The Glucose Measurement Industry and Hemoglobin A1c

Abstract

The MyStar Extra self-monitoring blood glucose (SMBG) system provides moving estimates of the patient’s hemoglobin A1c (HbA1c). There is a treasure trove of highly accurate glucose data available from highly accurate SMBG, CGM and FGM along with highly accurate HPLC HbA1c. If Nathan’s criteria are used to select subjects whose glucoses can be correlated to the HbA1c, then algorithms can be developed for robustly transforming glucose into HbA1c. These algorithms can then be implemented in any SMBG or with the CGM and FGM software. This calculated HbA1c would even be accurate with Nathan’s excluded population thus reducing the use of fructosamine and glycated protein. Finally, the developer of these new algorithms is advised to use a specific approach for testing her algorithm.

In this journal, Sieber et al evaluate a novel capability of the new MyStar Extra self-monitoring blood glucose (SMBG) system. ¹ In addition to measuring whole blood glucose, it also provides moving estimates of the patient’s hemoglobin A1c (HbA1c), eA1c, which are derived from a combination of the past eA1c and the current fasting SMBG reading. The eA1c calculations are based on the work of Kovatchev et al at the University of Virginia. ² For the purposes of initialization and calibration, the patient also undertakes monthly 7-point SMBG profiles. Presumably, this 7-point profile is also conducted with a new lot of test strips to mitigate any between lot shift in the glucose measurements. Usually the patient will document only 1 fasting glucose per day; as such, a large amount of glycemic information will not be incorporated into the eA1c.While requiring only 1 data point per day minimizes patient effort, it would seem that such a lean eA1c might compromise its clinical utility. In fact, the authors state that “eA1c is intended to raise patients’ A1c awareness by providing an estimated A1c value and trend in between health care professional visits. It is intended as an adjunctive tool to complement, not replace, a laboratory HbA1c test and is not intended to suggest changes in treatment decisions or to be used as a substitute for professional health care advice.” Despite this caution, 83% of the study patients agreed or strongly agreed that the eA1c might motivate them to manage their diabetes better.

Hemoglobin A1c, a Tired, Overused Test

In most Western nations, HbA1c test ordering is growing at least 8% to 10% annually. In Ontario, Canada, 3.4 million HbA1c were run in 2011 and 2012, a 55% increase over the period of 2006-2007. The role of HbA1c, recognized as the best indicator of long-term diabetes control, has become the preferred screening test for diabetes. The American Diabetes Association recommends screening for diabetes in adults aged 45 years or older and screening in persons with risk factors regardless of age. The American Association of Clinical Endocrinologists, American Academy of Family Physicians, Diabetes Australia, Diabetes UK, and the Canadian Task Force on Preventive Health Care all recommend screening in persons with risk factors. The US Preventive Services Task Force (USPSTF) will soon strongly sanction these recommendations in an upcoming update of the 2008 USPSTF recommendation on diabetes screening. Since 2008, 6 lifestyle intervention studies have demonstrated consistent benefit of lifestyle modifications to prevent or delay progression to diabetes, and longer-term follow-up has increased confidence that such interventions can improve clinical outcomes. The USPSTF has concluded that there is moderate net benefit to measuring blood glucose in adults at increased risk for diabetes. ³

Many health care organizations and governments are now using HbA1c to assess efficacy of individual physicians or physician group practices in treating diabetes. ⁴ For example, the US Affordable Care Act (ACA) requires performance measurements of diabetes care using HbA1c as a measure of diabetes control. The ACA requires enumeration of percentages of patients aged 18 through 75 years with diabetes mellitus who had most recent hemoglobin A1c greater than 9.0%.

Generally, tracking an actively treated diabetic patient’s serial HbA1c is effective in reducing the patient’s short- and long-term glycemic complications. We and others have demonstrated regional variations in the frequency of HbA1c. In the province of Alberta, approximately 500 000 HbA1c tests were extracted from laboratory information systems in the 3 study regions over a 2-year interval. ⁵ As we discovered too frequent HbA1c repeats (within 30 days) in several regions, we began cancelling any HbA1c request that was ordered within 28 days of a previous analysis. Our success in these 28 day cancellations was followed by attempts to cancel HbA1c ordered more frequently than every 90 days. Such efforts may be overzealous as a recent publication indicates that reduced testing frequency is associated with increased HbA1c and deteriorating diabetes control. ⁶

The use HbA1c to screen for hyperglycemic patients is more problematic in that a good number of patients will exhibit HbA1c values that are either too high or too low relative to their long-term glycemic state; that is, the HbA1c will not represent the long-term glycemic state. Radin has enumerated the reasons for falsely elevated HbA1c including reduced red cell turnover (iron deficiency, vitamin B12 deficiency, folate deficiency, and asplenia), severe hypertriglyceridemia, severe hyperbilirubinemia, uremia, lead poisoning, chronic ingestions of alcohol, salicylates, or opioids. ⁷ Analysis of National Health and Nutrition Examination Survey (NHANES) data demonstrates elevated HbA1c in the elderly (to higher degree than fasting and 2 hour glucose) as well as in non-Hispanic black individuals compared to whites and Mexican Americans, independent of fasting glucose levels. ⁸ These race-ethnic differences have been ascribed to genetics, red cell life span and differences in glycation.^9-12 Radin has also provided a differential for falsely reduced HbA1c including any state that increases red cell turnover (acute or chronic blood loss, hemolytic anemia and splenomegaly, end stage renal failure with decreased red cell survival, pregnancy), and some medications (vitamin E [reduces protein glycation], and Ribavirin or interferon-alpha [can induce hemolytic anemia]). ⁷ For patients whose HbA1c is artifactually increased or decreased, alternate glycemic markers have been recommended, including fructosamine and glycated albumin. Researchers now refer to glycation gaps, differences between HbA1c weighted fructosamine and HbA1c or between HbA1c and a weighted glycated albumin. Some medical investigators state these glycation gaps are associated with increased diabetic morbidity and mortality. ¹³

