Restoration of Hypoglycemia Awareness Alters Brain Activity in Type 1 Diabetes

A quarter of adults with type 1 diabetes (T1D) cannot sense falling plasma glucose reliably, increasing the risk of severe hypoglycemia (SH) (episodes in which cognitive function is so impaired that a person cannot self-treat [1]) three- to sixfold (1,2). Protective symptom and hormone responses are attenuated in impaired awareness of hypoglycemia (IAH), a condition associated with recurrent hypoglycemia exposure (3). Although SH risk is higher in IAH, studies have reported no difference in non-SH rates between hypoglycemia aware individuals and those with IAH (1,2), implying that there may be other important contributors to IAH pathogenesis. Neuroimaging data suggest that the brain plays a key role. We have shown that hypoglycemia aware individuals possess similar brain responses to hypoglycemia as those without diabetes (4); however, adults with IAH have attenuated thalamic and frontal responses, signaling disruption of arousal and decision-making pathways (5). IAH reversal has been demonstrated through strict avoidance of hypoglycemia (6–8); however, current therapeutic interventions do not seem to offer complete or permanent restoration of awareness in all. The cerebral mechanisms behind this high-risk condition are not clearly understood. The atypical brain and behavioral responses to hypoglycemia seen in IAH may underlie therapeutic resistance. Hypoglycemia avoidance and structured education in insulin management have restored symptom and hormone responses to hypoglycemia (6–9), but whether restoration of subjective awareness can alter IAH brain activity during hypoglycemia and potentially normalize some of the atypical responses seen with IAH is unknown. To better understand the central processes of IAH and hypoglycemia awareness restoration, we used three-dimensional pseudocontinuous arterial spin labeling (3DpCASL) functional MRI, a novel and sensitive imaging technique, to measure cerebral blood flow (CBF), a surrogate marker of brain activity (10). Participants with IAH had symptom, hormone, and CBF responses to hypoglycemia measured before and after a hypoglycemia avoidance intervention (11) that consisted of specialist contact, structured education in flexible insulin therapy (Dose Adjustment for Normal Eating [DAFNE] U.K. [9]), and sensor-augmented pump therapy (Medtronic MiniMed 640G Insulin Pump System).

Right-handed adults with T1D and IAH were recruited. Individuals with major psychological diagnoses, previous significant head injury, neurological conditions expected to produce MRI changes, contraindications to MRI, renal impairment (estimated glomerular filtration rate <60 mL/min/1.73 m²), and evidence of vascular disease were excluded. The Dulwich Research Ethics Committee (London, U.K.) approved the protocol, and participants gave written informed consent. Participants were classified as having IAH on the basis of a Gold score ≥4 (1), a validated single-question measure of hypoglycemia awareness whereby individuals rate their own ability to detect hypoglycemia on a scale of 1 (always able to detect hypoglycemia) to 7 (never able to detect hypoglycemia) (12,13). The Clarke score, an eight-item questionnaire assessing experience, frequency, and severity of hypoglycemia where a score of 4–7 signifies reduced awareness and a score of ≤2 signifies intact awareness, was also completed (14). Awareness status was validated by clinical history and lack of symptom response to the experimental hypoglycemia. Glycated hemoglobin (HbA_1c), a marker of medium-term glucose control, was measured pre- and postintervention.

