This Phase 1, single-center, two-period, crossover, randomized, double-blind study, conducted in Austria (Clinical Trials Unit, Medical University of Graz, Graz, Austria), assessed the secretion of counterregulatory hormones in response to a hypoglycemic stimulus and parameters of recovery from the nadir glucose levels to normoglycemia defined by plasma glucose (PG) 72 mg/dL (4.0 mmol/L), in tirzepatide-treated participants versus placebo. Following lead-in, eligible participants were randomized (1:1) to 12 weeks of tirzepatide (15 mg) followed by 12 weeks of placebo or vice versa ( Figure 1A ). A washout period of 8–10 weeks between the final dose of study drug in the first treatment period (on Week 12) and the first dose of study drug in the second treatment period was implemented based on the elimination half-life of tirzepatide (20). Key eligibility criteria included adults aged 18–70 years with T2D (HbA1c [6.5-9.0% [46–75 mmol/mol] at screening if taking metformin only or an HbA1c 6.0-8.5% [42–69 mmol/mol] at screening if taking metformin plus an additional oral antidiabetic medication [OAM]), and a body mass index (BMI) of 23.0-45.0 kg/m². Participants continued metformin treatment and those treated with any other OAM underwent an investigator-monitored washout of these additional drugs during a 4-week lead-in prior to randomization. A full list of eligibility criteria is in the Supplementary Appendix . The tirzepatide treatment period included stepwise dose escalation ( Supplementary Table S1 ).

This trial was conducted in accordance with the International Conference on Harmonisation Guidelines for Good Clinical Practice and the Declaration of Helsinki. All participants provided signed informed consent, and protocols were approved by local ethical review boards.

Key timepoints for the hypoglycemic clamp and a schematic are in Supplementary Table S2 ; Figure 1B . The last treatment dose for each period was given on Day 78 (± 2 days). The clamp procedure was initiated on Day 79 (clamp run-in period) and hypoglycemic induction was performed on Day 80 of each period using constant insulin i.v. infusion 2.5 mU/kg/min. The target PG level of 100 mg/dL (5.5 mmol/L) was maintained for 60 min before hypoglycemia induction, when PG was allowed to decline to 63 mg/dL (3.5 mmol/L) and subsequently to nadir (target PG 45 mg/dL [2.5 mmol/L]). At each PG level, a variable glucose infusion rate was used to maintain the target stable for 30 min and assessments of the hypoglycemic response were made, including blood sampling. PG <45 mg/dL was not permitted. If the target could not be reached within 30 min, a higher PG for the target plateau was acceptable for safety reasons. Once reached, the constant insulin infusion was terminated and the variable glucose restarted, if needed, to maintain the target nadir plateau for 30 min. The glucose infusion was then switched off to allow spontaneous recovery from the nadir to PG 72 mg/dL. If the PG had not reached ≥72 mg/dL 45 min after termination of the insulin infusion, PG was raised to 72 mg/dL using a constant glucose infusion. After recovery assessments were performed, PG was actively raised to 100 mg/dL by variable glucose infusion.

If hypoglycemia symptoms became intolerable during the hypoglycemic clamp procedure, glucose was infused to increase PG and PG concentrations above 45 mg/dL would be considered as the nadir.

