Continuous Glucose Monitoring for Hypoglycemia Avoidance and Glucose Counterregulation in Long-Standing Type 1 Diabetes

Abstract

Context

Patients with long-standing type 1 diabetes (T1D) are at increased risk for severe hypoglycemia because of defects in glucose counterregulation and recognition of hypoglycemia symptoms, in part mediated through exposure to hypoglycemia.

Objective

To determine whether implementation of real-time continuous glucose monitoring (CGM) as a strategy for hypoglycemia avoidance could improve glucose counterregulation in patients with long-standing T1D and hypoglycemia unawareness.

Design, Setting, Participants, and Intervention

Eleven patients with T1D disease duration of ∼31 years were studied longitudinally in the Clinical & Translational Research Center of the University of Pennsylvania before and 6 and 18 months after initiation of CGM and were compared with 12 nondiabetic control participants.

Main Outcome Measure

Endogenous glucose production response derived from paired hyperinsulinemic stepped-hypoglycemic and euglycemic clamps with infusion of 6,6-²H₂-glucose.

Results

In patients with T1D, hypoglycemia awareness (Clarke score) and severity (HYPO score and severe events) improved (P < 0.01 for all) without change in hemoglobin A_1c (baseline, 7.2% ± 0.2%). In response to insulin-induced hypoglycemia, endogenous glucose production did not change from before to 6 months (0.42 ± 0.08 vs 0.54 ± 0.07 mg·kg⁻¹·min⁻¹) but improved after 18 months (0.84 ± 0.15 mg·kg⁻¹·min⁻¹; P < 0.05 vs before CGM), albeit remaining less than in controls (1.39 ± 0.11 mg·kg⁻¹·min⁻¹; P ≤ 0.01 vs all).

Conclusions

Real-time CGM can improve awareness and reduce the burden of problematic hypoglycemia in patients with long-standing T1D, but with only modest improvement in the endogenous glucose production response that is required to prevent or correct low blood glucose.

CGM may help patients with T1D improve hypoglycemia awareness and experience less severe hypoglycemia but does not restore defective glucose counterregulation.

Patients with long-standing type 1 diabetes (T1D) are at increased risk for severe hypoglycemia because of defects in glucose counterregulation and recognition of hypoglycemia symptoms. This increased risk for severe hypoglycemia is related to the progressive development of compromised physiologic defense mechanisms against a falling plasma glucose concentration in the setting of therapeutic hyperinsulinemia (1). The near-total destruction of insulin-producing β cells produces an associated defect in glucagon secretion from α cells in response to hypoglycemia, which leaves activation of the sympathoadrenal system as the only defense against hypoglycemia; epinephrine secretion contributes to endogenous (primarily hepatic) glucose production (EGP), and autonomic symptoms alert the individual to ingest food. Unfortunately, recurrent episodes of hypoglycemia blunt sympathoadrenal activation and produce a syndrome of hypoglycemia unawareness, also known as hypoglycemia-associated autonomic failure (HAAF) (2). Once established, hypoglycemia unawareness in T1D is associated with a 20-fold increased risk for experiencing severe hypoglycemia (3), which contributes significantly to disease-related morbidity (4) and mortality (5).

Strict avoidance of hypoglycemia in T1D complicated by HAAF improves sympathoadrenal responses to hypoglycemia (6–9); however, previous studies involved patients with rather short disease duration after “conventional” insulin therapy regimens and did not assess EGP, the ultimate effector mechanism required to defend against hypoglycemia. Present glycemic control strategies for strict hypoglycemia avoidance include the use of “intensive” basal-bolus insulin delivery using currently available insulin analogs as multidose injection (MDI) or continuous subcutaneous insulin infusion (“pump”) therapy in conjunction with frequent (at least four times daily) glucose monitoring. Basal insulin doses or rates and bolus insulin components are adjusted according to glucose monitoring every 1 to 2 hours and/or diagnostic continuous glucose monitoring (CGM) (10). Insulin rates and doses often vary overnight and with activity to minimize sleep- and exercise-associated hypoglycemia. Despite such intensive attention to glycemic control, some patients remain unaware of hypoglycemia, continue to experience severe hypoglycemia, and can become dependent on others for functioning in a way that is both disabling and burdensome to loved ones.

