Prognostic value of N-terminal pro-B-type natriuretic peptide and C reactive protein testing in patients with acute ST-segment elevation myocardial infarction

Abstract

Background

Biomarkers could improve risk stratification in patients with acute ST-segment elevation myocardial infarction (STEMI), beyond left ventricular ejection fraction (LVEF). Our study evaluated the association between N-terminal pro-B-type natriuretic peptide (NT-proBNP), C reactive protein (CRP) and mortality in a cohort of patients with acute STEMI.

Methods

This prospective, observational cohort study included patients with reperfused acute STEMI admitted to a tertiary cardiovascular disease centre between July 2020 and October 2023. All patients underwent NT-proBNP and CRP testing. The association between NT-proBNP, CRP and all-cause mortality was evaluated in relation to predischarge LVEF.

Results

The cohort included 566 patients with a mean age of 63 years. After a median follow-up of 39 months, postdischarge all-cause mortality reached 13.4%. NT-proBNP was associated with mortality irrespective of LVEF (HR 2.34 per SD increment in log NT-proBNP; p<0.001 at LVEF <50% and HR 2.36; p=0.004 at LVEF ≥50%), but the association between CRP and mortality was significant only in patients with LVEF <50% (HR 1.55, p=0.003). Across the cohort, NT-proBNP remained associated with death after adjustment for age, sex, diabetes, baseline high-sensitivity cardiac troponin T (hs-cTnT), CRP, final Thrombolysis in myocardial infarction (TIMI) flow grade and reduced LVEF (HR 1.45, p=0.03). In patients with preserved LVEF, routine NT-proBNP testing (area under the curve (AUC) 0.753 (0.642–0.863), p<0.001) improved risk stratification compared with isolated LVEF assessment (AUC 0.592 (0.453–0.730), p=0.18).

Conclusions

In a cohort of stabilised acute STEMI survivors, NT-proBNP was associated with all-cause mid-term mortality independent of hs-cTnT and LVEF. The association between CRP and mortality was significant only in patients with LVEF <50%.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Neither N-terminal pro-B-type natriuretic peptide (NT-proBNP) nor C reactive protein (CRP) is routinely recommended for risk stratification in ST-segment elevation myocardial infarction (STEMI).

WHAT THIS STUDY ADDS

• In a cohort of stabilised acute STEMI survivors, we showed that NT-proBNP testing improved risk stratification both in patients with reduced and preserved left ventricular ejection fraction (LVEF), but CRP predicted mortality only in patients with LVEF <50%.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• In the setting of STEMI, the prognostic capacity of neurohormonal and inflammatory biomarkers may help identify a high-risk patient category that is not captured by widely assessed outcome predictors.

Introduction

Despite timely reperfusion and improved secondary prevention measures, ST-segment elevation myocardial infarction (STEMI) remains associated with significant morbidity and mortality.^{1 2} For this reason, better risk stratification is important to individually tailor postrevascularisation therapies and follow-up.

Left ventricular ejection fraction (LVEF) has long been considered the foundation for non-invasive imaging risk prediction in patients with acute STEMI.^{2 3} Routine predischarge echocardiographic evaluation of LVEF is performed to guide treatment intensity.² For many decades, heart failure (HF) with reduced ejection fraction (HFrEF) has been the primary type of HF following STEMI.⁴ However, the changes in STEMI epidemiology and the wide adoption of evidence-based therapies led to better short-term survival and to lower rates of left ventricular (LV) adverse remodelling.4,6 As a result, a shift in paradigm has occurred as the proportion of STEMI survivors presenting with HF with preserved ejection fraction (HFpEF) has increased.^{5 6} Since the recommendation for cardioprotective medication in patients with STEMI relies heavily on LVEF assessment,^{1 2} an important proportion of vulnerable STEMI survivors who do not meet the echocardiographic criteria might benefit from improved risk stratification beyond the assessment of LV function.⁷

