Prognostic values of fasting hyperglycaemia in non-diabetic patients with acute coronary syndrome: A prospective cohort study

Background:

Controversy remains regarding the prevalence of hyperglycaemia in non-diabetic patients hospitalised with acute coronary syndrome and its prognostic value for long-term outcomes.

Methods and results:

We evaluated the prevalence of hyperglycaemia (defined as fasting glycaemia ⩾10 mmol/l) among patients with no known diabetes at the time of enrolment in the prospective Special Program University Medicine-Acute Coronary Syndromes cohort, as well as its impact on all-cause death, myocardial infarction, stroke and incidence of diabetes at one year. Among 3858 acute coronary syndrome patients enrolled between December 2009–December 2014, 709 (18.4%) had known diabetes, while 112 (3.6%) of non-diabetic patients had hyperglycaemia at admission. Compared with non-hyperglycaemic patients, hyperglycaemic individuals were more likely to present with ST-elevation myocardial infarction and acute heart failure. At discharge, hyperglycaemic patients were more frequently treated with glucose-lowering agents (8.9% vs 0.66%, p<0.001). At one-year, adjudicated all-cause death was significantly higher in non-diabetic patients presenting with hyperglycaemia compared with patients with no hyperglycaemia (5.4% vs 2.2%, p=0.041) and hyperglycaemia was a significant predictor of one-year mortality (adjusted hazard ratio 2.39, 95% confidence interval 1.03–5.56). Among patients with hyperglycaemia, 9.8% had developed diabetes at one-year, while the corresponding proportion among patients without hyperglycaemia was 1.8% (p<0.001). In multivariate analysis, hyperglycaemia at presentation predicted the onset of treated diabetes at one-year (odds ratio 4.15, 95% confidence interval 1.59–10.86; p=0.004).

Conclusion:

Among non-diabetic patients hospitalised with acute coronary syndrome, a fasting hyperglycaemia of ⩾10 mmol/l predicted one-year mortality and was associated with a four-fold increased risk of developing diabetes at one year.

Acute coronary syndrome, diabetes, hyperglycaemia

Introduction

Type 2 diabetes mellitus is characterised by a state of chronic hyperglycaemia resulting from resistance to insulin-stimulated glucose uptake, whereby insulin secretion and/or action are impaired.¹ Diabetes is diagnosed either before patients develop overt cardiovascular disease or at the time of screening for risk factors in the context of an acute cardiovascular event, such as acute coronary syndrome (ACS). According to the European Society of Cardiology (ESC) guidelines² the presence of diabetes per se in an ACS setting is regarded as a marker of increased risk for heart failure, stroke, renal failure and bleeding as well as a predictor or recurrent ischaemic events after hospital discharge.^2,3 Controversy remains on how to interpret hyperglycaemia in patients presenting with ACS.^1,2,4,5 Observational studies suggest that hyperglycaemia in the context of ACS is associated with increased in-hospital mortality, but long-term data as well as standardization of hyperglycaemia cut-off levels are lacking.^6,,,–10 In addition, protocols of early aggressive glucose-lowering strategies in ACS patients have led to discordant results and were not performed in the contemporary era of drug-eluting stents and potent antiplatelet therapy.¹¹ Many patients with hyperglycaemia at the time of ACS will normalise their glycaemic state after the acute event,^1,4 but few data exist on the rate of subsequent confirmed diabetes in this patient population. This issue is especially important in avoiding potential over- or undertreatment.⁵

In the present study, we investigated data taken from a large prospective contemporary ACS cohort to assess the prevalence of fasting hyperglycaemia in patients with no known diabetes at the time of the ACS event leading to their inclusion in the cohort, and evaluate its impact on clinical outcomes and the likelihood of developing diabetes at one year.

Methods

Study population

We analysed data from the Special Program University Medicine-Acute Coronary Syndromes (SPUM-ACS; NCT01000701), in which patients were recruited at four academic centres in Switzerland with the aim of assessing the level of long-term control of secondary preventive targets following ACS.^12,13 For the present analysis, we considered all ACS patients who were enrolled between December 2009–December 2014. Inclusion criteria were age ⩾18 years, an index diagnosis of ST-elevation myocardial infarction (STEMI), non ST-elevation myocardial infarction (NSTEMI) or unstable angina.^12,13 Exclusion criteria were limited to severe physical disability or dementia, and a life expectancy of less than one year for non-cardiac events. The first available fasting plasma glucose measurement was collected within the first 24 h of hospital admission, as recommended by the American Heart Association (AHA) guidelines for key data elements and definitions for measuring outcomes in ACS registries.¹⁴ For this analysis, we excluded patients for whom glucose-lowering therapy and fasting glucose values had not been collected at presentation. All local ethical committees approved the protocol. All SPUM-ACS patients provided written informed consent.