Real-time CGM is increasingly being used as a tool to help avoid hypoglycemia in T1D. CGM use in adults with T1D has been associated with modest reductions in hemoglobin A_1c (HbA_1c) (∼0.5%) from baseline levels >7.5% (11–13), especially for patients >25 years of age who use their glucose sensor at least 6 days per week (12). Although the demonstrated reductions in HbA_1c seen with use of CGM have not been associated with increases in severe hypoglycemia, rates of severe hypoglycemic events were low in these trials, which were not designed to assess hypoglycemia in patients at increased risk for experiencing a severe episode (11–15). Nevertheless, time spent in the hypoglycemic range by CGM has been shown to be significantly shorter with real-time use compared with self-monitoring of blood glucose (SMBG) (14, 16), and so CGM may help achieve hypoglycemia avoidance in patients with T1D complicated by hypoglycemia unawareness. In the current study, we sought to determine whether implementation of real-time CGM as a strategy for hypoglycemia avoidance could improve glucose counterregulation measured by the EGP response to insulin-induced hypoglycemia in patients with long-standing T1D complicated by hypoglycemia unawareness.

Participants and Methods

Participants with T1D were selected for being age 25 to 70 years, having had T1D for >10 years with absent C-peptide and involvement in intensive diabetes management (including the use of carbohydrate ratios and correction factors for flexible bolus insulin dosing under the direction of an endocrinologist, diabetologist, or diabetes specialist), and having hypoglycemia unawareness (Clarke score ≥4) (17), severely problematic hypoglycemia [HYPO score ≥1047 (90th percentile)], marked glycemic lability [glycemic lability index ≥433 mmol/L²/h·wk⁻¹ (90th percentile)], or a composite of HYPO score ≥423 (75th percentile) and LI ≥329 mmol/L²/h·wk⁻¹ (75th percentile) (18), and either at least one episode of severe hypoglycemia in the past 12 months in which the patient was unable to treat himself or herself or presence of >5% of time spent at <60 mg/dL by 72-hour blinded CGM. Patients already using real-time CGM were excluded, as were those with active cardiovascular, liver, or kidney disease. Additional protocol details are available at ClinicalTrials.gov (NCT01474889). Nondiabetic control participants were enrolled to match the patients with T1D for sex, race, age, and body mass index; they served to provide internal validity to the measured response variables to insulin-induced hypoglycemia. The University of Pennsylvania Institutional Review Board approved the study protocol, and all participants provided written informed consent to participate.

Eligible participants were informed of their device choices for implementation of real-time CGM from among those currently Food and Drug Administration–approved as adjunctive tools to ongoing SMBG and completed a minimum 7-day run-in period to confirm their understanding of and compliance with CGM. During the run-in, participants were required to wear the CGM device for at least 6 of 7 days, with a minimum of 96 hours of glucose values including at least 24 hours overnight (12, 19). Participants were instructed to continue SMBG at least four times daily for all insulin bolus dose calculations and to maintain calibration of the glucose sensor. All sensors were equipped with adjustable high- and low-glucose and rate-of-change alarms that alert the user to the need for additional blood glucose monitoring in between routine checks (19). Although the alarm thresholds were individualized, the low-glucose alarm was not set at <70 mg/dL (20) to account for the ∼15% error in accuracy (due to a physiologic lag between blood and interstitial glucose of ∼7 minutes) and ensure the blood glucose would not drop to <60 mg/dL without activating the alarm.