As previously reported (4,5), participants avoided alcohol, caffeine, and strenuous activity 48 h before the study and were admitted to King’s College Hospital the evening before scanning. Participants omitted their evening dose of basal insulin or disconnected their insulin pumps when intravenous (IV) insulin was started. A variable-rate IV insulin infusion was used to maintain blood glucose between 5 and 8 mmol/L overnight, with venous sampling every 30–60 min. After 10:00 p.m., participants only had water or preemptive hypoglycemia treatment if blood glucose fell to <4.5 mmol/L. The study would be rescheduled if blood glucose fell to <3.0 mmol/L. In the morning, a hyperinsulinemic clamp was commenced for glucose stabilization 60 min before the scan, with a target glucose of 5 mmol/L. A primed IV insulin infusion (Actrapid; Novo Nordisk, London, U.K.) replaced the overnight insulin at a maintenance rate of 1.5 mU/kg/min with a variable rate 20% glucose solution (Baxter, Norfolk, U.K.) through an MRI-compatible pump. An IV cannula was inserted into a dorsal hand vein and a heated pack applied to warm the hand and arterialize venous blood (5). Blood was obtained every 5 min, plasma extracted, and glucose measured with a glucose oxidase analyzer (YSI 2300 Stat Plus; Yellow Springs Instruments, Yellow Springs, OH). Participants lay supine on the scanner table with head stabilization applied. Within the scanner, plasma glucose was maintained at 5 mmol/L for 30 min while two 3DpCASL scans were obtained. Thereafter, plasma glucose was lowered over ∼20 min. Once 2.6 mmol/L was achieved and maintained, two further 3DpCASL scans were performed. After each 3DpCASL scan, samples were collected for norepinephrine and epinephrine, and participants reported autonomic and neuroglycopenic symptoms on a seven-point visual analog scale using a button box. Participants rated perception of 11 hypoglycemic symptoms (1 = not at all, 7 = severely). These scores were aggregated to give a total symptom score. Seven symptoms were classified as autonomic (anxiety, pounding heart, shaking, tingling, sweating, hunger, nausea) and four as neuroglycopenic (drowsiness, irritability, visual disturbance, confusion) (15). Once symptom scoring was complete, the next 3DpCASL scan commenced. Participants were blinded to their plasma glucose throughout. After completion of the scanning protocol, IV insulin was stopped and normoglycemia restored. Following glucose stabilization, participants were discharged and provided with support to avoid hypoglycemia for 48 h.

Participants progressed to a hypoglycemia awareness restoration program that was based on published evidence and a meta-analysis (11,12,16). Participants attended structured education, such as DAFNE (9), a 5-day intensive course on optimum T1D management, if they had not already done so. Technological support was provided using a Medtronic MiniMed 640G insulin pump with low-glucose suspend according to clinical need and participant consent. In weekly virtual review and one to two monthly clinic visits, recent glucose data were uploaded onto web-based platforms and discussed throughout follow-up. Practical advice was given to reinforce principles of insulin dose adjustment according to the DAFNE curriculum, supporting self-adjustment of insulin doses on the basis of carbohydrate intake, lifestyle issues, and correction of out-of-range glucose. Self-management behavior was addressed, including frequency of glucose measurements, insulin boluses per day, pump setting changes, time of insulin bolus in relation to food, management of exercise, alcohol, and other activities, as well as prompt and appropriate treatment of out-of-range glucose, particularly readings <4.0 mmol/L (11,12). Gold and Clarke scores were repeated to reassess hypoglycemia awareness, and a follow-up scan occurred once participants were able to avoid exposure to a blood glucose <3.0 mmol/L without symptoms for at least 3 weeks (6) or if not achieved, 12 months after the preintervention scan.

Epinephrine and norepinephrine were analyzed by high-performance liquid chromatography with electrochemical detection (15). Automated immunoassay was used to analyze insulin (IMMULITE 2000 XPi; Siemens Healthineers).

Statistical analyses were performed using SPSS version 22 software (IBM Corporation). IAH preintervention and postintervention mean glucose, symptom, and hormonal responses were analyzed using paired t tests. Data are presented as mean ± SD unless otherwise stated. P < 0.05 was considered statistically significant.

Power calculations for this study were performed using the mean CBF value of chosen regions of interest (ROIs). A positron emission tomography study of nine individuals without diabetes detected a 26% reduction in regional CBF in the hippocampus and a 6.4–7.8% change in CBF in the cortex and brainstem (17). In a pulsed ASL study, Mangia et al. (18) described differences of between 8 and 10% in regional CBF between 12 individuals without diabetes and 11 with IAH in the thalamus and orbitofrontal cortex (OFC) at a significance level of P < 0.02. On the basis of these data, the current study had 80% power to detect an effect size of 0.6 or a 5% change in regional CBF within the thalamic, prefrontal cortex, and hippocampus in 12 patients.

