Nutrition Biomedical ResearchNetwork Consortium, Santiago de Compostela, A Corua, Spain.

Corresponding author: Santiago Rodrguez-Segade, [email protected] 16 December 2011 and accepted 2 June 2012.DOI: 10.2337/dc11-2450 2012 by the American Diabetes Association. Readers may use this article as long as the work is properly

cited,theuse iseducationaland notforprofit,and the workis notaltered.See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

See accompanying articles, pp. 2415, 2421, 2429, and 2674.

care.diabetesjournals.org DIABETES CARE, VOLUME 35, DECEMBER 2012 2447

C l i n i c a l C a r e / E d u c a t i o n / N u t r i t i o n / P s y c h o s o c i a l R e s e a r c hO R I G I N A L A R T I C L E
mailto:[email protected]://creativecommons.org/licenses/by-nc-nd/3.0/http://creativecommons.org/licenses/by-nc-nd/3.0/http://creativecommons.org/licenses/by-nc-nd/3.0/http://creativecommons.org/licenses/by-nc-nd/3.0/mailto:[email protected]


2/4

although gg appears to be fairly stable ingeneral (2,5,11), it is liable to alter if thereis a sudden change in average glycemiadue to, for example, a change in medica-tion. This is so because of the differentturnover times of A1C and fructosamine(812 weeks [12,13] and 13 weeks[14,15], respectively), which make the re-

sponse of A1C to such a change lag be-hind the response of fructosamine.

In this study, we investigated whethertwo determinations 30 days apart sufficefor a reliable estimateof theGG*ifbothA1Cand fructosamine show sufficient stability.

RESEARCH DESIGN ANDMETHODSdWe studied 311 diabeticpatients for whom data were available forsimultaneous measurements of A1C andfructosamine on dates 30 6 2 days apartwithin the past 5 years, none of whomhad any known hemoglobinopathy orerythrocyte disorder. Of the 311, 50(the volunteer group) were recruitedspecifically for the study; i.e., they wererequested, and agreed, to provide a sec-ond blood sample for research purposes 1month after the first was taken; the albu-min excretion rates of these 50 at entrywere all in the range of 414 mg/24 h.The other 261 were patients for whomsuitable data were already available inour database; at entry in the currentstudy, these 261 all had albumin excre-tion rates,100 mg/24 h. Age, sex, dura-

tion of diabetes, and type of treatmentwere recorded, as were the corresponding

A1C and fructosamine levels. In conso-nance with the racial make-up of ourarea, all participants were Caucasian.

Biochemical analysesA1C was determined by high-performanceliquid chromatography in a MenariniDiagnostics HA-8140 analyzer. All A1Cvalues were converted from Japanese Dia-betes Society (JDS)/Japanese Society forClinical Chemistry (JSCC)-referencedvalues to DCCT-aligned units using thefollowing equation: A1CNGSP = 0.985

A1CJDS/JSCC + 0.46%, where NGSP is Na-tional Glycohemoglobin StandardizationProgram (16). Fructosamine was assayedusing Genzyme GlyPro kits, which imple-ment an enzymatic method adapted to theRoche Diagnostics Cobas Mira analyzer.

SamplesWhole blood was collected in EDTA-containing tubes for determination of A1C.

A1C and serum fructosamine were bothdetermined on the day of collection in the

Clinical Biochemistry Laboratory of theUniversity Hospital Complex, Santiago deCompostela. Control materials were ana-lyzed to estimate between-run analyticalcoefficients of variation (CVA) for bothfructosamine (at 189 and 560 mmol/L)and A1C (at 5.4 and 9.3%); for each analyteand level, the control material was run in

duplicate twice a day for 10 days.

Statistical analysesReference change values (RCVs) werecalculated as RCV = 21/2 3 1.96 3[(CVA)

2 + (CVI)2]1/2, where CVI is the

within-subject biological CV (17); pairsof A1C and fructosamine values taken 1month apart were considered to indi-cate glycemic stability during this pe-riod if their differences, DA1 C an dDfructosamine (DFA), were both lessthan the corresponding RCV. That A1Cand fructosamine values were normallydistributed was verified by skewness andkurtosis tests. gg were calculated as de-scribed above, with the preliminary re-gression of A1C on fructosamine beingperformed with the pooled data from allpatients satisfying the glycemic stabilitycriteria; GG were calculated as the meanof each patients two gg values. Agreementbetween GG and GGnRl (an alternativecharacteristic glycation gap [see RESULTS])as regards classification of patients ashigh, medium, or low glycators was as-sessed as Cohen k (18). All statistical ana-

lyses were performed using SPSS 17software (SPSS, Chicago, IL).

