Sulphonylurea compared to DPP-4 inhibitors in
combination with metformin carries increased risk
of severe hypoglycemia, cardiovascular events, and
all-cause mortality
Jan W. Eriksson
a, Johan Bodegard
b,*, David Nathanson
c, Marcus Thuresson
d,
Thomas Nystro¨m
c, Anna Norhammar
ea
Department of Medical Sciences, Clinical Diabetes and Metabolism, Uppsala University, Uppsala, Sweden bAstraZeneca Nordic-Baltic, So¨derta¨lje, Sweden
cDepartment of Clinical Science and Education, Division of Internal Medicine, Unit for Diabetes Research, Karolinska Institute, So¨dersjukhuset, Stockholm, Sweden
dStatisticon AB, Uppsala, Sweden e
Cardiology Unit, Department of Medicine, Solna, Karolinska Institute, Stockholm, Sweden
A R T I C L E I N F O Article history:
Received 22 January 2016 Received in revised form 27 April 2016
Accepted 30 April 2016 Available online 6 May 2016
Keywords: Epidemiology Sulfonylurea DPP-4i Hypoglycemia Cardiovascular disease All-cause mortality A B S T R A C T
Aims: There are safety concerns related to sulphonylurea treatment. The objective of this nationwide study was to compare the risk of cardiovascular disease (CVD), all-cause mor-tality and severe hypoglycemia in patients with type 2 diabetes (T2D) starting second-line treatment with either metformin + sulphonylurea or metformin + dipeptidyl peptidase-4 inhibitor (DPP-4i).
Methods: All patients with T2D in Sweden who initiated second-line treatment with met-formin + sulphonylurea or metmet-formin + DPP-4i during 2006–2013 (n = 40,736 and 12,024, respectively) were identified in this nationwide study. The Swedish Prescribed Drug Regis-ter and the Cause of Death and National Patient RegisRegis-ters were used, and Cox survival models adjusted for age, sex, fragility, prior CVD, and CVD-preventing drugs were applied to estimate risks of events. Propensity score adjustments and matching methods were used to test the results.
Results: Of 52,760 patients; 77% started metformin + SU and 23% metformin + DPP-4i. Crude incidences for severe hypoglycemia, CVD, and all-cause mortality in the SU cohort were 2.0, 19.6, and 24.6 per 1000 patient-years and in the DPP-4i cohort were 0.8, 7.6, and 14.9 per 1000 patient-years, respectively. Sulphonylurea compared with DPP4i was associ-ated with higher risk of subsequent severe hypoglycemia, fatal and nonfatal CVD, and all-cause mortality; adjusted HR (95% CI): 2.07 (1.11–3.86); 1.17 (1.01–1.37); and 1.25 (1.02– 1.54), respectively. Results were confirmed by additional propensity-adjusted and matched
http://dx.doi.org/10.1016/j.diabres.2016.04.055
0168-8227/Ó 2016 The Authors. Published by Elsevier Ireland Ltd.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). * Corresponding author.
d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e1 1 7 ( 2 0 1 6 ) 3 9 – 4 7
Contents available atScienceDirect
Diabetes Research
and Clinical Practice
jo u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m /l o c a t e / di a br e sanalyses. Among the SU drugs, glibenclamide had the highest risks.
Conclusions: Metformin + SU treatment was associated with an increased risk of subse-quent severe hypoglycemia, cardiovascular events, and all-cause mortality compared with metformin + DPP4i. Results from randomized trials will be important to elucidate causal relationships.
Ó 2016 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1.
Introduction
The prevalence of type 2 diabetes treated with glucose-lowering drugs is increasing, and has now been estimated at 4.4% in Sweden[1]. It is recommended in the 2015 guideli-nes from the American Diabetes Association that glucose-lowering drug (GLD) therapy should be initiated as metformin monotherapy. When this fails, sulphonylurea or a dipeptidyl peptidase (DPP)-4 inhibitor is frequently added. Sulphony-lurea, which has been available for 50 years, is still primarily used because of its efficacy and low cost, but it has known side effects such as hypoglycemia and weight gain[2]. DPP-4 inhibitors have demonstrated a lower risk of hypoglycemia and weight gain than sulphonylurea[3–5], and is therefore recommended for patients at increased risk of hypoglycemia and with established obesity[6,7]. In addition, DPP-4 inhibitor shows a neutral effect on risk of cardiovascular death and myocardial infarction compared with placebo [3–5,8–10]
although there have been reports of increased numbers of non-fatal heart failure events[8].
Results from observational studies have raised safety con-cerns regarding increased risk of cardiovascular disease (CVD) and all-cause mortality in patients treated with sulphony-lurea[11–18]. Hypoglycemia shows a strong association with cardiovascular events, and this could partially explain the reported increased risk of CVD and mortality with sulphony-lurea [10,19]. Currently, no randomized trials have directly compared metformin combinations with sulphonylurea and DPP-4 inhibitor using CVD and mortality outcomes, but a lar-ger randomized study is currently ongoing (Cardiovascular outcome study of linagliptin versus glimepiride in patients with type 2 diabetes [CAROLINA]), which is expected to be completed in late 2018 [20,21]. Meanwhile, observational studies may add to our knowledge of the association between sulphonylurea use and cardiovascular events in patients with type 2 diabetes.
The aim of this observational, full-population study was to investigate the risk of severe hypoglycemia, fatal and nonfatal CVD (unstable angina, myocardial infarction, or stroke), and all-cause mortality associated with combination treatment with either sulphonylurea or DPP-4 inhibitor added to met-formin, using national healthcare registries in Sweden.
2.
Material and methods
2.1. Data sources
This observational registry study utilized data from Swedish national registries: the Prescribed Drug Register covering all
drug prescriptions filled since 2005 using Anatomical Thera-peutic Chemical (ATC) codes; the Cause of Death Register (established 1961); the National Patient Register covering all hospitalisations and discharge diagnoses since 1987 and all out-patient hospital visits since 2001. All three registers are held by the Swedish National Board of Health and Welfare (NBHW).