Hyperglycaemia and the definition of diabetes

At presentation, patients were considered to be diabetic if they were on any glucose-lowering treatment or on a low carbohydrate diet. In the absence of robust data on treatment thresholds and glucose targets in the context of STEMI and NSTEMI,^1,14 we defined hyperglycaemia at presentation in non-diabetic patients as fasting plasma glucose ⩾10 mmol/l recorded within the first 24 h of hospital admission.

Follow-up and study endpoints

The follow-up strategy of the SPUM-ACS cohort has been detailed elsewhere.¹² In brief, participants were first contacted by telephone by a trained study nurse at 30 days post-ACS, and at one year in the form of a clinical check-up visit. Among the 3858 SPUM-ACS participants included in our cohort, 3673 (95.2%) were reached for the one-year follow-up and 1952 (68.4%) had a fasting plasma glucose measurement during a clinical visit. Primary clinical outcome measures were defined as the incidence at one year of all-cause death. Secondary clinical outcomes were cardiac death, myocardial infarction and major cardiovascular events (MACCEs), defined as a composite of all-cause death, myocardial infarction and stroke. Secondary outcome measures were the incidence of new diabetes at one year and cardiac death and myocardial infarction. New onset diabetes was defined by the presence of a glucose-lowering treatment at one year in patients with no known diabetes at presentation. Clinical events were adjudicated using pre-specified forms by a panel of independent experts (three certified cardiologist) blinded to diabetes status.

Statistical analysis

At the time of hospitalization for ACS, we categorised patients into two groups, namely those with and those without pre-existing diabetes, and those non-diabetics were further stratified according to the presence or not of fasting hyperglycaemia at presentation. Two-tailed Fisher’s exact tests for dichotomous variables, larger Chi-squared test for Independence, and Mann-Whitney-U tests for continuous variables, were utilised as appropriate. We assessed the association between hyperglycaemia in non-diabetics and clinical outcomes at one year using Cox proportional hazard models; and for non-diabetics we compared clinical outcomes according to fasting hyperglycaemia status. The Kaplan-Meier cumulative hazard curves for an all-cause death outcome at one year were obtained for three groups: (a) non-diabetic patients with fasting glucose <10 mmol/l at presentation; (b) non-diabetic patients with fasting glucose ⩾10 mmol/l at presentation and (c) diabetic patients. We then conducted a univariate and multivariate Cox regression analysis adjusting for potentially confounding variables (age, sex, history of myocardial infarction, creatinine value at admission and a Killip score) in non-diabetic patients in order to assess the association between fasting hyperglycaemia at admission and the incidence at one-year of all-cause death using hazard ratio (HR) and 95% confidence interval (CI). In addition we studied the newly treated diabetes patients at the one-year visit using odds ratio (OR) and 95% CI, while adjusting for potentially confounding variables (age, body mass index, dyslipidaemia, hypertension, type of ACS (namely unstable angina, non-ST elevation myocardial infarction or STEMI) and any hypoglycaemic treatment at discharge). All hypothesis tests were two-sided with the significance level set at 5%. Statistical analyses were performed using Stata software (Version 13, Stata Corp, College Station, Texas, USA).

