When Are HBA1C Values Misleading?

In this issue of AACE Clinical Case Reports, Drs. Arakari and Chongcharoen have authored a very interesting article (1.Arakaki R.F. Changcharoen B. Glycemic assessment in a patient with Hb Leiden and type 2 diabetes.AACE Clinical Case Rep. 2016; 2: e307-e310Abstract Full Text Full Text PDF Scopus (3) Google Scholar) which describes how glycated hemoglobin (HbA1c) levels may be misleading as a measure of glycemic control. They provide a case report regarding a patient with diabetes who remained undiagnosed for several years and then, despite moderately elevated fasting glucose levels, received suboptimal therapy for his diabetes, probably because his HbA1c levels were still in the normal range. By the time the patient obtained endocrine consultation, his urine showed evidence of microalbuminuria, a complication that may have been preventable had treatment not been delayed. The patient who was the basis of the case report had a rare hemoglobin variant (hemoglobinopathy), hemoglobin Leiden, which is associated with a mild hemolytic anemia and splenomegaly, resulting in a shortened red cell survival and thus normal HbA1c values in the face of continuing hyperglycemia. The authors point out that other hemoglobin variants and hemolytic anemias may also contribute to lowering of HbA1c levels and lessen the value of HbA1c levels for both diagnosis and glycemic monitoring. The importance of this well-done report goes far beyond just the discussion of a fascinating but rare hemoglobin variant. There are many clinical situations in which the use of HbA1c levels alone for either diagnosis of diabetes or as a measure of glycemic control is unwise and may lead to serious errors. In fact, whenever a clinician notes that the glycemic control of their patient, as estimated either from continuous glucose monitoring system data, point-of-care (POC) glucose readings, or a glycated albumin, is discordant with their HbA1c level, the discordant data should prompt an investigation as to whether the patient has a condition in which HbA1c data are misleading. So where to start? One method is to first examine the source of the HbA1c value you are evaluating. Although the majority of HbA1c methods used in larger clinical chemistry laboratories are not falsely affected by hemoglobin variants (2.Bry L. Chen P.C. Sacks D.B. Effects of hemoglobin variants and chemically modified derivatives on assays for glycohemoglobin.Clin Chem. 2001; 42: 153-163Crossref Scopus (464) Google Scholar), some are, and a visit to the National Glycohemoglobin Standardization Program website (3.National Glycohemoglobin Standardization ProgramFactors that interfere with HbA1c test results. Available at: http://www.ngsp.org/factors.asp. Accessed August 25, 2016.Google Scholar) may identify those methods which may yield a falsely low or falsely elevated HbA1c value in the presence of one of the many hemoglobin variants. In addition, POC HbA1c methods are in wide use clinically and more often are less accurate. One problem is that several of the POC HbA1c methods in wide use show significant biases and may report falsely low HbA1c values and also are prone to interference with hemoglobin variants (4.Lenters-Westra E. Slingerland R.J. Three of 7 hemoglobin a1c point-of-care instruments do not meet generally accepted analytical performance criteria.Clin Chem. 2014; 60: 1062-1072Crossref PubMed Scopus (110) Google Scholar). An even more common issue is that although proficiency testing for those performing the POC testing is recommended by most experts, many of those who regularly perform the tests have not had adequate training or proficiency testing, and the results are more likely to be inaccurate. The problem of hemoglobin variants is a global one. About 7% of the world population may have a hemoglobin variant (5.Lorenzo-Medina M. De-La-Iglesia S. Ropero P. Nogueira-Salgueiro P. Santana-Benitez J. Effects of hemoglobin variants on hemoglobin A1c values measured using a high-performance liquid chromatography method.J Diabetes Sci Technol. 2014; 8: 1168-1176Crossref PubMed Scopus (24) Google Scholar), and although many of the more than 1,175 different variants (5.Lorenzo-Medina M. De-La-Iglesia S. Ropero P. Nogueira-Salgueiro P. Santana-Benitez J. Effects of hemoglobin variants on hemoglobin A1c values measured using a high-performance liquid chromatography method.J Diabetes Sci Technol. 