Despite its proven role in the diagnosis and management of HF, current guidelines do not recommend routine N-terminal pro-B-type natriuretic peptide (NT-proBNP) testing in patients with STEMI.^{1 2} The numerous factors influencing NT-proBNP levels (ie, age, chronic kidney disease and atrial fibrillation), the absence of a standardised measurement schedule, and the yet unclear impact of NT-proBNP-guided medical therapy on outcomes likely hindered the universal adoption of NT-proBNP testing in patients with acute myocardial infarction. Myocardial stretching, endocrine activation and myocardial hypoxia contribute to the increase in NT-proBNP values.⁸ However, in patients with STEMI, the increase in natriuretic peptides and the risk of adverse outcomes might also be related to other mechanisms like microvascular dysfunction, endothelial impairment or persistent inflammation.^{6 9 10} C reactive protein (CRP) is a recognised marker of low-grade systemic inflammation¹¹ and has been associated with cardiovascular events in patients with acute coronary syndromes.^{12 13} However, screening for inflammation with CRP in secondary prevention is not routinely performed.¹³ In this regard, complementary to non-invasive imaging, the combined assessment of neurohormonal and inflammatory biomarkers could improve risk stratification after reperfused STEMI and identify a vulnerable patient category that could benefit from more intensive monitoring and therapy.

The aim of this study was to evaluate the association between NT-proBNP, CRP and mid-term all-cause mortality in relation to LV systolic function in a contemporary cohort of patients with reperfused acute STEMI.

Methods

Study population

This was a prospective, observational, single-centre cohort study. Between July 2020 and October 2023, consecutive patients admitted for acute STEMI at ‘Niculae Stăncioiu’ Heart Institute, Cluj-Napoca, were included in the Romanian Registry for postmyocardial infarction left ventricular adverse remodelling (REDEFINE-HF). The study took place in a tertiary cardiovascular disease centre with 24/7 primary percutaneous coronary interventions (PCI) capabilities. Patients admitted with acute STEMI within 48 hours of symptom onset were screened for inclusion. The diagnosis of STEMI was established based on guideline recommendations.^{2 14 15} Patients with concomitant COVID-19 infection, missing LVEF measurement, NT-proBNP or CRP data were excluded from the analysis (online supplemental figure S1). The presence of mechanical complications or cardiogenic shock did not represent exclusion criteria, but most patients with such severe presentations were not included in the postdischarge survival analysis due to incomplete data or early in-hospital mortality. Total ischaemic time was defined as the time from the onset of symptoms to guidewire passage distal to the culprit lesion. The Killip classification was used to categorise patients based on the presence and severity of HF signs.¹⁶ Overt HF was defined as the presence of Killip class II, III or IV.

The investigation conforms with the principles outlined in the Declaration of Helsinki.

Treatment protocol

The treatment for the acute event followed the current standard of care.^{2 14} All patients received double antiplatelet therapy with aspirin and either ticagrelor or clopidogrel. Unfractionated or low-molecular-weight heparin was administered during the procedure. Drug-eluting stents were deployed in all patients who underwent PCI. The Thrombolysis in myocardial infarction (TIMI) flow and the percentage ST-segment resolution from admission to 60 min after the procedure were used to evaluate the success of reperfusion. Guideline-directed medical therapy and secondary prevention measures were applied in all patients according to current practice.^{2 14}

Laboratory analyses

Infarct size was estimated based on high-sensitivity cardiac troponin T (hs-cTnT) levels at the time of admission and 48 hours following PCI. During the first 24–72 hours of hospitalisation, patients underwent routine NT-proBNP and CRP testing. A complete blood count, serum creatinine, glycaemia and measured low-density lipoprotein (LDL) cholesterol values were available in all patients. The NT-proBNP and CRP were measured with Roche cobas chemistry analysers (Roche Diagnostics, Rotkreuz, CH). Estimated glomerular filtration rate (eGFR) was determined with the Modification of Diet in Renal Disease formula.¹⁷

Echocardiographic assessment

The echocardiographic evaluation was performed using Vivid E95 ultrasound systems (GE Healthcare, Horten, Norway). A comprehensive transthoracic echocardiogram was acquired following PCI, within 5 days of admission. The biplane Simpson method was used to measure LVEF. The LV end-systolic and end-diastolic volumes were computed by manually tracing the LV endocardial contours. The papillary muscles and trabeculations were left within the LV cavity.¹⁸ The Sonovue (Bracco, Milan, Italy) contrast agent was liberally used to improve endocardial border delineation whenever two or more contiguous LV segments were improperly visualised¹⁹ or whenever deemed necessary by the examiner. An LVEF <50% was categorised as reduced, and an LVEF ≥50% was categorised as preserved.

Outcomes

Patient follow-up was performed at 6 months and then annually until April 2025. Mortality data were collected from hospital or primary care physician records or interviews with patient relatives. Information on vital status was available for all patients through the electronic records of the national insurance system. Since the precise cause of death could not be established in all patients, all-cause mortality was defined as the clinical endpoint.