MRI images were acquired using a 3T Discovery MR750 scanner (GE Healthcare, Chicago, IL). Radiofrequency was transmitted with the scanner body coil, while signal was received with a 12-channel receive-only head coil. After an initial localizer scan, high-resolution anatomical images were acquired with an adapted 3D T1-weighted magnetization-prepared rapid gradient echo sequence using the following parameters: 1.2 mm isotropic resolution, repetition time 7.312 ms, echo time 3.01 ms, and inversion time 450 ms. CBF maps were acquired using a 3DpCASL pulse sequence to determine changes in regional resting perfusion. The sequence used four nonselective radiofrequency pulses for suppression of the static background signal, which increased sensitivity to the labeled arterial blood signal. The sequence used a labeling time of 1.5 s and a postlabeling delay of 1.5 s. Four control-label pairs were collected. Following the postlabeling delay, images were acquired using a multishot, segmented 3D fast-spin echo stack-of-spiral sequence with an effective resolution of 2 × 2 × 3 mm. A proton density image, acquired in the same series, enabled the computation of quantitative CBF maps (10). T1, T2-weighted, and fluid-attenuated inversion recovery scans were reported by a neuroradiologist, and participants were excluded if any pathology was observed.

CBF maps were analyzed using Statistical Parametric Mapping version 12 (SPM-12) (University College London). Maps were transformed to the standard space of the Montreal Neurological Institute (MNI) using a custom-built software package called Automatic Software for ASL Processing (Department of Neuroimaging, King’s College London) (19). The proton density image was coregistered to the T1 image after realigning the origin of both images. The transformation matrix of this coregistration step was then applied to the CBF map, transforming the CBF map to the space of the T1 image. Unified segmentation of the T1 image scan generated a “brain-only” binary mask. This mask was then multiplied with the CBF map in the space of the T1 image, eliminating extracerebral signal from the map. Normalization of the participant’s T1 and the skull-stripped CBF map was performed using the parameters of the unified segmentation matrices. For each participant, two normalized CBF maps obtained at euglycemia and two obtained during hypoglycemia were averaged. Finally, spatial smoothing of the averaged CBF maps was carried out using an 8-mm Gaussian smoothing kernel.

To assess the impact of hypoglycemia independent of intervention, regional differences in CBF between euglycemia and hypoglycemia were assessed using a voxelwise paired t test performed within the SPM-12 framework for preintervention and postintervention studies separately. Only clusters remaining statistically significant after familywise error (FWE) correction for multiple comparisons are reported (pFWE <0.05). Clusters of significant change were determined using the “cluster-extent” criterion (pFWE <0.05) from an uncorrected voxelwise cluster-forming threshold of P < 0.005.

To assess the effect of intervention on CBF response to hypoglycemia, a 2 × 2 flexible factorial ANOVA model within SPM-12 was used (glycemic state by intervention). Hypotheses-led analyses were applied to identify differences in CBF change pre- and postintervention; a priori–defined ROIs of the thalamus, anterior cingulate cortex (ACC), right orbitofrontal prefrontal cortex (OFC), and left hippocampus were preselected on the basis of published studies identifying key cerebral structures involved in the response to hypoglycemia (4,5,15,17,18,20). ROIs were created by applying 10-mm diameter spheres centered at independently defined, literature-based MNI coordinates (4,15) as follows: right thalamus [6, −12, 4], right OFC [46, 40, −16], right ACC [2, 38, 0], and left hippocampus [−24, −42, 2]. Statistical analyses to adjust for small volume were performed on all the voxels within each 10-mm sphere. Using the small volume correction option in SPM-12, the same familywise error–corrected statistical threshold was applied to each ROI. Peak level significance values were used, whose inference is derived from a correction for familywise error, including all the voxels of the ROI. Additional Bonferroni correction for the number of ROIs was applied, with critical α P < 0.0125. To identify regions beyond our prespecified ROIs, whole-brain analysis was performed. To identify regions commonly recruited pre- and postintervention in response to hypoglycemia, ROI conjunction analyses were performed. The null hypotheses of no hypoglycemia-induced activation pre- and postintervention were jointly rejected (21) to allow detection of hypoglycemic regional responses that remained consistent pre- and postintervention. Conjunction analyses were applied to a priori–selected ROIs (bilateral thalamus, bilateral ACC, right OFC, and left hippocampus) created through anatomical masks produced using the Wake Forest University School of Medicine PickAtlas. Additional Bonferroni correction for the number of ROIs was applied, with critical α P < 0.0125. Whole-brain conjunction analysis was also performed. All analyses were restricted to gray matter using a normalized binary mask, and global CBF was added to each model as a covariate to control for the effect of global perfusion.