Results

Patients’ characteristics

Among 5446 patients hospitalised for ACS, 174 (3.2%) were excluded because of an unknown diabetes status, 583 (10.7%) for missing values of glucose at hospitalization, 831 (15.3%) for random values of glucose, yielding a final sample of 3858 patients for the analysis (Figure 1). Among them, 709 (18.4%) were diabetic patients who received glucose-lowering agents or were following a low carbohydrate diet (baseline characteristics are reported in Supplementary Material Table 1). Non-diabetic patients (n=3149, 81.6%) compared with diabetics were younger (62.4±12.2 years vs 66.3±11.7 years, p<0.001) and more likely to present with STEMI (54.5% vs 42.0%, p<0.001). At baseline, non-diabetic patients were less frequently on aspirin, statins, angiotensin-converting enzyme (ACE)-inhibitors and beta-blockers (Supplementary Material Table 2), and less often on glucose-lowering therapies at discharge (insulin 0.3% vs 38.8, and oral glucose-lowering agents 0.6% vs 64.6%, p<0.001). Among non-diabetic patients, 112 (3.6%) had a fasting hyperglycaemia ⩾10 mmol/l at presentation (Table 1). Hyperglycaemic non-diabetic patients presented more often with STEMI (85.7% vs 53.3%, p<0.001), pulmonary oedema (Killip class III; 2.7% vs 1.6%, p<0.001) as well as cardiogenic shock (Killip class IV, 8.9% vs 1.9%, p<0.001) and required vasopressor therapy (12.5% vs 2.0%, p<0.001) more frequently compared with those without hyperglycaemia (Table 1). Moreover, patients with fasting hyperglycaemia had higher troponin I peak levels (8.3 mg/l vs 2.7 mg/l, p<0.001), higher creatinine-kinase (CK) peak levels (2398 U/l vs 798 U/l, p<0.001) and higher creatinine values at presentation (85.0 umol/l vs 77.0 umol/l, p=0.001).

Figure 1.

Flowchart of the studied cohort.

Table 1.

Baseline characteristics according to hyperglycaemia in nondiabetic patients hospitalised for acute coronary syndrome (ACS).

Demographic	Glycaemiaa<10 mmol/l (n=3037)	Glycaemia⩾10 mmol/l (n=112)	p-Value
Age, years	62.4±12.2	63.3±12.8	0.466
Women, n (%)	610 (20.1)	24 (21.4)	0.276
BMI, kg/m2	26.8±4.2	27.2±4.0	0.242
Hypertension, no. (%)	1593 (52.5)	59 (52.7)	0.522
Current smoking, no. (%)	2134 (70.3)	69 (61.6)	0.074
Dyslipidaemia, no. (%)	1816 (59.8)	65 (58.0)	0.382
Chronic CAD, no. (%)	793 (26.1)	33 (29.5)	0.212
History of MI, no (%)	398 (13.1)	8 (7.1)	0.037
PVD, no. (%)	171 (5.6)	4 (3.6)	0.629
Clinical presentation
ACS type			<0.001
STEMI	1620 (53.3)	96 (85.7)
NSTEMI	1198 (39.5)	16 (14.3)
Unstable angina	130 (4.3)	0 (0.0)
Killip classification at admissionb			<0.001
I, no. (%)	2434 (80.2)	73 (65.2)
II, no. (%)	246 (8.1)	16 (14.3)
III, no. (%)	47 (1.6)	3 (2.7)
IV, no. (%)	58 (1.9)	10 (8.9)
Heart rate, bpm	75.8±158	82.4±31.4	<0.001
SBP at admission, mm Hg	129.5±23.2	128.3±26.9	0.593
DBP at admission, mm Hg	76.2±14.5	76.1±14.9	0.935
Vasopressor therapy, no. (%)	62 (2.0)	14 (12.5)	<0.001
Biochemical
Troponin I at admission, mg/l	0.3 (0.1–1.1)	0.2 (0.1–1.1)	0.645
Peak troponin I, mg/l	2.7 (0.6–9.9)	8.3 (3.2–21.5)	<0.001
CK value at admission, U/l	223.0 (109.0–517.0)	246.0 (127.0–545.0)	0.306
CK peak value, U/l	798.0 (271.0–1926.5)	2398.0 (1170.0–4375.0)	<0.001
Hb at admission, gr/l	138.0 (123.0–149.0)	137.5 (84.0–149.0)	0.492
Creatinine at admission, umol/l	77.0 (66.0–91.0)	85.0 (71.0–101.0)	<0.001
Type of intervention
PCI (with or without stent)	2722 (89.6)	105 (93.8)	0.157
CABG	51 (1.7)	3 (2.7)	0.424
Conservative	264 (8.7)	4 (3.6)	0.056

BMI: body mass index; CABG: coronary artery by-pass grafting; CAD: coronary artery disease; CK: creatine-kinase; DBP: diastolic blood; Hb: haemoglobin; MI: myocardial infarction; NSTEMI: non ST-segment elevation myocardial infarction; PCI: percutaneous coronary intervention; PVD: peripheral vascular disease; SBP: systolic blood pressure; STEMI: ST-segment elevation myocardial infarction.