2.2. Study population
All patients with a filled prescription for an incident combina-tion treatment of either a sulphonylurea or a DPP-4 inhibitor together with metformin during the time period July 1 2006 to end of 2013 were identified. Patients had to be receiving monotherapy with one non-insulin antidiabetic drug (NIAD) prior to the start of combination treatment, and the index date was defined as the date of combination treatment initia-tion (sulphonylurea or metformin or DPP-4 inhibitor). Second-line treatment was defined as concomitant use of two or more NIADs, or as initiation of insulin treatment. Switching from NIAD to another was not considered as second-line treat-ment, unless a new prescription of the first-line drug was filled within 1 year after the switch, in which case the starting date of the second antidiabetic drug was the index date. Patients with a diagnosis of gestational diabetes (Interna-tional Classification of Diseases [ICD] code 10: O24.4) within 1 year of the index date and patients with type 1 diabetes were excluded. (See Online-OnlySupplementary Table 1for ICD diagnoses and ATC codes.) Patients with type 1 diabetes were arbitrarily defined as those with a registered type 1 dia-betes diagnosis (ICD-10 E10) and treated with insulin during their first year of GLD treatment, or aged under 30 years at the start of insulin medication, or aged under 15 years at the start of any diabetes medication. The main analyses were done according to an on-treatment approach, and patients were observed from the index date until: a gap of at least 6 months in filled prescription of metformin and sulphony-lurea or DPP-4 inhibitor; death; or December 31, 2013. In addi-tion, ITT (intention-to-treat) analyses were performed as above, but also including those with interrupted or changed treatment. The study cohort was formed among patients with type 2 diabetes moving from treatment with one NIAD to dual NIAD, as outlined inFig. 1.
Individual patient-level data from the national registers were linked using personal identification numbers, which are assigned at birth and whose use is mandatory in dealing with the public healthcare system. Data linkage was per-formed by the NBHW and the linked database was managed at Statisticon AB, Stockholm, Sweden. The study protocol
was approved by the Stockholm regional ethics committee (registration number 2013/2206-31). Baseline treatments, defined by ATC codes, were defined as any identified filled prescription of the treatment of interest during the year prior to the index date.
Three endpoints were defined as follows: (1) Severe hypo-glycemia a main or secondary diagnosis in the inpatient reg-ister of hypoglycemia (E16.0, E16.1, or E16.2) or diabetes with coma (ICD-10 E10.0, E11.0, E12.0, E13.0, or E14.0), as these codes are typically used for hypoglycemia requiring third party assistance. (2) Fatal and nonfatal CVD – a main diagno-sis in the inpatient register of myocardial infarction (I21), ischemic stroke (I63–I64), unstable angina pectoris (I20.0), or cardiovascular death (death with an ICD-10 code I diagnosis as primary cause of death). (3) Death of any cause.
2.3. Statistical analyses
The time from initiation of second-line treatment to a clinical event (severe hypoglycemia, fatal or nonfatal CVD, or all-cause mortality) was visualized using Kaplan–Meier graphs. Patients were censored at treatment discontinuation, death, or study period end.
Statistical analyses comparing treatments (metformin + DPP-4 inhibitor vs. metformin + sulphonylurea) was per-formed using Cox proportional hazards models. Two separate adjustment models were determined for risk of severe hypo-glycemia and for fatal and nonfatal CVD and all-cause mor-tality. Directed acyclic graphs [22] were used to minimize the risk of bias and identify the two primary adjustment mod-els (Online-OnlySupplementary Figs. 1 and 2).
The model for estimating the risk of severe hypoglycemia was adjusted for age and fragility (defined as at least 3 days of hospitalization during the year prior to the index date). The model for fatal and nonfatal CVD was adjusted for age, sex, fragility, prior CVD, and use of statins, low-dose aspirin, and antihypertensives. Prior history of CVD was assessed in the National Patient Register from 1987 until index date.
The main analysis was performed using an on-treatment approach, in which a treatment chain was regarded as unin-terrupted if a medication of interest had a prescription filled within 6 months of the prior time of a filled prescription. As a sensitivity analysis, an intention-to-treat (ITT) approach was adopted, in which patients were analyzed according to their index medication combination (with at least two dis-penses) irrespective of subsequent switches or treatment interruption or discontinuation.
A separate analysis, irrespective of treatment combina-tion, was performed to compare two patient groups, those with or without any event of severe hypoglycemia, and exam-ine the association with fatal and nonfatal CVD using a Cox proportional hazard model adjusted for age, sex, fragility, and prior CVD. This analysis is based on the same cohort used in the main analysis, and follow-up started at the date of ini-tiation of metformin + sulphonylurea or metformin + DPP-4
inhibitor. The proportional hazard assumptions were
assessed by examining Schoenfeld residuals.
Since this is a clinical effectiveness study to assess two dif-ferent treatment strategies a propensity score adjusted and a propensity score match model were additional used to test the results from the adjusted Cox proportional hazard model. Propensity scores were calculated using age, sex, fragility, prior CVD, and use of statins, low-dose aspirin, and antihy-pertensives. They were used in propensity score adjusted sur-vival models and in propensity score matching of SU and DPP-4 inhibitor treatment groups (2:1 using a caliper of 0.2 when identifying matches). The matching was performed using the Match function in the R package Matching[23].
A P value below 0.05 was considered significant, but because no adjustment for multiplicity was performed, P val-ues should be interpreted with caution. All analyses were conducted using R statistical software (R version 3.2.3)[24].
3.
Results
3.1. Patient characteristics and treatments
A total of 52,760 patients with type 2 diabetes initiated second-line treatment with metformin + sulphonylurea or metformin + DPP-4 inhibitor during 2006–2013. Of these, 77.2% initiated metformin + sulphonylurea and 22.8% met-formin + DPP-4 inhibitor. These regimens were the two most commonly seen second-line dual NIAD treatments, followed by a number of less frequently seen dual NIAD combinations (Fig. 1).