Patient and public involvement

The patients and the public were not involved in the design or conduct of the study.

Statistical methods

Data distribution was assessed using Kolmogorov-Smirnov and D’Agostino tests. Continuous data were summarised as mean±SD whenever data followed the normal distribution. Otherwise, the median and IQR were used. Continuous variables were compared with the Student’s t-test or Mann-Whitney test, as appropriate. Categorical data were presented as counts and proportions and compared with the χ² test.

Kaplan-Meier curves were constructed for survival analysis and compared with the log-rank test.

Univariable Cox proportional-hazards regression analysis was used to evaluate the association between variables of interest and mortality. The variables that were significantly associated with mortality in univariable analysis (p<0.05) were subsequently included in a multivariable model. HRs of mortality were calculated per 1 SD increment in logarithmically transformed NT-proBNP and CRP levels, together with their associated 95% CI and p values.

Receiver-operating characteristic (ROC) curves were constructed to evaluate the accuracy of NT-proBNP and CRP as predictors of mortality. The area under the ROC curves (AUC) was used as a scalar measure of performance, and the Youden index was used to derive optimal cut-off values. The AUCs were compared with the DeLong method. The Integrated Discrimination Improvement was calculated for each prediction model to reflect the incremental prognostic value of NT-proBNP and CRP testing beyond standard LVEF assessment.

Statistical analysis was performed with GraphPad Prism V.10.4.0 for Windows (GraphPad Software, San Diego, California, USA) and MedCalc (V.22.014, MedCalc Software). A two-sided p value <0.05 was considered statistically significant.

Results

Baseline patient characteristics

The baseline characteristics of the included (n=566) and excluded (n=519) patients are presented in online supplemental table S1. Excluded patients had a higher incidence of cardiogenic shock, mechanical complications and a higher in-hospital mortality as compared with the patients included in the analysis.

The mean±SD age of patients included in the analysis was 63±12 years. The median (IQR) total ischaemic time was 390 (230–720) min, and the average LVEF was 47±11%, measured at a median (IQR) of 3 (2–4) days after revascularisation. A history of active smoking at the time of admission was present in 61.5% of patients.

During the index hospitalisation, an LVEF <50% was present in 347 (61.3%) patients. Patients with reduced LVEF were more likely to be in Killip class II–IV, had higher hs-cTnT, NT-proBNP and CRP values and a higher rate of moderate or greater mitral regurgitation (MR) (p<0.001 for all). The clinical characteristics of the cohort stratified by LVEF category are presented in table 1.

Table 1. Patient characteristics according to LVEF.