Sensors were also equipped with rate-of-change indicators that continuously display whether the sensor glucose is stable, increasing, or decreasing. Participants were instructed to calibrate the sensor when the rate of change was stable (change <20 mg/dL) and to modify potential insulin boluses when the sensor glucose was increasing or decreasing according to the guidelines developed by the Juvenile Diabetes Research Foundation CGM Study Group (19). Accuracy of the sensor and compliance with CGM were assessed at each visit through downloads of the CGM and blood glucose meter devices (21); participants who could not maintain an average compliance with CGM of >70% were dropped because lower compliance has not been associated with any benefit of real-time CGM to glycemic control (12).

Participant visits occurred monthly until month 6, and then every 3 months until month 18. Participants also received weekly phone calls from the study team for at least the first month and then as necessary during the first 6 months, after which the study team only returned participant calls in between study visits. This schedule was designed to determine any possible benefit of real-time CGM on glucose counterregulation after 6 months of intensive attention and provider support, and then for assessment of the durability of any beneficial effects after another 12 months of more typical provider interaction occurring every 3 months, as in clinical practice. Both CGM and glucometer downloads, as well as insulin pump downloads and SMBG and insulin dosing logs, were used to adjust basal and bolus insulin dosing to maximize hypoglycemia avoidance.

CGM data from the every-3-months-visit downloads were analyzed for metrics of mean glucose, glucose standard deviation (SD), and percentage time spent hyperglycemic (glucose level >180 mg/dL) and hypoglycemic (glucose level <60 mg/dL) (22), including for the nocturnal period (defined as 00:00 to 06:00), using HypoCount software (version 1.1; PRECISE Center, University of Pennsylvania, Philadelphia, PA). Measures of reduced hypoglycemia awareness (Clarke score), hypoglycemia severity (HYPO score), and glycemic lability were determined every 6 months as previously described (23).

Assessment of glucose counterregulation

Patients with T1D underwent paired hyperinsulinemic stepped-hypoglycemic and euglycemic clamps before and 6 and 18 months after implementation of real-time CGM. Each pair of clamps was conducted with at least 1 week and not more than 1 month between studies, with the order of hypoglycemic vs euglycemic condition determined by block randomization. Patients with T1D were admitted to the University of Pennsylvania Clinical and Translational Research Center the afternoon before the study, fasted overnight after 2000 for 12 hours before testing, and were converted from subcutaneous insulin to a low-dose intravenous insulin infusion at 2100 the evening before the study to target blood glucose at 81 to 115 mg/dL overnight. By 700, one catheter was placed in an antecubital vein for infusions, and one catheter was placed in a hand or forearm vein for blood sampling, with the hand or forearm placed in a heating pad to promote arterialization of venous blood.

At t = −120 min a primed (5 mg/kg fasting plasma glucose in mg/dL/90 for 5 minutes) continuous (0.05 mg·kg⁻¹·min⁻¹ for 355 minutes) infusion of the stable glucose isotope tracer 6,6-²H₂-glucose (99% enriched; Cambridge Isotopes Laboratories, Andover, MA) was administered to assess EGP before and during the induction of hyperinsulinemia (24). After baseline blood sampling at −20, −10, and −1 min, at t = 0 minutes a continuous infusion of insulin was initiated at 1 mU·kg⁻¹·min⁻¹ for 240 minutes to produce hyperinsulinemia. Subsequently, a variable rate infusion of 20% glucose was administered according to the glycemic clamp technique to achieve hourly plasma glucose steps of ∼80, 65, 55, and 45 mg/dL. To reduce changes in plasma enrichment of 6,6-²H₂-glucose during the clamp, the 20% glucose solution was enriched to ∼2.0% with 6,6-²H₂-glucose (24). Plasma glucose was measured every 5 minutes at bedside with an automated glucose analyzer (YSI 2300; Yellow Springs Instruments, Yellow Springs, OH) to adjust the glucose infusion rate and achieve the desired plasma glucose concentration. Additional blood samples for biochemical analysis and an autonomic symptom questionnaire (25) were collected every 20 minutes.