The patients in the metformin + sulphonylurea group were slightly older than those in the metformin + DPP-4 inhibitor group, were less frequently men, had slightly more frequent history of cardiovascular- and microvascular disease (Table 1). Antihypertensive and low-dose aspirin treatments differed
GLD-treated T2DM patients
N = 417,691 Patients treated with:
• 2 NIADs, n = 37,681
• 3 NIADs, n = 4282
• Insulin, n = 78,170
Patients remaining on 1 NIAD alone throughout the study period, n = 195,473
Patients treated with: • 3 NIADs, n = 105
• Insulin, n = 33,633
1 NIAD as first-line treatment
n = 297,558
Second-line treatment
n = 102,085
Metformin + SU
n = 40,736 Metformin + DPP-4in = 12,024 Other combinationsn = 15,587
2 NIADs
n = 68,347
Fig. 1 – Flowchart on establishment of patient population. All patients with type 2 diabetes in Sweden who were on GLD treatment at the start of the study period (2006-07-01), as well as incident cases who initiated treatment thereafter, were surveyed for treatment changes until the end of the study period (2013-12-31). The resulting study cohort, i.e. patients initiating second-line treatment with metformin in combination with either a DPP-4 inhibitor or a
sulphonylurea, is indicated in bold. DPP, dipeptidyl peptidase; GLD, glucose-lowering drug; NIAD, non-insulin antidiabetic drug.
Table 1 – Baseline characteristics of patients initiated on either sulphonylurea (SU) or DPP-4 inhibitor (DPP-4i) in combination with metformin.
All patients Matched patients (2:1)
Metformin + SU Metformin + DPP-4i Total Metformin + SU Metformin + DPP-4i Total
Number of patients, N 40,736 12,024 52,760 23,846 11,923 35,769
Age, years, mean (SD) 65.3 (11.9) 61.1 (10.9) 64.4 (11.8) 61.3 (11.1) 61.3 (10.8) 61.3 (11.0) Sex, male, n (%) 24,044 (59.0) 7510 (62.5) 31,554 (59.8) 14,907 (62.5) 7431 (62.3) 22,338 (62.5) Comorbidities, n (%) CVD 13,143 (32.3) 3340 (27.8) 16,483 (31.2) 6698 (28.1) 3322 (27.9) 10,020 (28.0) Myocardial infarction 3456 (8.5) 895 (7.4) 4351 (8.2) 1731 (7.3) 890 (7.5) 2621 (7.3) STEMI 1579 (3.9) 417 (3.5) 1996 (3.8) 788 (3.3) 416 (3.5) 1204 (3.4) NSTEMI 2368 (5.8) 602 (5.0) 2970 (5.6) 1192 (5.0) 597 (5.0) 1789 (5.0) Unstable angina 1753 (4.3) 486 (4.0) 2239 (4.2) 929 (3.9) 485 (4.1) 1414 (4.0) Angina pectoris 7140 (17.5) 1915 (15.9) 9055 (17.2) 3719 (15.6) 1905 (16.0) 5624 (15.7) Heart failure 2468 (6.1) 560 (4.7) 3028 (5.7) 1132 (4.7) 555 (4.7) 1687 (4.7) Atrial fibrillation 3369 (8.3) 802 (6.7) 4171 (7.9) 1496 (6.3) 802 (6.7) 2298 (6.4) Stroke 3225 (7.9) 661 (5.5) 3886 (7.4) 1488 (6.2) 658 (5.5) 2146 (6.0) Hemorrhagic 2326 (5.7) 457 (3.8) 2783 (5.3) 1008 (4.2) 456 (3.8) 1464 (4.1) Ischemic 278 (0.7) 65 (0.5) 343 (0.7) 150 (0.6) 64 (0.5) 214 (0.6)
Transitory ischemic attack 716 (1.8) 154 (1.3) 870 (1.6) 357 (1.5) 153 (1.3) 510 (1.4) Peripheral artery disease 1432 (3.5) 343 (2.9) 1775 (3.4) 701 (2.9) 341 (2.9) 1042 (2.9) Kidney disease 430 (1.1) 134 (1.1) 564 (1.1) 240 (1.0) 131 (1.1) 371 (1.0) Microvascular disease 7429 (18.2) 1743 (14.5) 9172 (17.4) 3953 (16.6) 1731 (14.5) 5684 (15.9) Severe hypoglycemia 374 (0.9) 68 (0.6) 442 (0.8) 196 (0.8) 67 (0.6) 263 (0.7) Cancer 5528 (13.6) 1337 (11.1) 6865 (13.0) 2660 (11.2) 1336 (11.2) 3996 (11.2) Chronic obstructive pulmonary disease 1087 (2.7) 290 (2.4) 1377 (2.6) 541 (2.3) 290 (2.4) 831 (2.3) Treatments, n (%) CVD risk treatment 33,758 (82.9) 10,269 (85.4) 44,027 (83.4) 20,220 (84.8) 10,169 (85.3) 30,389 (85.0) Antihypertensives 29,064 (71.3) 8833 (73.5) 37,897 (71.8) 17,398 (73.0) 8746 (73.4) 26,144 (73.1) Statins 21,591 (53.0) 7269 (60.5) 28,860 (54.7) 14,308 (60.0) 7189 (60.3) 21,497 (60.1) Low-dose aspirin 14,645 (36.0) 4063 (33.8) 18,708 (35.5) 8005 (33.6) 4036 (33.9) 12,041 (33.7) Beta blockers 15,966 (39.2) 4654 (38.7) 20,620 (39.1) 9178 (38.5) 4618 (38.7) 13,796 (38.6) Lower limb amputations 89 (0.2) 13 (0.1) 102 (0.2) 43 (0.2) 13 (0.1) 56 (0.2)
CVD, cardiovascular disease (unstable angina, myocardial infarction, or stroke); DPP, dipeptidyl peptidase; NSTEMI, non-ST-segment-elevation myocardial infarction; SD, standard deviation; STEMI, ST-segment-elevation myocardial infarction.
d i a bet es rese ar ch a n d c li nica l p ra cti ce 11 7 ( 20 16) 3 9– 47
marginally between the two patient groups, while statins were substantially more often prescribed in the metformin + DPP-4 inhibitor group (Table 1).