Twelve participants with T1D and IAH were recruited (seven females and five males, aged 37.4 ± 9.0 years, nonobese [BMI 24.4 ± 3.6 kg/m²] with long-duration diabetes [23.6 ± 7.7 years]) and studied for 8.1 ± 2.7 months. Ten participants achieved asymptomatic hypoglycemia avoidance, as defined above, within 7.3 ± 2.2 months. Two were studied at 12 months because they did not achieve 3 consecutive weeks of hypoglycemia avoidance, despite reduced frequency of hypoglycemia and no episodes of SH during the study.

All participants were reviewed weekly by telephone, e-mail, or in person. Nine participants had completed structured education before commencing the study. Five underwent structured education within 2.1 ± 1.3 months of completing the initial scan. Ten used sensor-augmented pump therapy during the study for a mean of 5.7 ± 3.0 months each (Supplementary Table 1).

The intervention improved hypoglycemia awareness; Gold and Clarke scores reduced (preintervention 6.0 ± 1.0, postintervention 4.1 ± 1.6 [P = 0.0002] and preintervention 5.8 ± 1.5, postintervention 3.4 ± 1.8 [P = 0.0008], respectively) (Fig. 1A and B). SH rate fell (preintervention [SH rate per person in the year preceding the study] 1.5 ± 1.6, postintervention [annualized study SH rate per person per year] 0.3 ± 0.6, P = 0.03) (Fig. 1C). No difference was observed in HbA_1c (preintervention 7.8 ± 0.7% [61.6 ± 7.2 mmol/mol], postintervention 7.6 ± 0.6% [59.8 ± 6.7 mmol/mol], P = 0.48) (Fig. 1D).

Plasma glucose targets were achieved with no significant difference between pre- and postintervention (P = 0.8) (Fig. 2). Mean glucose concentrations during the euglycemic phase of ASL scan acquisition were 5.3 ± 0.3 and 5.2 ± 0.3 mmol/L for pre- and postintervention, respectively, and corresponding hypoglycemic concentrations were 2.6 ± 0.2 and 2.5 ± 0.1 mmol/L. Paired insulin concentrations were not different pre- and postintervention (P = 0.35).

Mean hypoglycemic total symptom scores increased postintervention (preintervention 18.2 ± 6.3, postintervention 25.6 ± 11.1, P = 0.02) (Fig. 1E). Postintervention, mean autonomic symptom score increased significantly (P = 0.047), while neuroglycopenic symptom score did not change (P = 0.16). Peak epinephrine hypoglycemic responses were low and did not change (preintervention 0.8 ± 0.7 nmol/L, postintervention 1.0 ± 0.9 nmol/L, P = 0.2) (Fig. 1F). Peak norepinephrine hypoglycemic responses also did not change (preintervention 1.0 ± 0.7 nmol/L, postintervention 1.0 ± 0.7 nmol/L, P = 0.8).

Preintervention, hypoglycemia induced left OFC and left dorsolateral prefrontal cortex (DLPFC) CBF increases, with decreases in the right temporal cortex (pFWE <0.05) (Fig. 3A). Independent analysis of postintervention scans demonstrated precuneus CBF increases bilaterally, with right temporal cortex decreases (pFWE <0.05) (Fig. 3B).

Postintervention ROI analysis demonstrated greater increases in right ACC CBF (pFWE = 0.011, peak coordinates [6, 40, −2], t score 3.14) and greater decreases in left hippocampal CBF (pFWE = 0.023, peak coordinates [−24, −38, 4], t score 2.76) in response to hypoglycemia (Table 1). ACC differences survived Bonferroni correction (critical α P < 0.0125). Whole-brain ANOVA demonstrated no significant difference postintervention (pFWE >0.05).