Characteristic	ALL (n=566)	LVEF <50% (n=347)	LVEF ≥50% (n=219)	P value*
Clinical data
Age (years), mean±SD	63±12	63±13	63±12	0.83
Male sex, n (%)	423 (74.7)	275 (79.2)	148 (68)	0.002
Body mass index (kg/m2), mean+SD	28.9±5.1	28.8±5.2	29.1±4.8	0.48
Active smoking, n (%)	348 (61.5)	210 (60.5)	138 (63)	0.55
Hypertension, n (%)	345 (60.9)	197 (56.8)	148 (67.6)	0.01
Diabetes (%)	145 (25.6)	93 (26.8)	52 (23.7)	0.42
Killip class, n (%)				0.003
I	416 (73.5)	240 (69.2)	176 (80.4)
II	103 (18.2)	70 (20.2)	33 (15.1)
III	22 (3.9)	19 (5.5)	3 (1.4)
IV	25 (4.4)	18 (5.2)	7 (3.2)
Prehospital thrombolysis, n (%)	8 (1.4)	3 (0.9)	5 (2.3)	0.16
Total ischaemic time (min), median (IQR)	390 (230–720)	400 (220–740)	375 (255–640)	0.73
Door-to-balloon time (min), median (IQR)	55 (30–107)	52 (30–105)	58 (31–110)	0.59
Procedural data
Three-vessel disease, n (%)	96 (16.9)	68 (19.6)	28 (12.8)	0.04
PCI, n (%)	547 (96.6)	337 (97.1)	210 (95.9)	0.43
Post-PCI thrombolysis in myocardial infarction <3 flow, n (%)	74 (13.1)	45 (13)	29 (13.2)	0.93
Mechanical complications, n (%)	3 (0.5)	2 (0.6)	1 (0.5)	0.85
ST-segment resolution (%), mean±SD	49.6±28.9	47.6+27.3	52.8±31.2	0.04
Acute kidney injury, n (%)	15 (2.6)	7 (2)	8 (3.6)	0.24
Laboratory tests
High-sensitivity cardiac troponin T (ng/mL), median (IQR)
Baseline	0.3 (0.1–0.9)	0.4 (0.1–1.2)	0.1 (0.04–0.5)	<0.001
Peak	2.1 (1–3.4)	2.8 (1.4–4.4)	1.4 (0.8–2.4)	<0.001
N-terminal pro-B-type natriuretic peptide (pg/mL), median (IQR)	1179 (417–3154)	1581 (609–4562)	754 (239–1650)	<0.001
Estimated glomerular filtration rate (mL/min/1.73 m2), mean±SD
Baseline	83.4±25.1	82.3+25.4	85.3±24.5	0.17
48 hours	77.8±25.2	77.1+26	79±24	0.39
Haemoglobin (g/L), mean±SD
Baseline	146±17	147±18	145±17	0.29
48 hours	139±18	139±19	138±18	0.57
Glycaemia (mg/dL), mean±SD
Baseline	164±68	167+68	157.7±68.3	0.11
48 hours	133±48	136+49	129.5±45.8	0.18
Neutrophil/lymphocyte ratio, median (IQR)
Baseline	5.3 (3.4–8.1)	5.7 (3.7–8.5)	4.7 (3.1–7.2)	0.003
48 hours	3.2 (2.5–4.9)	3.3 (2.5–5.4)	3.1 (3.4–4.3)	0.02
C reactive protein (mg/dL), median (IQR)	9.2 (2.9–26.7)	12.9 (4.6–39.4)	5.6 (1.6–14.4)	<0.001
Low-density lipoprotein-cholesterol (mg/dL), mean+SD	110+43.4	110+43	110+45	0.85
Echocardiographic data
Moderate or greater mitral regurgitation, n (%)	128 (22.6)	98 (28.2)	30 (13.7)	<0.001
Treatment
Beta-blockers, n (%)	491 (87.8)	305 (89.2)	186 (85.7)	0.22
ACE inhibitor/angiotensin receptor-blocker, n (%)	481 (86.2)	288 (84.5)	193 (88.9)	0.13
Statin, n (%)	541 (96.8)	328 (95.9)	213 (98.2)	0.14
Outcomes
In-hospital mortality, n (%)	12 (2.1)	10 (2.9)	2 (0.9)	0.11
All-cause mortality, n (%)
Including in-hospital	88 (15.5)	67 (19.3)	21 (9.5)	0.002
From discharge	76 (13.4)	57 (16.9)	19 (8.7)	0.006

^*The p values refer to the comparison between the LVEF groups.

LVEF, left ventricular ejection fraction; PCI, percutaneous coronary intervention.

Outcome

The in-hospital mortality reached 2.1%. After a median (IQR) follow-up time of 39 (24–46) months, the postdischarge all-cause mortality was 16.9% in patients with reduced LVEF and 8.7% in patients with preserved LVEF (p=0.006). The survival curves are presented in figure 1. Patients who died were older, more commonly diabetic and had higher rates of postprocedural TIMI flow below 3 (p<0.001 for all). Higher hs-cTnT, NT-proBNP, CRP, neutrophil/lymphocyte (N/Ly) ratio and baseline glycaemia, as well as the presence of moderate or greater MR, were also associated with an adverse outcome (p<0.001 for all). The absence of beta-blocker or ACE inhibitor/angiotensin receptor-blocker therapy was associated with worse survival (p=0.03 and 0.05, respectively). The associations between clinical, laboratory, procedural characteristics and mortality in patients with preserved and reduced LVEF are presented in online supplemental table S2.

Figure 1. Kaplan-Meier survival curves by LVEF category. LVEF, left ventricular ejection fraction; STEMI, ST-segment elevation myocardial infarction.

The association between biomarkers and patient characteristics

In the overall cohort, the median NT-proBNP and CRP values were 1179 (417–3154) pg/mL, and 9.2 (2.9–26.7) mg/dL, respectively. Most patients underwent biomarker testing within 24 hours after admission. Patients with NT-proBNP above the median were older, less commonly males (p<0.001 for both), had higher rates of hypertension (p<0.001), diabetes (p=0.01), lower eGFR and higher CRP (p<0.001 for both). Older age (p<0.001) and diabetes (p=0.02) were also associated with higher CRP. Patients with NT-proBNP and CRP above the median more commonly presented in Killip class II–IV, had significantly longer total ischaemic times, higher hs-cTnT values, lower percentage of ST-segment resolution, lower LVEF and higher death rates at follow-up (p<0.001 for all) (table 2).