Patients were followed from initiation of dual NIAD treat-ment (occurring during the period July 1, 2006 to December 31, 2013) until death or end of the study period (December 31, 2013). Thus, the follow-up time ranged up to 7.5 years. The median follow-up time for the metformin + sulphony-lurea and metformin + DPP-4 inhibitor groups was 3.4 and 2.5 years, respectively, and the number of patient-years was 143,344 and 32,133, respectively. The most frequently filled prescriptions for sulphonylureas were glibenclamide (41.7%), glipizide (34.9%), and glimepiride (23.4%), while among the DPP-4 inhibitors, prescriptions for sitagliptin, saxagliptin, and vildagliptin were filled for 80.3%, 5.4%, and 2.1% of patients with type 2 diabetes, respectively (Online-Only Sup-plementary Table 2a). The metformin dose at index was 21% higher in the metformin + DPP-4 inhibitor group than in the metformin + sulphonylurea group (Online-Only Supplemen-tary Table 2a).
3.2. Risk of events in the two treatment groups
The main results were obtained by on-treatment analyses. In the metformin + sulphonylurea group, the crude numbers (incidence per 1000 patient-years) of severe hypoglycemia, fatal and nonfatal CVD, and all-cause death were 120 (2.0), 1153 (19.6), and 1420 (24.6), respectively. The corresponding results for the metformin + DPP-4 inhibitor group were 11 (0.8), 105 (7.6), and 204 (14.9). As illustrated by the Kaplan– Meier curves (Fig. 2A–C), the increased incidence in the met-formin + sulphonylurea group of all types of events can be observed already in the first 6 months, with a continued increase in separation between the curves with follow-up time.
3.3. Adjusted analyses
In analyses adjusting for known risk factors, metformin + sulphonylurea compared with metformin + DPP-4 inhibitor as second-line treatment was associated with a higher risk of severe hypoglycemia, fatal and nonfatal CVD and all-cause mortality. Adjusted hazard ratios (95% CI) in on-treatment analyses were as follows: 2.07 (1.11–3.86), 1.17 (1.01–1.37), and 1.25 (1.02–1.54), respectively (Fig. 2 and
Table 2). The ITT analyses yielded similar results with hazard ratios (95% CI) of 1.71 (1.19–2.44), 1.11 (1.01–1.21), and 1.25 (1.12–1.38) for severe hypoglycemia, fatal/ nonfatal CVD and all-cause mortality, respectively. Details are shown in Supple-mentary Table 3.
3.4. Propensity score adjusted analyses
Propensity score adjustments survival analyses resulted in slightly greater differences between SU and DPP4 inhibitor treatment for severe hypoglycemia, CVD and all-cause mortality compared to the adjusted survival analyses, 2.08 (1.11–3.89), 1.21 (1.04–1.40) and 1.35 (1.10–1.65), respectively (Table 2).
3.5. Propensity matched analyses
Propensity matching decreased the total number of patients from 52,760 to 35,769, i.e. 16,991 (32%) patients could not be matched. The resulting groups were very similar in essen-tially all baseline parameters (Table 1). Hazard ratios were 1.88 (0.98–3.62), 1.23 (1.05–1.44) and 1.43 (1.55–1.77) for severe hypoglycemia, CVD and all-cause mortality, respectively (Table 2). This was very similar to the primary adjusted sur-vival analysis. Severe hypoglycemia n at risk MET + SU, 40,736 MET + DPP-4i, 12,024 3
Cumulative proportion of hypoglycemia (%)
2 1 0 0 1 2 3 MET + SU MET + DPP-4i Adjusted HR (95% CI): 2.07 (1.11–3.86)
Time from second-line therapy (years)
4 5 6 19,573 4764 9625 1924 4933 817 2399 345 1169 132 445 27
Fatal and nonfatal CVD
n at risk MET + SU, 40,736 MET + DPP-4i, 12,024 15 Cumulative proportion of CVD (%) 10 5 0 0 1 2 3 MET + SU MET + DPP-4i Adjusted HR (95% CI): 1.17 (1.01–1.37)
Time from second-line therapy (years)
4 5 6 19,304 4723 9391 1878 4781 795 2312 338 1116 128 442 24 All-cause mortality
A
B
C
n at risk MET + SU, 40,736 MET + DPP-4i, 12,024 15 Cumulative proportion of deaths (%) 10 5 0 0 1 2 3 MET + SU MET + DPP-4i Adjusted HR (95% CI): 1.25 (1.02–1.54)Time from second-line therapy (years)
4 5 6 19,605 4767 9656 1924 4954 817 2407 345 1172 132 447 27
Fig. 2 – Kaplan–Meier curves and adjusted HRs for (A) severe hypoglycemia, (B) fatal and nonfatal CVD, and (C) all-cause mortality. CVD, cardiovascular disease; DPP-4i, dipeptidyl peptidase-4 inhibitor; HR, hazard ratio; MET, metformin; SU, sulphonylurea. These analyses were performed in patients remaining on the initiated 2nd-line treatment.
3.6. Severe hypoglycemia and CVD
In patients with type 2 diabetes, regardless of treatment group, who had an episode of severe hypoglycemia during follow-up compared with those who did not, the risk of fatal and nonfatal CVD was increased 1.5-fold (adjusted hazard ratio 1.51, 95% CI 1.21–1.88, P < 0.001).