ROI conjunction analyses demonstrated common responses pre- and postintervention within the left thalamus (pFWE = 0.01, peak voxel coordinates [−12, −26, −2]); conjunction survived Bonferroni correction for the number of ROI tests (critical α pFWE <0.0125) (Table 1). Conjunction analyses across the whole brain demonstrated common left OFC and left DLPFC recruitment pre- and postintervention during hypoglycemia (pFWE = 0.019, cluster size 1,956 voxels) (Table 1).

IAH is a major risk factor for SH in T1D (1,2). Associated with attenuated symptom and hormone responses to hypoglycemia, IAH cerebral responses to hypoglycemia are also different from their counterparts with hypoglycemia awareness and without diabetes (5,18,20,22). IAH has been reversed by hypoglycemia avoidance, with previous studies showing recovery of symptoms but variable recovery of counterregulatory hormone responses (6–8,16). To our knowledge, this is the first study to investigate whether the cerebral responses of IAH are also reversible. We have shown that an intervention of structured education, technology, and close clinical contact improves subjective awareness of hypoglycemia without deterioration of glycemic control. Hypoglycemic symptom scores increased postintervention, despite an equivalent hypoglycemic stimulus. The key difference in neuroimaging postintervention was a greater CBF increase within the ACC, a region involved in autonomic control, interoception (awareness of internal homoeostatic state), and complex decision making (23–25). However, responses within the thalamus, involved in arousal, sensory and motor relay, and cortical signal transmission (26–28), and left OFC and DLPFC, involved in executive function (24,29), were not altered.

Participants had enhanced hypoglycemia perception postintervention, as shown by significantly lower Gold and Clarke scores; higher symptom scores during experimental hypoglycemia; and significantly reduced SH. Our data fit with published studies demonstrating that education, clinical contact, and technology are effective at reducing SH (9,12), improving hypoglycemia awareness (9,12), and restoring symptom responses to experimental hypoglycemia (6,7). A significant increase in the hypoglycemic autonomic symptom response was seen; however, epinephrine secretion was not different postintervention. These findings are consistent with Dagogo-Jack et al. (7), who demonstrated restoration of symptoms and awareness but not hormonal responses after hypoglycemia avoidance in people with IAH. Although hormone response restoration has been described (6,30), we have previously shown that hypoglycemia aware participants exhibit robust autonomic symptoms despite significantly reduced epinephrine responses compared with those without diabetes (4), and individuals who have undergone adrenalectomy have been shown to have intact symptom responses to hypoglycemia (31). Taken together, these data suggest that although robust counterregulatory responses are a critical factor in preventing hypoglycemia, they are not a prerequisite for subjective awareness of hypoglycemia. Parenthetically, HbA_1c did not increase, providing further evidence that glycemic targets need not be relaxed to prevent problematic hypoglycemia.

Preintervention, hypoglycemia induced left OFC and DLPFC CBF increases, regions involved in executive function. Increases in right prefrontal CBF are seen in individuals without diabetes during hypoglycemia (4); activity within the left prefrontal cortex in IAH may be another diversion from the usual response to hypoglycemia. The left DLPFC has been shown to be less activated than the right during executive function tasks (32), and participants with left DLPFC lesions demonstrated no deficit in working memory in contrast to those with right lesions (33). Hypoglycemia also induced CBF reductions in the right temporal cortex, an area involved in processing of sensory input and memory, a response also seen in individuals without diabetes and hypoglycemia aware individuals (4,15,17). These changes may underlie the memory loss known to occur during hypoglycemic episodes in those with and without diabetes, independent of awareness status (34). Postintervention, a reduction in right temporal CBF was again seen during hypoglycemia; however, increases in precuneus CBF, a region involved in attention, memory, and self-reflection (35), were also observed. Despite being an isolated finding not detected in our formal pre- and postintervention comparison, this region may play a part in cerebral responses to hypoglycemia, particularly because the precuneus has connections with higher association regions such as the prefrontal cortex and thalamus (35), structures known to be activated during hypoglycemia (15,20). We used cluster-extent criterion to identify clusters of significant CBF change in response to hypoglycemia within the preintervention and postintervention data separately. Because these paired t test maps could not be compared directly, a 2 × 2 flexible factorial ANOVA model was used to compare pre- and postintervention data.

