Impaired Awareness of Hypoglycemia in Type 1 Diabetes: A Report of An NIDDK Workshop in October 2021

Hypoglycemia remains a limiting factor in the optimal treatment of type 1 diabetes. Repeated episodes of hypoglycemia result in impaired awareness of subsequent hypoglycemic events, inducing a vicious feed-forward cycle and increasing the risk of morbidity and mortality. New technologies, including continuous glucose monitors (CGM) and artificial pancreas devices, alert patients to declining levels of blood glucose, raising hopes that they can reduce time spent in the hypoglycemic range and improve patient awareness of their hypoglycemic state. However, some data suggest that impaired awareness of hypoglycemia (IAH) persists even with CGM use. Because individuals with IAH often are excluded from clinical trials in which technologies are tested, our understanding of the factors that contribute to the development of IAH and restoration of awareness of hypoglycemia remains limited. Furthermore, animal and healthy human models that were used to investigate the mechanisms associated with IAH may or may not accurately reflect the human pathophysiology. To address the obstacles and barriers that have hindered progress in this clinically important area, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) convened a workshop on 14–15 October 2021. This perspective offers a summary of this outstanding meeting that brought clinical and basic scientists from the fields of diabetes, neuroscience, psychology, psychiatry, and imaging together to consider how to best advance this field. The speakers and moderators who participated in the meeting are listed in Supplementary Table 1.

Dr. Simon Heller from the University of Sheffield opened the workshop with a discussion of how to best define the syndrome of IAH in patients with diabetes. He reminded us that hypoglycemia has been a known adverse event of insulin treatment since the 1920s. It constitutes a real danger to patients, and the fact that a given patient’s response to hypoglycemia changes over time was known as early as 1941, when R.D. Lawrence, the famous diabetologist who founded the British Diabetes Association, described his personal and professional experience with how symptoms change over time (1). In a research setting, patients with diabetes are classified as impaired, indeterminant, or normal awareness by using a number of questionnaires that ask about their personal experiences with hypoglycemia. Investigators also make this categorization based on an individual’s response to experimental hypoglycemia. Patients with IAH fail to generate an adequate counterregulatory response to restore euglycemia, patients with normal awareness generate a robust epinephrine response that increases endogenous glucose production, and patients with an indeterminant status have a response that is in between (Fig. 1).

The fact that a single episode of experimental hypoglycemia can reduce the counterregulatory response to hypoglycemia the next day in healthy control subjects has been known for many years (2). We have also known that patients with longstanding type 1 diabetes have an impaired counterregulatory response compared with healthy control subjects (3). Such patients are unable to generate a glucagon response to hypoglycemia and become dependent on the epinephrine response to restore glycemia. Because individuals with type 1 diabetes experience frequent hypoglycemia, and since hypoglycemia can reduce the counterregulatory response to subsequent hypoglycemia, the belief has been that the recurrent hypoglycemia in patients with type 1 diabetes is the cause of impaired awareness. Autonomic neuropathy has also been proposed to contribute to the development of this syndrome (4), but this has been difficult to prove (5).

Dr. Elizabeth Seaquist from the University of Minnesota spoke next about heterogeneity of the response to hypoglycemia exposure and how this heterogeneity makes it difficult to categorize the awareness status of individuals with type 1 diabetes. When researchers are trying to understand the hypoglycemia awareness status of a subject, they usually use a questionnaire that asks about the person’s experience with hypoglycemia and then total up the points the person accumulated to determine if they are aware, unaware, or indeterminant. Several questionnaires are in use, including the Gold (6), the Clarke (7), and the Pedersen-Bjergaard (8). Researchers also will expose study participants to experimental hypoglycemia and then determine whether their epinephrine, metabolic, and symptom responses are normal or reduced.