RESULTSdThe CVAs were 1.4% forfructosamine and 1.1% for A1C. The cor-responding CVIs for nondiabetic patients,

taken from the literature, are 3.4% forfructosamine (19) and 2.5% for A1C(20). Thus, the RCVs used were 10.2%for fructosamine and 7.6% for A1C.

Table 1 summarizes relevant clinicaland biochemical characteristics of the pa-tients studied. Of the 311, 144 (46.3%)fulfilled the glycemic stability criteria. Of

these 144, 34 were volunteers(68% of thevolunteer group) and 110 nonvolunteers(42% of the nonvolunteer group); similarproportions were observed in subgroupsof the volunteer and nonvolunteer groupsdefined by sex or type of diabetes. Table 2shows that, on average, both DA1C andDFA were smaller among both type 1 andtype 2 diabetic volunteers than amongother patients with the same type of di-abetes, although the differences only at-tained statistical significance in the case ofDA1C (P, 0.001). Moreover, the A1Cand fructosamine levels of the 16 volun-teers who did not fulfill the glycemic sta-bility criteria were all lower at the seconddetermination than at the first.

Among the nonvolunteers, 151 failedto fulfill the glycemic stability criteria. In90% of these cases, thepatients treatmenthad been changed in the month before thefirst measurement in a way that was plau-sibly responsible for nonfulfillment (in 74cases between 10 and 15 days earlier, in55 between 15 and 20 days earlier, and in22 between 20 and 30 days earlier). Noneof the 144 patients fulfilling the stability

criteria had had their treatment changedin the month before thefirst measurement.

Regression of A1C on fructosamineusing the 288 measurements from the144 stable patients produced the follow-ing relationship:

Table 1dBaseline characteristics of the study groups

Volunteers Others

Type 1

diabetic

Type 2

diabetic

Type 1

diabetic

Type 2

diabetic

n 24 26 38 223Male 54.2 57.7 39.5 54.3

Age (years) 40 6 10 646 11 32 6 13 61 6 13

Duration of diabetes (years) 16 (821) 13 (521) 15 (623) 7 (314)

Treatment

Insulin 100 27 100 30.5

OADs 73 39.5

Insulin and OAD 5.8

Diet 24.2

Fructosamine (mmol/L) 409 6 84 3366 98 428 6 132 326 6 128

A1C (%) 7.9 6 0.9 7.66 1.1 8.9 6 2.2 7.8 6 1.9

Data are means 6 SD, percent, or median (interquartile range) unless otherwise indicated. OAD, oral anti-diabetes drug.

2448 DIABETES CARE, VOLUME 35, DECEMBER 2012 care.diabetesjournals.org

Estimation of the glycation gap


3/4

A1C % 0:013 fructosamine mmol=L

4:25

r2 0:57

The SD of the 144 differences between thefirst and second gg values was 0.28% andthat of the GG values 1.1%. These patientswere classified as high, medium, or lowglycators depending on whether their GGvalues were .0.5%, between 20.5 and0.5%, or below20.5%.

During the past 5 years, our labora-tory has maintained the same analyticalprocedures for both fructosamine and

A1C. For 129 of the 144 stable patients(23 volunteers and 106 nonvolunteers),

simultaneous A1C and fructosaminemeasurements were made during this

time on at least three occasions beforethe first measurement used in the aboveregression (the average number of prior

measurements per patient was eight).Without regard to glycemic stability, ggvalues were calculated for each of theseprior measurements using the above re-gression equation, and a nonRCV-limited characteristic glycation gap(GGnRl) was calculated for each patientas the mean of his or her prior gg values.Regression ofGG on GGnRl showed goodcorrelation between the two (r2 = 0.902)(Fig. 1), and a t test showed no significantdifference between mean GG and meanGGnRl; the SD of the 129 differences be-

tween GG and GGnRl was 0.34%. Some86% of patients were placed in the same

category byGGnRl asbyGG,andinnocasewas a patient a high glycator according toGG but a low glycator according toGGnRldor vice versa. Cohen k was 0.79(95% CI 0.750.84), indicating goodagreement between GG and GGnRl.

CONCLUSIONSdAs argued above,

estimation of a patients GG* requiresmore than one determination of the gg be-cause of the different kinetics of A1C andfructosamine: A1C reflects weighted meanplasma glucose over the past 100 days, andfructosamine reflects it over the past 30 days(15,21). However, since ;50% of A1C isproduced during the 3035 days prior tomeasurement(15,21,22),both fructosamineand A1C are affected by any significantalterationin plasma glucose during thispe-riod. It therefore seems reasonableto estab-lish the stability of fructosamine and A1Cover this period as a criterion for the prob-able validity of theassociatedgg values. Thiswas the criterion adopted in this study,with the stability of the analytes identifiedby comparison of the 30-day differ-ences with the corresponding RCVs. Theadequacy of this criterion is reflectedpartly by the small SD of thegg differencesfor the 144 pairs of determinations bywhich it wassatisfiedand partly by the closecorrelation between the GG and GGnRl val-ues of the 129 patients for whom the latterwere available. Definitive validation wouldconsist in showing that GG, determined

from twogg values as in this study, remainsstable throughout the course of the dis-ease (or that it variesin a well-defined man-ner); pending long-term studies designedto test this possibility, the procedure testedhere seems reasonably adequate.