Data are expressed as mean (±standard deviation), median (interquartile range) or number (no.), percentages (%).

a

Intended as fasting glycaemia at admission.

b

38 (6.5%) missing data in the diabetes patients, 378 (9.7%) missing data in the ‘Glycaemia⩽10 mmol/l group’ and 19 (7.8%) missing data in the ‘Glycaemia⩾10 mmol/l group’.

Table 2.

Medications at admission and at discharge according to hyperglycaemia in nondiabetic patients hospitalised for acute coronary syndrome.

	Glycaemiaa<10 mmol/l (n=3037)	Glycaemia⩾10 mmol/l (n=112)	p-Value
Medications at admission
Aspirin, no. (%)	843 (27.8)	19 (17.0)	0.006
P2Y12 inhibitors, no. (%)	101 (3.3)	7 (6.3)	0.495
Oral anticoagulation, no. (%)	85 (2.8)	3 (2.7)	0.617
ACE therapy, no. (%)	413 (13.6)	15 (13.4)	0.553
AT-II therapy, no. (%)	526 (17.3)	18 (16.2)	0.419
B-blocker therapy, no. (%)	641 (21.1)	14 (12.5)	0.016
Calcium antagonists, no. (%)	293 (9.6)	5 (4.5)	0.039
Statin, no. (%)	784 (25.8)	15 (13.4)	0.001
Amiodarone, no. (%)	23 (0.8)	1 (0.9)	0.510
Digoxin, no. (%)	13 (0.4)	0 (0.0)	0.667
Antiarrythmics others, no. (%)	19 (0.6)	0 (0.0)	0.313
Nitrate, no. (%)	105 (3.5)	1 (0.9)	0.131
Diuretic therapy, no. (%)	400 (13.2)	10 (8.9)	0.118
NSAID, no. (%)	109 (3.6)	7 (6.3)	0.116
Proton pump inhibitors, no. (%)	387 (12.7)	10 (8.9)	0.146
Medications at discharge
Aspirin, no. (%)	3009 (99.1)	111 (99.1)	0.724
P2Y12 inhibitors, no. (%)	2860 (94.2)	105 (93.8)	0.281
Oral anticoagulation, no. (%)	214 (7.1)	10 (8.9)	0.272
ACE therapy, no. (%)	2325 (76.6)	94 (83.9)	0.040
AT-II therapy, no. (%)	398 (13.1)	11 (9.8)	0.194
B-blocker therapy, no. (%)	2446 (80.5)	81 (72.3)	0.032
Calcium antagonists, no. (%)	197 (6.5)	5 (4.5)	0.264
Statin, no. (%)	2978 (98.1)	109 (97.3)	0.380
Amiodarone, no. (%)	58 (1.9)	2 (1.8)	0.640
Digoxin, no. (%)	14 (0.5)	1 (0.9)	0.420
Antiarrythmics others, no. (%)	6 (0.2)	0 (0.0)	0.805
Nitrate, no. (%)	178 (5.9)	9 (8.0)	0.218
Diuretic therapy, no. (%)	585 (19.3)	30 (26.8)	0.030
NSAID, no. (%)	42 (1.4)	1 (0.9)	0.544
Proton pump inhibitors, no. (%)	795 (26.2)	39 (34.8)	0.029
Insulin therapy, no. (%)	4 (0.1)	7 (6.3)	<0.001
Oral antidiabetics, no. (%)	16 (0.5)	4 (3.8)	0.005

ACE: angiotensin-converting enzyme; AT-II: angiotensin receptor II antagonist; NSAID: non-steroidal anti-inflammatory drug.

Data are expressed as number (no.), percentages (%). Two-tailed Fisher’s exact tests for dichotomous

variables for independence were employed.

a

Intended as fasting glycaemia at admission.

Baseline and discharge therapies of non-diabetic patients classified for the presence of hyperglycaemia are presented in Table 2. At admission, the main differences in therapy were a lower use of aspirin in the hyperglycaemia group (17.0% vs 27.8%, p=0.006), statins (13.4% vs 25.8%, p=0.001), beta-blockers (12.5% vs 21.1%, p=0.016) and calcium antagonists (4.5% vs 9.6%, p=0.039). Discharge treatments were almost similar, except for a broader use in hyperglycaemic patients of ACE-inhibitors (83.9% vs 76.6%, p=0.040), diuretics (26.8% vs 19.3%, p=0.030), proton pump inhibitors (34.8% vs 26.2%, p=0.029), oral glucose-lowering agents (3.8% vs 0.5%, p<0.005), as well as insulin therapy (6.3% vs 0.1%, p<0.001).