The primary objective was assessment of glucagon response during induced hypoglycemia from a target PG 100 mg/dL to a nadir 45 mg/dL in people with T2D following 12 weeks of tirzepatide 15 mg versus placebo. Key secondary objectives included change in glucagon response from a target PG 100 to 63 mg/dL and to recovery from the nadir (PG 72 mg/dL), changes in glucagon response from ambient PG (measured in the pre-prandial state prior to clamp initiation) to target PG 100, 63, 45 mg/dL, and to recovery from the nadir (PG 72 mg/dL). Additional secondary objectives included assessments of changes in mean adrenaline, noradrenaline, cortisol, growth hormone (GH), insulin, and C-peptide concentrations from a target PG 100, 63, and 45 mg/dL, and to recovery from the nadir (PG 72 mg/dL), and the area under the curve for glucose infusion rate (AUC_GIR) at each PG target. Symptoms of hypoglycemia questionnaires were applied at 0 min and 30 min of each PG plateau and at recovery from the nadir, using the 7-point Edinburgh Hypoglycemia Symptom scale (21). Thirteen questions were used to measure neuroglycopenic symptoms (i.e., cognitive dysfunction including inability to concentrate, confusion, blurred vision, difficulty speaking, anxiety, and double vision, and neuroglycopenia including drowsiness, tiredness, hunger, and weakness) and autonomic symptoms (i.e., sweating, trembling and warmness). This is a subjective, validated questionnaire that measures the intensity of commonly experienced hypoglycemic symptoms where 1 represents no symptoms and 7 represents severe symptoms (i.e., a lower score indicates less severe symptoms). Sub-scores for neuroglycopenic and autonomic symptoms and overall scores were calculated. Hypoglycemic awareness was evaluated at 0 min and 30 min of each PG plateau and at recovery from the nadir based on the participant’s response to the question “Do you feel hypoglycemic” (yes/no).

Blood samples were collected for in-protocol biomarkers including glucagon, insulin and C-peptide, GH, cortisol, adrenaline, and noradrenaline at fasting state and during the hypoglycemic plateaus, were measured at a central laboratory (Labcorp Central Laboratory Services, Geneva, Switzerland). PG concentrations during hypoglycemic clamp were analyzed using SUPER GL compact (Dr. Müller Gerätebau GmbH, Freital, Germany).

Safety assessments, including evaluation of adverse events (AEs), clinical laboratory parameters, vital signs, electrocardiograms, injection site and hypersensitivity reactions, were performed throughout the hypoglycemic clamps and pharmacodynamic assessments were performed during each plateau. Outside of the hypoglycemic clamp, PG levels were regularly monitored at a minimum of once daily (fasting PG) or in case of any hypoglycemic symptoms or required by participant. Hypoglycemic events are defined in the Supplementary Appendix .

A sample size of 30 completers was determined based on the criterion that the width of the 95% confidence interval (CI) for the treatment comparison of the primary endpoint was within ±23.65 pg/mL with ~80% probability, provided that within-subject variability was 40 pg/mL (10).

Safety analyses were conducted on all randomized participants exposed to ≥1 dose of study drug (safety population), and pharmacodynamic analyses were conducted on evaluable pharmacodynamic data from all exposed participants who complete both hypoglycemic clamp procedures (pharmacodynamic analysis set). All tests of treatment effects were conducted at a 2-sided alpha=0.05.

The primary pharmacodynamic analysis assessed the treatment difference on the change in mean glucagon concentration from the target PG plateau of 100 mg/dL to the nadir target PG plateau of 45 mg/dL, using a linear mixed-effects model with treatment, treatment period, and treatment sequence as fixed effects and participants as a random effect, using a restricted maximum likelihood method. Statistical analyses for secondary continuous pharmacodynamic variables were conducted in a similar manner. A nonparametric McNemar test at each target PG plateau and at recovery from the nadir was used to assess differences in percent of participants treated with tirzepatide versus placebo who were aware of hypoglycemia.

Sensitivity analyses were conducted for the primary endpoint using only data from those participants who achieved the target PG nadir plateau of 45 mg/dL.

This trial was conducted between February 25, 2020 and January 25, 2022. Forty-two randomized participants received ≥1 dose of study drug ( Supplementary Figure S1 ). Nine participants did not complete the trial due to the COVID-19 pandemic (tirzepatide: n=2; placebo: n=2), withdrawal by participant (tirzepatide: n=2), and AEs (tirzepatide: n=2, placebo: n=1). Participants in the pharmacodynamic population (N=33) had a mean baseline age of 57.5 years, 84.8% were male with a mean duration of diabetes of 8.2 years, HbA1c 7.5% (58.8 mmol/mol), and BMI 30.9 kg/m² ( Table 1 ). Similar ranges were observed in the safety population (N=42).