Hypoglycemia-induced increases within the ACC were greater after our intervention. The ACC has been associated with autonomic control, interoception, and complex decision making (23–25). With strong reciprocal connections to the OFC and DLPFC, the ACC is pivotal in evaluating events that require behavioral modification and monitoring performance and outcome (36). We have previously reported a trend toward increased ACC CBF during hypoglycemia in individuals without diabetes compared with those with hypoglycemia awareness (4); ACC activity of postintervention participants with IAH appears to reflect that observed in those with no prior hypoglycemia exposure. We speculate that a greater ACC response contributes to reestablishing appropriate hypoglycemia perception, appraisal, and action. A signal toward a greater reduction in hippocampal CBF was noted postintervention. Although this did not survive Bonferroni correction, the fall in hippocampal CBF seen in participants with IAH postintervention reflects that seen in individuals without diabetes and hypoglycemia awareness (4,5,15), implying partial recovery of typical hippocampal responses to hypoglycemia. Whole-brain ANOVA demonstrated no significant differences postintervention. This was not unexpected because we had anticipated that postintervention changes would be subtle and require more focused ROI analysis for detection.

ROI conjunction analyses demonstrated a significant common response within the thalamus during hypoglycemia pre- and postintervention, while across the whole brain, the left OFC and DLPFC were commonly recruited pre- and postintervention. The thalamus is believed to mediate corticocortical signaling through multiple reciprocal and nonreciprocal cortical connections, including prefrontal networks (26), while the OFC and DLPFC are part of a network with the ACC, temporal association cortex, hippocampus, and amygdala recruited for motivation and emotional processing (24). As described in a neuroanatomical model for decision making by Wallis et al. (36), the OFC evaluates anticipated reward outcomes on the basis of information held in working memory, the DLPFC selects and prioritizes the behavioral response on the basis of OFC appraisal, while the ACC evaluates the likelihood of success. We have previously shown that there are diminished responses to hypoglycemia within the thalamus and frontal cortex of those with IAH compared with those with hypoglycemia awareness (5). We speculate that disruption of these networks may be a factor in maladaptive behaviors and beliefs noted in some with IAH associated with reduced ability to achieve hypoglycemia avoidance, such as inappropriately low concern despite high comorbid risk, hypoglycemia risk minimization, and preoccupation with hyperglycemia avoidance (37–40).

Our intervention did not change thalamic and OFC responses to hypoglycemia, despite restoring awareness and increasing ACC responses. Despite the success of structured education in reducing SH, improving glycemic control, and easing psychological distress, Hopkins et al. (9) reported that only 43% of participants with IAH demonstrated an improvement in hypoglycemia recognition. Cranston et al. (6) successfully restored hypoglycemia awareness in 12 study participants through education and professional support; however, 1 participant later died because of SH complications (41). Frontothalamic pathway disruption that persists despite reducing IAH risk, as in our intervention, may underlie persistent, difficult-to-treat IAH. The lack of epinephrine change despite an increase in symptom response to hypoglycemia seen here, together with noted dissociation between the adrenal medulla and adrenergic symptoms (31), also supports the hypothesis that top-down control is a key factor in hypoglycemia awareness.

Acquired, restored awareness of hypoglycemia, as seen in our participants postintervention, rather than preserved, inherent awareness, as seen in hypoglycemia aware individuals who have never experienced IAH, may preferentially recruit regions not only involved in arousal and autonomic function but also involved in decision making and emotion processing. These data imply that the hypoglycemic cerebral mechanisms of these two groups are different. We have previously shown hypoglycemia-induced increases in CBF within the thalamus of those with intact hypoglycemia awareness (5), whereas in the current study of restored awareness, a difference in ACC CBF was seen. We speculate that heightened ACC activity may compensate for the loss of thalamic action and contribute to symptom control in restored awareness. Indeed, the ACC has been implicated in multiple cognitive functions, including impulse control and complex decision making (23–25).