The hyperinsulinemic euglycemic clamp was conducted as described for the hypoglycemic clamp noted previously but with the target plasma glucose level at ∼90 mg/dL for the entire 240-minute study. Participants were blinded to the hypoglycemic vs euglycemic conditions of testing.

All samples were collected on ice into chilled tubes containing EDTA, with Protease Inhibitor Cocktail (Sigma-Aldrich, St. Louis, MO) added to the tubes for peptide hormones, centrifuged at 4°C, separated, and frozen at −80°C for subsequent analysis. Plasma glucose was verified in duplicate by the glucose oxidase method using an automated glucose analyzer (YSI 2300). Plasma insulin, glucagon, and pancreatic polypeptide were measured in duplicate by double-antibody radioimmunoassays (for insulin and glucagon: Millipore, Billerica, MA; for pancreatic polypeptide: ALPCO Diagnostics, Salem, NH). Plasma epinephrine and norepinephrine were measured by high-performance liquid chromatography with electrochemical detection. Enrichment of 6,6-²H₂-glucose was measured by gas chromatography/mass spectrometry.

Calculations and statistics

The rate of appearance (R_a) of glucose during the clamps was calculated by using the Steele non–steady-state equation modified for the use of stable isotopes, as previously described (26). EGP was calculated from the difference between the R_a of glucose in the plasma and the infusion rate of exogenous glucose. The magnitude of each hormonal, incremental symptom, and EGP response was assessed as the mean of values obtained during the last 60 minutes of hypoglycemia (27).

Data are expressed as mean ± standard error of the mean unless otherwise noted. Comparison of results within the patients with T1D from pre- to postintervention times of assessment was performed by Friedman analysis of variance; when these findings were significant, pairwise comparisons were performed by using the Wilcoxon matched-pairs test, whereas comparison of results between each T1D time of assessment and nondiabetic controls was performed with the Mann-Whitney U test using Statistica software (StatSoft Inc., Tulsa, OK). Significance was considered at P < 0.05 (two-tailed).

Results

Participant characteristics, CGM, and measures of glycemic control

Thirteen patients met eligibility criteria, enrolled in the study, and completed the run-in period and baseline clamp procedures. Two patients, one female and one male, were withdrawn from the study without completing the 6-month postintervention assessment and so were not included in the analysis. Eleven participants completed the 18-month intervention and all study visits. The sex and race distribution, age, and body mass index of these completers were similar to those of the nondiabetic controls (Table 1). The patients with T1D on average were above the American Diabetes Association target HbA_1c value of <7.0%, with long disease duration (>30 years) and requiring ∼0.5 U of insulin per kg of body weight daily delivered via an insulin pump or MDI using modern insulin analogs (Table 1).

Table 1.

Participant Characteristics

Characteristic	Patients With T1D	Nondiabetic Controls
Men/women (n/n)	5/6	5/7
White/African American (n/n)	9/2	10/2
Age (y)	44 ± 4	46 ± 2
BMI (kg/m2)	25 ± 1	25 ± 1
HbA1c (%)	7.2 ± 0.2	5.4 ± 0.1
Disease duration (y)	31 ± 4	—
Insulin requirement (U·kg−1·d−1)	0.5 ± 0.1	—
Insulin delivery (CSII/MDI)	8/3	—

Unless otherwise noted, data are mean ± standard error of the mean. Abbreviations: BMI, body mass index; CSII, continuous subcutaneous insulin infusion.

Seven participants elected to initiate CGM with the Dexcom Seven Plus or G4 device (Dexcom, San Diego, CA) and four with the Medtronic Sof-Sensor (Medtronic MiniMed, Northridge, CA); all participants completed the study using the sensor device with which they initiated the intervention. Sensor compliance was maintained at a median 100% at 6, 12, and 18 months. During the 18 months of the intervention period, insulin delivery method, insulin dose requirements, and sensor mean glucose values did not change; there was a trend toward reduced glucose variability as assessed by sensor glucose SD (P = 0.07) (Table 2). There were no differences in time spent in ranges of hyper- or hypoglycemia, including for the nocturnal period (Table 2). The incidence of severe hypoglycemia events decreased significantly (P < 0.01) (Table 2).

