Hypoglycemia and Cardiovascular Outcomes in the CARMELINA and CAROLINA Trials of Linagliptin: A Secondary Analysis of Randomized Clinical Trials

Nikolaus Marx; Ahmed A Kolkailah; Julio Rosenstock; Odd Erik Johansen; Mark E Cooper; John H Alexander; Robert D Toto; Christoph Wanner; Mark A Espeland; Michaela Mattheus; Sven Schnaidt; Vlado Perkovic; Nicholas D Gollop; Darren K McGuire

doi:10.1001/jamacardio.2023.4602

. 2024 Jan 3;9(2):134–143. doi: 10.1001/jamacardio.2023.4602

Hypoglycemia and Cardiovascular Outcomes in the CARMELINA and CAROLINA Trials of Linagliptin

A Secondary Analysis of Randomized Clinical Trials

Nikolaus Marx ¹, Ahmed A Kolkailah ^2,^✉, Julio Rosenstock ³, Odd Erik Johansen ⁴, Mark E Cooper ⁵, John H Alexander ⁶, Robert D Toto ⁷, Christoph Wanner ⁸, Mark A Espeland ⁹, Michaela Mattheus ¹⁰, Sven Schnaidt ¹¹, Vlado Perkovic ¹², Nicholas D Gollop ¹³, Darren K McGuire ^2,¹⁴

¹Department of Internal Medicine I, University Hospital Aachen, RWTH Aachen University, Aachen, Germany

²Division of Cardiology, Department of Internal Medicine, The University of Texas Southwestern Medical Center, Dallas

³Dallas Diabetes Research Center at Medical City, Dallas, Texas

⁴Therapeutic Area Cardiometabolism, Boehringer Ingelheim KS, Asker, Norway

⁵Department of Diabetes, Central Clinical School, Monash University, Melbourne, Victoria, Australia

⁶Duke Clinical Research Institute, Duke Health, Durham, North Carolina

⁷Department of Internal Medicine, Division of Nephrology, The University of Texas Southwestern Medical Center, Dallas

⁸Division of Nephrology, Department of Medicine, University Hospital Würzburg, Würzburg, Germany

⁹Division of Gerontology and Geriatric Medicine, Wake Forest School of Medicine, Winston-Salem, North Carolina

¹⁰Biostatistics and Data Sciences, Boehringer Ingelheim Pharma GmbH & Co KG, Ingelheim am Rhein, Germany

¹¹Biostatistics and Data Sciences, Boehringer Ingelheim Pharma GmbH & Co KG, Biberach an der Riß, Germany

¹²Renal and Metabolic Division, The George Institute for Global Health, University of New South Wales Sydney, Newtown, New South Wales, Australia

¹³Therapeutic Area Cardiometabolism, Boehringer Ingelheim International GmbH, Ingelheim am Rhein, Germany

¹⁴Parkland Health, Dallas, Texas

Accepted for Publication: October 6, 2023.

Published Online: January 3, 2024. doi:10.1001/jamacardio.2023.4602

^✉

Corresponding Author: Ahmed A. Kolkailah, MD, MSc, Division of Cardiology, Department of Internal Medicine, The University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390-8830 (ahmed.kolkailah@utsouthwestern.edu).

Author Contributions: Drs Marx and Kolkailah had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Marx and Kolkailah equally contributed as co–first authors.

Concept and design: Marx, Kolkailah, Johansen, Alexander, Toto, Wanner, Mattheus, Schnaidt, Gollop, McGuire.

Acquisition, analysis, or interpretation of data: Marx, Kolkailah, Rosenstock, Cooper, Wanner, Espeland, Mattheus, Schnaidt, Perkovic, Gollop, McGuire.

Drafting of the manuscript: Marx, Kolkailah, McGuire.

Critical review of the manuscript for important intellectual content: Kolkailah, Rosenstock, Johansen, Cooper, Alexander, Toto, Wanner, Espeland, Mattheus, Schnaidt, Perkovic, Gollop, McGuire.

Statistical analysis: Johansen, Mattheus, Schnaidt.

Obtained funding: Johansen.

Administrative, technical, or material support: Rosenstock, Toto, Wanner, Gollop.

Supervision: Marx, Kolkailah, Wanner, Perkovic, Gollop, McGuire.

Conflict of Interest Disclosures: Dr Marx reported receiving grants from German Research Foundation (TRR 219; Project-ID 322900939 [M03, M05]) and other contract support paid to University Hospital Aachen from Amgen, Boehringer Ingelheim, Sanofi-Aventis, Merck Sharp & Dohme, Bristol Myers Squibb, AstraZeneca, Eli Lilly and Company, Novo Nordisk, and Bayer during the conduct of the study. Dr Kolkailah was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under award T32HL125247 during the conduct of the study. Dr Rosenstock reported receiving grants and personal fees from Boehringer Ingelheim during the conduct of the study as well as grants and personal fees from Boehringer Ingelheim, Eli Lilly and Company, Novo Nordisk, Sanofi, and Oramed; grants from Merck, Novartis, and Pfizer; and personal fees from Zealand outside the submitted work. Dr Johansen reported being an employee of Boehringer Ingelheim during the conduct of the study. Dr Cooper reported receiving honoraria from Boehringer-Ingelheim and Astra Zeneca and grants from Boehringer-Ingelheim and NovoNordisk. Dr Alexander reported receiving grants from Boehringer Ingelheim during the conduct of the study as well as grants from Bayer, CSL Behring, and Humacyte; personal fees from AbbVie, Akros, AtriCure, Bayer, GlaxoSmithKline, Janssen, Pfizer, Portola, Novostia, and Veralox; and grants and personal fees from Artivion/Cryolife, Bristol Myers Squibb, and Ferring outside the submitted work. Dr Toto reported receiving personal fees from Boehringer Ingelheim during the conduct of the study and outside the submitted work as well as grants from the Mary M. Conroy Professorship in Kidney Diseases and the Houston J and Florence A Dosswell Center for Development of New Approaches for the Treatment of Hypertension during the conduct of the study. Dr Wanner reported receiving personal fees from Boehringer Ingelheim, AstraZeneca, Eli Lilly and Company, MSD, and Bayer during the conduct of the study. Dr Espeland reported receiving personal fees from Boehringer Ingleheim during the conduct of the study, grants from the National Institutes of Health and Alzheimer's Association outside the submitted work, and personal fees from Nestle Corporation outside the submitted work. Ms Mattheus reported being an employee of Boehringer Ingelheim Pharma GmbH & Co KG during the conduct of the study. Mr Schnaidt reported being an employee of Boehringer Ingelheim Pharma GmbH & Co KG during the conduct of the study. Professor Perkovic reported receiving other from Boehringer Ingelheim during the conduct of the study and personal fees from AbbVie, Amgen, Astellas, AstraZeneca, Janssen, Novo Nordisk, Otsuka, Novartis, Bayer, Chinook Therapeutics, Dimerix, George Clinical, Gilead, GlaxoSmithKline, Travere, Medimmune, Mitsubishi Tanabe, Mundipharma, and Vifor Pharma outside the submitted work. Dr Gollop reported being an employee of Boehringer Ingelheim International GmbH during the conduct of the study. Dr McGuire reported personal fees and nonfinancial support from Boehringer Ingelheim during the conduct of the study and personal fees from AstraZeneca, Boehringer Ingelheim, Esperion, Pfizer IDMC, Novo Nordisk, Eli Lilly and Company, CSL Behring IDMC, New Amsterdam Pharma, Bayer, Lexicon, Altimmune, Intercept, and Applied Therapeutics outside the submitted work. No other disclosures were reported.

