Patient-specific model takes variability out of blood glucose average

Researchers have identified a patient-specific correction factor based on the age of a patient’s red blood cells that can improve the accuracy of average blood glucose estimations from hemoglobin A_1c measures in patients with diabetes.

“The true average glucose concentration of a nondiabetic and a poorly controlled diabetic may differ by less than 15 mg/dL, but patients with identical HbA_1c values may have true average glucose concentrations that differ by more than 60 mg/dL,” wrote Roy Malka, PhD, a mathematician and research fellow at Massachusetts General Hospital, Boston, and his associates.

In an article published in the Oct. 5 online edition of Science Translational Medicine, the researchers reported the development of a mathematical model for HbA_1c formation inside the red blood cell, based on the chemical contributions of glycemic and nonglycemic factors, as well as the average age of a patient’s red blood cells.

Using existing continuous glucose monitoring (CGM) data from four different groups – totaling more than 200 diabetes patients – the researchers then personalized the parameters of the model to each patient to see the impact of patient-specific variation in the relationship between average concentration of glucose in the blood and HbA_1c.

Finally, they applied the personalized model to data from each patient to determine its accuracy in estimating future blood glucose from future HbA_1c measurements, and compared the blood glucose estimates with those made using the current standard method (Sci Transl Med. 2016, Oct 5. http://stm.sciencemag.org/lookup/doi/10.1126/scitranslmed.aaf9304).

This showed that their patient-specific model reduced the median absolute error in estimated average blood glucose from more than 15 mg/dL to less than 5 mg/dL, representing an error reduction of more than 66%.

“The model would require one pair of CGM-measured AG [average blood glucose] and an HbA_1c measurement that would be used to determine the patient’s [mean red blood cell count (MRBC)],” the researchers wrote. “MRBC would then be used going forward to refine the future [average glucose (AG)] calculated on the basis of HbA_1c.”

The researchers pointed out that work was still needed to calculate the duration of CGM that would allow sufficient calibration of MRBC. However, their analysis suggested that no more than 30 days would be needed, and significant improvements could be achieved within just 21 days.

In patients with stable monthly glucose averages, the prior month would be enough for calibration, and even a single week of stable weekly glucose averages might be sufficient.

“The improvement in AG calculation afforded by our model should improve medical care and provide for a personalized approach to determining AG from HbA_1c.”

The ADAG study was supported by the American Diabetes Association, the European Association for the Study of Diabetes, Abbott Diabetes Care, Bayer Healthcare, GlaxoSmithKline, Sanofi-Aventis Netherlands, Merck, LifeScan, Medtronic MiniMed, and HemoCue. Two authors were supported by the National Institutes of Health, and Abbott Diagnostics. The authors are also listed as inventors on a patent application related to this work submitted by Partners HealthCare.

[email protected]

Researchers have identified a patient-specific correction factor based on the age of a patient’s red blood cells that can improve the accuracy of average blood glucose estimations from hemoglobin A_1c measures in patients with diabetes.

“The true average glucose concentration of a nondiabetic and a poorly controlled diabetic may differ by less than 15 mg/dL, but patients with identical HbA_1c values may have true average glucose concentrations that differ by more than 60 mg/dL,” wrote Roy Malka, PhD, a mathematician and research fellow at Massachusetts General Hospital, Boston, and his associates.

In an article published in the Oct. 5 online edition of Science Translational Medicine, the researchers reported the development of a mathematical model for HbA_1c formation inside the red blood cell, based on the chemical contributions of glycemic and nonglycemic factors, as well as the average age of a patient’s red blood cells.

Using existing continuous glucose monitoring (CGM) data from four different groups – totaling more than 200 diabetes patients – the researchers then personalized the parameters of the model to each patient to see the impact of patient-specific variation in the relationship between average concentration of glucose in the blood and HbA_1c.

Finally, they applied the personalized model to data from each patient to determine its accuracy in estimating future blood glucose from future HbA_1c measurements, and compared the blood glucose estimates with those made using the current standard method (Sci Transl Med. 2016, Oct 5. http://stm.sciencemag.org/lookup/doi/10.1126/scitranslmed.aaf9304).