Impact of diabetes and fasting hyperglycaemia on outcomes

Non-diabetic patients had significantly better clinical outcomes than diabetics at one year (Table 3 and Supplementary Material Table 3): all-cause death (2.2% vs 4.2%, p=0.004), MACCE (6.1% vs 11.4% vs 6.1%, p<0.001), myocardial infarction (3.1% vs 5.2%, p=0.005) as well as stroke (1.4% vs 2.8% vs 1.4%, p=0.010). However, as shown in the Kaplan-Meyer analysis (Figure 2), all-cause death at one year in non-diabetic patients with hyperglycaemia was closest to diabetic patients and worse compared with normoglycaemic patients (p=0.024). 4. The univariate Cox regression showed an association between hyperglycaemia and all-cause death at one year in non-diabetic patients with an unadjusted HR of 2.53 (95% CI 1.10–5.83, p=0.03). The significant association persisted in multivariate analysis (adjusted HR 2.39; 95% CI 1.03–5.56, p=0.04) (Table 4). No significant association was found for other clinical outcomes (Table 3).

Table 3.

Univariate analysis of diabetes and presence of hyperglycaemia in nondiabetic patients for all-cause death, cardiac death, myocardial infarction and major adverse cardiovascular and cerebrovascular events (MACCEs) at one-year.

	Diabetic patients		Nondiabetic patients (glycaemia<10 mmol/l vs ⩾10 mmol/l )
	Univariate analysis HR (95% CI)	p-Value	Univariate analysis HR (95% CI)	p-Value
Variables at 365 days
All-cause death	1.87 (1.22–2.86)	0.004	2.53 (1.10–5.83)	0.030
Cardiac death	1.89 (1.01–3.53)	0.046	1.77 (0.42–7.39)	0.435
Myocardial infarction	1.80 (1.22–2.67)	0.003	0.64 (0.16–2.61)	0.535
MACCEs	1.72 (1.28–2.32)	<0.001	1.20 (0.59–2.45)	0.614

CI: confidence interval; HR: hazard ratio.

MACCEs represent the composite outcome of all-cause death, myocardial infarction and stroke.

Figure 2.

One year Kaplan-Meier survival curves stratified by diabetic status and presence of hyperglycaemia (i.e. glycaemia⩾10 mmol/l) at admission.

Table 4.

Univariate and multivariate Cox regression analysis of independent baseline variables (age, sex, Killip score, renal insufficiency at admission and presence of hyperglycaemia (i.e. glycaemia ⩾10 mmol/l at admission) in nondiabetic patients, with regard to all-cause mortality at one year.

Independent variables	Univariate analysis		Multivariate analysis
	HR (95% CI)	p-Value	HR (95% CI)	p-Value
Hyperglycaemia at admission	2.53 (1.10–5.83)	0.030	2.39 (1.03–5.56)	0.043
Agea	1.08 (1.06–1.11)	<0.001	1.07 (1.05–1.10)	<0.001
Sex	1.23 (0.71–2.12)	0.456	n.a.	n.a.
History of MI	2.63 (1.57–4.40)	<0.001	1.71 (1.00–2.95)	0.052
Renal insufficiency~ at admission	1.01 (1.00–1.01)	<0.001	1.00 (1.00–1.01)	<0.001
Killip score (I vs II–IV)	3.36 (2.04–5.54)	<0.001	2.39 (1.44–3.97)	0.001

CI: confidence interval; HR: hazard ratio; MI: myocardial infarction; n.a.: not adjusted for.

a

Intended as continuous variable.

Fasting hyperglycaemia is associated with the onset of diabetes at one year

A total of 11 non-diabetic patients with hyperglycaemia at presentation (9.8%) developed diabetes requiring glucose-lowering treatment at one year, compared with 50 non-diabetic patients without hyperglycaemia (1.6%) (p<0.001). Logistic regression analysis showed a significant association between hyperglycaemia at admission and confirmed diabetes requiring treatment at one year (OR 5.10, 95% CI 2.44–10.66, p<0.001) (Table 5). The association remained significant after adjustment for potentially confounding factors (OR 4.15, 95% CI 1.59–10.86, p=0.004). Among the 1651 non-diabetic patients tested for hyperglycaemia at one year, we detected a fasting plasma glucose ⩾7 mmol/l in 100 patients (6.1%). This proportion was significantly higher (10/42, 23.8%) in patients with hyperglycaemia at admission compared with patients without hyperglycaemia at admission (90/1609, 5.6%, p<0.001).