Funding/Support: This study was sponsored by the Boehringer Ingelheim and Eli Lilly and Company Diabetes Alliance.

Role of the Funder/Sponsor: Boehringer Ingelheim, through their employees and coauthors of this study, had a role in the design and conduct of the study; management, analysis, and interpretation of the data; and preparation, review, and approval of the manuscript. Other funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.

Additional Contributions: Giles Brooke, PhD, Elevate Scientific Solutions, was contracted and compensated by Boehringer Ingelheim to provide medical writing assistance.

^✉

Corresponding author.

PMCID: PMC10765314 PMID: 38170502

This secondary analysis evaluates antihyperglycemic treatments, hypoglycemic episodes, and risk for cardiovascular events and mortality in 2 randomized clinical trials involving adults with type 2 diabetes.

Key Points

Question

Is the association between hypoglycemia and cardiovascular (CV) events in type 2 diabetes causal or reflective of individuals susceptible to both outcomes?

Findings

In this secondary analysis of 2 randomized clinical trials (CARMELINA [involving advanced diabetes] and CAROLINA [involving early-stage diabetes]) of linagliptin, an antihyperglycemic drug, associations were found between hypoglycemia and subsequent CV events as well as between nonfatal CV events and subsequent hypoglycemia in the CARMELINA trial but not in the CAROLINA trial.

Meaning

The findings of this study challenge the notion that hypoglycemia is associated with adverse CV events and suggest that hypoglycemia and CV events most likely identify a patient phenotype susceptible to or at high risk for both.

Abstract

Importance

Previous studies have reported an association between hypoglycemia and cardiovascular (CV) events in people with type 2 diabetes (T2D), but it is unclear if this association is causal or identifies a high-risk patient phenotype.

Objective

To evaluate the associations between hypoglycemia and CV outcomes.

Design, Setting, and Participants

This secondary analysis was a post hoc assessment of the multinational, double-blind CARMELINA (Cardiovascular and Renal Microvascular Outcome Study With Linagliptin; 2013-2016) and CAROLINA (Cardiovascular Outcome Trial of Linagliptin vs Glimepiride in Type 2 Diabetes; 2010-2018) randomized clinical trials of the antihyperglycemic drug, linagliptin, a dipeptidyl peptidase 4 inhibitor. Participants were adults with T2D at high CV risk with or without high kidney risk. By design, participants in the CARMELINA trial had longer duration of T2D and had a higher CV risk than participants in the CAROLINA trial. Data analyses were conducted between June 2021 and June 2023.

Intervention

Linagliptin or placebo in the CARMELINA trial, and linagliptin or glimepiride in the CAROLINA trial.

Main Outcomes and Measures

The primary outcome for both trials was CV death, myocardial infarction (MI), or stroke (3-point major adverse CV events [3P-MACE]). For the present analyses, hospitalization for heart failure (HF) was added. Hypoglycemia was defined as plasma glucose less than 54 mg/dL or severe hypoglycemia (episodes requiring the assistance of another person). Associations between the first hypoglycemic episode and subsequent CV events and between nonfatal CV events (MI, stroke, hospitalization for HF) and subsequent hypoglycemic episodes were assessed using multivariable Cox proportional hazards regression models. Sensitivity analyses explored the risk of CV events within 60 days after each hypoglycemic episode.

Results

In the CARMELINA trial (6979 patients; 4390 males [62.9%]; mean [SD] age, 65.9 [9.1] years), there was an association between hypoglycemia and subsequent 3P-MACE plus hospitalization for HF (hazard ratio [HR], 1.23; 95% CI, 1.04-1.46) as well as between nonfatal CV events and subsequent hypoglycemia (HR, 1.39; 95% CI, 1.06-1.83). In the CAROLINA trial (6033 patients; 3619 males (60.0%); mean [SD] age, 64.0 [9.5] years), there was no association between hypoglycemia and subsequent 3P-MACE plus hospitalization for HF (HR, 1.00; 95% CI, 0.76-1.32) and between nonfatal CV events and subsequent hypoglycemia (HR, 1.44; 95% CI, 0.96-2.16). In analyses of CV events occurring within 60 days after hypoglycemia, there was either no association or too few events to analyze.

Conclusions and Relevance

This study found bidirectional associations between hypoglycemia and CV outcomes in the CARMELINA trial but no associations in either direction in the CAROLINA trial, challenging the notion that hypoglycemia causes adverse CV events. The findings from the CARMELINA trial suggest that both hypoglycemia and CV events more likely identify patients at high risk for both.

Trial Registration

ClinicalTrials.gov Identifier: NCT01897532 (CARMELINA) and NCT01243424 (CAROLINA)

Introduction

Hypoglycemia is a feared adverse effect of antihyperglycemic therapy due to its unpleasant symptoms and the associated medical risk of each episode, which interfere with daily activities and impose psychosocial stress on both patients and their caregivers. Beyond these concerns, numerous studies have found an association between hypoglycemia episodes and hypoglycemia burden with subsequent risk for adverse cardiovascular (CV) outcomes, leading to the hypothesis that avoidance of hypoglycemia in the treatment of patients with type 2 diabetes (T2D) may mitigate CV risk.^{1,2,3,4,5,6,7} However, it remains a matter of debate whether the observed associations represent a direct causal link between hypoglycemia and increased risk for adverse CV events or mortality or merely highlight a frail phenotype of patients who are susceptible to both events.^1,8,9

Two CV outcome trials designed to assess the CV safety of the antihyperglycemic medication linagliptin, a dipeptidyl peptidase 4 (DPP-4) inhibitor, provide a unique opportunity to assess the interrelationship between antihyperglycemic treatment, hypoglycemia, and risk for CV events and mortality in a broad spectrum of patients with T2D. The Cardiovascular and Renal Microvascular Outcome Study With Linagliptin (CARMELINA) trial in patients with a long duration of T2D and prevalent chronic kidney disease demonstrated the CV noninferiority of linagliptin vs placebo, and the Cardiovascular Outcome Trial of Linagliptin vs Glimepiride in Type 2 Diabetes (CAROLINA) trial in patients with relatively early T2D demonstrated the CV noninferiority of linagliptin vs glimepiride.¹⁰ This secondary analysis aimed to evaluate the associations between hypoglycemia and CV outcomes among patients in the CARMELINA and CAROLINA trials.