Dr. Seaquist’s group recently reviewed the data they collected over the years in 75 patients with type 1 diabetes (9). The awareness status of these volunteers was defined using the Gold and Clarke questionnaires, and they then underwent a 2-h hyperinsulinemic hypoglycemic clamp (target 50 mg/dL) in which samples for later measurement of epinephrine and symptom scores were collected. The data shown in Fig. 2 demonstrate that a wide range of responses of both epinephrine and symptom responses are seen in these subjects. They then performed a clustering analysis to determine how well the questionnaires predicted the epinephrine and symptom responses. They found one cluster of subjects who had a high epinephrine response and a high symptom response to the hypoglycemia, a second cluster of subjects who had low epinephrine and low symptom responses, and a third cluster that had a low epinephrine response and a moderate to high symptom score. Subjects who were categorized as hypoglycemia unaware by both Gold and Clarke questionnaires were more likely to fall into the cluster with low epinephrine and symptom responses, and subjects who were categorized as hypoglycemia aware by both questionnaires were more likely to fall into the cluster with high epinephrine and symptom responses. There was much more discordance between the awareness status categorization by the questions in those subjects with low epinephrine and moderate to high symptom scores. This discordance between hypoglycemia-induced epinephrine and symptom responses also has been shown in adrenalectomized humans exposed to hypoglycemia and in subjects with type 1 diabetes and impaired awareness after they participated in an intervention to avoid hypoglycemia. The discussion that followed Dr. Seaquist’s talk raised the question of whether the sympathetic nervous system is responsible for the symptoms associated with hypoglycemia and whether epinephrine secretion from the adrenal gland is responsible for the metabolic response. Could it be that subjects who do not have symptoms during hypoglycemia but do have an epinephrine response have a form of autonomic neuropathy?

Dr. Seaquist then presented work from Lontchi-Yimagou et al. (10) and her own group (11) where normal control subjects were exposed to recurrent hypoglycemia three times (2 h in the morning of day 1, 2 h in the afternoon of day 1, 2 h in the morning of day 2) to determine if the response during the first episode predicts whether the subject will develop blunting of the response during the third. Both groups found that the patients who had the highest epinephrine responses during the first episode of hypoglycemia were the ones who had the greatest blunting on day three. Dr. Seaquist questioned if there were differences in the experiences between those that blunted and those that did not in the days before the experiment. Does stress contribute to the heterogeneity of the counterregulatory response to hypoglycemia?

The final speaker in this session was Dr. Raimund Herzog, from Yale School of Medicine, who reviewed animal models in hypoglycemia research. He pointed out that multiple different approaches (selection of the animal species, mode and frequency of insulin delivery, and the outcomes to be measured) are available and should be selected depending on the scientific questions being addressed and the objectives of the study. Dr. Herzog noted that one of the critical decisions in choosing a model is whether the model is designed to test a specific hypothesis regarding individual physiological components of the condition or whether the objective of the model is to replicate the totality of the human experience. An interesting depiction of how the selected model may define the outcome was that related to a hypothetical study designed to look at central hypoglycemia. Central hypoglycemia can be achieved by a peripheral insulin injection, resulting in global hypoglycemia by insertion of both central and peripheral cannulas, such that glucose is infused peripherally to maintain normoglycemia and insulin is perfused centrally to induce central hypoglycemia, or by injection of 2-deoxyglucose, which causes severe neuroglycopenia but also peripheral hyperglycemia. While the latter two approaches allow the investigator to tease out the differential effects of central versus peripheral hypoglycemia, they do not replicate the physiological consequences of iatrogenic hypoglycemia. Other considerations include the species selected, whether genetic manipulations are required, and the frequency and rate of insulin administration to induce hypoglycemia.

While many of the features of recurrent hypoglycemia exposure can be replicated in common animal models, for example, impaired counterregulatory responses and decreased endogenous glucose production, measurement of IAH is not straightforward, since it relies on behavioral testing that may or may not be directly related to the experience in humans. One approach to address the question has been the learning paradigm conditioned place preference, in which the animal learns to associate a specific place with a positive reward compared with a different place with a neutral award and then is given the same choice during repeated hypoglycemia (12). A different approach to measuring impaired awareness in rodents is the use of food intake as a biomarker for IAH and was discussed by Dr. Simon Fisher later in the meeting. In summary, Dr. Herzog pointed out the insightful data generated by CGM in rodent models and the importance of selecting the appropriate animal models and study design depending on the specific question being addressed.