Absolute glucagon concentrations were significantly lower with tirzepatide versus placebo at ambient glucose concentrations and all target PG plateaus, except PG 45 mg/dL ( Figure 2A ). From mean values of 4.48 pmol/L in the tirzepatide group and 7.45 pmol/L in the placebo group at a target PG 100 mg/dL, the primary analysis showed that increases in mean glucagon concentration in response to induced hypoglycemia from the target PG 100 mg/dL to nadir did not differ between treatments ( Table 2 ). Sensitivity analyses including only those participants who reached the target PG 45 mg/dL (tirzepatide: N=30, placebo: N=22) showed similar results (18.15 versus 20.98 pmol/L; treatment difference [95% CI], -2.83 pmol/L [-6.45, 0.80], p=0.120).

Secondary analysis showed that increases in mean glucagon concentration with tirzepatide versus placebo did not differ in response to induced hypoglycemia from the target PG 100 mg/dL to PG 63 mg/dL and to recovery from the nadir (PG 72 mg/dL) ( Table 2 ).

Absolute GH concentrations were higher with tirzepatide versus placebo at target PG 45 mg/dL ( Figure 2B ). From mean values of 1.74 ug/L in the tirzepatide group and 0.57 ug/L in the placebo group at target PG 100 mg/dL, changes in GH did not differ versus placebo at any other PG target ( Table 2 ). Absolute cortisol concentrations were lower with tirzepatide versus placebo at nadir and at recovery from the nadir (PG 72 mg/dL) ( Figure 2C ). From mean values of 328 nmol/L in the tirzepatide group and 353 nmol/L in the placebo group at target PG 100 mg/dL, increases in cortisol to nadir were smaller with tirzepatide versus placebo ( Table 2 ). Absolute adrenaline concentrations were lower with tirzepatide versus placebo at target PG 100 mg/dL and at target PG 63 mg/dL ( Figure 2D ). From mean values of 111 pmol/L in the tirzepatide group and 178 pmol/L in the placebo group at target PG 100 mg/dL, increases in adrenaline to target PG 63 mg/dL were smaller with tirzepatide versus placebo. The changes in mean adrenaline concentrations at target PG nadir of 45 mg/dL and recovery from the nadir did not differ when receiving tirzepatide versus placebo ( Table 2 ). Absolute noradrenaline concentrations did not differ between treatments at any target PG plateau ( Figure 2E ). From mean values of 2171 pmol/L in the tirzepatide group and 2013 pmol/L in the placebo group at target PG 100 mg/dL, increases in noradrenaline to target PG 63 mg/dL and to 45 mg/dL were smaller with tirzepatide versus placebo, whereas changes in mean noradrenaline concentrations from PG 100 mg/dL to recovery from the nadir did not differ between treatments ( Table 2 ).

Mean absolute insulin concentration at the PG 100 mg/dL was higher with tirzepatide versus placebo ( Figure 2F ). From mean values of 1402 pmol/L in the tirzepatide group and 1270 pmol/L in the placebo group at target PG 100 mg/dL, decreases in insulin from a target PG 100 mg/dL to a target PG 63 mg/dL and to nadir were greater with tirzepatide versus placebo ( Table 2 ). Mean absolute C-peptide concentrations at all target PG plateaus were higher with tirzepatide versus placebo ( Figure 2G ). From mean values of 967 pmol/L in the tirzepatide group and 298 pmol/L in the placebo group at target PG 100 mg/dL, decreases in C-peptide from a target PG 100 mg/dL to all target PG plateaus were greater with tirzepatide versus placebo ( Table 2 ).

Mean time to recovery from the nadir to 72 mg/dL was 47.7 min with tirzepatide and 43.3 min with placebo (estimated treatment difference [ETD] [95% CI]: 4.39 min [1.69, 7.08]; p=0.002). However, nadir PG was lower with tirzepatide (2.47 versus 2.63 mmol/L; p=0.015). Additional sensitivity analysis showed that when the nadir PG achieved was considered, the significant difference in the time taken to recover from hypoglycemia when receiving tirzepatide versus placebo was reduced (47 versus 44 min; ETD: 3.09 min [0.02, 6.17]; p=0.049).

The AUC_GIR was higher with tirzepatide during each of the PG plateaus of the hypoglycemic induction versus placebo, indicating improvement in insulin sensitivity ( Supplementary Figure S2 ).