Methods

Study Design

The rationale, design, and results of the CARMELINA and CAROLINA trials have been previously described in detail.^10,11,12,13 In brief, both CARMELINA (placebo comparator) and CAROLINA (active comparator) were multinational, double-blind, randomized clinical trials designed to evaluate CV outcomes with linagliptin vs comparator in adults with T2D, with the composite of CV death, nonfatal myocardial infarction (MI), or nonfatal stroke (3-point major adverse CV events [3P-MACE]) as the primary outcome. The institutional review board or independent ethics committee at each site of the CARMELINA and CAROLINA trials approved the study protocols, which included secondary analyses. All patients provided written informed consent.

The CARMELINA trial enrolled individuals with a hemoglobin A_1c (HbA_1c) of 6.5% to 10.0% (48-86 mmol/mol) who had a history of CV disease together with a urine albumin-to-creatinine ratio (UACR) higher than 30 mg/g and/or an estimated glomerular filtration rate (eGFR) of 45 to 75 mL/min/1.73 m² or greater with a UACR higher than 200 mg/g or an eGFR of 15 to less than 45 mL/min/1.73 m² regardless of UACR. Both medication-naive individuals and those receiving antihyperglycemic medications were permitted to participate, except individuals using DPP-4 inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, and/or sodium-glucose cotransporter 2 inhibitors. Participants were randomized to receive linagliptin or placebo added to existing therapy.^11,12

The CAROLINA trial enrolled individuals with an HbA_1c of 6.5% to 8.5% (48-69 mmol/mol) and elevated CV risk (established CV disease or CV risk factors). Medication-naive individuals and those receiving antihyperglycemic medications were permitted to participate, except individuals using insulin, DPP-4 inhibitors, GLP-1 receptor agonists, and/or thiazolidinediones. Participants were randomized to receive linagliptin or glimepiride.^10,13

Assessment of Hypoglycemia and Clinical Outcomes

Consistent with contemporary consensus definitions, episodes of hypoglycemia occurring during the CARMELINA and CAROLINA trials were captured using common processes and definitions across the trials.^14,15 Hypoglycemia events reported by trial investigators were classified into prespecified categories, including plasma glucose less than 54 mg/dL (to convert to millimoles per liter, multiply by 0.0555; ie, <3.0 mmol/L) and severe hypoglycemia (ie, episodes requiring the assistance of another person to actively administer carbohydrate, glucagon, or other resuscitative actions). In addition to the primary CV outcome of 3P-MACE, other CV and mortality outcomes were prospectively captured, and all CV events were centrally adjudicated in both trials by clinical events committees who were blind to treatment assignment. For the present analyses, the following outcomes were evaluated: composite of 3P-MACE plus hospitalization for heart failure (HF), 3P-MACE, CV death, non-CV death, all-cause mortality, fatal or nonfatal MI, fatal or nonfatal stroke, and hospitalization for HF.

Statistical Analysis

Time to first hypoglycemic episode was analyzed with a Cox proportional hazards regression model, with treatment group and geographical region (in the CARMELINA trial) as factors. In addition, Kaplan-Meier estimates were presented. Analyses were conducted in randomized patients who received 1 or more doses of the study drug. Data for patients who did not have an event were censored on the last day they were known to be free of the outcome.

Each trial data set was analyzed separately for associations between hypoglycemia events and selected clinical outcomes using multivariable Cox proportional hazards regression models that included a time-varying covariate for hypoglycemia, a factor for treatment, and the following baseline covariables: region; age; sex; smoking status; albuminuria and previous macrovascular disease; history of HF; atrial fibrillation; T2D duration; body mass index; HbA_1c; eGFR; and treatment with sulfonylurea, glinide, and (for the CARMELINA trial) insulin. Any episodes of hypoglycemia prior to the first event of an outcome, or end of follow-up for an outcome, were considered.

For the time-dependent analyses, the patient’s status changed from nonexposed to hypoglycemia to exposed to hypoglycemia at the first time of a hypoglycemic episode and was kept until the first occurrence of a CV event or censoring for a CV event. For sensitivity analyses, the time-dependent covariate was introduced such that the patient was considered to be exposed to hypoglycemia at the time of each episode until 60 days thereafter.¹⁶ Risk-interval plots were used to demonstrate the respective temporal associations.

Evidence of modification of hypoglycemia on the treatment effect of linagliptin vs glimepiride for CV events was assessed with the same multivariable Cox proportional hazards regression model used for analysis of the primary outcome, which included a term for treatment assignment,¹⁰ and a term for time-varying covariate for hypoglycemia episode was added. To investigate the reverse associations between the time-dependent covariate of nonfatal CV events (primary and secondary outcomes) and hypoglycemia to indirectly assess the temporality of the associations, similar methods were used. The Cox proportional hazards regression models were replicated with substitution of the outcome event as time to the first event of hypoglycemia and included the time-dependent variable of the respective nonfatal CV events of nonfatal MI or stroke or hospitalization for HF in the model as a covariate. In exploratory analyses, we also evaluated the pooled data from the 2 trials, but given the extreme heterogeneity of the 2 trials, these results should be interpreted with caution.

All analyses were defined post hoc and were conducted using SAS, version 9.4 (SAS Institute Inc). A nominal 2-sided α = .05 was considered the criterion for statistical significance without adjustment for multiplicity. Data were analyzed between June 2021 and June 2023.

Results

Associations Between Hypoglycemia and CV Outcomes in the CARMELINA Trial

Baseline characteristics of participants, stratified by treatment assignment and hypoglycemia occurrence during the CARMELINA trial, are presented in Table 1. Of the 6979 patients included, 2589 were females (37.1%) and 4390 were males (62.9%), the mean (SD) age was 65.9 (9.1) years, and 5596 (80.2%) were White and 640 (9.2%) were Asian individuals (eTable 3 in Supplement 1). Participants had a median (IQR) duration of T2D of 13.6 (7.4-20.3) years and a mean (SD) HbA_1c of 8.0% (1.0%) and eGFR of 54.6 (25.0) mL/min/1.73 m², with 5584 (80.0%) having a UACR of 30 mg/g or greater.¹¹ At baseline, patients with hypoglycemia during the trial had had T2D for longer than patients without hypoglycemia (mean [SD] duration, 19.58 [9.47] years vs 13.78 [9.14] years), were more commonly insulin users (84.9% [992] vs 52.7% [3062]), and had worse kidney function (mean [SD] eGFR, 45.6 [20.8] mL/min/1.73 m² vs 56.4 [25.4] mL/min/1.73 m²; median [IQR] UACR, 190.3 [48.7-935.4] mg/g vs 158.4 [43.4-692.0] mg/g). These differences were consistent across treatment groups. The median (IQR) follow-up in the CARMELINA trial was 2.2 (1.6-3.0) years.

Table 1. Baseline Characteristics of Participants by Hypoglycemia Occurrence During the Cardiovascular and Renal Microvascular Outcome Study With Linagliptin (CARMELINA) Trial^a.