2014; 8: 1168-1176Crossref PubMed Scopus (24) Google Scholar) are innocuous, some, as for example hemoglobin Leiden, are not. Also, some hemoglobin variants, such as hemoglobins S, C, F, E, and D, are found in many patients worldwide. In Southeast Asia, up to 30% of the population may have at least a heterozygous hemoglobin variant, often S, E, and D, but also F. About 10% of the African American population has a hemoglobin variant. Consequently, when caring for a patient with an ethnic or racial background which is associated with a higher incidence of hemoglobin variants, it is reasonable to consider whether the patient has a hemoglobin variant that needs to be factored into the evaluation of their HbA1c data. In the case study, the mechanism by which a hemoglobin variant caused a falsely low HbA1c in the case reported was by causing hemolysis and a shortened red cell survival. In fact, any condition that shortens red cell survival or necessitates blood transfusions will falsely lower HbA1c levels, as for example, hemolytic anemias, blood loss, or treatment with multiple transfusions. But not all anemias will lower HbA1c levels. Both iron deficiency anemia and pernicious anemias, for example, are associated with a longer red blood cell survival, which is associated with an increase in HbA1c levels. In addition, in the case of iron deficiency, malondialdehyde, which is increased in iron deficiency anemia, enhances the glycation of hemoglobin, and the combination of these two mechanisms regularly results in a false increase in HbA1c levels in patients with iron deficient anemia. Chronic renal failure (CRF) is also associated with a shortened red cell survival and falsely low HbA1c levels, but in CRF, other mechanisms are also in play that may instead raise the HbA1c. These include decreased levels of erythropoietin, increased glycation, higher levels of carbamylated hemoglobin, and variable exposure to higher levels of glucose during dialysis. Overall, however, HbA1c tends to be falsely lowered during CRF. Other conditions to consider that diminish the utility of HbA1c levels include chronic liver disease, which is associated with anemia and shortened red cell survival. When jaundice occurs, however, the bilirubin elevations may artifactually increase HbA1c levels. Other diseases to be considered that may cause shortened red cell survival include splenomegaly due to any cause, rheumatoid arthritis (6.Bernstein R.M. Freedman D.B. Liyanage S.P. Dandona P. Glycosylated haemoglobin in rheumatoid arthritis.Ann Rheum Dis. 1982; 41: 604-606Crossref Scopus (5) Google Scholar), or any of the many drugs or diseases that cause hemolysis. The mechanisms may be diverse. High doses of vitamins C and E are reported to inhibit glycation. Hydroxyurea is associated with lower HbA1c levels but by another completely different mechanism. It is reported that hydroxyurea induces higher levels of hemoglobin F, and when the hemoglobin is measured, some analytic methods will give a lower HbA1c than would be expected by their average glucose levels. Chronic alcoholism also may be associated with falsely low HbA1c levels, although the mechanism is unclear. There are also genetic variances in the speed of glycation, one of the explanations given for the observation that in the African American population, the HbA1c may be higher than expected from the levels of glycemia (7.Herman W.H. Yong M. Uwaifo G. et al.Differences in A1C by race and ethnicity among patients with impaired glucose tolerance in the Diabetes Prevention Program.Diabetes Care. 2007; 30: 2453-2457Crossref PubMed Scopus (427) Google Scholar). One should not take an HbA1c at face value, since in some clinical settings, the assay is misleading and cannot be used either for diagnosis or as a true measure of glucose control. Also, the low sensitivity of HbA1c in the diagnosis of early diabetes, which is approximately 0.5, further limits its usefulness as a diagnostic tool. The high specificity of the test (0.95) makes the HbA1c test unlikely to be a false positive, but it may often provide a result that is falsely negative. Recently, however, a new problem has emerged. While HbA1c levels are quite useful for the evaluation of glucose control in both individuals and large populations, when used in an iterative fashion, measuring progress of glycemic control across periods of time and to follow the progress of a patient, it must be done keeping in mind the limitations of the HbA1c test. But when HbA1c levels are used by payers uncritically to reward or punish health care providers, perverse consequences may ensue. The problem is most acute when HbA1c levels are misused by payers at the individual provider level, because the HbA1c level of the patient may not reflect the average glycemic control of the patient but instead any of the other nonglycemic related factors previously described. As an example, Provider A has a high proportion of patients with diabetes with CRF and liver disease. Provider B has many patients with diabetes who also have iron deficiency anemia and other nutritional anemias. Even if the two providers have patients with equal levels of glycemic control, the HbA1c levels would tend to be higher in the second group than the first for reasons that are unrelated to glycemic control. But in pay-for-performance settings, the physician may be prodded to overtreat the patients with falsely elevated HbA1c levels and to undertreat those with falsely lower glucose levels, thereby injuring each group of patients. Not only physicians but policymakers as well need to understand the limitations of HbA1c measurement and why uncritical acceptance of the HbA1c values as an accurate measure of glycemic control can lead to poor care and poor clinical outcomes.