Prognostic nutritional index in the prediction of adverse cardiac and cerebrovascular events in ST-segment elevation myocardial infarction patients with type 2 diabetes mellitus in Urumqi, China: a retrospective cohort study

Abstract

Objective

Prognostic nutritional index (PNI) is an index for assessing nutritional and immune status. The aim of this study is to investigate the predictive value of PNI for long-term major adverse cardiac and cerebrovascular events (MACCE) in ST-segment elevation myocardial infarction (STEMI) patients with type 2 diabetes mellitus (T2DM).

Design, setting and participants

This retrospective cohort study analysed 1582 STEMI patients with T2DM who underwent percutaneous coronary intervention from January 2015 to June 2023 in Urumqi, China. Patients were followed up for MACCE.

Primary and secondary outcome measures

The primary endpoint was new-onset MACCE including all-cause death, non-fatal MI and non-fatal stroke.

Results

This study ultimately included 1582 patients for analysis with a median follow-up period of 48 months (IQR: 24–84 months) and 282 patients (17.8%) developed MACCE. Of them, 138 (8.7%), 84 (5.3%) and 60 (3.8%) patients developed all-cause death, a non-fatal MI and a non-fatal stroke, respectively. Incidences of MACCE and all-cause death conversely correlated with PNI. Kaplan-Meier curves showed a significant difference in all components of MACCE between PNI quartiles (p<0.001). The multivariate Cox regression analysis revealed that PNI was an independent predictor of MACCE (adjusted HR 0.95, 95% CI 0.93 to 0.97, p<0.001) and all-cause death (adjusted HR 0.93, 95% CI 0.90 to 0.97, p<0.001). The optimal PNI cut-off for predicting MACCE and all-cause death was 45.10 and 45.09, respectively. Moreover, the addition of PNI to the traditional prognostic model for MACCE improved the C-statistic value (p<0.001).

Conclusions

PNI, a simple and easily obtainable index, was independently associated with MACCE and all-cause death in this study. Lower PNI levels were significantly linked to an increased risk of long-term MACCE, especially in male, elderly patients and those with higher glycosylated haemoglobin and low- density lipoprotein cholesterol levels.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study used a large-sample retrospective cohort design, enhancing the statistical power and reliability of the results.

• Multiple adjusted Cox regression models, combined with Kaplan-Meier analysis, restricted cubic splines, time-dependent receiver operating characteristic curves and C-statistics, provided a relatively systematic and comprehensive assessment of prognostic nutritional index’s (PNI’s) independent predictive value.

• Data derived from historical medical records may be subject to unmeasured confounding factors (eg, dietary specifics or undocumented comorbidities).

• Post-discharge PNI levels were not recorded, and their impact on clinical outcomes could not be assessed.

Introduction

According to the latest Global Burden of Disease 2021 analysis, ischaemic heart disease (IHD) remains the leading cause of age-standardised death worldwide.¹ By 2050, the incidence, prevalence, death and disability-adjusted life years of global IHD are expected to reach 67.3 million, 510 million, 16 million and 302 million, respectively, representing increases of 116%, 106%, 80% and 62% compared with 2021.² Acute coronary syndrome (ACS) is a severe manifestation of IHD, mainly caused by thrombosis/embolism due to the rupture of vulnerable plaques in atherosclerotic lesions. ACS includes unstable angina and acute myocardial infarction (AMI).³ Based on the ECG results within 10 min of patients who may present with ACS,⁴ AMI can be further classified into non-ST-segment elevation MI (NSTEMI) and ST-segment elevation MI (STEMI), the latter accounting for about 30% of ACS cases.⁴ Due to the uneven global economic and technological development, changes in lifestyle, environment and an ageing population, the global burden of IHD has been shifting, with Asia becoming the epicentre of this disease. The China Patient-centred Evaluative Assessment of Cardiac Events indicates that the number of patients hospitalised for STEMI is increasing year by year.⁵ Despite the widespread and advanced percutaneous coronary intervention (PCI) and other revascularisation procedures, STEMI patients still have a higher risk of death.