Characteristic	Participants with on-trial hypoglycemia, No. (%)		Participants without on-trial hypoglycemia, No. (%)
Characteristic	Linagliptin (n = 576)	Placebo (n = 592)	Linagliptin (n = 2918)	Placebo (n = 2893)
Age, mean (SD), y	66.9 (8.66)	66.2 (8.97)	65.8 (9.09)	65.4 (9.14)
<65	215 (37.3)	249 (42.1)	1252 (42.9)	1252 (43.3)
≥65	361 (62.7)	343 (57.9)	1666 (57.1)	1641 (56.7)
Sex
Female	232 (40.3)	224 (37.8)	1114 (38.2)	1019 (35.2)
Male	344 (59.7)	368 (62.2)	1804 (61.8)	1874 (64.8)
Race and ethnicity^b
American Indian or Alaska Native	22 (3.8)	27 (4.6)	137 (4.7)	129 (4.5)
Asian	38 (6.6)	56 (9.5)	269 (9.2)	277 (9.6)
Black or African American	51 (8.9)	55 (9.3)	143 (4.9)	162 (5.6)
Hawaiian or Pacific Islander	1 (0.2)	4 (0.7)	6 (0.2)	6 (0.2)
White	464 (80.6)	450 (76.0)	2363 (81.0)	2319 (80.2)
eGFR, mean (SD), mL/min/1.73 m²	45.7 (20.9)	45.4 (20.8)	56.4 (25.5)	56.4 (25.3)
<60	460 (79.9)	463 (78.2)	1740 (59.6)	1685 (58.2)
≥60	116 (20.1)	129 (21.8)	1178 (40.4)	1208 (41.8)
UACR, median (IQR), mg/g	185.8 (45.1-996.5)	198.7 (51.3-853.1)	158.4 (43.4-671.7)	158.0 (42.5- 725.7)
UACR, mg/g
<30	110 (19.1)	110 (18.6)	586 (20.1)	586 (20.3)
30-300	219 (38.0)	233 (39.4)	1244 (42.6)	1198 (41.4)
>300	245 (42.5)	249 (42.1)	1088 (37.3)	1108 (38.3)
Missing data	2 (0.3)	0	0	1 (0.0)
BMI, mean (SD)	31.21 (5.25)	31.50 (5.51)	31.24 (5.30)	31.27 (5.34)
HbA_1c, mean (SD), %	7.96 (0.94)	7.98 (1.02)	7.93 (1.00)	7.96 (1.01)
T2D duration, mean (SD), y	19.40 (9.56)	19.76 (9.39)	14.10 (9.41)	13.46 (8.85)
T2D duration, y
≤5	31 (5.4)	29 (4.9)	490 (16.8)	524 (18.1)
>5 to <10	66 (11.5)	50 (8.4)	630 (21.6)	638 (22.1)
≥10	479 (83.2)	513 (86.7)	1798 (61.6)	1731 (59.8)
SBP, mean (SD), mm Hg	142.2 (19.71)	141.6 (19.98)	140.0 (17.26)	140.4 (17.61)
DBP, mean (SD), mm Hg	75.0 (11.17)	75.3 (10.70)	78.3 (10.31)	78.4 (10.26)
CV risk stratum at randomization
Macrovascular disease, albuminuria, normal eGFR	135 (23.4)	126 (21.3)	1226 (42.0)	1241 (42.9)
Reduced eGFR only^c	305 (53.0)	334 (56.4)	1157 (39.7)	1125 (38.9)
Macrovascular disease, reduced eGFR, albuminuria^c	133 (23.1)	123 (20.8)	514 (17.6)	492 (17.0)
Missing data	3 (0.5)	9 (1.5)	21 (0.7)	35 (1.2)
Baseline medications
Antihyperglycemic therapy	576 (100.0)	590 (99.7)	2821 (96.7)	2811 (97.2)
Metformin	261 (45.3)	242 (40.9)	1610 (55.2)	1673 (57.8)
Sulfonylurea	138 (24.0)	118 (19.9)	964 (33.0)	1022 (35.3)
Insulin	486 (84.4)	506 (85.5)	1576 (54.0)	1486 (51.4)
Antihypertensives	558 (96.9)	584 (98.6)	2779 (95.2)	2770 (95.7)
ACE inhibitor or ARB	489 (84.9)	496 (83.8)	2371 (81.3)	2302 (79.6)
Statin	439 (76.2)	460 (77.7)	2056 (70.5)	2063 (71.3)

Open in a new tab

Abbreviations: ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CV, cardiovascular; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate (calculated using the Modification of Diet in Renal Disease equation); HbA_1c, hemoglobin A_1c; SBP, systolic blood pressure; T2D, type 2 diabetes; UACR, urine albumin-to-creatinine ratio.

^{^a}

Hypoglycemia was defined as plasma glucose less than 54 mg/dL (to convert to millimoles per liter, multiply by 0.0555; ie, <3.0 mmol/L) or severe hypoglycemia (ie, an episode requiring the assistance of another person to actively administer carbohydrate, glucagon, or other resuscitative actions).

^{^b}

Race and ethnicity were self-identified by trial participants and obtained from electronic case record forms (fixed categories) after written informed consent.

^{^c}

Reduced eGFR was defined as eGFR of 45 to 75 mL/min/1.73 m² and UACR higher than 200 mg/g or equivalent or as eGFR of 15 to 45 mL/min/1.73 m², regardless of UACR.

As previously reported,¹¹ the adjusted mean placebo-subtracted difference in HbA_1c with linagliptin was –0.51% (95% CI, –0.55% to –0.46%) after 12 weeks of treatment and –0.36% (95% CI, –0.42% to –0.29%) over the full study duration, in the context of earlier and more frequent addition of other antihyperglycemic medications in the placebo group (adjusted hazard ratio [HR], 0.76; 95% CI, 0.69-0.84). During the trial, hypoglycemia with plasma glucose less than 54 mg/dL or severe hypoglycemia occurred in 576 participants (16.5%) in the linagliptin group and 592 participants (17.0%) in the placebo group (adjusted HR, 0.97; 95% CI, 0.86-1.08) (Figure 1A).

Figure 1. — Hypoglycemia was defined as blood glucose less than 54 mg/dL (to convert to millimoles per liter, multiply by 0.0555; ie, <3.0 mmol/L) or severe hypoglycemia (ie, an episode requiring the assistance of another person to actively administer carbohydrate, glucagon, or other resuscitative actions). HR indicates hazard ratio.