Table 5.

Univariate and multivariate analysis of age, sex, body mass index (BMI), hypertension history at admission, smoking status, Killip status, type of acute coronary syndrome, renal insufficiency at admission and presence of hyperglycaemia (i.e. glycaemia⩾10 mmol/l at admission) in patients without a history of diabetes at admission.

Association between hyperglycaemia and treated diabetes at one year.
Independent variables	Univariate analysis		Multivariate analysis
	OR (95% CI)	p-Value	OR (95% CI)	p-Value
Hyperglycaemia	5.10 (2.44–10.66)	<0.001	4.15 (1.59–10.86)	0.004
Age	1.00 (0.98–1.02)	0.718	0.99 (0.97–1.02)	0.637
BMI	1.11 (1.06–1.17)	<0.001	1.11 (1.05–1.18)	<0.001
Type of ACS	2.00 (1.19–3.36)	0.009	1.11 (1.05–1.18)	0.003
HTA at admission	1.46 (0.72–2.46)	0.150	1.26 (0.69–2.30)	0.449
Dyslipidaemia at admission	1.20 (0.71–2.03)	0.509	1.01 (0.57–1.80)	0.967
OADs at discharge	69.13 (24.80–192.70)	<0.001	78.84 (25.70–241.82)	<0.001
Insulin at discharge	39.27 (10.28–150.08)	<0.001	15.60 (2.88–84.40)	0.001

ACS: acute coronary syndrome; CI: confidence interval; HTA: hypertension; OAD: oral antidiabetic drug; OR: odds ratio.

Discussion

In this multicentre, prospective real world ACS cohort, we found that 3.6% non-diabetic patients presented with a fasting hyperglycaemia of ⩾10 mmol/l, a threshold for which glucose-lowering treatment should be considered according to ESC guidelines.² Such patients more frequently presented in high-risk settings such as STEMI, heart failure as well as cardiogenic shock, suggesting that hyperglycaemia may be have been related to neurohumoral activation. As expected, diabetic patients had higher event rates at one year compared to non-diabetics. Hyperglycaemia among non-diabetics was associated with increased mortality, as well as a four-fold increased risk of being treated for diabetes at one year among non-diabetics. Approximately one-quarter of patients with hyperglycaemia at baseline developed new onset diabetes or maintained a persistent glucose disorder with a fasting plasma glucose ⩾7 mmol/l at one year. However, plasma glucose levels at one year were available for only 53.9% of non-diabetic patients, implying that the true prevalence of new cases of diabetes may have been underestimated.

Several studies have shown that patients with ACS and diabetes have a worse prognosis compared to patients with no pre-existing diabetes. As an example, in the Swiss Acute Myocardial Infarction in Switzerland (AMIS) registry that includes 3565 diabetics and 15,531 non-diabetic patients, mortality was two-fold higher (12.1% vs 6.1%, p<0.001) in patients with diabetes compared to those without it.¹⁵ In a registry of 11,232 patients with myocardial infarction in the USA, overall mortality after a follow-up of 3.5 years was significantly higher in patients with diabetes with a relative risk (RR) of 1.68 (95% CI 1.44–1.95, p<0.001) compared with non-diabetics. The same was true for cardiovascular events (RR 1.47, 95% CI 1.27–1.70) and heart failure (RR 1.89, 95% CI 1.34–2.67).¹⁶ Among the 13,526 patients included in the Global Registry of Acute Coronary Events (GRACE), elevated fasting glucose levels were associated with increased in-hospital and six-month mortality.¹⁷ Some studies also suggested that newly diagnosed hyperglycaemia could be considered an equivalent of pre-existing diabetes. Indeed, among 8795 patients with NSTEMI, the presence of previously unknown diabetes (1069 or 12% of patients with glucose ⩾7.0 mmol/l at presentation) was associated with a higher risk of 30-day mortality with an OR of 1.65 (95% CI 1.09–2.48), as was previously known diabetes (2860 or 32.5% of patients) with an OR of 1.40 (95% CI 1.01–1.93).¹⁸

However, so far none of the available studies systematically validated glucose levels with a second assessment during follow-up, nor did they report anti-diabetic treatment initiation in the long-term. Based on our findings, the incidence of newly treated diabetes at one year after ACS remains the minority including in those with hyperglycaemia.