Table 2. Patient characteristics according to median NT-proBNP and CRP value.

Characteristic	NT-proBNP ≤median (n=284)	NT-proBNP >median (n=282)	P value*	CRP ≤median (n=283)	CRP >median (n=283)	P value†
Clinical data
Age (years), mean±SD	59±11	67±12	<0.001	61±12	65±12	0.001
Male sex, n (%)	231 (81.3)	192 (68.1)	<0.001	210 (74.2)	213 (75.3)	0.77
Body mass index (kg/m2), mean+SD	29.1±5.2	28.7±4.9	0.41	28.5±4.7	29.4±5.4	0.05
Active smoking, n (%)	205 (72.2)	143 (50.7)	<0.001	187 (66.1)	161 (56.9)	0.02
Hypertension, n (%)	157 (55.3)	188 (66.6)	0.005	168 (59.4)	177 (62.5)	0.44
Diabetes (%)	60 (21.1)	85 (30.1)	0.01	60 (21.2)	85 (30)	0.02
Killip class, n (%)			<0.001			<0.001
I	242 (85.2)	174 (61.7)		236 (83.4)	180 (63.6)
II	30 (10.6)	73 (25.9)		35 (12.4)	68 (24)
III	1 (0.4)	21 (7.4)		5 (1.8)	17 (6)
IV	11 (3.9)	14 (4.9)		7 (2.5)	18 (6.4)
Prehospital thrombolysis, n (%)	5 (1.7)	3 (1)	0.48	5 (1.8)	3 (1)	0.47
Total ischaemic time (min), median (IQR)	330 (212–590)	460 (270–808)	<0.001	331 (206–600)	468 (270 – 780)	<0.001
Door-to-balloon time (min), median (IQR)	50 (30 – 100)	60 (30 – 120)	0.32	50 (30 – 111)	60 (30 – 100)	0.67
Procedural data
Three-vessel disease, n (%)	44 (15.5)	52 (18.4)	0.35	38 (13.4)	58 (20.5)	0.03
PCI, n (%)	276 (97.2)	271 (96.1)	0.47	272 (96.1)	275 (97.2)	0.48
Post-PCI thrombolysis in myocardial infarction <3 flow, n (%)	38 (13.4)	36 (12.8)	0.83	32 (11.3)	42 (14.8)	0.21
Mechanical complications, n (%)	0 (0)	3 (1)	0.08	1 (0.4)	2 (0.7)	0.56
ST-segment resolution (%), mean±SD	54±29	46±28	0.001	53±30	46±27	0.005
Acute kidney injury, n (%)	4 (1.4)	11 (3.9)	0.07	4 (1.4)	11 (3.9)	0.07
Laboratory tests
Baseline high-sensitivity cardiac troponin T (ng/mL), median (IQR)	0.2 (0.04–0.4)	0.5 (0.1–1.4)	<0.001	0.1 (0.05–0.4)	0.5 (0.1–1.3)	<0.001
NT-proBNP (pg/mL), median (IQR)	420 (130 – 779)	3159 (1898–6181)	<0.001	779 (221–1704)	1977 (773–5132)	<0.001
Estimated glomerular filtration rate (mL/min/1.73 m2), mean±SD	90±23	77±25	<0.001	85±23	82±27	0.16
Haemoglobin (g/L), mean±SD	150±15	143±19	<0.001	148±16	145±18	0.04
Glycaemia (mg/dL), mean±SD	152±63	175±72	<0.001	153±56	174±77	<0.001
Neutrophil/lymphocyte ratio, median (IQR)	4.9 (3.1–7.2)	5.8 (3.7–8.6)	0.002	4.9 (3–7.4)	5.8 (3.8–8.6)	<0.001
CRP (mg/dL), median (IQR)	5.9 (2–13.8)	15.3 (5.1–52.1)	<0.001	2.9 (1.4–5.7)	26.6 (14.7–75.6)	<0.001
Low-density lipoprotein-cholesterol (mg/dL), mean+SD	115±44	105±42	0.007	114±42	106±44	0.02
Echocardiographic data
LVEF (%), mean±SD	50±9	44±11	<0.001	49±10	44±10	<0.001
Moderate or greater mitral regurgitation, n (%)	36 (12.7)	92 (32.6)	<0.001	55 (19.4)	73 (25.8)	0.07
Treatment
Beta-blockers, n (%)	252 (90)	239 (85.7)	0.12	248 (88.3)	243 (87.4)	0.76
ACE inhibitor/angiotensin receptor-blocker, n (%)	249 (89)	232 (83.4)	0.06	241 (86.1)	240 (86.3)	0.93
Statin, n (%)	274 (97.9)	267 (95.7)	0.15	276 (98.2)	265 (95.3)	0.05
Outcomes
In-hospital mortality, n (%)	4 (1.4)	8 (2.8)	0.24	4 (1.4)	8 (2.8)	0.24
All-cause mortality, n (%)—from discharge	22 (7.7)	54 (19.1)	<0.001	24 (8.5)	52 (18.4)	<0.001