This showed that their patient-specific model reduced the median absolute error in estimated average blood glucose from more than 15 mg/dL to less than 5 mg/dL, representing an error reduction of more than 66%.

“The model would require one pair of CGM-measured AG [average blood glucose] and an HbA_1c measurement that would be used to determine the patient’s [mean red blood cell count (MRBC)],” the researchers wrote. “MRBC would then be used going forward to refine the future [average glucose (AG)] calculated on the basis of HbA_1c.”

The researchers pointed out that work was still needed to calculate the duration of CGM that would allow sufficient calibration of MRBC. However, their analysis suggested that no more than 30 days would be needed, and significant improvements could be achieved within just 21 days.

In patients with stable monthly glucose averages, the prior month would be enough for calibration, and even a single week of stable weekly glucose averages might be sufficient.

“The improvement in AG calculation afforded by our model should improve medical care and provide for a personalized approach to determining AG from HbA_1c.”

The ADAG study was supported by the American Diabetes Association, the European Association for the Study of Diabetes, Abbott Diabetes Care, Bayer Healthcare, GlaxoSmithKline, Sanofi-Aventis Netherlands, Merck, LifeScan, Medtronic MiniMed, and HemoCue. Two authors were supported by the National Institutes of Health, and Abbott Diagnostics. The authors are also listed as inventors on a patent application related to this work submitted by Partners HealthCare.

[email protected]

Researchers have identified a patient-specific correction factor based on the age of a patient’s red blood cells that can improve the accuracy of average blood glucose estimations from hemoglobin A_1c measures in patients with diabetes.

“The true average glucose concentration of a nondiabetic and a poorly controlled diabetic may differ by less than 15 mg/dL, but patients with identical HbA_1c values may have true average glucose concentrations that differ by more than 60 mg/dL,” wrote Roy Malka, PhD, a mathematician and research fellow at Massachusetts General Hospital, Boston, and his associates.

In an article published in the Oct. 5 online edition of Science Translational Medicine, the researchers reported the development of a mathematical model for HbA_1c formation inside the red blood cell, based on the chemical contributions of glycemic and nonglycemic factors, as well as the average age of a patient’s red blood cells.

Using existing continuous glucose monitoring (CGM) data from four different groups – totaling more than 200 diabetes patients – the researchers then personalized the parameters of the model to each patient to see the impact of patient-specific variation in the relationship between average concentration of glucose in the blood and HbA_1c.

Finally, they applied the personalized model to data from each patient to determine its accuracy in estimating future blood glucose from future HbA_1c measurements, and compared the blood glucose estimates with those made using the current standard method (Sci Transl Med. 2016, Oct 5. http://stm.sciencemag.org/lookup/doi/10.1126/scitranslmed.aaf9304).

This showed that their patient-specific model reduced the median absolute error in estimated average blood glucose from more than 15 mg/dL to less than 5 mg/dL, representing an error reduction of more than 66%.

“The model would require one pair of CGM-measured AG [average blood glucose] and an HbA_1c measurement that would be used to determine the patient’s [mean red blood cell count (MRBC)],” the researchers wrote. “MRBC would then be used going forward to refine the future [average glucose (AG)] calculated on the basis of HbA_1c.”

The researchers pointed out that work was still needed to calculate the duration of CGM that would allow sufficient calibration of MRBC. However, their analysis suggested that no more than 30 days would be needed, and significant improvements could be achieved within just 21 days.

In patients with stable monthly glucose averages, the prior month would be enough for calibration, and even a single week of stable weekly glucose averages might be sufficient.

“The improvement in AG calculation afforded by our model should improve medical care and provide for a personalized approach to determining AG from HbA_1c.”

The ADAG study was supported by the American Diabetes Association, the European Association for the Study of Diabetes, Abbott Diabetes Care, Bayer Healthcare, GlaxoSmithKline, Sanofi-Aventis Netherlands, Merck, LifeScan, Medtronic MiniMed, and HemoCue. Two authors were supported by the National Institutes of Health, and Abbott Diagnostics. The authors are also listed as inventors on a patent application related to this work submitted by Partners HealthCare.

[email protected]