The session on mechanisms involved in IAH was opened by Dr. Rory McCrimmon, from the University of Dundee, who discussed habituation. Habituation is the where a psychological, behavioral, or physiological response to a stimulus as a result of repeated or prolonged exposure is different from the one generated during the first exposure to the stimulus (Fig. 3). This applies to the counterregulatory response to hypoglycemia, where a robust response is present the first time a person experiences hypoglycemia and the response is attenuated when the hypoglycemia is recurrent. In a habituated response, if you withhold the stimulus for a long enough period of time, the response will normalize when the original stimulus is presented. In addition, if you increase the stimulus frequency, you can further suppress that response or achieve greater habituation, something that we also see in response to recurrent episodes of hypoglycemia. In a habituated response, you can pair a different stimulation that prompts the same response as the original during exposure to the original stimulus and see that there is a return of the response. Dr. McCrimmon is now addressing this question in his laboratory. He presented one study where rodents were exposed to recurrent hypoglycemia or euglycemia in the days before an experimental episode of hypoglycemia (13). During this experimental hypoglycemia, the animals that had antecedent hypoglycemia were exposed to low-intensity exercise or the more stressful, dishabituating stimulus of high-intensity exercise. As expected, the animals who were exposed to euglycemia had abundant epinephrine and glucagon responses to experimental hypoglycemia. The animals who were exposed to recurrent hypoglycemia before the study and then had low to moderate exercise during the experiment exhibited attenuated counterregulatory responses. The animals with high-intensity exercise during the hypoglycemia had very robust epinephrine and glucagon responses. These data imply that counterregulatory responses to hypoglycemia can be restored if a dishabituating stimulus like high-intensity exercise is applied.

Dr. Matthew Whim from the Louisiana State University School of Medicine next discussed autonomic plasticity in response to hypoglycemia and the possibility that peripheral mechanisms were important. He presented data that revealed that a single episode of hypoglycemia causes both epinephrine and neuropeptide Y (NPY) to be released from the adrenal medulla. During recurrent exposure to hypoglycemia, the epinephrine response attenuates as expected but the NPY response remained unchanged (14). He explained that NPY can inhibit tyrosine hydroxylase, the enzyme responsible for synthesis of epinephrine, and he hypothesizes that recurrent hypoglycemia allows repeated release of NPY, which modifies epinephrine synthesis in the adrenal medulla such that it is no longer present in response hypoglycemia. This mechanism could be contributing to the development of IAH.

The next talk was by Dr. Paul Kenny from the Icahn School of Medicine at Mount Sinai. He discussed the habenula–pancreatic axis that regulates nicotine-induced abnormalities in blood glucose homeostasis. He presented data that demonstrated the neural connectivity between the habenula and the pancreas and that stimulation of nicotinic acetylcholine receptors in the habenula increases blood glucose (15). This elevated blood glucose level inhibits the nicotinic acetylcholine receptor function in the habenula, which leads to nicotine aversion and protects against addiction. He suggests that it is possible that hypoglycemia contributes to the increased addictive properties of nicotine in patients with type 2 diabetes and that perhaps this axis is important in regulating awareness of hypoglycemia.

The final talk of the session was by Dr. Conor Liston from Weill Cornell Medicine. He discussed allostatic load and adaptations to glucocorticoids, chronic stress, and neural plasticity. Glucocorticoids play a role in modifying and preserving newly formed neuronal spines that develop in response to learning (16). He questioned if the stress of hypoglycemia (and subsequent release of cortisol) affects the impact of the circadian rhythm on synaptic connectivity in such a way that recurrent hypoglycemia may impair synaptic connectivity and contribute to some of the persistence of the counterregulatory deficit that we see in IAH.

In the first talk on assessments and outcomes, Sylvia Mangia from the University of Minnesota reviewed how neuroimaging during hypoglycemia can provide insights into how the brain responds to recurrent hypoglycemia. Cerebral blood flow (CBF) in the thalamus and frontal cortex have been reported to increase during hypoglycemia in healthy control subjects and in patients with type 1 diabetes with normal awareness, but how hypoglycemia alters CBF in these regions in patients with type 1 diabetes and IAH has not been consistently reported. She believes this may be because of the different approaches taken to analyze imaging data.