Hypoglycemia preceded 3P-MACE plus hospitalization for HF in 177 (89 linagliptin, 88 placebo) of the 1082 participants (16.4%) who experienced 3P-MACE plus hospitalization for HF. Among these 177 participants, the median (IQR) time between the last hypoglycemic episode and 3P-MACE plus hospitalization for HF was 24 (12-58) weeks in participants receiving linagliptin and 27 (10-60) weeks in those receiving placebo. Overall, hypoglycemia was associated with higher adjusted risks for subsequent 3P-MACE plus hospitalization for HF (adjusted HR, 1.23; 95% CI, 1.04-1.46). Similarly, hypoglycemia was associated with higher adjusted risk for subsequent CV death (adjusted HR, 1.29; 95% CI, 1.02-1.62), MI (adjusted HR, 1.48; 95% CI, 1.10-2.00), and 3P-MACE (adjusted HR, 1.32; 95% CI, 1.10-1.59). Hypoglycemia preceded death from any cause in 132 (67 linagliptin, 65 placebo) of the 740 participants (17.8%) who died. The median (IQR) time between last hypoglycemic episode and death was 44 (23-75) weeks in participants receiving linagliptin and 32 (17-53) weeks in those receiving placebo. Overall, there was no difference in the adjusted risk for all-cause mortality after a hypoglycemia episode (adjusted HR, 1.16; 95% CI, 0.95-1.41) nor for the individual outcomes of non-CV death, stroke, or hospitalization for HF (Figure 2A). Within 60 days after each hypoglycemic episode, patients were not at evident increased risk for 3P-MACE plus hospitalization for HF, hospitalization for HF, 3P-MACE or other events, with too few events for some outcomes to support formal statistical analyses; risk-interval plots for time from last hypoglycemia are provided in eFigures 1 to 8 in Supplement 1.

Figure 2. — Data were based on multivariable Cox proportional hazards regression models. 3P-MACE indicates 3-point major adverse CV events (CV death, nonfatal myocardial infarction [MI], or nonfatal stroke); HHF, hospitalization for heart failure; HR, hazard ratio.

^aHypoglycemia was defined as blood glucose less than 54 mg/dL (to convert to millimoles per liter, multiply by 0.0555; ie, <3.0 mmol/L) or severe hypoglycemia (ie, an episode requiring the assistance of another person to actively administer carbohydrate, glucagon, or other resuscitative actions).

^bHR was not calculated for fewer than 14 events.

When assessing the reverse association, incident nonfatal CV events (ie, composite of nonfatal MI, nonfatal stroke, or hospitalization for HF) were associated with higher adjusted risk for subsequent hypoglycemia (adjusted HR, 1.39; 95% CI, 1.06-1.83) (Figure 2B).

Associations Between Hypoglycemia and CV Outcomes in the CAROLINA Trial

Baseline characteristics are presented in Table 2, stratified by treatment assignment and hypoglycemia occurrence during the CAROLINA trial. Of the 6033 patients included, 2414 were females (40.0%) and 3619 were males (60.0%). Patients had a mean (SD) age of 64.0 (9.5) years, and 4407 (73.0%) were White and 1061 (17.6%) were Asian individuals (eTable 3 in Supplement 1). Participants had a median (IQR) duration of T2D of 6.3 (3.0-11.0) years and a mean (SD) HbA_1c of 7.15% (0.57%) and eGFR of 76.8 (19.8) mL/min/1.73 m², with 1533 (25.6%) having a UACR of 30 mg/g or greater. Participants randomized to linagliptin who experienced hypoglycemia during the trial at baseline had a longer duration of T2D (mean [SD] duration, 10.0 [7.0] years vs 7.6 [6.1] years), worse kidney function (mean [SD] eGFR, 72.4 [19.7] mL/min/1.73 m² vs 76.7 [19.7] mL/min/1.73 m²; median [IQR] UACR, 11.5 [5.3-42.9] mg/g vs 9.7 [5.3-30.9] mg/g), and higher prevalence of albuminuria (31.2% [34] vs 25.6% [745]) and microvascular disease (35.8% [39] vs 27.8% [808]), compared with those without hypoglycemia. This pattern was less apparent with glimepiride. The median (IQR) follow-up in the CAROLINA trial was 6.3 (5.9-6.6) years.

Table 2. Baseline Characteristics of Participants by Hypoglycemia Occurrence During the Cardiovascular Outcome Trial of Linagliptin vs Glimepiride in Type 2 Diabetes (CAROLINA) Trial^a.

Characteristic	Participants with on-trial hypoglycemia, No. (%)		Participants without on-trial hypoglycemia, No. (%)
Characteristic	Linagliptin (n = 109)	Glimepiride (n = 537)	Linagliptin (n = 2905)	Glimepiride (n = 2463)
Age, mean (SD), y	64.6 (9.8)	64.4 (9.4)	63.9 (9.5)	64.1 (9.5)
≥70	42 (38.5)	191 (35.6)	963 (33.1)	834 (33.9)
≥75	16 (14.7)	84 (15.6)	394 (13.6)	352 (14.3)
Sex
Female	41 (37.6)	234 (43.6)	1140 (39.2)	990 (40.2)
Male	68 (62.4)	303 (56.4)	1765 (60.8)	1473 (59.8)
Race and ethnicity
American Indian or Alaska Native	6 (5.5)	16 (3.0)	100 (3.4)	92 (3.7)
Asian	17 (15.6)	77 (14.3)	514 (17.7)	453 (18.4)
Black or African American	9 (8.3)	34 (6.3)	146 (5.0)	135 (5.5)
Hawaiian or Pacific Islander	0	0	5 (0.2)	3 (0.1)
White	77 (70.6)	410 (76.4)	2140 (73.7)	1780 (72.3)
eGFR, mean (SD), mL/min/1.73 m²	72.4 (19.7)	74.3 (20.2)	76.7 (19.7)	77.6 (19.7)
≥60	79 (72.5)	420 (78.2)	2340 (80.6)	2042 (82.9)
30-<60	30 (27.5)	114 (21.2)	546 (18.8)	411 (16.7)
<30	0	3 (0.6)	16 (0.5)	10 (0.4)
Missing data	0	0	3 (0.1)	0
UACR, median (IQR), mg/g	11.5 (5.3-42.9)	9.7 (5.3-28.3)	9.7 (5.3-30.9)	9.7 (5.3-30.1)
UACR, mg/g
<30	74 (67.9)	404 (75.2)	2154 (74.1)	1830 (74.3)
30-300	22 (20.2)	103 (19.2)	623 (21.4)	527 (21.4)
>300	12 (11.0)	27 (5.0)	122 (4.2)	97 (3.9)
Missing data	1 (0.9)	3 (0.6)	6 (0.2)	9 (0.4)
BMI, mean (SD)	29.6 (5.1)	29.6 (4.8)	30.2 (5.2)	30.1 (5.1)
HbA_1c, mean (SD), %	7.0 (0.5)	7.1 (0.5)	7.2 (0.6)	7.2 (0.6)
T2D duration, mean (SD), y	10.0 (7.0)	7.8 (6.2)	7.6 (6.1)	7.5 (6.1)
T2D duration, y
≤5	30 (27.5)	219 (40.8)	1194 (41.1)	993 (40.3)
>5-10	30 (27.5)	152 (28.3)	818 (28.2)	709 (28.8)
>10	49 (45.0)	166 (30.9)	893 (30.7)	761 (30.9)
SBP, mm Hg	136 (19)	135 (17)	136 (16)	136 (16)
DBP, mm Hg	80 (9)	78 (10)	79 (10)	79 (9)
CV risk stratum at randomization
Vascular disease	37 (33.9)	191 (35.6)	1014 (34.9)	847 (34.4)
Microvascular disease	23 (21.1)	73 (13.6)	377 (13.0)	345 (14.0)
Age ≥70 y	42 (38.5)	191 (35.6)	963 (33.1)	834 (33.9)
Multiple CV risk factors	90 (82.6)	424 (79.0)	2326 (80.1)	1970 (80.0)
Background medications
No antihyperglycemic therapy	9 (8.3)	40 (7.4)	265 (9.1)	232 (9.4)
Metformin	95 (87.2)	468 (87.2)	2415 (83.1)	2042 (82.9)
Sulfonylurea	51 (46.8)	145 (27.0)	818 (28.2)	701 (28.5)
Antihypertensives	97 (89.0)	484 (90.1)	2557 (88.0)	2184 (88.7)
ACE inhibitor or ARB	83 (76.1)	408 (76.0)	2133 (73.4)	1787 (72.6)
Statin	67 (61.5)	410 (76.4)	1836 (63.2)	1566 (63.6)