Mean overall hypoglycemic symptom scores at the target PG plateaus of 63 and 45 mg/dL were 1.17 and 1.49 with tirzepatide and 1.35 and 1.68 with placebo, respectively ( Figure 3 , Supplementary Table S3 ). Mean overall hypoglycemic symptom scores were lower with tirzepatide versus placebo (ETD: at target PG 63 mg/dL: -0.18 [-0.31, -0.05]; p=0.010; ETD at target PG 45 mg/dL: -0.19 [-0.32, -0.06]; p=0.007). Cognitive dysfunction symptoms were significantly lower with tirzepatide at target PG 63 mg/dL (ETD -0.11 [-0.21, 0.00]; p=0.046) ( Supplementary Table S3 ), which was mainly driven by higher scores for the symptom “inability to concentrate” in the placebo group ( Supplementary Table S4 ). Neuroglycopenia symptoms were significantly lower with tirzepatide at targets PG 63 mg/dL and PG 45 mg/dL (ETD -0.31 [-0.56, -0.06] p=0.017 and -0.28 [-0.52, -0.04]; p=0.022, respectively) ( Supplementary Table S3 ), which was mainly driven by higher scores for the symptom “drowsiness” and “hunger” in the placebo group ( Supplementary Table S4 ). Finally, autonomic symptoms were significantly lower with tirzepatide at target PG 45 mg/dL (ETD -0.36 [-0.60, -0.12]; p=0.005) ( Supplementary Table S3 ), which was mainly driven by higher scores for the symptom “sweating” and “warmness” in the placebo group ( Supplementary Table S4 ). Symptoms were transient and were not associated with any adverse events.

The proportions of participants aware of hypoglycemia during induced hypoglycemia and recovery from the nadir (PG 72 mg/dL) were similar between treatments ( Supplementary Figure S3 ). All tirzepatide-treated participants answered “no” to the question “Do you feel hypoglycemic?” at the target levels of 100 mg/dL, whereas 3 (9.1%) placebo-treated participants answered “yes”.

At Week 12, mean HbA1c change from baseline was -1.5% (-16.4 mmol/L) with tirzepatide versus +0.5% (+5.8 mmol/L) with placebo. Mean body weight changes from baseline were -7.6 kg versus +1.3 kg, respectively.

Geometric mean tirzepatide concentrations ranged from 296.4 ng/mL at the end of the second week of dosing with tirzepatide 5 mg to 947.3 ng/mL at the end of the third week of dosing with tirzepatide 15 mg.

No deaths were reported, and four participants reported single serious adverse events (SAEs) during tirzepatide treatment (see Supplementary Table S5 for details on individual cases of SAEs and study discontinuation due to AEs).

The incidence of treatment-emergent AEs related to study drug was higher with tirzepatide versus placebo, and mostly gastrointestinal in nature (nausea, vomiting, diarrhea) and decreased appetite ( Supplementary Table S5 ).

No severe or clinically significant hypoglycemic events occurred outside of the hypoglycemic clamp. Nine hypoglycemic events were reported with tirzepatide (15 mg, n=7; 10 mg, n=2). Eight of these events were documented asymptomatic hypoglycemia with a PG ≤70 mg/dL (3.9 mmol/L) but ≥54 mg/dL and one was symptomatic with a PG >70 mg/dL. One placebo-treated participant reported one documented symptomatic hypoglycemic event with a PG 67 mg/dL (3.7 mmol/L). All participants promptly recovered, the majority following food intake.

Mean changes in clinical laboratory values and safety vital signs were similar between the treatment periods. Pulse rate (PR) was higher at all PG plateaus by up to 11 bpm, systolic blood pressure (BP) at the target PG 100 mg/dL and 45 mg/dL was lower by up to 8 mmHg, and diastolic BP at the target PG 100 mg/dL was lower by ~4 mmHg with tirzepatide versus placebo ( Supplementary Figure S4 ). Changes in PR and BP in response to induced hypoglycemia and recovery from the nadir (PG 72 mg/dL) did not differ between treatments.