Table 2.

CGM Intervention and Metrics of Glycemic Control

Variable	Run-in	3 mo	6 mo	9 mo	12 mo	15 mo	18 mo
Dexcom/Medtronic (n/n)	7/4	7/4	7/4	7/4	7/4	7/4	7/4
Insulin requirement (U·kg−1·d−1)	0.5 ± 0.1	—	0.5 ± 0.0	—	0.5 ± 0.0	—	0.5 ± 0.0
Mean glucose (mg/dL)	157 ± 6	159 ± 8	165 ± 10	170 ± 12	166 ± 11	158 ± 7	157 ± 8
Glucose SDa (mg/dL)	71 ± 5	68 ± 6	66 ± 4	68 ± 6	69 ± 6	64 ± 5	63 ± 4
Time with glucose > 180 mg/dL (%)	34 ± 4	32 ± 5	36 ± 5	37 ± 6	35 ± 5	32 ± 4	32 ± 5
Time with glucose < 60 mg/dL (%)	6.5 ± 1.6	4.3 ± 1.2	4.1 ± 1.0	3.6 ± 0.8	3.9 ± 0.8	4.0 ± 0.7	4.0 ± 0.7
Dayb time with glucose < 60 mg/dL (%)	6.2 ± 1.9	3.9 ± 1.1	3.5 ± 0.7	3.6 ± 0.8	3.5 ± 0.7	3.9 ± 0.7	3.6 ± 0.6
Nocturnalc time with glucose < 60 mg/dL (%)	6.8 ± 1.6	5.7 ± 1.7	6.2 ± 2.0	4.1 ± 1.6	5.1 ± 1.7	4.8 ± 1.6	5.2 ± 1.7
No. of severe hypoglycemic events per yd	2.2 ± 0.7	—	1.4 ± 0.6	—	1.2 ± 0.7	—	0.9 ± 0.5

Unless otherwise noted, data are mean ± standard error of the mean.

^a P = 0.07.

^bDay hours defined as 0600 to 0000.

^cNocturnal hours defined as 0000 to 0600.

^d P < 0.01.

Average glucose as assessed by HbA_1c did not change (Fig. 1A). The patients with T1D experienced improvement in hypoglycemia awareness (P < 0.01) (Fig. 1B) and severity (P < 0.001) (Fig. 1D). There was also a trend for a reduction in glycemic lability (P = 0.1) (Fig. 1C).

Figure 1.

Clinical measures of average glycemic control, (A) HbA_1c, hypoglycemia unawareness, (B) Clarke score, temporal glucose variability, (C) glycemic lability index, and (D) hypoglycemia severity (HYPO score), before and throughout intervention with implementation of real-time CGM. The dotted lines give the thresholds for (A) target glycemic control, (B) reduced awareness of hypoglycemia (17), (C and D) and the 90th percentile for glycemic lability and hypoglycemia severity derived from a population of 100 patients with T1D (18). The box plots represent the median, upper and lower quartiles, mean (□), and range (error bars).

Counterregulatory responses during the hypoglycemic clamp

Insulin administration during the hypoglycemic clamp resulted in similar hyperinsulinemia in the patients with T1D before and 6 and 18 months after implementation of real-time CGM and in the nondiabetic controls; this was also not different in any group from the hyperinsulinemia achieved during their respective euglycemic control experiments (Fig. 2A). During the hypoglycemic clamp, plasma glucose by 60 minutes was near 80 mg/dL in all groups and overlapped thereafter during the 65-, 55-, and 45-mg/dL hourly steps, whereas during the euglycemic clamp, plasma glucose remained between 85 and 90 mg/dL (Fig. 2B).