Open in a new tab

^{^a}

As previously reported,¹⁰ initially the adjusted mean change in HbA_1c favored glimepiride over linagliptin, but overall there was no significant difference between the groups. While the number of introductions of additional antihyperglycemic therapies did not differ across study groups, there was a pattern of shorter time to introduction in the linagliptin group than in the glimepiride group.

Overall, hypoglycemia occurred in 646 participants (10.7%). Those randomized and treated with linagliptin vs glimepiride had a lower incidence of hypoglycemia (3.6% and 17.9%, respectively; HR, 0.19; 95% CI, 0.15-0.23) (Figure 1B). The reduced risk of hypoglycemia with linagliptin vs glimepiride was consistent across baseline characteristics. Severe hypoglycemia occurred in 15 patients randomized to linagliptin vs 72 randomized to glimepiride (HR, 0.21; 95% CI, 0.12-0.36).

Hypoglycemia preceded 3P-MACE plus hospitalization for HF in 57 (8 linagliptin, 49 glimepiride) of the 810 participants (7.0%) who experienced 3P-MACE plus hospitalization for HF. Among these 57 participants, the median (IQR) time between last hypoglycemic episode and first 3P-MACE plus hospitalization for HF was 92 (16-124) weeks in participants receiving linagliptin and 85 (32-156) weeks in those receiving glimepiride. There was no association between the time-dependent covariate of hypoglycemia and 3P-MACE plus hospitalization for HF (adjusted HR, 1.00; 95% CI, 0.76-1.32). There were no associations between hypoglycemia and subsequent CV outcomes, including 3P-MACE, CV death, MI, or hospitalization for HF, and there were too few patients with events (<14) to conduct this analysis for stroke. Hypoglycemia was associated with significantly higher adjusted risk for all-cause mortality (adjusted HR, 1.49; 95% CI, 1.16-1.92) and a 2-fold higher adjusted risk for non-CV mortality (adjusted HR, 2.16; 95% CI, 1.57-2.97) (Figure 2A). Hypoglycemia preceded death from any cause in 74 (7 linagliptin, 67 glimepiride) of the 643 participants (11.5%) who died. Hypoglycemia preceded non-CV death in 49 (5 linagliptin, 44 glimepiride) of the 307 participants (16.0%) who died from non-CV causes. There were too few patients with events (<14) within 60 days after each hypoglycemic episode to perform Cox proportional hazards regression analyses for all outcomes studied. eFigures 9 to 16 in Supplement 1 provide risk-interval plots demonstrating the absence of temporal associations between hypoglycemia and all outcomes.

There was no association between incident nonfatal CV events and subsequent hypoglycemia (adjusted HR, 1.44; 95% CI, 0.96-2.16) (Figure 2B). Antecedent hypoglycemia did not influence the relative association of linagliptin vs glimepiride with any CV or mortality outcome (Figure 3).

Figure 3. — Data are shown with and without adjustment for time-dependent occurrence of hypoglycemia. Data were based on Cox proportional hazards regression models (eMethods in Supplement 1). The model evaluated 6033 participants for the primary outcome analysis, whereas the model with the additional term for time-dependent hypoglycemia evaluated 6014 participants, omitting the 19 individuals enrolled and treated at multiple sites. Hypoglycemia was defined as blood glucose less than 54 mg/dL (to convert to millimoles per liter, multiply by 0.0555; ie, <3.0 mmol/L) or severe hypoglycemia (ie, an episode requiring the assistance of another person to actively administer carbohydrate, glucagon, or other resuscitative actions). 3P-MACE indicates 3-point major adverse CV events (CV death, nonfatal myocardial infarction [MI], or nonfatal stroke); HHF, hospitalization for heart failure; HR, hazard ratio.

^aHR was not calculated for fewer than 14 events.

Pooled Analyses

Analyses of pooled data from the 2 trials yielded results that were largely consistent with the analyses of individual trial data. Additional details are provided in eTables 1 and 2 in Supplement 1.

Discussion

Results of these analyses of the CARMELINA and CAROLINA trials provide insights into the associations between hypoglycemia and CV or mortality outcomes in individuals with T2D, after adjustments for clinically relevant covariates. In patients with advanced T2D and established CV and/or kidney disease in the CARMELINA trial, hypoglycemia was associated with a 23% higher adjusted risk for subsequent 3P-MACE plus hospitalization for HF (adjusted HR, 1.23), while the reverse association was also observed where a nonfatal CV event was associated with a 39% higher adjusted risk of subsequent hypoglycemia (adjusted HR, 1.39). In patients with relatively early T2D and elevated CV risk in the CAROLINA trial, hypoglycemia was not associated with higher risk for subsequent 3P-MACE plus hospitalization for HF, and similarly, nonfatal CV events were not associated with a higher risk for subsequent hypoglycemia. In both trials, either there was no association between hypoglycemia and risk of subsequent 3P-MACE plus hospitalization for HF within 60 days of the hypoglycemic episodes or there were too few patients with events for meaningful assessment. In the CAROLINA trial, hypoglycemia was associated with higher risk for subsequent non-CV mortality but not for CV mortality.

These observations challenge, to some degree, the widely held viewpoint that hypoglycemia has a direct, causal association with CV events and mortality.¹ First, within the 60 days after each hypoglycemia event, no evident risk increase for CV events was detected in either trial, as depicted in eFigures 1 to 16 in Supplement 1. Second, in the CAROLINA trial, there was an association between hypoglycemia and adjusted risk for non-CV death and consequently all-cause mortality, but not for CV death or any other key CV outcome. Third, while the CAROLINA trial demonstrated numerically fewer hypoglycemic events with linagliptin vs glimepiride, it did not demonstrate favorable association with any CV outcome. Thus, we speculated that both hypoglycemia and CV events occur in patients at highest risk for subsequent events, but most likely neither event causes the other. This hypothesis is not intended to undermine the importance of avoiding hypoglycemia with its myriad adverse consequences, including on patients’ quality of life.