3.7. Risk of events within sulphonylurea class
In comparison with metformin + DPP-4 inhibitor all sulpho-nylurea agents in combination with metformin had numeri-cally increased risk of fatal-/nonfatal CVD and all-cause mortality, i.e. glibenclamide (25% and 37% respectively), glip-izide (11% and 17%) and glimepiride (11% and 11%), however only glibenclamide use reached statistical significance (Online-OnlySupplementary Table 2b). Based on the limited number of events (81, 19 and 20 for the glibenclamide, glip-izide and glimepiride groups, respectively), risk of severe hypoglycemia was numerically increased with glibenclamide and glimepiride, where the former reached statistical signifi-cance (Online-OnlySupplementary Table 2b).
4.
Discussion
In this nationwide register-based study metformin + sulpho-nylurea and metformin + DPP-4 inhibitor were the most com-mon non-insulin second-line treatment strategies in patients with type 2 diabetes. Using three different statistical meth-ods, sulphonylurea compared to DPP-4 inhibitors, in combi-nation with metformin, was associated with a 88–107%, 17– 23%, and 25–43% increased risk of severe hypoglycemia, fatal and nonfatal CVD, and all-cause mortality, respectively. This was seen after adjustment for comorbidities and other risk factors. Propensity score adjusted and matched analyses yielded similar results compared with the primary adjusted survival analyses, and this consistency supports the
robust-ness of the findings. The increased risk associated with SU can be observed early after initiation of second-line treatment and continues to increase with follow-up time. We also found a strong association between severe hypoglycemia and CVD, in support of the above-reported findings. These results sup-port the association of increased CVD risk with metformin + sulphonylurea treatment described in previous reports
[16,25].
In a Danish nationwide study of patients with type 2 dia-betes without a prior history of CVD, Mogensen et al. found
that metformin + sulphonylurea was associated with
increased risk of cardiovascular and all-cause mortality com-pared with DPP-4 inhibitor: 1.8-fold and 1.5-fold, respectively
[25]. Their use of a different statistical method, Poisson regression, allowed patients to switch groups and contribute risk time on the new treatment, whereas we used an on-treatment approach with censoring at on-treatment discontinu-ation. The consequences of these different statistical approaches are difficult to assess, but it may be speculated that the increased on-treatment time in the Danish study increased the number of registered events, thereby resulting in a numerically increased risk estimation. Still, the numeric direction for CVD mortality related to metformin + sulphony-lurea treatment was similar to our fatal and nonfatal CVD rate, and estimated risks of all-cause mortality were also sim-ilar in both studies.
Using data from the Clinical Practice Research Datalink in the UK [26], Morgan et al. found similar associated risks of CVD and all-cause mortality with metformin + sulphonylurea versus metformin + DPP-4 inhibitor: 1.4-fold and 1.7-fold, when using a similar statistical method (adjusted Cox regres-sion) (15). To compare statistical methods, they used three dif-ferent statistical methods: full- (adjusted Cox regression), direct-, and propensity-matched cohorts. Small numerical differences were found between the methods, but the two lat-ter methods seemed to generate higher risk estimates. In their full population design, they used only laboratory mea-Table 2 – Adjusted hazard ratios (HRs) in patients treated with metformin + sulphonylurea (SU) versus metformin + dipeptidyl peptidase-4 inhibitors (DPP-4i) using multivariate adjusted Cox proportional hazard analyses, propensity-adjusted analyses and propensity-matched patients (2:1). Univariate analyses of single variables used in the multivariate adjusted survival models are also presented.
Severe hypoglycemia Fatal and nonfatal CVD All-cause mortality
HR 95% CI p HR 95% CI p HR 95% CI p
Multivariate adjusted (SU vs DPP-4i) 2.07a 1.11–3.86 0.022 1.17b 1.01–1.37 0.035 1.25b 1.02–1.54 0.030
Propensity-adjusted (SU vs DPP-4i) 2.08a (1.11–3.89) 0.021 1.21b (1.04–1.40) 0.014 1.35b (1.10–1.65) 0.004
Propensity-matched (SU vs DPP-4i) 1.88c (0.98–3.62) 0.057 1.23c (1.05–1.44) 0.010 1.43c (1.15–1.77) 0.001
Age (per year) 1.05 1.03–1.06 <0.001 1.05 1.05–1.06 <0.001 1.11 1.10–1.12 <0.001
Fragile (yes) 2.02 1.30–3.15 0.002 1.94 1.72–2.19 <0.001 2.15 1.89–2.46 <0.001
Female sex – – – 0.76 0.68–0.84 <0.001 0.80 0.71–0.89 <0.001
Prior CVD (yes) – – – 2.52 2.24–2.83 <0.001 1.46 1.27–1.67 <0.001
Statins (yes) – – – 0.74 0.66–0.82 <0.001 0.63 0.56–0.71 <0.001
Antihypertensives (yes) – – – 1.11 0.97–1.27 0.137 0.92 0.80–1.06 0.250
Low-dose aspirin (yes) – – – 1.27 1.13–1.42 <0.001 1.10 0.97–1.24 0.126
CVD, cardiovascular disease; DPP-4i, dipeptidyl peptidase; HR, hazard ratio.
a Adjusted for age and fragility.
bAdjusted for age, sex, fragility, prior CVD, and use of statins, low-dose aspirin, and antihypertensives.
c Matched by age, sex, fragility, prior CVD, and use of statins, low-dose aspirin, and antihypertensives. Other HRs for single variables are
surements to adjust for cardiovascular risk at baseline, whereas we used only history of CVD and cardiovascular pre-ventive treatment. When comparing metformin in combina-tion with sulphonylurea versus DPP-4 inhibitor, irrespective of statistical method, all their results were of the same direc-tion and magnitude as those in our study. Recently Seong et al. published results from the Korean National Health Insurance claims database, also showing that metformin + sulphonylurea was associated with a 1.2-fold higher risk of nonfatal CVD compared with metformin + DPP-4 inhibitor (17). They used an on-treatment approach and their risk esti-mates are similar to our results. A nationwide study from the Taiwan National Health Insurance database by Chang et al. reported numerically increased 1.1-fold (1.11 [95% CI 0.98– 1.27]) risk of nonfatal CVD with metformin + sulphonylurea versus metformin + DPP-4 inhibitor (16). This study used an intention-to-treat approach, and their nonsignificant finding is identical with our significant intention-to-treat result of 1.11 (95% CI 1.01–1.21). The low number of patients treated with metformin + DPP-4 inhibitor (n = 2242) might very well have resulted in too low power for CVD risk estimations.
