Portal de Periódicos da CAPES

How reliable is your HbA1c test? Revisiting the use of HbA1c in cystic fibrosis-related diabetes (CFRD) screening

2019; Elsevier BV; Volume: 18; Issue: 2 Linguagem: Inglês

10.1016/j.jcf.2019.02.001

1873-5010

Grace Y. Lam, Shelby Sissons, Maeve P. Smith, Neil E. Brown, Winnie M. Leung, Mathew P. Estey,

Diabetes Management and Research

HbA1c has historically been considered unreliable in screening for cystic fibrosis-related diabetes (CFRD). However, recent work from Burgess et al. puts this paradigm into question, suggesting that screening with HbA1c could eliminate the need for the onerous oral glucose tolerance test (OGTT) in a significant proportion of patients [[1]Burgess J.C. Bridges N. Banya W. Gyi K.M. Hodson M.E. Bilton D. et al.HbA1c as a screening tool for cystic fibrosis related diabetes.J Cyst Fibros. 2016; 15: 251-257Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar]. Follow up studies yielded conflicting results, with one supporting the use of HbA1c in CFRD screening [[2]Gilmour J.A. Sykes J. Etchells E. Tullis E. Cystic fibrosis related diabetes screening in adults: a gap analysis & evaluation of A1C accuracy.Can J Diabetes. 2019; 43 ([in press]): 13-18Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar], and others arguing that HbA1c has limited utility in this setting (3; 4). There remains great mystery surrounding these discrepant findings. There are many different HbA1c assays used by clinical laboratories, which have undergone steady improvement in analytical reliability since the early 2000's [[5]Little R.R. Rohlfing C.L. Sacks D.B. National glycohemoglobin standardization program steering C: status of hemoglobin A1c measurement and goals for improvement: from chaos to order for improving diabetes care.Clin Chem. 2011; 57: 205-214Crossref PubMed Scopus (247) Google Scholar]. We therefore hypothesized that the controversy surrounding the utility of HbA1c in CFRD screening may be explained by improvements in HbA1c assay performance over time. To address this hypothesis, we first reviewed HbA1c proficiency testing data from the College of American Pathologists (CAP) (available on the National Glycohemoglobin Standardizaton Program website at http://www.ngsp.org/CAPdata.asp). This body provides standardized samples with known HbA1c values at regular intervals to laboratories around the world, in order to assess test accuracy and reproducibility. The variability of HbA1c results for each sample is determined by calculating the coefficient of variation (CV; calculated as standard deviation of HbA1c results/known value *100%). We identified all CAP proficiency testing samples from 2002 to 2018 with HbA1c values between 4 and 6%, and plotted the average CV across all HbA1c methods as a function of time. As shown in Fig. 1A , the average CV improved steadily from ~6% in 2002 to ~3% in 2016, demonstrating a clear reduction in HbA1c test imprecision over this timeframe. Interestingly, this improvement in CV appears to plateau in 2016, suggesting that current HbA1c tests may be maximally optimized. We then plotted the specificity and sensitivity of HbA1c for identifying CFRD in previously published studies [1Burgess J.C. Bridges N. Banya W. Gyi K.M. Hodson M.E. Bilton D. et al.HbA1c as a screening tool for cystic fibrosis related diabetes.J Cyst Fibros. 2016; 15: 251-257Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar, 2Gilmour J.A. Sykes J. Etchells E. Tullis E. Cystic fibrosis related diabetes screening in adults: a gap analysis & evaluation of A1C accuracy.Can J Diabetes. 2019; 43 ([in press]): 13-18Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar, 3Boudreau V. Coriati A. Desjardins K. Rabasa-Lhoret R. Glycated hemoglobin cannot yet be proposed as a screening tool for cystic fibrosis related diabetes.J Cyst Fibros. 2016; 15: 258-260Abstract Full Text Full Text PDF PubMed Scopus (33) Google Scholar, 4Schnyder M.A. Benden C. Schmid C. HbA1c: an effective screening tool for cystic fibrosis related diabetes?.J Cyst Fibros. 2016; 15: 261-262Abstract Full Text Full Text PDF PubMed Scopus (13) Google Scholar] as a function of the start year of data collection. The HbA1c screening cutoff with optimal sensitivity was used for this analysis. As shown in Fig. 1B, the specificity of HbA1c in CFRD screening increased from 2002 to 2010, mirroring the reduction in HbA1c test imprecision. Next, we performed a retrospective analysis of more recent HbA1c and OGTT results from both the adult and pediatric CF clinics in Edmonton, Alberta, Canada. All HbA1c and OGTT results from January 2012 to February 2018 were extracted from the laboratory information system, and receiver operating characteristic (ROC) curve analysis was performed to assess the ability of HbA1c to identify CFRD. OGTT results with a corresponding HbA1c measured within +/− 100 days were included in the analysis (since HbA1c reflects glycemic control over the preceding 3 months). In the adult population (N = 106), an HbA1c cutoff of >5.5% identified CFRD with 95% sensitivity and 46% specificity (area under the curve (AUC) = 0.841, p < 0.001) (Fig. 1C). Similar results were obtained in the pediatric population (N = 54), where the same HbA1c cutoff of >5.5% identified CFRD with 100% sensitivity and 47% specificity (AUC = 0.843, p < 0.001) (Fig. 1D). A slightly lower HbA1c cutoff of >5.3% also identified impaired glucose tolerance (IGT) with high sensitivity but lower specificity in both the adult (94% sensitivity and 40% specificity) and pediatric (94% sensitivity and 32% specificity) populations. Interestingly, two different HbA1c assays with different analytical performance were used during the study period. The BioRad Variant II Turbo 2.0 high performance liquid chromatography assay was employed from 2012 to December 2016 (CV = 2.4%), whereas the Roche c513 immunoassay was used from December 2016 to February 2018 (CV = 1.9%). Subgroup analysis in the adult population demonstrated that the newer Roche HbA1c assay (AUC = 0.924; optimal HbA1c cutoff >6.0%; sensitivity 100%; specificity 86%) outperformed the older BioRad HbA1c assay (AUC = 0.836; optimal HbA1c cutoff >5.5%; sensitivity 94%; specificity 51%) in its ability to identify CFRD. This further illustrates the reduction in CV and consequently the improvement in screening performance of HbA1c assays over time. Collectively, our results support the notion that better HbA1c analytical performance over time is one key reason why there has been a major controversy in the field of HbA1c screening for CFRD. This is likely a unique challenge in the CF population since HbA1c values for the non-CFRD versus the CFRD populations are not as distinct as the difference between type 1 or type 2 diabetics and the normal population. Our findings are not only in keeping with observations made by Burgess et al. [[1]Burgess J.C. Bridges N. Banya W. Gyi K.M. Hodson M.E. Bilton D. et al.HbA1c as a screening tool for cystic fibrosis related diabetes.J Cyst Fibros. 2016; 15: 251-257Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar] and Gilmour et al. [[2]Gilmour J.A. Sykes J. Etchells E. Tullis E. Cystic fibrosis related diabetes screening in adults: a gap analysis & evaluation of A1C accuracy.Can J Diabetes. 2019; 43 ([in press]): 13-18Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar], but also argue that HbA1c may now be a viable screening modality for CFRD in both the adult and pediatric populations. Thus, we are in the midst of a large multi-center study to confirm these observations and to determine a harmonized HbA1c screening cutoff across a number of currently available assays. This work was supported by funds from the University Hospital Foundation, Canada.