To consider the causality of observed associations, we systematically applied the Bradford Hill criteria for causation, a series of 9 principles, in assessing the probability of causality.¹⁷ A key criterion is temporality; that is, the exposure precedes the event of interest. We analyzed the association between hypoglycemia and CV outcomes bidirectionally specifically to indirectly assess temporality; that is, if the association is bidirectional, it much more likely represents common vulnerability for each outcome rather than one causing the other. In the present study, the bidirectional associations of similar magnitude observed in the CARMELINA trial suggest that these 2 events (hypoglycemia and CV events) were associated in a temporally independent fashion. This association was complemented by the observed lack of increased risk for CV events within the 60-day risk interval after a hypoglycemia episode, as presented in eFigures 1 to 8 in Supplement 1. While such results did not exclude a causal association, taken together, they did suggest a low likelihood for causality.

Another key Bradford Hill principle is plausibility, defined as the presence of a biologically plausible mechanism that can link cause and effect. While much has been written about plausible ways that hypoglycemia might affect CV risk,^1,5,18 in the present analyses, there was no apparent, biologically plausible mechanism that can explain how, for instance, an episode of hypoglycemia can increase the risk for MI a median of 6 to 23 months later. It is the absence of plausible causality that contributes to our interpretation of the observed higher risk for hypoglycemia events after nonfatal CV events in the CARMELINA trial as supporting an association most likely attributable to common vulnerability for both outcomes rather than one causing the other.

These observations add to prior results that challenged a causal association between hypoglycemia and CV risk. For example, in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial, which randomized participants with T2D to standard vs more intensive glucose control and was terminated early due to excess mortality in the intensive strategy arm, hypoglycemia was associated with higher CV risk, yet no temporal association was observed.¹⁹ In the Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation (ADVANCE) trial and in the Veterans Affairs Diabetes Trial (VADT), which randomized participants to treatment strategies similar to the ACCORD trial, no association between hypoglycemia and CV outcomes was observed.^20,21 In the Trial Comparing Cardiovascular Safety of Insulin Degludec vs Insulin Glargine in Patients With Type 2 Diabetes at High Risk of Cardiovascular Events (DEVOTE),⁹ there was no association between hypoglycemia and subsequent risk for 3P-MACE, yet there was a significantly higher adjusted risk of all-cause mortality in those with hypoglycemia without temporal association, with a higher risk evident for up to 1 year after the hypoglycemic episode. Similar to observations in the present analyses, in the Trial Evaluating Cardiovascular Outcomes With Sitagliptin (TECOS), another DPP-4 inhibitor, severe hypoglycemia was associated with a higher crude risk of subsequent CV events, while CV events were associated with a higher crude risk of subsequent hypoglycemia, with only the latter association remaining after adjusting for relevant clinical characteristics.⁸ Additionally, bidirectional associations between severe hypoglycemia and CV events were found in the Exenatide Study of Cardiovascular Event Lowering (EXSCEL) of the GLP-1 receptor agonist, exenatide.²²

The CARMELINA and CAROLINA trials were primarily analyzed separately, as opposed to pooling the data for analyses given the extreme heterogeneity of risks and outcomes between them (eTables 3-5 in Supplement 1). First, the trials had different comparators (placebo and sulfonylurea, respectively). Second, the trials had different durations (median [IQR] follow-up, 2.2 [1.6-3.0] years vs 6.3 [5.9-6.6] years, respectively). Third, the trials had different patient mixes. The CARMELINA trial enrolled patients at high CV risk enriched for chronic kidney disease; with a median (IQR) T2D duration of 13.6 (7.4-20.3) years; and who were intentionally recruited for being at or having high risk of kidney disease, which further amplified the risk for both CV events and hypoglycemia. In contrast, participants in the CAROLINA trial had lower CV risk, had a median (IQR) T2D duration of 6.3 (3.0-11.0) years, and included fewer patients with or at risk for kidney disease. Therefore, the results from the pooled analyses (eTables 1-2 in Supplement 1) should be interpreted cautiously due to the extreme differences between the trials that challenge the validity of such pooling of the data.

Strengths and Limitations

The strengths of this study include the use of large databases (including approximately 13 000 individuals) from trials that systematically captured hypoglycemic episodes as well as CV and mortality outcomes, with the CV outcomes centrally adjudicated. The study limitations are that hypoglycemia or CV events cannot be randomized, restricting the analyses to observational associations and making any causal inference challenging, and that all analyses presented were defined post hoc. Although statistical modeling adjusted for numerous covariates known to affect CV and mortality risk, some unknown degree of residual confounding inevitably persisted. Furthermore, given that most patients enrolled in each trial were of White or Asian race, the generalizability of the findings to other races is uncertain.

Conclusions

This secondary analysis found bidirectional adjusted associations between hypoglycemia and CV events in the CARMELINA trial, but no association in either direction in the CAROLINA trial. In both trials, there was no evident temporal association between hypoglycemic episodes and CV events. These findings support the theory that the observed associations between hypoglycemia and adverse CV events and mortality that have been iteratively reported most likely represent reverse confounding. The associations identify a phenotype of patients with T2D most susceptible to both hypoglycemia and CV events or mortality, in which case we speculated that hypoglycemia is a risk marker rather than a risk factor for CV events. Thus, while efforts to minimize hypoglycemia in patients with T2D are warranted due to its unpleasant symptoms, acute risk for serious adverse outcomes of severe hypoglycemia, and the overall burden of hypoglycemia on quality of life, avoiding hypoglycemia may not mitigate CV risk.

Supplement 1.

eTable 1. Risk for CV Event After Hypoglycemia in CARMELINA, CAROLINA and in Pooled Analyses

eTable 2. Risk for Hypoglycemia After Non-Fatal CV Event in CARMELINA, CAROLINA and in Pooled Analyses

eTable 3. Demographics and Baseline Characteristics by Clinical Trial

eTable 4. Incidence Rates of CV Events and Hypoglycemia by Clinical Trial

eTable 5. Major Design Characteristics by Clinical Trial

eFigure 1. Time From Last Hypoglycemia to First 3P-MACE or Hospitalization for Heart Failure in the CARMELINA Trial

eFigure 2. Time From Last Hypoglycemia to First 3P-MACE in the CARMELINA Trial

eFigure 3. Time From Last Hypoglycemia to CV Death in the CARMELINA Trial

eFigure 4. Time From Last Hypoglycemia to Non-CV Death in the CARMELINA Trial

eFigure 5. Time From Last Hypoglycemia to All-Cause Death in the CARMELINA Trial

eFigure 6. Time From Last Hypoglycemia to First Non-Fatal or Fatal MI in the CARMELINA Trial

eFigure 7. Time From Last Hypoglycemia to First Non-Fatal or Fatal Stroke in the CARMELINA Trial

eFigure 8. Time From Last Hypoglycemia to First Hospitalization for Heart Failure in the CARMELINA Trial

eFigure 9. Time From Last Hypoglycemia to First 3P-MACE or Hospitalization for Heart Failure in the CAROLINA Trial

eFigure 10. Time From Last Hypoglycemia to First 3P-MACE in the CAROLINA Trial

eFigure 11. Time From Last Hypoglycemia to CV Death in the CAROLINA Trial

eFigure 12. Time From Last Hypoglycemia to Non-CV Death in the CAROLINA Trial

eFigure 13. Time From Last Hypoglycemia to All-Cause Death in the CAROLINA Trial

eFigure 14. Time From Last Hypoglycemia to First Non-Fatal or Fatal MI in the CAROLINA Trial

eFigure 15. Time From Last Hypoglycemia to First Non-Fatal or Fatal Stroke in the CAROLINA Trial

eFigure 16. Time From Last Hypoglycemia to First Hospitalization for Heart Failure in the CAROLINA Trial

Click here for additional data file.^{(1.1MB, pdf)}

Supplement 2.