In this study, we have demonstrated a markedly increased risk of severe hypoglycemia associated with sulphonylurea, compared with DPP-4 inhibitors. It is also already well known that sulphonylurea in comparison with DPP-4 inhibitor treat-ment is associated with increased risk of hypoglycemia, which in turn has been reported to be strongly associated with CVD and mortality [3–5,10,19,27–29]. Hypoglycemic responses, such as sympathoadrenergic activation, have been suspected to trigger CVD and might act as a causal pathway between sulphonylurea treatment and risk of CVD and mor-tality, and may contribute to serious cardiac arrhythmias
[19]. We also found an association between occurrence of sev-ere hypoglycemia and CVD events, similar to another study
[19], in support of this suggested causal mechanism. In addi-tion, the early separation of event curves for CVD and all-cause death, between the two treatment groups, is compati-ble with hypoglycemia being a contributing factor. The hypo-glycemia risk is known to be increased, very early after the start of sulphonylurea medication [30], as noticed in our study.
Myocardial ischemic preconditioning has also been dis-cussed as a potential explanation for the harmful CVD effects observed with sulphonylurea. Sulphonylurea promotes the release of insulin from beta cells by binding to the sulphony-lurea receptor, inhibiting the ATP-sensitive potassium chan-nels[31]. However, sulphonylurea receptors are also present in cardiac muscle cells, and inhibition of ATP-sensitive potas-sium channels impairs ischemic preconditioning[32,33]. Con-sequently, the mechanism for cardiac muscle cells to survive brief ischemic events is disturbed, which may translate to increased CVD risk[34].
Numerically increased risk of fatal-/nonfatal CVD and all-cause mortality was observed with all sulphonylureas + met-formin, compared to DPP-4 inhibitors + metformin. In the cur-rent study, treatment with glibenclamide was associated with the highest mortality-, and CVD risks compared to the other sulphonylureas. This finding might be explained by its higher unselective binding to cardiac muscle cells [34], leading to impaired cardiac preconditioning [32,33], and also by its
higher risk of hypoglycemia[35]compared to other sulphony-lureas. The increased all-cause mortality associated with glibenclamide is supported by a recently published meta-analysis by Simpson et al.[36], demonstrating variations in all-cause mortality risks within the sulphonylurea class. Notably, gliclazide, which never was introduced on the Swed-ish market, has been reported to have the lowest cardiovascu-lar risk in mono- or combination therapy compared to other sulphonylureas[14,37]. In our study, severe hypoglycemia risk was increased with glibenclamide and glimepiride compared to DPP-4 inhibitors in combination with metformin. Glipizide, on the other hand, did not display any significant increase of risk, albeit this has been reported in randomized controlled trials[4,5]. Taken together, current and previous results sug-gest that short-acting glipizide might carry less risk of severe hypoglycemia when compared to long-acting sulphonylureas. Importantly, interpretation of severe hypoglycemia risk for the different sulphonylureas should be made with caution because of the very limited number of events in our study.
Another partial explanation for the increased CVD risk shown with sulphonylurea could be that treatment with DPP-4 inhibitor is not associated with any weight gain, in con-trast to the association of sulphonylurea treatment with mod-erate weight gain. Modmod-erate weight gain has been strongly associated with increased risk of cardiovascular death and all-cause mortality[38,39]. Furthermore, in subgroup analyses of the Trial to Evaluate Cardiovascular Outcomes after Treat-ment with Sitagliptin (TECOS), a positive CVD-preventive effect of DPP-4 inhibitor was reported in obese patients, sug-gesting that body mass index might be an important clinical measurement when considering blood glucose-lowering drug treatment [9]. This reasoning is obviously speculative, and further studies are needed to explore such associations.
A strength of the present work is the population-based, nationwide and unselected real-world study design, which provides a high external validity. In addition, this is a study with full (100%) register coverage for hospitalizations, filled drug prescriptions, and cause of death in a country with an established and complete public healthcare system. Diag-noses in the Swedish Patient Registry have been reported to have high validity [40], and few patients are lost to follow-up. Moreover, the directed acyclic graphs used to create the optimal adjustments of hazard models should provide mini-mal bias[22].
Observational studies such as this obviously have limita-tions. The present study has no information on laboratory measurements, lifestyle parameters, primary healthcare data, or socioeconomic data, and consequently there may be remaining confounding factors. Moreover, glucose-lowering drugs such as metformin could potentially have been used for reasons other than type 2 diabetes treatment (such as polycystic ovarian syndrome or prediabetes), which could influence the results. However, such usage should be very small compared with the type 2 diabetes indication. Specifically, the combination treatments addressed in this study (i.e. metformin together with either a DPP4 inhibitor or a sulphonylurea) would have been used to a negligible extent for conditions other than type 2 diabetes. In the propensity score -matched analyses, the numerical risk of severe hypoglycemia reflected the unmatched analyses, but
without showing statistically significant differences. This was likely due to lack of power, with low number of events and a substantially reduced study cohort (by 32%). We have no information on emigration, which could result in loss to follow-up. However, the on-treatment analyses used in this study should minimize the effects of patients moving out of Sweden. Furthermore, our assessment of severe hypo-glycemia is crude, including only events leading to hospital admission. It was not possible to evaluate other hypoglycemic events in this register-based study. Another limitation is that patients with a recorded hypoglycemia had to survive until this occasion and, if anything, this would underestimate the total mortality rate in these patients compared to those with-out a hypoglycemic event.