It is striking that the proportion ofpatients satisfying the glycemic stabilitycriterion was much larger in the volunteergroup than among the other participants.This suggests that patients aware that thefirst and second determinations will beconsidered jointly may exercise bettercontrol than others, and this notion issupported by the finding that the volun-teers who did not fulfill the glycemicstability criteria all had lower A1C andfructosamine levels at the second deter-mination than at the first. The exercise ofbetter control does not in itself invalidatemeasurement of the glycation gap, sincethe gap is in principle independent ofglucose level; what is important is that

A1C and fructosamine both reflect the sameglucose level, and it is this that is favored byrequiring fulfillment of the RCV criterion byboth analytes over the 30-day period.

Table 2dDifferences between initial and final values of fructosamine and A1C(% of initial values) and compliance with the RCV criteria

Volunteers Others

Type 1

diabetic

Type 2

diabetic

Type 1

diabetic

Type 2

diabetic

n 24 26 38 223

Fructosamine% difference 6.0 (2.612.9) 5.5 (2.314.8) 8.2 (6.017.9) 9.0 (4.219.1)

,RCV 15 (62.5) 19 (73.1) 20 (52.6) 118 (52.9)

A1C

% difference 1.5 (1.33.0) 1.5 (1.12.6) 5.6 (2.28.5)* 4.1 (1.87.9)*

,RCV 24 (100) 25 (96.2) 28 (73.7) 170 (76.2)

Data are n (%) of patients complying with the RCV criterion or median (interquartile range) unless otherwiseindicated. RCV (fructosamine) = 10.2%; RCV (A1C) = 7.6%. * P, 0.001 with respect to the correspondingvolunteer group.

Figure 1dScatterplot ofGG (%) vs. GGnRl (%) with the estimated regression line and correlation.

care.diabetesjournals.org DIABETES CARE, VOLUME 35, DECEMBER 2012 2449

Rodrguez-Segade and Associates


4/4

The equation obtained in this studyfor the regression of A1C on fructosaminepredicts A1C levels ;0.5% higher thanthose predicted by the equation obtainedin our earlier study (5). The cause of thisdiscrepancy appears to be the fact that forboth A1C and fructosamine, the earlierstudy included data obtained by a variety

of different methods (though standardlinear conversions were used to minimizethe influence of this heterogeneity). Anal-ysis of 5,987 A1C/fructosamine pairs thatwere included in the earlier study and hadbeen obtained by the methods used in thecurrent study produces an equation thatis very similar to the one obtained in thecurrent study.In short, as onemightexpect,the crucial regression equation, like the an-alyte measurements it relates, depends onthe methods used for the measurements,and this should be taken into account byphysicians interpreting glycation gap data.

In summary, GG, the average of twogg values determined 1 month apart,agrees reasonably well with long-term av-erage gg if neither A1C nor fructosaminechanges by more than their RCV duringthe month in question. GG so measuredconstitutes an adequate estimation of thepatients GG*.

AcknowledgmentsdThis work was sup-ported in part by grants from the SecretariatGeneral for Research and Development of theXuntade Galicia,Spain(PGIDIT04BTF203016PRand PGIDIT06BTF20302PR and 10SCA203008PR) and from the Spanish Ministry of Edu-cation and Science (SAF2004-07602).

This work was also supported by MenariniDiagnostics and Siemens Healthcare Diag-nostics. No other potential conflicts of interestrelevant to this article were reported.

S.R.-S. designed, researched, and wrote themanuscript and approved the final version ofthe manuscript. J.R., J.M.G.L., and F.F.C. re-searched data, contributed to discussion, andapproved the final version of the manuscript.F.C. researched data, contributed to discus-sion, edited the manuscript, and approved thefinal version of the manuscript. S.R.-S. is theguarantor of this work and, as such, had fullaccess to all the data in the study and takesresponsibility for the integrity of the data andthe accuracy of the data analysis.

The authors acknowledge the statisticalguidance of Dr. Juan Manuel Paz and ofDr. Francisco Gude from the Clinical Epide-miology Unit, University Hospital, Universityof Santiago de Compostela.

References1. McCarter RJ, Hempe JM, Gomez R,

Chalew SA. Biological variation in HbA1cpredicts risk of retinopathy and nephropa-thy in type 1 diabetes. Diabetes Care 2004;27:12591264

2. Cohen RM, HolmesYR, Chenier TC, JoinerCH. Discordance between HbA1c andfructosamine: evidence for a glycosylationgap and its relation to diabetic nephropa-thy. Diabetes Care 2003;26:163167

3. Wilson DM, Kollman; Diabetes Researchin Children Network (DirecNet) StudyGroup. Relationship of A1C to glucose con-centrations in children with type 1 diabetes:assessments by high-frequency glucose de-terminations by sensors. Diabetes Care

2008;31:3813854. Soros AA, Chalew SA, McCarter RJ,

Shepard R, Hempe JM. Hemoglobin gly-cation index: a robust measure of hemo-globin A1c bias in pediatric type 1 diabetespatients. Pediatr Diabetes 2010;11:455461

5. Rodriguez-Segade S, Rodriguez J, Cabezas-Agricola JM, Casanueva FF, Camia F.Progression of nephropathy in type 2 dia-betes: the glycation gap is a significantpredictor after adjustment for glyco-hemoglobi n (HbA1c). Clin Chem 2011;57:264271

6. Yudkin JS, Forrest RD, Jackson CA, RyleAJ, Davie SJ, Gould BJ. Unexplained var-iability of glycated haemoglobin in non-diabetic subjects not related to glycaemia.Diabetologia 1990;33:208215

7. Gould BJ, Davie SJ, Yudkin JS. In-vestigation of the mechanism underlyingthe variability of glycatedhaemoglobin innon-diabetic subjects not related to gly-caemia. Clin Chim Acta 1997;260:4964

8. Kilpatrick ES, Maylor PW, Keevil BG. Bi-ological variation of glycated hemoglobin.Implications for diabetes screening andmonitoring. DiabetesCare 1998;21:261264

9. RohlfingC, Wiedmeyer HM, LittleR, etal.Biological variation of glycohemoglobin.Clin Chem 2002;48:11161118

10. Sacks DB,Nathan DM,Lachin JM.Gaps inthe glycation gap hypothesis. Clin Chem2011;57:150152

11. Nayak AU, Holland MR, Macdonald DR,Nevill A, Singh BM. Evidence for con-sistency of the glycation gap in diabetes.Diabetes Care 2011;34:17121716

12. Goldstein DE, Little RR, Lorenz RA, et al.Tests of glycemia in diabetes. DiabetesCare 2004;27:17611773

13. Rohlfing CL, Wiedmeyer HM, Little RR,England JD, Tennill A, Goldstein DE.

Defining the relationship between plasmaglucose and HbA(1c): analysis of glucoseprofiles and HbA(1c) in the DiabetesControl and Complications Trial. Di-abetes Care 2002;25:275278

14. Baker JR, Metcalf PA, Holdaway IM,Johnson RN. Serum fructosamine con-centration as measure of blood glucosecontrol in type I (insulin dependent) di-abetes mellitus. Br Med J (Clin Res Ed)1985;290:352355

15. Tahara Y, Shima K. Kinetics of HbA1c,glycated albumin, and fructosamine andanalysis of their weight functions againstpreceding plasma glucose level. Diabetes

Care 1995;18:44044716. Hoelzel W, Weykamp C, Jeppsson JO,

et al.; IFCC Working Group on HbA1cStandardization. IFCC reference systemfor measurement of hemoglobin A1c inhuman blood and the national standardiza-tion schemes in theUnited States, Japan,andSweden: a method-comparison study. ClinChem 2004;50:166174

17. Fraser CG, Harris EK. Generation andapplication of data on biological variationin clinical chemistry. Crit Rev Clin Lab Sci1989;27:409437

18. Fleiss JL. Statistical Methods for Rates andProportions. New York, Wiley, 1981

19. Rics C, Alvarez V, Cava F, et al. Currentdatabases on biological variation: pros,cons andprogress. ScandJ Clin LabInvest1999;59:491500

20. Braga F, Dolci A, Montagnana M, et al.Revaluation of biological variation of gly-cated hemoglobin (HbA(1c)) using an ac-curately designed protocol and an assaytraceable to the IFCC reference system. ClinChim Acta 2011;412:14121416

21. Tahara Y, Shima K. The response of GHbto stepwise plasma glucose change overtime in diabetic patients. Diabetes Care1993;16:13131314

22. Cohen RM, Franco RS, Khera PK, et al.Red cell life span heterogeneity inhematologically normal people is suffi-cient to alter HbA1c. Blood 2008;112:42844291

2450 DIABETES CARE, VOLUME 35, DECEMBER 2012 care.diabetesjournals.org

Estimation of the glycation gap

gg clinchem

Documents