So far there is not a broadly accepted cut-off value for the diagnosis nor consensus on the initiation of glucose-lowering treatment in ACS patients with hyperglycaemia. The ESC guidelines for the management of ACS² and those for diabetes¹ consider any glucose value>10 mmol/l (>180 mg/dl) as an indication for glucose-lowering therapy, without mentioning whether this concerns only diabetics or all ACS patients. The French Society of Cardiology states that admission levels of glucose in ACS patients cannot be used to diagnose hyperglycaemia or diabetes, and does not predict glucose intolerance after the acute phase.⁵ According to these guidelines, admission glucose levels should therefore not be used to classify ACS patients as glucose-intolerant, but rather to initiate insulin treatment in the acute setting.⁵ Thus, there is still a need to define the target glucose levels that are associated with the best outcomes in hospitalised ACS patients, and determine whether these targets differ in patients with or without pre-existing diabetes mellitus.¹⁹ In addition, more data are needed to determine whether persisting hyperglycaemia during ACS would have a greater negative impact than solely transient hyperglycaemia at admission in terms of survival, shorter length of stay in hospital, in-hospital complications or left ventricular systolic function. Based on our data, non-diabetic patients presenting with fasting hyperglycaemia⩾10 mmol/l should indeed be screened for diabetes after the acute phase, but may not need upfront glucose-lowering treatment. Indeed, only 10% of patients received anti-diabetic treatment at one year and one-quarter fulfilled criteria of diabetes based on fasting plasma glucose levels in our study. Therefore, the initiation of treatment based solely on fasting glycaemic values at hospital admission would result in both overdiagnosis and overtreatment. Aggressive glucose-lowering strategies during the hospital stay have been tested mainly in patients with STEMI with diverging results.¹¹ In the Diabetes Mellitus Insulin-Glucose In Acute Myocardial Infarction (DIGAMI)-I trial including 620 diabetic patients with hyperglycaemia values above 11 mmol/l (198 mg/dl), an insulin-based glucose lowering treatment was superior in terms of one-year mortality (18.6% vs 26.1%, p=0.027) compared to standard care.²⁰ In the DIGAMI-2 trial, 1253 diabetic patients were randomised to an intense short-term (intravenous dextrose-insulin infusion) plus a long-term (subcutaneous insulin injection) regimen or usual care. However, there was no significant difference in the primary endpoint of all-cause mortality after a mean follow-up of two years.²¹ In the Immediate Myocardial Metabolic Enhancement During Initial Assessment and Treatment in Emergency Care (IMMEDIATE) trial involving 852 ACS patients, a glucose-insulin-potassium infusion (GIK) or 5% glucose was initiated in the out-of hospital emergency medical service by paramedics and continued for 12 h. The primary endpoint was progression of ACS to myocardial infarction within 24 h, pre-specified secondary endpoints were pre- or in-hospital cardiac arrest.²² No significant differences were found for the primary endpoint (progression to myocardial infarction): 48.7% vs 52.6% (OR 0.88, 95% CI 0.66–1.13, p=0.28). However, GIK was associated with a significant 52% reduction in rates of cardiac arrest and in-hospital mortality with an OR of 0.48 (4.4 % vs 8.7%, 95% CI 0.27–0.85, p=0.01).¹⁷ In the BIOMarker Study to Identify the Acute Risk of a Coronary Syndrome-2 (BIOMArCS-2), the authors tested the infusion of insulin to restore normoglycaemia in 294 ACS patients with glucose levels at admission between 140–288 mg/dl (12.8–25 mmol/l). The primary endpoint was the concentration of high-sensitivity troponin 72 h after admission. The infusion of insulin five hours after symptom onset and 2.3 h after percutaneous coronary intervention (PCI) did not significantly reduce the 72-hour troponin or creatinine-kinase levels, while a significantly increased in-hospital mortality was observed in the insulin infusion group (5.7% vs 0.7%% p=0.04).²³ Those findings suggest that the management of hyperglycaemia in the acute phase of ACS in non-diabetic patients is uncertain and, if at all, may only be beneficial in the very early phase of ACS.

