In this article, I make the argument that the HbA1c test, despite its popularity, is a blunt and inadequate instrument for diagnosing diabetes. By the time an individual is diagnosed with diabetes based on an A1c greater than 6.5%, they are already at risk for complications like nephropathy and retinopathy.
A serious approach to longevity requires early detection and prevention of disease, not identification of disease when it’s already reached an advanced state. To this end, I explore the utility of measuring fasting insulin levels to detect insulin resistance before an individual develops diabetes.
I discuss the challenges in interpreting fasting insulin levels, but ultimately advocate for its use within a comprehensive approach to the assessment of metabolism, insulin resistance, and glucose tolerance.
Topics covered within the article:
- A brief description of the nature of type II diabetes and the history of its diagnostic criteria
- An explanation of hemoglobin A1c, how it’s measured, what it represents, how its use is recommended by the guidelines, and its shortcomings
- A discussion of insulin resistance as the precursor to type II diabetes
- A brief history of the insulin assay, the utility of measuring fasting insulin levels, and how to interpret the results
- An indication of a comprehensive strategy for recognizing and reversing insulin resistance
What is Type II Diabetes?
Type II diabetes is a disease of impaired glucose metabolism and is characterized by hyperglycemia. The most common cause of type II diabetes is the consumption of excess energy beyond what an individual is able to use or safely store. As such, it’s common for type II diabetes to be accompanied by obesity, fatty liver, dyslipidemia, and a host of other related biochemical abnormalities. Even absent other derangements, over time, hyperglycemia itself is sufficient to damage essential organs like the heart, eyes, kidney, and brain, and is toxic to endothelium and peripheral nerves. Diabetes is associated with an increased risk for nearly every chronic disease; e.g. heart attack, stroke, dementia, and many cancers. Thus, any serious approach to preventing chronic disease must include a plan to identify the earliest indicators of diabetes.
The First Diagnostic Criteria for Type II Diabetes
The first diagnostic criteria for type II diabetes were offered by an expert committee who convened in Geneva in 1965. At the time, the diagnosis was based on oral glucose tolerance tests and fasting glucose levels. To be diagnosed with diabetes, a patient would need to have met one of the following criteria:
- 2-hour oral glucose tolerance test (OGTT) > 140 mg/dl OR
- Fasting blood glucose > 130 mg/dl
In addition to these tests, the expert committee also discussed the importance of clinical judgment. Judgment is always important when considering diagnostic tests with dichotomous thresholds. The importance of judgment is illustrated by considering the case of a patient with a fasting blood glucose of 129 mg/dl. In such a case, the patient would not have met the diagnostic criteria for diabetes based on the above criteria. But, you can easily imagine if this patient’s blood glucose had been tested an hour earlier, or on a different day, it might have been a few points higher, thus exceeding the diagnostic threshold and triggering a diagnosis of diabetes.
Consider a case in which a patient’s blood sugar is even further below the diagnostic threshold. They may still be developing diabetes. Type II diabetes develops over a time-course of years or decades starting with hyperinsulinemia and insulin resistance. Only after compensatory mechanisms are overwhelmed will a patient exhibit hyperglycemia. So, hyperglycemia is a late sign of the process that leads to a diagnosis of diabetes.
Even in 1965, the expert committee understood these facts.
“Many persons are in the border-line state between the values given in the accompanying table for normality and diabetes. The Committee urged all physicians and health authorities to pay special attention to this group, who will yield a high proportion of diabetics and for whom preventive measures will be most rewarding.”
Diagnostic Criteria and Approach are Largely Unchanged
It’s clear that even in the 60s, experts were aware that diabetes was a major problem and that it should be diagnosed early before overt disease.
Despite these prescient observations and recommendations, diagnostic criteria for diabetes have been essentially unchanged for over 57 years. Only in 2021 did the United States Preventive Services Taskforce recommend screening all individuals at age 35 (the previous recommendation was age 45). This is in spite of the fact that the CDC estimates that 5% of US adults age 18-44 have diabetes and that almost half of them (2.1 million people) are undiagnosed. Besides this failure to screen people at younger ages, the diagnostic criteria still depend on similar dichotomous thresholds for fasting glucose, OGTT results, and now, hemoglobin A1c.