^*The p values refer to the comparison between the NT-proBNP groups.

^†The p values refer to the comparison between the CRP groups.

CRP, C reactive protein; LVEF, left ventricular ejection fraction; PCI, percutaneous coronary intervention.

Biomarkers as predictors of mortality

In univariable Cox proportional-hazards regression analysis, NT-proBNP (HR 2.51 per 1 SD increment in log-transformed NT-proBNP; 95% CI 1.90 to 3.34 p<0.001) and CRP (HR 1.53 per 1 SD increment in log-transformed CRP, 95% CI 1.21 to 1.95, p<0.001) were associated with all-cause mortality. Analyses according to LVEF showed a significant association between NT-proBNP and mortality irrespective of LVEF <50% (HR 2.34, 95% CI 1.69 to 3.29, p<0.001 in patients with LVEF <50% and HR 2.36, 95% CI 1.33 to 4.31, p=0.004 in patients with LVEF ≥50%), while the association between CRP and mortality was significant only in patients with LVEF <50% (HR 1.55, 95% CI 1.16 to 2.09, p=0.003). Patient mortality according to LVEF category and median NT-proBNP and CRP values is presented in online supplemental table S3. Throughout the cohort, NT-proBNP remained associated with all-cause mortality after adjustment for age, sex, diabetes, post-PCI TIMI flow grade, baseline hs-cTnT, CRP and reduced LVEF (HR 1.45, 95% CI 1.03 to 2.05, p=0.03) (table 3).

Table 3. Unadjusted and multivariable adjusted HRs of mortality.

	HR	95% CI	P value
NT-proBNP*
Overall cohort	2.51	1.90 to 3.34	<0.001
Subgroups
LVEF<50%	2.34	1.69 to 3.29	<0.001
LVEF ≥50%	2.36	1.33 to 4.31	0.004
CRP*
Overall cohort	1.53	1.21 to 1.95	<0.001
Subgroups
LVEF <50%	1.55	1.16 to 2.09	0.003
LVEF ≥50%	1.07	0.66 to 1.75	0.78
Multivariable model (overall cohort)†
Age (year)	1.06	1.03 to 1.08	<0.001
Sex (men vs women)	1.36	0.81 to 2.35	0.25
Diabetes (yes/no)	1.95	1.20 to 3.14	0.007
Post-percutaneous coronary intervention thrombolysis in myocardial infarction <3 flow (yes/no)	2.53	1.43 to 4.33	0.002
Baseline high-sensitivity cardiac troponin T (ng/mL)	1.06	0.89 to 1.21	0.44
NT-proBNP*	1.45	1.03 to 2.05	0.03
CRP*	1.08	0.85 to 1.39	0.52
LVEF <50% (yes/no)	1.86	1.07 to 3.39	0.03

^*HRs are reported per SD increment on log scale of CRP and NT-proBNP.

^†HRs are mutually adjusted for the listed variables.

CRP, C reactive protein; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; TIMI, thrombolysis in myocardial infarction.

In patients discharged alive after STEMI, the optimal cut-off value for the association between NT-proBNP and mortality was 2282 pg/mL (AUC 0.720, 95% CI 0.681 to 0.757, p<0.001), while the optimal cut-off value for the association between CRP and mortality was 16.5 mg/dL (AUC 0.654, 95% CI 0.613 to 0.694, p<0.001). A worse mid-term survival was observed in patients with an increase above the cut-off in either NT-proBNP or both NT-proBNP and CRP values (figure 2). Notably, a significant association between an NT-proBNP >2282 pg/mL and mortality was observed irrespective of the presence of overt HF (signs of HF: HR 3.87, 95% CI 1.72 to 10.22, p=0.003; no signs of HF: HR 2.59, 95% CI 1.38 to 4.80, p=0.002). In fact, a subgroup analysis in patients with Killip class I on admission showed that NT-proBNP remained associated with all-cause death after adjustment for baseline hs-cTnT and reduced LVEF (HR 1.63, 95% CI 1.10 to 2.47, p=0.01).