To conclude, to our knowledge, this is the first study to examine neuroimaging correlates of restoration of hypoglycemia awareness. Previous work demonstrating differences in brain responses to hypoglycemia between IAH and hypoglycemia awareness could not address whether these differences were part of the inducible syndrome of IAH or were preexisting differences that predispose some individuals to IAH. Our finding that ACC responses can be altered in parallel with restoration of subjective awareness demonstrates plasticity within the ACC response and highlights the importance of this region in brain responses to hypoglycemia. Lack of restoration of thalamic and frontal responses to hypoglycemia may still represent a predisposition to IAH or that meticulous avoidance of hypoglycemia through education and technology is not sufficient to address cognitive or behavioral drivers of hypoglycemia awareness. These data suggest that subjective awareness of hypoglycemia is a composite of functioning symptomatic, hormonal, thalamocortical, and behavioral pathways. Loss of one or more of these elements may be characteristic of IAH. We speculate that to achieve robust resolution of IAH in those at risk, each factor, namely loss of symptom responses, attenuation of hormonal action, cerebral adaptations, and behavioral barriers, requires careful optimization.

Acknowledgments. The authors thank the participants; the King’s College diabetes research team and nursing staff Megan L. Byrne, Bula M. Wilson, and Andrew Pernet; Sally M. Cordon, the research technician at Nottingham University Medical School, Queen’s Medical Centre; the clinical research staff at the National Institute for Health Research and Wellcome Trust King’s Clinical Research Facility Louisa Green and John Lord Villajin; the laboratory staff at Viapath, King’s College Hospital, Tracy Dew, Gemma Cross, Andrew Given, and Joseph Molloy; and the radiographers and administrative staff at the Centre of Neuroimaging Sciences, King’s College London.

Funding and Duality of Interest. This study was funded by Diabetes UK registered charity project grant 13/0004653. Investigator-initiated study approval was granted for this study by Medtronic UK (A 2265657), who provided Medtronic MiniMed 640G insulin pumps, Guardian 2 Link transmitters, and Enlite glucose sensors for use in this study. M.N. has received travel grant support from Sanofi, Janssen, and Eli Lilly and reports prior committee membership of the Young Diabetologists’ and Endocrinologists’ Forum in the U.K., which uses unrestricted sponsorship from industry partners to deliver educational programs for health care professionals. S.A.A. is a member of the International Hypoglycaemia Study Group, is co-investigator on the European Union Innovative Medicines Initiative project HypoRESOLVE (Hypoglycaemia–Redefining Solutions for Better Lives), and has served on advisory panels for Abbott UK, Medtronic, and Novo Nordisk. I.A.M. is a member of the scientific advisory boards of Nestlé Research, Novozymes, AIJN (European Fruit Juice Consortium), International Life Sciences Institute Europe Dietary Carbohydrate Task Force, Mars Inc., and Waltham Pet Healthcare Research Institute and a member of the Scientific Advisory Committee on Nutrition, including a working group with NHS England and Diabetes UK on high-fat diets in diabetes management and the Medical Research Council nutrition grants panel. P.C. reports consultant fees and speaker honoraria from Medtronic; consultant fees from Dexcom, Roche Diabetes Care, Novo Nordisk, Sanofi Diabetes, Lilly Diabetes, Novartis, Ascensia Diabetes, and Insulet; and committee membership of the Diabetes Technology Network in the U.K., which uses unrestricted sponsorship from industry partners to deliver educational programs for health care professionals. No other potential conflicts of interest relevant to this article were reported.

Medtronic UK had no role in the study design; collection, analysis, and interpretation of data; writing of the report; and decision to submit the article for publication.

Author Contributions. M.N. recruited the participants, collated and analyzed study data, and wrote the manuscript. M.N., S.A.A., and P.C. performed the study. M.N., O.O., and F.O.Z. analyzed the neuroimaging data. I.A.M. analyzed the catecholamine data. M.N., S.A.A., O.O., I.A.M., F.O.Z., and P.C. contributed to the research concept and design, interpreted the data, and reviewed and edited the manuscript. M.N. and P.C. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 55th European Association for the Study of Diabetes Annual Meeting, Barcelona, Spain, 16–20 September 2019.