Limitations

Our study has several limitations. First, some patients had missing fasting glucose data during hospitalization and we therefore excluded them from the analysis. Second, we did not collect a second glucose measurement during hospitalization to confirm abnormal values. Thus, at least in some patients, hyperglycaemia in the acute setting may just reflect sympathetic overactivation due to pain and/or hypotension. Accordingly, an American College of Cardiology (ACC)/American Heart Association (AHA) and ESC consensus paper recommends confirming the first available glucose values during hospitalization in ACS.⁹ However, in this analysis we included patients who had available data for fasting plasma glucose. Those who had hyperglycaemia in a non-fasting condition were excluded. Although, we have adjusted hyperglycaemia for potential confounding factors and found similar HRs in this prospective cohort, we cannot make any inference regarding the causality with clinical outcomes. Third, a baseline glycated haemoglobin was not measured to confirm or reject long-standing hyperglycaemia before the acute event. Fourth, the indication for initiating glucose-lowering therapy was not predefined not standardised, but rather left to the discretion of the treating physician. Fifth, the patients with newly treated diabetes at one year probably include unknown diabetic patients who were not treated at baseline, yielding a possible bias of classification as suggested by the introduction of glucose-lowering therapies already from hospital discharge. Finally, fasting plasma glucose at one year was not systematically collected, as some patients did not return to the one-year clinical visit at one of the involved hospitals for a blood sample.

Conclusion

Among non-diabetic patients hospitalised for ACS, fasting hyperglycaemia of ⩾10 mmol/l at presentation predicted one-year mortality and was associated with a four-fold increased risk of developing diabetes requiring treatment. However, this does not imply that those patients should have upfront glucose-lowering treatment, as at one year less than one-quarter of patients had abnormal glucose levels.

The authors wish to acknowledge the work of the clinical event committee for SPUM-ACS: Matthias Pfisterer, University of Basel (chair), Tiziano Moccetti, CardioCentro Lugano, Lukas Kappenberger, Lausanne University, Switzerland. They thank the local study nurses, the core lab technicians, the central data monitors, the electronic data conducting system (2mt GmbH Ulm, Jürgen Nagler-Ihlein, Torsten Illmann), the research coordinator, Lambertus J van Tits, and the members of the local catheter teams for their invaluable work. Special gratitude is expressed to Aliki Buhayer (Prism Scientific Sàrl) for medical writing support.

Conflict of interest

TFL reports receiving research grants to the institution from Abbot, Biosensors, Biotronik, Boston Scientific, Daichi Sankyo, Eli Lilly and Medtronic, and consultant payments from Amgen, AstraZeneca, Boehringer Ingelheim, Bayer, Merck, and Pfizer, MSD, Roche, and Servier. CMM reports receiving grants from MSD, AstraZeneca, and Roche, and having patents from Mabimmune, CH. SW reports receiving research contracts to the institution from Abbott, Biotronik, Boston Scientific, Biosensors, Cordis, Medtronic, St Jude Medical, and speaker fees from Abbott, Biotronik, Boston Scientific, Biosensors, Medtronic, Eli Lilly, and AstraZeneca. FM has received research grants to the institution from Amgen, AstraZeneca, Boston Scientific, Biotronik, Medtronic, MSD, Eli Lilly, Sanofi, Pfizer, and St Jude Medical including speaker of consultant fees. MR has received institutional research grants from Abbott Vascular, Medtronic, Boston Scientific, Terumo, Biotronik as well as speaker fees from Astra Zeneca and Cordis.

Funding

The work was supported by the Swiss National Science Foundation (SPUM 33CM30-124112 and SPUM 33CM30-140336, Inflammation and acute coronary syndromes (ACS)-Novel strategies for prevention and clinical management, and 32473B-163271, Long-term Benefit of the Multi-Center, Multi-Dimensional Secondary Prevention Program in Patients With Acute Coronary Syndromes). BG’s research is supported by grants from the Geneva University Hospitals, Swiss Heart Foundation, de Reuter Foundation, Gerbex-Bourget Foundation, Gustave-Prevot and Schmidheiny Foundation. NR’s research is supported by grants from the Swiss National Science Foundation (SNSF 320030-150025). RA and NR’s research on cardiovascular prevention is supported by grants from the Swiss Heart Foundation. The SPUM consortium was also supported by Roche Diagnostics, Eli Lilly, AstraZeneca, Medtronic, Merck Sharpe and Dome, Sanofi-Aventis; St Jude Medical as well as the Zurich Heart House - Foundation for Cardiovascular Research, Zurich, Switzerland. None of the funding institutions had any role in the design and conduct of the study, collection, management, analysis and interpretation of the data, as well as preparation, review or approval of the manuscript.

Author notes

*

Both authors have contributed as first authors.