Figure 2.

(A) Plasma insulin and (B) glucose during the hyperinsulinemic hypoglycemic clamp in patients with T1D before (■) and at 6 months (●) and 18 months (▲) after implementation of real-time (RT) CGM (n = 11), and in nondiabetic controls (▼, n = 12). The shaded area represents the 95% confidence interval for data derived from the hyperinsulinemic euglycemic control experiments (n = 44).

There was no difference in glucagon levels during the final hour of hypoglycemia from before to 6 and 18 months after CGM; these remained lower in patients with T1D than in controls (P < 0.01 for all comparisons) (Fig. 3A; Table 3). There was also no difference in epinephrine, which was lower in patients with T1D than in controls (P < 0.001 for all comparisons) (Fig. 3B; Table 3). EGP did not change from before to 6 months after CGM but improved by 18 months (P < 0.05 compared with before intervention), although it remained less than in controls (P ≤ 0.01 for all comparisons) (Fig. 3C; Table 3). The autonomic symptom response to hypoglycemia did not change significantly from before to 6 and 18 months after CGM, but only before the intervention was it less in the patients with T1D than in controls (P < 0.05 vs before CGM only) (Fig. 3D; Table 3). In addition, although the autonomic symptom response to hypoglycemia did not differ than under euglycemic conditions before CGM (3.7 ± 0.9 vs 2.5 ± 0.3), autonomic symptoms during hypoglycemia were greater than during euglycemia at both 6 and 18 months after CGM (5.1 ± 1.0 vs 1.5 ± 0.7 and 5.6 ± 1.2 vs 2.2 ± 0.6, respectively; P < 0.05 for both comparisons). There was no difference in norepinephrine between patients with T1D and controls (Supplemental Fig. 1; Table 3), consistent with previously published work (28). Pancreatic polypeptide did not change significantly from before to 6 and 18 months after CGM in patients with T1D and remained lower than in controls (P < 0.01 for all comparisons) (Supplemental Fig. 1; Table 3).

Figure 3.

Counterregulatory hormone [(A) glucagon, (B) epinephrine], (C) EGP, and (D) autonomic symptom responses during the hyperinsulinemic hypoglycemic clamp in patients with T1D before (■) and at 6 months (●) and 18 months (▲) after implementation of real-time (RT) CGM (n = 11), and in nondiabetic controls (▼, n = 12). The shaded area represents the 95% confidence interval for data derived from the hyperinsulinemic euglycemic control experiments (n = 44).

Table 3.

Magnitude of Counterregulatory Responses

Time From CGM	Patients With T1D			P Value a	Nondiabetic Controls
	Before	6 mo After	18 mo After
Glucagon (pg/mL)	41 ± 5	44 ± 5	53 ± 8	NS	96 ± 10b
Epinephrine (pg/mL)	152 ± 37	204 ± 37	152 ± 36	NS	568 ± 61b
EGP (mg·kg−1·min−1)	0.42 ± 0.08	0.54 ± 0.07	0.84 ± 0.15c	<0.05	1.39 ± 0.11b
Autonomic symptoms (Δ)	3.7 ± 0.9	5.1 ± 1.0	5.6 ± 1.2	NS	8.0 ± 1.9c
Norepinephrine (pg/mL)	378 ± 44	317 ± 39	362 ± 60	NS	360 ± 29
Pancreatic polypeptide (pmol/L)	31 ± 6	47 ± 12	57 ± 16	NS	139 ± 15b

Unless otherwise noted, data are mean ± standard error of the mean. The magnitude of each hormonal, EGP, and incremental symptom response to insulin-induced hypoglycemia was assessed as the mean of values obtained during the last 60 minutes of each hypoglycemic clamp.

^a P value for Friedman analysis of variance comparison of results within the patients with T1D from pre- to post-CGM intervention times of assessment.