When collecting neuroimaging using MRI, Dr. Mangia said that one of the first things an investigator must do is look carefully through all of the data to determine if any of it needs to be corrected for motion (Fig. 4). Many subjects find it challenging to remain still in the magnetic resonance scanner, and if they move too much, the data will be distorted and should be removed from the data set. However, the methodology to do this motion correction must be very rigorous so as to not compromise the quality of the data. After performing motion correction, the correct analytical approach must be selected. One common method is to create a CBF map that displays the differences in blood flow during euglycemia versus hypoglycemia. To create such a map requires that the data be smoothed and consolidated into a single standardized space, which is a very powerful method. This powerful method allows differences to be observed but minimizes the interindividual differences among patients, which can be a problem if you have a small data set and a response that has large interindividual variation. A different approach is using parcellation and looking at specific regions of interest. This gives you a higher signal-to-noise ratio than consolidating things into a standard space, but you can miss the overall brain relationships between the different regions. Dr. Mangia suggested that one of the reasons that different laboratories report different findings using the same experimental design may be that they did not use the same approach to data analysis. When interpreting imaging data, it is important to understand the strengths and limitations of each methodology.

Dr. Jane Speight from the Australian Centre for Behavioral Research in Diabetes spoke next about developments in the assessment of awareness of hypoglycemia in human subjects. She reminded us that awareness of hypoglycemia is a subjective condition, and to assess it accurately you must focus on patient-reported outcome measures. She said instruments like the Gold (6), Clarke (7), and Pedersen-Bjergaard (8) instruments have been useful but do not meet U.S. Food and Drug Administration standards, which require validation and reproducibility of the data. The strengths of these surveys are that they are used the most and are sufficiently brief to be used in clinical practice as well as research. Their weakness is that the anchored response options can lead to ambiguous and possibly unreliable midpoints, which really confounds the classification of impaired awareness with severe hypoglycemia. To address these limitations Dr. Speight has developed the Hypo-AQ tool (17). It meets the rigorous standards of the Food and Drug Administration in that it has strong content validity and is highly reliable. To assure it is useful in defining clinically relevant metrics important in impaired awareness, she is now performing the Hypoglycemia—Measurement, Thresholds, and Impacts (Hypo-METRICS) study to generate new knowledge about adults with diabetes who use CGM.

Dr. Pratik Choudhary from the University of Leicester spoke next about the value and limitation of CGMs. Dr. Chaudhary reviewed data and studies from CGM metrics collected in individuals without diabetes as well as those with type 1 diabetes and type 2 diabetes. He noted that there have been improvements in CGM accuracy and that multiple studies have demonstrated a reduction in severe hypoglycemic events with CGM use (18). He also discussed whether CGM can be used to predict a hypoglycemic event or whether the patterning and timing of the event can be informative in treatment. He ended his talk on a new metric being used in the ongoing Hypo-METRICS study that generates a heat map of hypoglycemic symptoms versus levels of glucose and patterning of the event.

Dr. Michael Rickels from the University of Pennsylvania was the next speaker. He discussed approaches that can be used to categorize the response to hypoglycemia. To do this he presented data collected from subjects with type 1 diabetes and IAH who were given a real-time CGM for 18 months as an intervention to restore awareness of hypoglycemia (19). IAH was defined by questionnaires about experience with hypoglycemia and by the symptom and epinephrine responses collected during experimental hypoglycemia. He found that real-time CGM reduced time spent in hypoglycemia and restored some symptoms of hypoglycemia but did not restore the epinephrine responses to experimental hypoglycemia

Dr. Stephanie Amiel from King’s College opened the session on interventions with a discussion of patients who continue to experience severe hypoglycemia despite undergoing intensive hypoglycemia avoidance education. She has observed many of these patients who have enrolled in structured educational programs designed to reduce hypoglycemia and believes their perception of hypoglycemia differs from that of their doctors or other patients with diabetes. Patients who continue to experience severe hypoglycemic events despite intensive education generally worry more about high rather than low blood glucose and have little fear of hypoglycemia or its consequences despite having negative experiences with hypoglycemia. In addition, these patients were found to have brain imaging patterns that display greater activation of brain areas associated with reward functions than do patients with awareness of hypoglycemia (20). A pilot study has indicated that a psychoeducational group intervention in which abnormal cognitions were challenged by nurse and dietitian educators appeared to reduce rates of severe hypoglycemia. The results of a multicenter controlled trial that tested this approach (Hypoglycemia Awareness Restoration Program [HARPdoc]) against a psychoeducational intervention (Blood Glucose Awareness Training [BGAT]), which does not address cognitions, have recently been published (21). The results showed that both interventions significantly reduced rates of severe hypoglycemia but that HARPdoc also significantly reduced both unhelpful cognitions and adverse psychological outcomes, including anxiety, depression, and diabetes distress.