What is Hemoglobin A1c?
Hemoglobin is the protein within red blood cells that carries oxygen. There are several subtypes of hemoglobin molecules, one of which is hemoglobin A. Hemoglobin A can be further subdivided into types A1 and A2. During the lifetime of a red blood cell, some of its hemoglobin A1 molecules become covalently bound to glucose molecules in a non-enzymatic process called glycation. Glycated hemoglobin A1 is known as hemoglobin A1c or HbA1c or just A1c. The ‘c’ designation comes from the process by which HbA1c is identified in the lab. Scientists use liquid chromatography to measure glycated hemoglobin A1 in a sample of blood. This process yields a signal with several peaks. The third of these peaks corresponds to glycated hemoglobin A1. The peaks are labeled in order using the alphabet. Thus, the third peak is designated by the letter ‘c.’
What is the Relationship Between Hemoglobin A1c and Blood Glucose?
Red blood cell membranes allow glucose to freely enter the cytoplasm where it can bind to hemoglobin A1 to form what we call hemoglobin A1c. Higher levels of glucose in the blood will cause hemoglobin A1c levels to increase over the lifetime of a red blood cell. The lifespan of a red blood cell from its birth in the bone marrow to degradation in the spleen is roughly 120 days such that at any given time, the mean age of all red blood cells is 38-60 days. In theory, therefore, HbA1c represents average blood glucose levels over a three to four month period. In reality, HbA1c reflects a weighted average blood glucose with more recent blood glucose levels affecting HbA1c more than more remote levels. Fifty percent of the HbA1c is determined by average glucose levels from just the preceding month which is why one month of significant glucose lowering can have a significant effect on the A1c.
Labs often report estimated average glucose (eAG) along with A1c. For example, an A1c of 6.5% correlates with an eAG of 140 mg/dl.
A1c results can be affected by hemoglobin variants, hemolysis, anemia, medications, and other physiologic variations and medical conditions. Despite these complexities, for most people, A1c correlates well with average glucose.
Even when A1c is accurate, it cannot tell us what our glucose curve looks like, what our peaks or troughs look like, and how exactly we respond to different foods. It also doesn’t tell us how much carbohydrate or glucose we can dispose of before hyperglycemia occurs. While A1c tells us about average glucose, only glucose monitoring or glucose tolerance tests can give us more granular information.
How is the A1c Used?
In 2009, the International Expert Committee recommended the A1c be used as a diagnostic test for diabetes. Prior to that, only fasting plasma glucose and an oral glucose tolerance test were accepted, and in those cases, careful interpretation of borderline cases was encouraged.
The A1c is a controversial screening test because it is well known to be insensitive for detecting diabetes. The ADA says this in their own guideline document:
“The A1C test, with a diagnostic threshold of 6.5% (48 mmol/mol), diagnoses only 30% of the diabetes cases identified collectively using A1C, FPG, or 2-h PG, according to National Health and Nutrition Examination Survey (NHANES) data. Despite these limitations with A1C, in 2009 the International Expert Committee added A1C to the diagnostic criteria with the goal of increased screening.”
Here are the 2022 diagnostic criteria for type II diabetes according to The American Diabetes Association (ADA):
- Fasting plasma glucose greater than 126 mg/dl OR
- 2-hour OGTT greater than 200 mg/dl OR
- Hemoglobin A1c greater than 6.5% OR
- In a patient with classic symptoms of hyperglycemia or hyperglycemic crisis, a random plasma glucose greater than 200 mg/dL
Of the above tests, the A1c has gained popularity for several reasons:
Convenience – An A1c can be evaluated from a standard tube of blood. It’s a relatively inexpensive test. The results come back quickly.
Ease of Interpretation – The guidelines for interpreting A1c are straightforward. An individual with an A1c under 6.5% does not have diabetes while someone with an A1c of 6.5% or above does have diabetes. While other factors could theoretically affect diagnosis, in practice, the approach is frequently this simple.