Although a major concern with the present analyses is the non-randomized approach with risk for confounding by indi-cation, the data are an in agreement with several other obser-vational studies reporting data raising concerns about the use of sulphonylureas compared with DPP-4 inhibitor, both as
monotherapy and in combinations with metformin
[13,14,16,18,25]. Of these studies, four papers from Denmark, South Korea, Sweden, and the UK have compared metformin combinations with sulphonylurea versus DPP-4 inhibitor using different analytic approaches and data sources. In spite of this, the resulting risk estimates are similar and show increased clinical risks for metformin + sulphonylurea com-pared with metformin + DPP-4 inhibitor[16,25]. These concor-dant study results may thus demonstrate the robustness of epidemiological methods using large register datasets to assess specific research questions. Despite unknown factors potentially confounding the results, the results may raise concerns regarding the cardiovascular safety of sulphony-lurea while awaiting results from the CAROLINA study[21].
5.
Conclusion
This nationwide register-based study shows that type 2 dia-betes patients that received a combination of metformin and sulphonylurea as second-line treatment had an increased risk of subsequent severe hypoglycemia, cardiovascular events and all-cause mortality compared with those receiving metformin in combination with a DPP-4 inhibitor. The causal relationships remain to be further elucidated, but results from this and other observational studies should be consid-ered in the choice of treatment for patients with type 2 dia-betes, until results from future randomized trials are available.
Funding
This study received financial support by AstraZeneca AB.
Conflicts of interest statement
JWE was previously employed by AstraZeneca R&D and has received honoraria or research grants from AstraZeneca, NovoNordisk, Bristol-Myers-Squibb and MSD. JB holds a full-time position at AstraZeneca as epidemiologist. DN has received consultancy fees from Novo Nordisk, Astra Zeneca
and Eli Lilly. MT is employed by an independent statistical consultant company, Statisticon AB, Uppsala, Sweden, for which AstraZeneca Nordic-Baltic is a client. TN is on the national board of NovoNordisk, Sanofi-Aventis and Eli Lilly. AN has honoraria from expert group participation (MSD, Astra Zeneca, Eli Lilly). Editorial support funded by AstraZeneca was provided by Oxford PharmaGenesis.
Author contributions
All authors participated in the study design. MT performed the data collection and statistical analyses after discussion with all authors. All authors participated in data interpreta-tion and in writing the manuscript. All authors took final responsibility in the decision to submit for publication.
Acknowledgments
We are grateful to Urban Olsson at Statisticon for support with data merging and management of the database, and to Susanna Jerstro¨m and Helena Goike at AstraZeneca for logis-tic support and valuable comments on the manuscript. Jan W. Eriksson is the guarantor of the manuscript. Editorial support
funded by AstraZeneca was provided by Oxford
PharmaGenesis.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.diabres. 2016.04.055.
R E F E R E N C E S
[1]Norhammar A, Bodegard J, Nystrom T, Thuresson M, Eriksson JW, Nathanson D. Incidence, prevalence, and mortality of type 2 diabetes requiring glucose-lowering treatment, and associated risks of cardiovascular complications: a nationwide study in Sweden, 2006–2013. Diabetologia 2016. Accepted.
[2]Mishriky BM, Cummings DM, Tanenberg RJ. The efficacy and safety of DPP4 inhibitors compared to sulfonylureas as add-on therapy to metformin in patients with Type 2 diabetes: a systematic review and meta-analysis. Diabetes Res Clin Pract 2015;109:378–88.
[3]Gallwitz B, Rosenstock J, Rauch T, et al. 2-Year efficacy and safety of linagliptin compared with glimepiride in patients with type 2 diabetes inadequately controlled on metformin: a randomised, double-blind, non-inferiority trial. Lancet 2012;380:475–83.
[4]Goke B, Gallwitz B, Eriksson J, Hellqvist A, Gause-Nilsson I, Investigators DC. Saxagliptin is non-inferior to glipizide in patients with type 2 diabetes mellitus inadequately controlled on metformin alone: a 52-week randomised controlled trial. Int J Clin Pract 2010;64:1619–31. [5]Seck T, Nauck M, Sheng D, et al. Safety and efficacy of
treatment with sitagliptin or glipizide in patients with type 2 diabetes inadequately controlled on metformin: a 2-year study. Int J Clin Pract 2010;64:562–76.
[6] The management of type 2 diabetes. London: National Institute for Health and Care Excellence; 2009.
[7] Swedish National Guidelines for Diabetes Care. In: Welfare NBoHa, (ed.). 2015; p. 120.
[8]Scirica BM, Bhatt DL, Braunwald E, et al. Saxagliptin and cardiovascular outcomes in patients with type 2 diabetes mellitus. N Engl J Med 2013;369:1317–26.
[9]Green JB, Bethel MA, Armstrong PW, et al. Effect of sitagliptin on cardiovascular outcomes in type 2 diabetes. N Engl J Med 2015.
[10]White WB, Cannon CP, Heller SR, et al. Alogliptin after acute coronary syndrome in patients with type 2 diabetes. N Engl J Med 2013;369:1327–35.
[11]Garratt KN, Brady PA, Hassinger NL, Grill DE, Terzic A, Holmes Jr DR. Sulfonylurea drugs increase early mortality in patients with diabetes mellitus after direct angioplasty for acute myocardial infarction. J Am Coll Cardiol 1999;33:119–24. [12]Mogensen UM, Andersson C, Fosbol EL, et al. Sulfonylurea in
combination with insulin is associated with increased mortality compared with a combination of insulin and metformin in a retrospective Danish nationwide study. Diabetologia 2015;58:50–8.