Dr. Simon Fisher from the University of Kentucky then discussed the use of a translational model for screening potential pharmacological agents for the treatment of IAH. In his model he measures food intake as a marker of hypoglycemic awareness in rodents, because animals will increase their food intake the first time they experience experimental hypoglycemia but will not increase their intake if they have been exposed to repeated hypoglycemia before the episode of experimental hypoglycemia. When he gave rats the drug metoclopramide, a dopamine receptor 2 antagonist, he found that it prevented the blunting of food ingestion and partially restored the counterregulatory responses of glucagon and epinephrine seen in animals exposed to recurrent hypoglycemia. Dr. Fisher now has a clinical trial testing this compound as a treatment for IAH in humans.

At the end of the day, Dr. Jim Shaw (22) from Newcastle University talked about My HypoCOMPASS, a learning tool that has been used to prevent and treat impaired awareness in patients with type 1 diabetes. This tool optimizes hypoglycemia recognition and action using a structured personal curriculum. It comprises 3 h of structured education, delivered either individually or in small groups, including four key elements of the Comparison of Optimized MDI Versus Pumps With or Without Sensors in Severe Hypoglycemia (HypoCOMPaSS) trial (Fig. 4): never delay treatment; recognize an individual’s unique periods of increased risk; aid hypoglycemia recognition by focusing on the onset of subtle symptoms; and prioritize detecting and preventing nocturnal episodes. My HypoCOMPASS has been used in clinical trials where it proved comparable to both pumps and CGM in improving impaired awareness and reducing severe hypoglycemic episodes (23).

The scope and views of the workshop were inevitably constrained by the need to shorten the program, originally envisaged as a 3-day, face-to-face meeting, to a 2-day virtual gathering due to the coronavirus disease 2019 pandemic. This reduced the numbers of speakers and meant abandoning plans to hold breakout sessions and opportunities for early career researchers to present their research. However, early career investigators ranging from graduate students to assistant professors were invited to give 5-min lightning talks focused on their current research. It was also felt that there would be insufficient time to include contributions from patients with diabetes, so the powerful voice of those with lived experience was also missing. Nevertheless, the virtual format enabled a large global audience of delegates who contributed to a rich discussion after each session, which we have not included for reasons of space.

The workshop highlighted the many gaps in knowledge and questions that remain regarding the mechanisms and factors that contribute to IAH and its relationship to the counterregulatory responses. The workshop also underscored that despite advances in diabetes technologies, diabetes treatments, and education, many individuals still experience repeated episodes of hypoglycemia and the associated complications, which increase morbidity and mortality. Thus, the clinical needs of patients with type 1 diabetes and IAH are still not being adequately addressed. Patients with type 1 diabetes and control subjects without diabetes have a heterogeneous response to recurrent hypoglycemia. Understanding the factors contributing to the heterogeneity will allow us to develop personalized approaches to improve treatment. IAH arises in part through a process of habituation and autonomic plasticity. There are novel assessments that involve measurement of CBF, and use of validation tools will further help us in the future. Finally, glucose monitoring, patient education, and potentially dishabituation are key to preventing hypoglycemia and modifying impaired awareness. While there has been substantial research on IAH in recent years, there are still many unanswered questions that can be addressed with basic animal models as well as clinical research. This is an area ripe for development.

Acknowledgments. The authors acknowledge the work of Six Degrees Academy in preparing Figs. 3–5 and of Sarah Chapeau (Department of Medicine, University of Minnesota, Minneapolis, MN), who assisted with formatting.

Funding. The workshop was supported by the NIKKD, but no specific grant funding was assigned.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.