The Use of A1c Alone is Supported by Professional Guidelines – To quote directly from the 2022 ADA guidelines, “Generally, FPG, 2-h PG during 75-g OGTT, and A1C are equally appropriate for diagnostic screening.”
The 6.5% Threshold
The original use of the A1c test was to monitor glycemic control in patients with diabetes. Now, A1c is used for screening and diagnosis of diabetes. An A1c of 6.5% or greater constitutes a diagnosis of diabetes. Where does this threshold come from?
Atkin et al. in 2021 explain:
“The International Expert Committee (IEC) put forward the recommendation of an HbA1c level of 6.5% (48 mmol/mol) as the threshold for diabetes diagnosis; notably, their recommendation was predicated upon the probability that individuals with HbA1c levels of 6.5% or more have a markedly increased risk of retinopathy relative to those whose HbA1c falls below that threshold. Endorsed by both the American Diabetes Association (ADA) and the World Health Organization, this recommendation was accepted based solely upon evidence of diabetic retinopathy risk from several key studies.”
This means that one of the most common screening tests for diabetes is not predicated upon a threshold for early detection, but upon a threshold above which one becomes at increasing risk for retinopathy. Atkin et al. go on to show that an A1c not only predicts increased rates of retinopathy, but also nephropathy. In other words, 6.5% is a threshold above which patients are at risk for serious complications. This is the antithesis of early detection and prevention.
How widespread is the practice of screening with A1c alone?
According to Evron et al. 2019, out of nearly 10,000 patients in a major health insurance network in Michigan who were eligible to be screened for diabetes, only 11 (patients, not percent) were screened for diabetes with an OGTT over a 3 year time period. The rest were screened using either a fasting plasma glucose (FPG) level (12%), an A1c (14%), or not at all (74%). Overall, this study demonstrated that screening was inadequate and that A1c and FPG were used while OGTT was hardly used at all.
The Pathophysiology that Precedes Diabetes
The initial pathophysiology of type II diabetes is not hyperglycemia; it’s hyperinsulinemia. Insulin is the hormone that is released by the pancreas in response to energy consumption. In particular, when carbohydrates are consumed, the pancreas responds by releasing insulin. Insulin has a multitude of effects on different organs and cell-types, but its primary effects include:
- signaling the liver to stop gluconeogenesis and glycogenolysis so that the liver does not release any additional glucose into circulation
- signaling the liver, muscle, and other cell types to open channels to allow glucose to pass from the bloodstream into the cells
- stimulating glycogen synthesis within the liver, muscle, and other cells in order to store glucose for later use
As a person consumes more energy than they can use or store, higher levels of insulin are required to stimulate liver and muscle cells to take in glucose despite the fact that they are overloaded. Hyperinsulinemia can be present for years or decades prior to an individual meeting the diagnostic criteria for diabetes.
You might be wondering now why we don’t simply test insulin levels instead of glucose levels in order to screen for type II diabetes. The reasons are numerous.
The History of the Insulin Assay
It was not possible to measure insulin levels until a genius named Rosalyn Yalow pioneered the radioimmunoassay in 1956. She went on to win the Nobel Prize for this work in 1977. Her story is amazing and worth reading in its own right. The radioimmunoassay allowed scientists to measure hormones and other molecules in blood which exist in low concentrations. Yalow’s test was the first which could reliably measure insulin levels. Using her test, she observed that people with type II diabetes produce more insulin than their non-diabetic counterparts. Amazingly, she correctly interpreted this to mean that type II diabetes is a disease marked by a decreased ability to use insulin.
Why Don’t We Test Insulin Levels?
Today, it’s easy and inexpensive to order insulin levels. So, why isn’t this test used more often?
There are two reasons it’s not commonly used to screen for diabetes. First, guidelines do not recommend using insulin levels.