[13]Morgan CL, Mukherjee J, Jenkins-Jones S, Holden SE, Currie CJ. Association between first-line monotherapy with sulphonylurea versus metformin and risk of all-cause mortality and cardiovascular events: a retrospective, observational study. Diabetes Obes Metab 2014;16:957–62. [14]Schramm TK, Gislason GH, Vaag A, et al. Mortality and
cardiovascular risk associated with different insulin secretagogues compared with metformin in type 2 diabetes, with or without a previous myocardial infarction: a nationwide study. Eur Heart J 2011;32:1900–8.
[15]Simpson SH, Majumdar SR, Tsuyuki RT, Eurich DT, Johnson JA. Dose-response relation between sulfonylurea drugs and mortality in type 2 diabetes mellitus: a population-based cohort study. CMAJ: Canadian Medical Association Journal = journal de l’Association medicale canadienne
2006;174:169–74.
[16]Morgan CL, Mukherjee J, Jenkins-Jones S, Holden SE, Currie CJ. Combination therapy with metformin plus
sulphonylureas versus metformin plus DPP-4 inhibitors: association with major adverse cardiovascular events and all-cause mortality. Diabetes Obes Metab 2014;16:977–83. [17]Chang YC, Chuang LM, Lin JW, Chen ST, Lai MS, Chang CH.
Cardiovascular risks associated with second-line oral antidiabetic agents added to metformin in patients with type 2 diabetes: a nationwide cohort study. Diabetic Med: J Br Diabetic Assoc 2015.
[18]Seong JM, Choi NK, Shin JY, et al. Differential cardiovascular outcomes after dipeptidyl peptidase-4 inhibitor,
sulfonylurea, and pioglitazone therapy, all in combination with metformin, for type 2 diabetes: a population-based cohort study. PLoS ONE 2015;10:e0124287.
[19]Zoungas S, Patel A, Chalmers J, et al. Severe hypoglycemia and risks of vascular events and death. N Engl J Med 2010;363:1410–8.
[20]Nissen SE. Cardiovascular effects of diabetes drugs: emerging from the dark ages. Ann Intern Med 2012;157:671–2. [21]Rosenstock J, Marx N, Kahn SE, et al. Cardiovascular outcome
trials in type 2 diabetes and the sulphonylurea controversy: rationale for the active-comparator CAROLINA trial. Diabetes Vasc Dis Res 2013;10:289–301.
[22]Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol 2008;8:70.
[23]Sekhon and Jasjeet. Multivariate and propensity score matching software with automated balance optimization. J Stat Softw 2011;7:1–52.
[24] CT. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2015.
[25] Mogensen UM, Andersson C, Fosbol EL, et al. Cardiovascular safety of combination therapies with incretin-based drugs and metformin compared with a combination of metformin and sulphonylurea in type 2 diabetes mellitus–a retrospective nationwide study. Diabetes Obes Metab 2014;16:1001–8. [26] Herrett E, Gallagher AM, Bhaskaran K, et al. Data resource
profile: clinical practice research datalink (CPRD). Int J Epidemiol 2015.
[27] Monami M, Dicembrini I, Kundisova L, Zannoni S, Nreu B, Mannucci E. A meta-analysis of the hypoglycaemic risk in randomized controlled trials with sulphonylureas in patients with type 2 diabetes. Diabetes Obes Metab 2014;16:833–40. [28] Action to Control Cardiovascular Risk in Diabetes Study G, Gerstein HC, Miller ME, et al. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008;358:2545–59. [29] Zhao Y, Campbell CR, Fonseca V, Shi L. Impact of
hypoglycemia associated with antihyperglycemic medications on vascular risks in veterans with type 2 diabetes. Diabetes Care 2012;35:1126–32.
[30] Thule´ PM, Umpierrez G. Sulfonylureas: a new look at old therapy. Curr Diab Rep 2014;14:473.
[31] Boyd 3rd AE, Aguilar-Bryan L, Nelson DA. Molecular mechanisms of action of glyburide on the beta cell. Am J Med 1990;89:3S–10S. discussion 51S–3S.
[32] Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998; 352: 837–53.
[33] Leibowitz G, Cerasi E. Sulphonylurea treatment of NIDDM patients with cardiovascular disease: a mixed blessing? Diabetologia 1996;39:503–14.
[34] Abdelmoneim AS, Hasenbank SE, Seubert JM, Brocks DR, Light PE, Simpson SH. Variations in tissue selectivity amongst insulin secretagogues: a systematic review. Diabetes Obes Metab 2012;14:130–8.
[35] Gangji AS, Cukierman T, Gerstein HC, Goldsmith CH, Clase CM. A systematic review and meta-analysis of hypoglycemia and cardiovascular events: a comparison of glyburide with other secretagogues and with insulin. Diabetes Care 2007;30:389–94.
[36] Simpson SH, Lee J, Choi S, Vandermeer B, Abdelmoneim AS, Featherstone TR. Mortality risk among sulfonylureas: a systematic review and network meta-analysis. Lancet Diabetes Endocrinol 2015;3:43–51.
[37] Chan SP, Colagiuri S. Systematic review and meta-analysis of the efficacy and hypoglycemic safety of gliclazide versus other insulinotropic agents. Diabetes Res Clin Pract 2015;110:75–81.
[38] Bodegard J, Sundstrom J, Svennblad B, Ostgren CJ, Nilsson PM, Johansson G. Changes in body mass index following newly diagnosed type 2 diabetes and risk of cardiovascular mortality: a cohort study of 8486 primary-care patients. Diabetes Metabol 2013;39:306–13.
[39] Grundvold I, Bodegard J, Nilsson PM, et al. Body weight and risk of atrial fibrillation in 7,169 patients with newly diagnosed type 2 diabetes; an observational study. Cardiovasc Diabetol 2015;14:5.
[40] Ludvigsson JF, Andersson E, Ekbom A, et al. External review and validation of the Swedish national inpatient register. BMC Public Health 2011;11:450.