^b P ≤ 0.01 for comparison with patients with T1D before and 6 months, and 18 months after CGM.

^c P < 0.05 for comparison with patients with T1D before CGM.

Discussion

These results demonstrate that use of real-time CGM may be associated with reduction of problematic hypoglycemia in patients with long-standing T1D complicated by hypoglycemia unawareness, without deterioration in glycemic control. Reductions in hypoglycemia unawareness, hypoglycemia severity, and the incidence of severe hypoglycemia events were documented after 6 months of intensive implementation of CGM and remained lower, in fact, during the subsequent 12 months of standard follow-up. There were trends for reductions in glycemic lability assessed by SMBG and glycemic variability assessed by sensor glucose SD, which might work together with improved awareness of hypoglycemia in limiting hypoglycemia severity (29, 30). Although time spent in the hypoglycemia range of <60 mg/dL was not significantly reduced, this small cohort study was not powered to detect the ∼40% less time on average documented in this range after implementation of real-time CGM. Nevertheless, continued exposure to hypoglycemia would account for the lack of improvement in the epinephrine response, small differences in autonomic symptoms during hypoglycemic clamp when compared with euglycemic clamp after 6 and 18 months on CGM, and only modest improvement in the EGP response to insulin-induced hypoglycemia seen by 18 months. Thus, physiologic defenses against the development of hypoglycemia remained compromised, and so the near-complete adherence to CGM use was likely responsible for most of the observed reduction in clinical hypoglycemia.

Although most studies of CGM have excluded patients experiencing problematic hypoglycemia or hypoglycemia unawareness, in clinical practice CGM is most often implemented in hopes of reducing hypoglycemia. In a retrospective clinic-based analysis, implementation of CGM for 1 year in patients with problematic hypoglycemia at baseline was associated with a reduction, but not elimination, of severe hypoglycemic events, but it did not improve hypoglycemia awareness (31). Similar to the intensive protocol for participant contact reported here, in the prospective HypoCOMPaSS trial, patients with long-standing T1D and hypoglycemia unawareness received an educational intervention and were randomly assigned to receive insulin delivery by MDI or insulin pump and glucose monitoring by SMBG with or without adjunctive CGM. After 6 months hypoglycemia awareness modestly improved and severe hypoglycemia events decreased significantly regardless of intervention arm (32). Compliance with sensor use in the CGM arm of HypoCOMPaSS was low (median, 57%), however, raising the question of whether more consistent use would have further benefited participants in reducing hypoglycemia over the benefit that has been demonstrated with education alone (33). A randomized crossover trial comparing 16 weeks of CGM with SMBG in patients with long-standing T1D and hypoglycemia unawareness demonstrated reductions in time spent with hypoglycemia, glycemic variability, and severe hypoglycemia events with CGM, but without an effect on hypoglycemia awareness (34). Taken together with the current study, these results support an effect of CGM itself on alleviating the burden of problematic hypoglycemia in long-standing T1D and further point out that at least 6 months of intervention may be required to document improvement in hypoglycemia unawareness.