In fact, professional associations like the American Association for Clinical Chemistry and American Diabetes Association have explicitly recommended against using fasting insulin levels to screen for insulin resistance as recently as 2011. I apologize for the upcoming long quote, but this is so egregious in my opinion that I’d like to include the whole thing:
“In the last several years, interest has increased in the possibility that measurements of the concentrations of plasma insulin and its precursors might be of clinical benefit. In particular, published evidence reveals that increased concentrations of insulin and/or proinsulin in nondiabetic individuals predict the development of coronary artery disease. Although this possibility may be scientifically valid, its clinical value is questionable. An increased insulin concentration is a surrogate marker that can be used to estimate resistance to insulin-mediated glucose disposal, and it can identify individuals at risk for developing syndrome X, also known as the insulin resistance syndrome or the metabolic syndrome. Accurate measurement of insulin sensitivity requires the use of complex methods, such as the hyperinsulinemic euglycemic clamp technique, which are generally confined to research laboratories. Because of the critical role of insulin resistance in the pathogenesis of type 2 diabetes, hyperinsulinemia would also appear to be a logical risk predictor for incident type 2 diabetes.”
“Earlier studies may not have controlled well for glycemic status and other confounders. More-recent analyses suggest that insulin values do not add significantly to diabetes risk prediction carried out with more traditional clinical and laboratory measurements and that measures of insulin resistance (that include insulin measurements) predict the risk of diabetes or coronary artery disease only moderately well, with no threshold effects. Consequently, it seems of greater clinical importance to quantify the consequences of the insulin resistance and hyperinsulinemia rather than the hormone values themselves, i.e., by measuring blood pressure, the degree of glucose tolerance, and plasma lipid/lipoprotein concentrations. It is these variables that are the focus of clinical interventions, not plasma insulin or proinsulin concentrations.”
In other words, these authors suggest we should look only for consequences of disease, not its cause or precursors. They fail to acknowledge the idea that if you’re able to detect insulin resistance before it manifests as hyperglycemia, hypertension, and dyslipidemia, you can alter the course of the disease so that those complications never occur.
The second reason that insulin levels aren’t frequently measured is that they’re difficult to interpret. There are no guidelines for interpreting fasting insulin levels. The threshold for the upper limit of normal on the Labcorp insulin test is 24.9 µIU/mL. But, this test is used for multiple purposes, not just for screening for insulin resistance. For example, this test might be ordered to look for an insulin secreting tumor called an insulinoma, or to determine whether a patient is injecting insulin or not. The lab’s normal range is calibrated for a variety of situations, not specifically for insulin resistance.
Finally, there are technical considerations for insulin assays. The test cannot detect insulin which is bound to insulin autoantibodies, which some patients may produce unbeknownst to themselves and their physician.
Even if insulin is measured at the right time in the right person and there are no complicating factors, it can still be difficult to determine if a certain fasting insulin level is predictive of insulin resistance.
Fasting Insulin Levels Can be Useful
A 1993 study by Laakso et al. Indicates how fasting insulin levels might be used. The authors tested a total of 132 subjects; 50 subjects with normal glucose tolerance, 28 with impaired glucose tolerance, and 54 with non-insulin dependent diabetes (aka type II diabetes). Within their study, 100% of participants who had normal glucose tolerance but a fasting insulin level > 18 µIU/mL were in the most insulin resistant tertile of participants as determined by euglycemic hyperinsulinemic clamp, which is the gold standard laboratory test for determining insulin resistance.
In other words, based on this study, if an individual’s fasting insulin level is > 18 µIU/mL, he or she almost certainly has insulin resistance.
Here is the key graphic for this study:
Not only is there a threshold at 18, but there is a clear increase in the 9-13 µIU/mL range whereby subjects with fasting insulin levels above 13 µIU/mL have at least a 75% chance of being in the most insulin-resistant tertile of subjects.
This graph also demonstrates the limitations of the fasting insulin level. Focus your attention on the lines that are identified with triangles and boxes. These lines represent subjects who are known to have impaired glucose tolerance and type II diabetes respectively. These individuals are likely to be insulin resistant at any level of fasting insulin.
The fact that a person with type II diabetes can have a normal fasting insulin is not a diagnostic disaster, because this individual is highly likely to be revealed using other tests. They are likely to have elevated fasting plasma glucose, post-prandial glucose, and an elevated A1c, all of which are easy to detect. So, while it’s true that an isolated fasting insulin level without any additional context might be difficult to interpret, it’s still of great value in a patient with normal glucose tolerance.
Modeling Insulin Resistance
In order to ease the interpretation of fasting insulin, it’s useful to model its relationship to fasting glucose.
The most well validated and frequently used model is the HOMA-IR, or the Homeostatic Model Assessment for Insulin Resistance. This model was developed by Dr. David Matthews and is based on the idea that there is a feedback loop between the liver and pancreas whereby hepatic glucose output, hepatic insulin sensitivity, and pancreatic insulin secretion exist in homeostasis.
In the fed state, the pancreas releases insulin which signals the liver to decrease glucose output. In the fasting state, insulin levels decrease and the pancreas releases glucagon which stimulates the liver to increase glucose output. However, an insulin resistant liver will persist in releasing glucose even in the fed state and may release too much in the fasting state. The increased blood glucose causes the pancreas to secrete higher levels of insulin. Higher levels of insulin may temporarily compensate for the insulin resistance in the liver and help maintain normoglycemia. Eventually, the liver becomes so insulin resistant that it cannot respond, and/or the pancreas becomes unable to secrete enough insulin. At this point, the homeostasis is disturbed and fasting plasma glucose begins to climb.
It turns out that calculating a HOMA-IR score is quite simple.
If the resulting score is less than 1, it’s unlikely that an individual has insulin resistance. Numbers above 1.9 are suggestive of early insulin resistance and above 2.9 signal significant insulin resistance.
Another model that aims to accomplish a similar goal is called the QUICKI, or the Quantitative Insulin Sensitivity Check Index. This is similar to the HOMA-IR, except that the fasting insulin and fasting glucose levels are inverted and transformed logarithmically which gives the results more granularity and a finer ability to identify insulin resistance.
- A score greater than 0.45 indicates likely normal insulin sensitivity.
- A score between 0.3-0.45 suggests that insulin resistance is likely.
- A score less than 0.3 suggests that diabetes is likely.
HOMA-IR and QUICKI are computational methods of interpreting fasting plasma insulin in its proper physiologic context so that insulin resistance can be detected and treated early.
Achieving Prevention and Reversal
Progressive experts in the field are pushing the idea that we shouldn’t simply manage and treat type II diabetes. Rather, we should aim to prevent and reverse it. I wholeheartedly agree. A 2019 literature review published by Hallberg et al. found “99 original articles containing information pertaining to diabetes reversal or remission” and concluded, “evidence exists that T2D [type II diabetes] reversal is achievable using bariatric surgery, low-calorie diets, or carbohydrate restriction.” It’s heartening to see this proactive mindset.
Even better than reversal is prevention. Insulin resistant individuals can benefit from early detection and targeted interventions in order to reverse insulin resistance and to surveil for and prevent progression to diabetes. There is strong evidence that insulin resistance alone even without dysglycemia is detrimental to long-term health aside from the fact that it’s a harbinger for diabetes.
Obtaining the most accurate and specific diagnosis possible is always the first step in any treatment plan. In order to diagnose insulin resistance, we should utilize the full battery of available tests and the patient’s full clinical context. This means taking a detailed family history, a detailed clinical history, reviewing nutrition logs, exercise logs, and quantifying body composition, fasting insulin, fasting plasma glucose, A1c, continuous glucose monitoring, relevant genetics, lipid phenotyping, and an OGTT with insulin levels if the individual is willing.
No single test or piece of information is enough on its own, but with all of this information, it becomes possible to obtain a better picture of an individual’s metabolic health and risk for insulin resistance. Obtaining this information early in life is also useful so that subsequent tests can be judged in comparison to past tests. The efficacy of interventions, or lack-thereof can also be better interpreted when more information is available.
The A1c has its place in broadly identifying individuals with long-standing hyperglycemia and diabetes, but it doesn’t identify individuals with insulin resistance who have normal glycemic control. I recommend a more detailed diagnostic approach that offers the chance to detect insulin resistance before it becomes a danger and when it’s easier to reverse.
Featured Photo by Alexander Grey