Figure 2. Kaplan-Meier survival curves by combined cut-offs of NT-proBNP and CRP. CRP, C reactive protein; NT-proBNP, N-terminal pro-B-type natriuretic peptide.

In ROC curve analysis, the models combining NT-proBNP and CRP testing with standard LVEF assessment predicted postdischarge all-cause mortality both in patients with preserved and reduced LVEF. While on direct comparison, the AUCs were not significantly different between models; the absolute increase in AUC values showed that the addition of NT-proBNP and CRP testing to LVEF assessment improved risk stratification in patients with reduced LVEF (figure 3A). In patients with preserved LVEF, the addition of NT-proBNP testing improved risk stratification compared with isolated LVEF assessment (p=0.06), but no additive predictive value of CRP testing was observed (figure 3B). Throughout the cohort, the combined NT-proBNP and LVEF model improved mortality prediction by 6.9%, while the combined NT-proBNP, CRP and LVEF model improved mortality prediction by 7.3% compared with isolated LVEF measurement.

Figure 3. ROC curve analysis of mortality according to LVEF category. (A) An increase in absolute AUC values shows improved risk stratification with combined NT-proBNP and CRP testing in addition to LVEF assessment in patients with reduced LVEF. On direct comparison, the AUCs were not significantly different between models. (B) Routine NT-proBNP testing in addition to LVEF assessment improves risk stratification in patients with preserved LVEF (p=0.06 for ROC curves comparison), but the additive effect of CRP testing is null. AUC, area under the curve; CRP, C reactive protein; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; ROC, receiver operating characteristic.

Discussion

In a cohort of patients with mechanically reperfused STEMI and guideline-directed medical therapy, we showed that (1) NT-proBNP measured in the first 24–72 hours was associated with mid-term all-cause mortality, (2) whereas NT-proBNP was associated with all-cause mortality irrespective of LVEF and troponin T, CRP was a predictor of mortality only in patients with LVEF <50% and (3) NT-proBNP testing improved risk stratification both in patients with reduced and preserved LVEF.

The association between NT-proBNP and mortality may be explained by its capacity to identify an older, sicker population at risk of adverse outcomes.⁹ Similarly to other reports, our study showed that patients with high NT-proBNP values were older and had more extensive comorbidities, like hypertension, diabetes and chronic kidney disease.^{9 20 21} However, in our study, the association between NT-proBNP and mortality remained significant after adjustment for age, baseline comorbidities, reduced LVEF and troponin T. Also, NT-proBNP was associated with mortality irrespective of the presence of overt HF. In this regard, our study brings evidence on the value of routine NT-proBNP measurement for the risk stratification of patients with acute STEMI, beyond two of the most powerful outcome predictors: predischarge LVEF and overt HF.^{5 22 23} Most importantly, we showed that combined NT-proBNP and LVEF assessment improved mortality prediction in STEMI patients with both reduced and preserved LVEF.

The median NT-proBNP concentration observed in our study (1179 pg/mL) was lower than the one reported by the Prospective ARNI versus ACE Inhibitor Trial to Determine Superiority in Reducing Heart Failure Events after Myocardial Infarction (PARADISE-MI) investigators (1757 pg/mL)⁹ and may reflect the different clinical structure of the cohort, since the PARADISE-MI trial only included patients with acute myocardial infarction complicated by LV systolic dysfunction, pulmonary congestion or both. In the group of patients with LVEF <50%, the observed median NT-proBNP concentration of 1581 pg/mL was closer to the value reported by the PARADISE-MI investigators. Nevertheless, it must be underlined that the timing of blood sample collection was different: 24–72 hours after admission in our study versus 0.5–7 days after myocardial infarction in the PARADISE-MI trial.⁹

The high mortality observed in patients with preserved LVEF and high NT-proBNP concentrations (16.9%) is unnerving, as it was close to that of patients with high NT-proBNP in the setting of low LVEF (20.7%). Nonetheless, this result is in accordance with a recent analysis from the Norwegian Myocardial Infarction Registry, which showed that almost 23% of patients with symptoms of HF and preserved LVEF died after 1 year.²⁴ In this real-world cohort study, Jortveit and colleagues also showed that the presence of symptomatic HF led to significantly worse outcomes in patients with preserved LVEF, while it conferred limited additive risk in patients with reduced LVEF.²⁴ While their definition of HF was entirely clinical, the assessment of NT-proBNP and CRP allowed us to explore potential mechanisms responsible for the association. As such, an important observation was that patients with high NT-proBNP and CRP concentrations had significantly longer total ischaemic times, higher troponin T and less ST-segment resolution, despite achieving similar rates of post-PCI TIMI 3 flow. In patients with reperfused STEMI, the ST-segment resolution is a surrogate marker of tissular reperfusion²⁵ and is well correlated with clinical outcomes.²⁶ The presence of inflammation in the setting of ischaemia-reperfusion injury contributes to the extent of microvascular damage and determines final infarct size.^{27 28} Previous studies showed that the inflammatory burden might be more extensive in patients with larger infarct size and reduced LVEF,²⁸ a finding also supported by our observation that the association between CRP and mortality was only significant in patients with LVEF <50%.

Our cohort study with prospective data collection provides important new insights into the association between neurohormonal and inflammatory biomarkers and outcomes in a population of patients with acute STEMI. We demonstrate that combined NT-proBNP and LVEF assessment improves outcome prediction irrespective of the presence of impaired LV systolic function, and we show an additive predictive value of combined NT-proBNP and CRP testing for the risk stratification of patients with reduced LVEF. In this regard, the prognostic capacity of neurohormonal and inflammatory biomarkers may help identify a high-risk patient category that is not captured by widely assessed outcome predictors. Whether these patients may benefit from more intensive antithrombotic therapies, neurohormonal antagonism or potent anti-inflammatory therapies remains to be demonstrated by future studies.^{20 29} The findings of our study should be interpreted in the context of a high-risk patient population, with a high prevalence of active smoking, hypertension and a relatively long total ischaemic time.

The main limitation of our study is related to the smaller sample size, which led to a decrease in statistical power when stratifying the cohort according to LVEF groups. Although a separate analysis of patients with mildly reduced (40%–49%) LVEF was not performed, a previous study showed that both the LVEF cut-off of <50% and the traditional one of LVEF <40% perform similarly in predicting long-term mortality.²³ Furthermore, a 6-month follow-up comprehensive echocardiogram would have provided important information on the degree of functional recovery after successful reperfusion, since the presence of myocardial stunning might have impacted early LVEF assessment.³⁰ The use of all-cause mortality as an endpoint must also be mentioned, since an analysis based on the cause of death would have provided additional insight into the association with the tested biomarkers. While the prospective design of the study with predefined time points for biomarker and echocardiographic assessment supports the validity of our results, the absence of serial and peak NT-proBNP and CRP levels must be acknowledged, as it might have provided better prognostic information. Also, data on pre-existing LV systolic function, left atrial myopathy and HF were not available. While unavoidable for most patients presenting with acute STEMI, this limitation represents a confounding factor since, in some patients, the elevated NT-proBNP levels might have preceded the acute event. The use of baseline hs-cTnT values in the multivariable analysis must also be acknowledged as a limitation, since the 48-hour value would have been a better surrogate marker for infarct size. Nevertheless, it is important to note that the elevated baseline hs-cTnT values might also reflect a late presentation and, therefore, a higher-risk patient category. The registry did not include information on concomitant respiratory, urinary tract infections or sepsis, and their impact on CRP levels cannot be established. However, regardless of the underlying pathology, elevated CRP concentrations lead to a significant increase in cardiovascular risk and should not be overlooked.³¹ Finally, the results of this study should be interpreted in the context of a relatively stable population with acute STEMI, since most patients with severe presentations (ie, cardiogenic shock and mechanical complications) were excluded from the analysis because of incomplete data or early in-hospital death.

Conclusions

In a cohort of stabilised acute STEMI survivors, elevated NT-proBNP was associated with increased mid-term mortality independent of LVEF and troponin T level. While NT-proBNP was associated with all-cause mortality irrespective of LVEF, CRP was a predictor of mortality only in patients with LVEF <50%. Our results suggest an additive prognostic value of NT-proBNP and CRP testing for the risk stratification of patients with acute STEMI, beyond commonly assessed outcome predictors.

Data availability statement

Data are available upon reasonable request.