Data Sharing Statement

Click here for additional data file.^{(17.2KB, pdf)}

References

1.International Hypoglycaemia Study Group . Hypoglycaemia, cardiovascular disease, and mortality in diabetes: epidemiology, pathogenesis, and management. Lancet Diabetes Endocrinol. 2019;7(5):385-396. doi: 10.1016/S2213-8587(18)30315-2 [DOI] [PubMed] [Google Scholar]
2.Desouza C, Salazar H, Cheong B, Murgo J, Fonseca V. Association of hypoglycemia and cardiac ischemia: a study based on continuous monitoring. Diabetes Care. 2003;26(5):1485-1489. doi: 10.2337/diacare.26.5.1485 [DOI] [PubMed] [Google Scholar]
3.Adler GK, Bonyhay I, Failing H, Waring E, Dotson S, Freeman R. Antecedent hypoglycemia impairs autonomic cardiovascular function: implications for rigorous glycemic control. Diabetes. 2009;58(2):360-366. doi: 10.2337/db08-1153 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Desouza CV, Bolli GB, Fonseca V. Hypoglycemia, diabetes, and cardiovascular events. Diabetes Care. 2010;33(6):1389-1394. doi: 10.2337/dc09-2082 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Frier BM, Schernthaner G, Heller SR. Hypoglycemia and cardiovascular risks. Diabetes Care. 2011;34(suppl 2):S132-S137. doi: 10.2337/dc11-s220 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Cryer PE. Death during intensive glycemic therapy of diabetes: mechanisms and implications. Am J Med. 2011;124(11):993-996. doi: 10.1016/j.amjmed.2011.08.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Tsujimoto T, Yamamoto-Honda R, Kajio H, et al. Vital signs, QT prolongation, and newly diagnosed cardiovascular disease during severe hypoglycemia in type 1 and type 2 diabetic patients. Diabetes Care. 2014;37(1):217-225. doi: 10.2337/dc13-0701 [DOI] [PubMed] [Google Scholar]
8.Standl E, Stevens SR, Armstrong PW, et al. ; TECOS Study Group . Increased risk of severe hypoglycemic events before and after cardiovascular outcomes in TECOS suggests an at-risk type 2 diabetes frail patient phenotype. Diabetes Care. 2018;41(3):596-603. doi: 10.2337/dc17-1778 [DOI] [PubMed] [Google Scholar]
9.Pieber TR, Marso SP, McGuire DK, et al. ; DEVOTE Study Group . DEVOTE 3: temporal relationships between severe hypoglycaemia, cardiovascular outcomes and mortality. Diabetologia. 2018;61(1):58-65. doi: 10.1007/s00125-017-4422-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Rosenstock J, Kahn SE, Johansen OE, et al. ; CAROLINA Investigators . Effect of linagliptin vs glimepiride on major adverse cardiovascular outcomes in patients with type 2 diabetes: the CAROLINA randomized clinical trial. JAMA. 2019;322(12):1155-1166. doi: 10.1001/jama.2019.13772 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Rosenstock J, Perkovic V, Johansen OE, et al. ; CARMELINA Investigators . Effect of linagliptin vs placebo on major cardiovascular events in adults with type 2 diabetes and high cardiovascular and renal risk: the CARMELINA randomized clinical trial. JAMA. 2019;321(1):69-79. doi: 10.1001/jama.2018.18269 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Rosenstock J, Perkovic V, Alexander JH, et al. ; CARMELINA® investigators . Rationale, design, and baseline characteristics of the Cardiovascular Safety and Renal Microvascular Outcome Study with Linagliptin (CARMELINA): a randomized, double-blind, placebo-controlled clinical trial in patients with type 2 diabetes and high cardio-renal risk. Cardiovasc Diabetol. 2018;17(1):39. doi: 10.1186/s12933-018-0682-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Marx N, Rosenstock J, Kahn SE, et al. Design and baseline characteristics of the Cardiovascular Outcome Trial of Linagliptin Versus Glimepiride in Type 2 Diabetes (CAROLINA). Diab Vasc Dis Res. 2015;12(3):164-174. doi: 10.1177/1479164115570301 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.International Hypoglycaemia Study Group . Glucose concentrations of less than 3.0 mmol/l (54 mg/dl) should be reported in clinical trials: a joint position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia. 2017;60(1):3-6. doi: 10.1007/s00125-016-4146-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Seaquist ER, Anderson J, Childs B, et al. Hypoglycemia and diabetes: a report of a workgroup of the American Diabetes Association and the Endocrine Society. Diabetes Care. 2013;36(5):1384-1395. doi: 10.2337/dc12-2480 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Collett D. Modelling Survival Data in Medical Research. 3rd ed. Chapman and Hall/CRC; 2014. [Google Scholar]
17.Hill AB. The environment and disease: association or causation? Proc R Soc Med. 1965;58(5):295-300. doi: 10.1177/003591576505800503 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Zoungas S, Patel A, Chalmers J, et al. ; ADVANCE Collaborative Group . Severe hypoglycemia and risks of vascular events and death. N Engl J Med. 2010;363(15):1410-1418. doi: 10.1056/NEJMoa1003795 [DOI] [PubMed] [Google Scholar]
19.Bonds DE, Miller ME, Bergenstal RM, et al. The association between symptomatic, severe hypoglycaemia and mortality in type 2 diabetes: retrospective epidemiological analysis of the ACCORD study. BMJ. 2010;340:b4909. doi: 10.1136/bmj.b4909 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Patel A, MacMahon S, Chalmers J, et al. ; ADVANCE Collaborative Group . Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med. 2008;358(24):2560-2572. doi: 10.1056/NEJMoa0802987 [DOI] [PubMed] [Google Scholar]
21.Duckworth W, Abraira C, Moritz T, et al. ; VADT Investigators . Glucose control and vascular complications in veterans with type 2 diabetes. N Engl J Med. 2009;360(2):129-139. doi: 10.1056/NEJMoa0808431 [DOI] [PubMed] [Google Scholar]
22.Standl E, Stevens SR, Lokhnygina Y, et al. ; EXSCEL Study Group . Confirming the bidirectional nature of the association between severe hypoglycemic and cardiovascular events in type 2 diabetes: insights from EXSCEL. Diabetes Care. 2020;43(3):643-652. doi: 10.2337/dc19-1079 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials