| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
General Clinical Chemistry |
1 Department of Laboratory Medicine and Pathology, University of Minnesota, Minneapolis, MN.
2 Biostatistics Center, the George Washington University, Rockville, MD.
3 Departments of Pathology and Anatomical Sciences and Child Health, University of Missouri, Columbia, MO.
4 Fairview University Medical Center, Minneapolis, MN.
aAddress correspondence to this author at: University of Minnesota, Department of Laboratory Medicine and Pathology, MMC 609, 420 Delaware St. S.E., Minneapolis, MN 55454. Fax 612-273-3489; e-mail steff001{at}umn.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: The Diabetes Control and Complications Trial (DCCT) and its follow-up study, the Epidemiology of Diabetes Interventions and Complications (EDIC), required 20 years of consistent HbA1c results, measured by three different ion-exchange HPLC procedures. To maintain and document consistent HbA1c results measured in the DCCT and EDIC Central Biochemistry Laboratory, a backup laboratory used frozen hemolysates as long-term calibrators and a HPLC method with a single lot of Bio-Rex 70 resin.
Results: Over 20 years, long-term quality-control values have remained constant. Four studies of nondiabetic ranges produced nearly identical values [mean (SD), 5.1 (0.5)%, 4.9 (0.3)%, 5.0 (0.4)%, and 5.0 (0.3)%].
Conclusion: The overall consistency of the HbA1c assays during the 20-year course of the DCCT and EDIC has been critical in establishing the benefits of intensive therapy and in understanding the relationship between long-term glycemia and the development and progression of the complications of diabetes.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
laboratories
The Central Hemoglobin Laboratory (CHL) and Backup Hemoglobin Laboratory (BHL) were established in 1983. The CHLs have analyzed all routine HbA1c specimens for the DCCT and EDIC. The BHL has provided external quality control (QC) and backup laboratory services for the CHL. The CHL was initially located at the Joslin Diabetes Center (Boston, MA); the BHL has been continuously located at the University of Missouri-Columbia School of Medicine (6). In 1986, all measurements of HbA1c were temporarily moved to the BHL for 6 months, until the CHL was moved to the University of Minnesota (Minneapolis, MN). New instruments with automated capacity [initially Diamat HPLC (Bio-Rad Laboratories) and subsequently Tosoh A1c 2.2 Plus HPLC (Tosoh Medics, Inc.)] were introduced in Minnesota. For all CHL assays, calibrators were assigned values based on comparisons between the CHL and the BHL (see below). Separate QC aliquots were assayed in the CHLs to determine acceptance of each assay run.
reference method at the bhl
The BHL has performed the reference method for HbA1c during the entire course of the DCCT and EDIC, using a HPLC assay with Bio-Rex 70 ion-exchange resin, as described in detail previously (6). Whole blood was transported at 4 °C on wet ice by overnight courier from clinical centers to the BHL. The labile glycated fraction was removed by incubation of washed, packed erythrocytes in saline at 37 °C for at least 5 h, with subsequent lysis of the erythrocytes in water. Cellular debris was removed by mixing with CCl4 followed by centrifugation; the hemolysate sample was injected on the HPLC column. The HbA1c fraction eluted after HbA1a and HbA1b and before the HbA0 fractions. Percentage HbA1c was calculated by dividing the area under the HbA1c peak by the total area under all peaks.
Four lots of calibrators have been prepared at the BHL (for the BHL method only) since 1978 with blood from single nondiabetic donors according to preparation methods similar to those for patient samples. Aliquots of calibrator hemolysates for single use were stored at –70 °C or in liquid nitrogen until the day of analysis. A long-term QC sample was first prepared at the Joslin Diabetes Center laboratory (Boston, MA) in 1983, by use of pooled blood collections from patients with diabetes mellitus (6). The hemolysate preparation was similar to that prepared at the BHL. In addition to the long-term (QC) sample, the Joslin CHL also prepared low, medium, and high pooled controls, which were used at both Joslin and the BHL between 1983 and 1987. The BHL then prepared additional control materials, using single diabetic and nondiabetic donors; these control materials were used from 1987 to present.
first chl method
The first CHL, based at the Joslin Diabetes Center, used a HPLC method identical to the method used at the BHL until early 1986, with calibration and internal QC procedures as described previously (6).
second chl method
At the University of Minnesota (Minneapolis, MN), a Diamat HPLC dedicated ion-exchange system was used from November 1986 until April 2002 (1). The Diamat instrument was calibrated daily by use of a single calibrator provided by the manufacturer (with a value of 8–9%) standardized to the BHL reference system. Nominal values for this calibrator were adjusted based on comparisons between the CHL and BHL for the samples assayed in both the CHL and BHL. With each new lot of calibrator, values were checked and adjusted based on analysis with the previous calibrator. Occasionally (e.g., with changes in columns at the CHL), in-house whole-blood materials were exchanged between the CHL and BHL to verify CHL calibration. These comparisons were in addition to the QC specimens used by the CHL to monitor the assay with each run and to accept or reject the run. During the DCCT and EDIC, different lots of columns and reagents were provided by Bio-Rad, some with apparently different performance characteristics; these are noted when appropriate.
third chl method
In April 2002, a Tosoh A1c 2.2 Plus dedicated ion-exchange HPLC system replaced the Diamat system at the CHL. The Tosoh 2.2 Plus was calibrated weekly by use of two calibrators provided by the manufacturer (usually with values of 
5% and 11% for each calibrator), with values adjusted based on comparisons with the BHL and with consistency checked as described above for the Diamat. All data reduction was performed automatically; analysis time was 3 min. The transition from the Diamat to the Tosoh instrument was monitored with 178 samples, showing consistency (r2 = 0.992; y = 1.036x – 0.236, where x is the Diamat and y is the Tosoh 2.2 Plus) between the instruments.
external qc for chl analyses
For external QC throughout the DCCT, 10% of the specimens were prepared as split duplicates at the clinical centers; samples were sent masked to the BHL and CHL (1). On a quarterly basis, some centers shipped additional split-duplicate specimens to both the CHL and BHL. After 1993, 10 samples were exchanged monthly between the CHL and BHL. Beginning in May 1996, this monthly 10-sample exchange became part of the National Glycohemoglobin Standardization Program (NGSP) network monitoring program. Differences between the BHL and CHL were used as needed to recalibrate the assays in the CHL, as described above. In addition, 5–10% of specimens were split and sent as duplicates, to check on reliability of the system from collection through shipping, analysis in the laboratory, and reporting to the Biostatistics Center at George Washington University.
reproducibility of the nondiabetic range
As an additional check to ensure that the HbA1c assay did not drift over the course of the DCCT and EDIC studies, four nondiabetic range studies were performed with the assumption that the true nondiabetic range should not change over time (1). The first study was performed in 1983 at the onset of the DCCT; the second study in 1988 after the CHL changed from Joslin to Minnesota; the third study in 1992 at the end of the DCCT; and the fourth study in 2000, during EDIC. For each study in 1983, 1985, and 1992, nondiabetic individuals were enrolled, selected to match the age and gender distributions of the DCCT/EDIC groups at baseline; in 2000, the age and gender distribution matched the contemporary age of the EDIC participants. Thus, the nondiabetic individuals in 2000 were older than the previously measured nondiabetic groups. All were recruited from all participating clinical sites in the United States and Canada. Before 2000, local glucose values were used to assure nondiabetic status. For the 2000 study, fasting glucose was measured centrally with a glucose oxidase method [on a Vitros analyzer (Ortho Clinical Diagnostics)] at the CHL to assure that no participants had impaired fasting glucose or diabetes mellitus. HbA1c was analyzed by both the CHL and BHL.
statistical analyses
Values are reported as the mean (SD). The coefficient of reliability was determined as: (between-specimen variance – within-specimen variance)/(between-specimen variance + within-specimen variance) (11)(12). Comparisons over time and among values were completed by use of standard parametric procedures with the SAS statistical package, Ver. 8.2, (SAS Inc.).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. The overall CV for this specimen was <2% from 1984 through 2003, and results remained very close to the value originally established in 1984 [8.28 (0.10)%]. However, beginning in 2001, this lot of hemolysate demonstrated a small reduction in measured values, with clear evidence in 2003 (presumably attributable to deterioration of the material). Thus, we relied on the other control samples to demonstrate the consistency of the assay, with the overlapping lots demonstrating equally good stability over time (Fig. 1
).
|
The nondiabetic range studies, postulated to demonstrate no changes in mean values in a healthy nondiabetic population over time, also confirmed the stability of the assay and the frozen hemolysates used for QC. The values of HbA1c and selected characteristics of the nondiabetic populations over the course of the DCCT and EDIC are shown in Table 1
. There were no appreciable differences over the time span evaluated, with mean HbA1c values from 1983, 1992, and 2000 not statistically significantly different from one another. The 1988 mean was slightly but statistically significantly lower than in 1983 (P <0.01), in 1992 (P <0.02), and in 2000 (P <0.01). Also shown in Table 1
are comparisons within different age groups (reading horizontally). The values across the different years were nearly identical, with the values from 1988 slightly lower for the <20 year and 20–29 year groups (Table 1
), mirroring the results of the whole populations sampled (Table 1
). HbA1c results from all years studied for the 30–39 year group were not significantly different.
|
monitoring of HBA1c values in the central and reference laboratories over time
The split duplicate samples (i.e., aliquots of the same collection) sent separately to the CHLs showed consistency, with a coefficient of reliability >0.99 spanning the 20 years of the DCCT/EDIC. Overall HbA1c values in the CHL vs the BHL differed slightly throughout the DCCT and for most of EDIC. However, despite efforts to sustain similar values, there were small differences between the CHL Diamat and the BHL, especially in the high HbA1c concentration range (>10% HbA1c), which increased beginning in 1995 until the CHL Diamat was replaced by the Tosoh 2.2 Plus in 2002. We addressed the timing (and the possible causes) for the differences between the methods that occurred during EDIC by examining the differences over time and within each lot of columns and reagents supplied by Bio-Rad, believing that most of the differences occurred with changes in columns and reagents. In doing so we found that the performance of column Lot M (which was used in the laboratory March 1993 to July 1996) deteriorated over the last half of its life (data not shown). To determine the relative magnitude of the differences post-DCCT, we used the mean values of the differences from 1987 to mid-1993 (covering the performance of the Diamat during the DCCT) as the baseline. We then subtracted the 1987–1993 values from each of the mean values in EDIC, with the results presented in Fig. 2
. Only for HbA1c values >10% were the absolute differences in mean values >0.2%.
|
In April 2002, the Tosoh A1c 2.2 Plus replaced the Bio-Rad Diamat in the CHL. The performance of the method was similar to the performance of the Diamat method during the DCCT (1987–1993) at HbA1c concentrations <10% and was only slightly higher (0.2%) in the >10% HbA1c range (Fig. 2
).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
By maintaining the same assay method over time at the BHL, we were able to monitor and achieve consistent HbA1c results throughout the DCCT and EDIC. Calibration to the BHL reference method provided stable results at the CHL, despite several changes in HbA1c methodology. Moreover, the stability and precision of the HbA1c assay have enabled the DCCT/EDIC Research Group to assess HbA1c as a risk factor for complications for a decade after the DCCT ended (8)(15)(20).
The challenge of maintaining consistent values over two decades was compounded by the absence of a true standard for the assay. We were not able to precisely match the values from the CHL and BHL. The Diamat with single-point calibration was suboptimal in maintaining consistent values between the CHL and BHL at higher HbA1c concentrations in both the DCCT and EDIC. The Tosoh A1c 2.2 Plus method also demonstrated somewhat greater variance with the BHL at the highest HbA1c concentrations, although to a lesser degree than the Diamat, possibly because of the two-point calibration of the 2.2 Plus. Whether some of these differences may relate to properties of the Bio-Rex resin used in the BHL method and/or to the single-point calibration of the Bio-Rex assay is not clear. In addition, the values of calibrators may not have been optimal to produce accurate values in the high ranges. Thus, even with the consistent exchange of samples between the BHL and the CHL, at higher HbA1c concentrations, the differences between the methods increased during the last years of Diamat use. Perhaps this was attributable to the Diamat calibrators having values of 8%, below the values where the BHL and CHL differed to the greatest extent. However, it should be noted that only 12% of DCCT patients and 7.5% of EDIC patients had HbA1c concentrations >10%. Thus, there was little impact on the conclusions from these studies over nearly two decades of the DCCT and EDIC, despite changes in HbA1c methodology.
The success of the NGSP in standardizing HbA1c testing over the past several years (21) should allow continued consistency in results over time in many laboratories. This will facilitate prospective studies of relationships between glycemia and complications and ensure comparability among studies. It will also ensure optimal use of HbA1c as a tool in patient care, especially with the American Diabetes Association stating a target value of 7% using methods traceable to DCCT values (10). New reference procedures have been proposed (22), and these may be used in conjunction with the NGSP to base the HbA1c assays on definitive standards and methods. In addition, methods continue to improve with regard to precision and specificity. For these reasons, we expect that variation of results over time and among laboratories will progressively diminish, thereby providing even more consistency of HbA1c results.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: DCCT, Diabetes Control and Complications Trial; HbA1c, hemoglobin A1c; EDIC, Epidemiology of Diabetes Interventions and Complications; CHL, Central Hemoglobin Laboratory; BHL, Backup Hemoglobin Laboratory; QC, quality control; NGSP, National Glycohemoglobin Standardization Program; and UKPDS, United Kingdom Prospective Diabetes Study. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  S.-C. Shiesh, H.-M. Wiedmeyer, J.-T. Kao, S. D. Vasikaran, J. B. Lopez, and the Laboratory Management Committee for the Asian- Proficiency Testing of Hb A1c: A 4-Year Experience in Taiwan and the Asian Pacific Region Clin. Chem., October 1, 2009; 55(10): 1876 - 1880. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Diabetes Control and Complications Trial/Epidemiol Modern-Day Clinical Course of Type 1 Diabetes Mellitus After 30 Years' Duration: The Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications and Pittsburgh Epidemiology of Diabetes Complications Experience (1983-2005) Arch Intern Med, July 27, 2009; 169(14): 1307 - 1316. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Enzlin, R. Rosen, M. Wiegel, J. Brown, H. Wessells, P. Gatcomb, B. Rutledge, K.-L. Chan, P. A. Cleary, and the DCCT/EDIC Research Group Sexual Dysfunction in Women With Type 1 Diabetes: Long-term findings from the DCCT/ EDIC study cohort Diabetes Care, May 1, 2009; 32(5): 780 - 785. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. K. Van Den Eeden, A. V. Sarma, B. N. Rutledge, P. A. Cleary, J. W. Kusek, L. M. Nyberg, K. T. McVary, H. Wessells, and for the DCCT/EDIC Research Group Effect of Intensive Glycemic Control and Diabetes Complications on Lower Urinary Tract Symptoms in Men With Type 1 Diabetes: Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) study Diabetes Care, April 1, 2009; 32(4): 664 - 670. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. D. Bash, E. Selvin, M. Steffes, J. Coresh, and B. C. Astor Poor Glycemic Control in Diabetes and the Risk of Incident Chronic Kidney Disease Even in the Absence of Albuminuria and Retinopathy: Atherosclerosis Risk in Communities (ARIC) Study Arch Intern Med, December 8, 2008; 168(22): 2440 - 2447. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Diabetes Control and Complications Trial/Epidemiol Prolonged Effect of Intensive Therapy on the Risk of Retinopathy Complications in Patients With Type 1 Diabetes Mellitus: 10 Years After the Diabetes Control and Complications Trial Arch Ophthalmol, December 1, 2008; 126(12): 1707 - 1715. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Musen, A. M. Jacobson, C. M. Ryan, P. A. Cleary, B. H. Waberski, K. Weinger, W. Dahms, M. Bayless, N. Silvers, J. Harth, et al. Impact of Diabetes and Its Treatment on Cognitive Function Among Adolescents Who Participated in the Diabetes Control and Complications Trial Diabetes Care, October 1, 2008; 31(10): 1933 - 1938. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. H. de Boer, B. Kestenbaum, T. C. Rue, M. W. Steffes, P. A. Cleary, M. E. Molitch, J. M. Lachin, N. S. Weiss, J. D. Brunzell, and for the Diabetes Control and Complications Trial ( Insulin Therapy, Hyperglycemia, and Hypertension in Type 1 Diabetes Mellitus Arch Intern Med, September 22, 2008; 168(17): 1867 - 1873. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A H Berg and D B Sacks Haemoglobin A1c analysis in the management of patients with diabetes: from chaos to harmony J. Clin. Pathol., September 1, 2008; 61(9): 983 - 987. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. D. Paterson, B. N. Rutledge, P. A. Cleary, J. M. Lachin, R. S. Crow, and for the Diabetes Control and Complications Trial/E The Effect of Intensive Diabetes Treatment on Resting Heart Rate in Type 1 Diabetes: The Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications Study Diabetes Care, August 1, 2007; 30(8): 2107 - 2112. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Selvin, C. M. Crainiceanu, F. L. Brancati, and J. Coresh Short-term Variability in Measures of Glycemia and Implications for the Classification of Diabetes Arch Intern Med, July 23, 2007; 167(14): 1545 - 1551. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 The Diabetes Control and Complications Trial/Epide Long-Term Effect of Diabetes and Its Treatment on Cognitive Function N. Engl. J. Med., May 3, 2007; 356(18): 1842 - 1852. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. S. Kilpatrick, A. S. Rigby, and S. L. Atkin Variability in the Relationship between Mean Plasma Glucose and HbA1c: Implications for the Assessment of Glycemic Control Clin. Chem., May 1, 2007; 53(5): 897 - 901. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Fox, M. Dontchev, K. Ruedy, R. Beck, C. Kollman, L. Messer, J. Coffey, D. Wilson, E. Doyle, W. Tamborlane, et al. Relative Inaccuracy of the A1cNow in Children With Type 1 Diabetes Diabetes Care, January 1, 2007; 30(1): 135 - 137. [Full Text] [PDF]
|
|
|
|
 |
|
 P. A. Cleary, T. J. Orchard, S. Genuth, N. D. Wong, R. Detrano, J.-Y. C. Backlund, B. Zinman, A. Jacobson, W. Sun, J. M. Lachin, et al. The Effect of Intensive Glycemic Treatment on Coronary Artery Calcification in Type 1 Diabetic Participants of the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Study Diabetes, December 1, 2006; 55(12): 3556 - 3565. [Abstract] [Full Text] [PDF]
|
|
Read all eLetters
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