Translating Measured HbA1c Into Glucose

Nathan et al, in their highly cited article “Translating the A1c Assay Into Estimated Average Glucose Values,” derived a robust equation to relate average glucose (from CGM) to HbA1c. ¹⁴ This equation could then be used to transform patient HbA1c to estimated glucose which could be added to the laboratory report detailing the HbA1c. Of course, the authors were highly selective in patient selection and needed to exclude any subjects with factiously elevated or reduced HbA1c. Naturally, men and women with low hematocrits < 39% and <36%, respectively, were excluded as were subjects who were pregnant, had a hemoglobinopathy or were on steroid therapy or had increased red cell turnover (indicated by reticulocyte count), blood loss, and or transfusion, chronic renal or liver disease, high dose vitamin C or erythropoietin treatment. In addition, to be a study subject, all diabetic subjects required stable HbA1c within 1% of each other in the prior 6 months. The social media is replete with inquiries about the disagreement between HbA1c-derived central laboratory estimates of average glucose and SMBG meter’s calculated average glucose. It is estimated that about 15% of HbA1c values may be misleading. ¹⁵

Translating Measured Glucose Into HbA1c

There is a notable secular trend in diabetes testing, and it is represented by the increasing accuracy of patient SMBG, CGM, and the very new flash glucose monitoring (FGM) systems.^16,17 Over the past decade certain SMBG devices have become industry leaders in terms of high reproducibility and high accuracy in the presence of anemia and specific interfering agents. All SMBG systems are not alike and if a highly precise and accurate SMBG system is operated by a trained and motivated patient, it will almost always produce a highly accurate glucose result. In north-central Alberta, approximately 10 000 nurses operate over 800 hospital meters supplied by a single manufacturer and hundreds of emergency medical services personnel operate the matching consumer meter which utilizes the same strip lot. The introduction of new strip lots is now seamless and the analytical imprecision’s of the SMBG are uniformly 2.5% at high and low glucoses, much lower than the patient’s biologic variation.

Think of the richness of the highly accurate glucose data stored in patient SMBG systems that utilize the reagent strips identical to those used in Alberta. A patient with type 1 diabetes typically measures 8 or more glucoses and the motivated patient with type 2 diabetes might uses 3 to 5 glucose measurements for monitoring and hypoglycemia prevention.

The perceived value of the laboratory run HbA1c has been diminished for all of the reasons stated above. The prevalence of iron deficiency anemia is 2% in adult men, 9% to 12% in non-Hispanic white women, and nearly 20% in black and Mexican American women. ¹⁸ Anemia prevalence increases in lower socioeconomic classes. Interestingly, high waist circumference males will tend to have lower hemoglobins and even anemia, probably due to the increased abdominal fat and the resulting inflammatory reaction which evokes the hepatic secretion of hepcidin. ¹⁹ Just as in the case of glucose, all HbA1c assays are not equivalent and as a rule, the high-performance liquid chromatography (HPLC) assays are significantly better than the immunoassay varieties. ²⁰

Today’s mega laboratories will often house multiple, identical immunoassay systems for HbA1c analysis and the analysis of other common analytes. It requires significant effort to minimize the between instrument bias of these multiple analyzers; as such, determining whether a patient’s HbA1c is significantly different from the previous HbA1c incorporates not only the patient’s biologic variation of HbA1c (which increases over time) and the intrinsic imprecision of the instrument but also the between instrument error component (propagation error) which arises from the analysis of the HbA1c on an alternate analyzer in the same mega laboratory.

Conclusion

Many of today’s SMBG manufacturers have been developing CGM and FGM that offer multiple sensors, multiple measurements, and the ability to monitor continuously for several weeks. Today’s manufacturers of highly accurate glucose measurement systems are at an important juncture. They are poised to take over the HbA1c testing for patients with type 1 diabetes and perhaps the HbA1c testing of diligent type 2 patient. There is finally an abundance of highly accurate glucose data and this abundance will grow. When these data are linked with highly accurate (HPLC) HbA1c data of subjects resembling Nathan et al’s abbreviated population sample, numerical analysts will be able to develop powerful algorithms for deriving the best HbA1c, one that will be representative of the time averaged glucose in the body’s internal environment, the “milieu intérieur,” which will be superior to the concentration of glycosylated hemoglobin measured in a distant commercial laboratory. This HbA1c will be able to be related to all the risks of diabetes; in addition, it will obviate the need for those HbA1c wannabes, the fructosamine, the glycated albumin, and so on. Unlike the eA1c calculated by the MyStar Extra this new calculated HBA1c will replace the laboratory HbA1c and can be used to suggest changes in treatment decisions and may sometimes be used as a substitute for professional health care advice.

It will be interesting if these superior, calculated HbA1c will first be derived from the CGM/FGM data or from a cohort of type 1 patients’ SMBG data. I have some advice directed to the creators of these transformative algorithms. Please test the intrapatient sequences of calculated HbA1c with a unique algorithm that extrapolates between patient differences to time zero and allows the calculation of either biologic or analytic variation.^21,22 If the y-intercept of the extrapolated data is close to that derived from intrapatient sequences of specimens analyzed on a HPLC HbA1c system, then you and the glucose measurement industry will be in the money.