The current study of prospective implementation of CGM is limited by the absence of a control arm, which prevents separation of the effect of education imparted during frequent contact from that of CGM. However, participants here were required to have a combination of marked hypoglycemia severity and glycemic lability in addition to hypoglycemia unawareness, despite already receiving intensive insulin therapy at enrollment; this placed them at substantially increased risk for severe hypoglycemia. Randomly assigning such patients to a condition without use of commercially available CGM would have been unethical. In addition to improving hypoglycemia awareness and reducing the incidence of severe hypoglycemia events with implementation of real-time CGM, this study demonstrated a significant reduction in hypoglycemia severity as measured by the HYPO score. The HYPO score includes the frequency of experiencing clinically significant hypoglycemia [<54 mg/dL (3.0 mmol/L) (35)] derived from a prospective diary and adds weight for the presence of neuroglycopenia and requirement for varying degrees of assistance. Thus, it better reflects the severity of hypoglycemia than does reporting on only the occurrence of low blood glucose and/or whether symptom awareness is impaired according to retrospective questioning. The HYPO score has previously been demonstrated to be serially consistent when assessed every 6 months over >1 year of monitoring in patients with long-standing T1D (23), and so the significant reduction reported here is best explained by the CGM intervention. Nevertheless, the resulting HYPO score after implementation of real-time CGM remained higher than that reported in a general population of patients with T1D (18), reflecting the fact that although significantly improved, hypoglycemia remained problematic for some patients; this is further evidenced by severe hypoglycemia being reduced but not abolished, as seen with other interventions, such as islet transplantation (27). Thus, the current study supports recent guidelines to consider use of real-time CGM subsequent to educational intervention to address problematic hypoglycemia in T1D, and referring for consideration of pancreas or islet transplantation those patients in whom hypoglycemia unawareness and severe hypoglycemia events persist (36).

The epinephrine response to insulin-induced hypoglycemia did not improve after implementation of real-time CGM, although participants' reporting of autonomic symptoms during the stepped-hypoglycemic clamp was higher than during the euglycemic clamp only after intervention. This finding suggests some recovery of hypoglycemia symptom recognition, as was also reported in patients with long-standing T1D and hypoglycemia unawareness after 6 months in the HypoCOMPaSS clamp substudy (37). The HypoCOMPaSS clamp substudy also demonstrated increases in plasma metanephrine levels in responses to insulin-induced hypoglycemia (37); this metabolite of epinephrine was not assessed here. Our study further suggests that up to 18 months may be required to change glucose counterregulation in this population when we first detected an improvement in the EGP response to insulin-induced hypoglycemia. The modest increase in EGP was observed without corresponding improvement in the defective glucagon and epinephrine responses and suggests partial recovery of neural or hepatic autoregulatory responses that provide hormone-independent counterregulation in defense against hypoglycemia (38–40). Although we did not see a significant increase in pancreatic polypeptide in response to insulin-induced hypoglycemia, a marker of parasympathetic activation, we did not assess sympathetic activation directly as other groups have done using microneurography (41). Nevertheless, further reduction of hypoglycemia than can be achieved by CGM alone will be necessary to more substantially improve glucose counterregulation in patients with long-standing T1D.

A limitation to instituting hypoglycemia avoidance by real-time CGM as conducted in the current study was the patients' dependence on responding appropriately to device alerts (vibration) and alarms to ingest carbohydrate and/or decrease or suspend insulin delivery to prevent or correct low blood glucose; this is particularly challenging during sleep, when nocturnal hypoglycemia contributes importantly to the development and maintenance of HAAF (2). Elimination of nocturnal hypoglycemia may be particularly amenable to automated suspension of insulin delivery with a sensor-augmented pump (42, 43). Whether this or more advanced (44) strategies may similarly benefit patients with long-standing T1D complicated by hypoglycemia unawareness and lead to clinically meaningful improvement of glucose counterregulation remains unknown but is the topic of ongoing investigation (ClinicalTrials.gov #NCT03215914).

In conclusion, patients with long-standing T1D experiencing problematic hypoglycemia despite current intensive insulin management may benefit from effects of real-time CGM on improving awareness while reducing the burden and severity of hypoglycemia. These effects may in part be related to improvement in the EGP response to insulin-induced hypoglycemia, which is required to prevent or correct low blood glucose, but substantial residual defects in glucose counterregulation remain and will require alternative approaches to achieve more thorough recovery of defense mechanisms against the development of low blood glucose.

Abbreviations:

CGM

continuous glucose monitoring

EGP

endogenous glucose production

HAAF

hypoglycemia-associated autonomic failure

HbA_1c

hemoglobin A_1c

MDI

multidose injection

SD

standard deviation

SMBG

self-monitoring of blood glucose

T1D

type 1 diabetes.