Diabetes is a common cardiovascular risk factor, and its induced macrovascular complications can lead to AMI, a common cause of death in diabetic patients.⁶ Compared with patients without diabetes, ACS patients with diabetes, despite receiving the best treatment according to current guidelines, still have a higher risk of recurrent ischaemic cardiovascular events.^{7 8} Furthermore, diabetes was associated with higher 30-day and 1-year readmission rates after PCI, highlighting the detrimental impact of diabetes on outcomes in STEMI patients.⁹

Prognostic nutritional index (PNI) is a valuable tool in medical research, first proposed by Buzby et al in 1980.¹⁰ However, its original calculation method incorporated complex and difficult-to-access clinical parameters. In 1984, Onodera et al constructed a simpler and more convenient formula for calculating PNI based on serum albumin levels and peripheral lymphocyte counts. This revised method was derived from a linear predictive model of the risk of gastrointestinal surgical complications, mortality or both, related to the nutritional status of malnourished patients with cancer.¹¹ Albumin is an indicator of the body’s nutritional status with antioxidant and anti-inflammatory functions,¹² while lymphocytes, a subtype of white blood cells, play a crucial role in the immune system, resisting infections and regulating immune responses.¹³ Therefore, PNI can assess the nutritional and immune status of patients. Subsequently, due to its ease of acquisition, the PNI has increasingly become a focus of clinical research in recent years, and its application has expanded. Previous attention to PNI has not been widely targeted at cardiovascular diseases, particularly in the context of diabetes. This study aims to address this gap by investigating the predictive value of PNI for long-term major adverse cardiac and cerebrovascular events (MACCE) in STEMI patients with type 2 diabetes mellitus (T2DM) following PCI.

Methods

Study population

This was a retrospective cohort study enrolled 1834 STEMI patients (≥18 years) with T2DM who underwent PCI for the first time in the First Affiliated Hospital of Xinjiang Medical University (Urumqi, China) from January 2015 to June 2023. According to the Fourth Universal Definition of Myocardial Infarction (2018)¹⁴ and Classification and Diagnosis of Diabetes: Standards of Care in Diabetes (2023),¹⁵ the inclusion criteria were set as (1) STEMI was diagnosed as the presence of typical chest pain lasting more than 20 min, associated with ECG changes indicating elevation of ST-segment >1 mm or a new pathological Q wave in two or more contiguous leads, and subsequent elevation of cardiac biomarkers, such as cardiac high sensitive-troponin T (hs-TnT) >0.1 µg/mL; (2) T2DM was diagnosed with fasting plasma glucose of 126 mg/dL or higher, 2-hour plasma glucose of 200 mg/dL or higher during a 75 g oral glucose tolerance test, or glycosylated haemoglobin (HbA1c) of 6.5% or higher. The exclusion criteria were as follows: (1) patients were <18 years old at the time of first admission; (2) with severe hepatic or renal insufficiency or suffering from other diseases that may lead to an abnormal decrease in albumin; (3) suffering from blood disorders, serious infections, malignant tumours or other diseases that may cause an abnormal elevation of inflammatory indices; (4) with incomplete clinical data.

Previous studies have shown that the incidence of MACCE in STEMI patients with T2DM who underwent PCI was 25.8%.¹⁶ The HR for MACCE in patients with geriatric nutritional risk index <92 (malnutrition), a nutrition-related indicator, was 2.30,¹⁷ while the HR for MACCE in patients with low neutrophil to lymphocyte ratio, an inflammation-related indicator, was 0.77.¹⁸ Based on these data, we used PASS V.11.0 software to calculate that the relevant studies would require 540 (nutrition-related) and 1081 (inflammation-related) subjects, with a significance level α=0.05 and power β=0.9. According to the inclusion and exclusion criteria, this study ultimately included 1582 participants, which meets the required sample size. All participants were informed in writing to notify the project purpose, data collection and telephone interview for the follow-up procedure and a signed consent form was obtained. The follow-up was carried out by trained medical staff through telephone interview or outpatient clinical visit. The primary endpoint was new-onset of MACCE including all-cause death (death due to cardiovascular or non-cardiovascular reasons), non-fatal MI and non-fatal stroke.

Finally, 1582 patients were divided into four groups according to the quartiles of PNI for analysis. The flow chart shows the study design and patient recruitment (figure 1).

Figure 1. Flow diagram of the study design and patient recruitment. MACCE, major adverse cardiac and cerebrovascular events; PCI, percutaneous coronary intervention; PNI, prognostic nutritional index; STEMI, ST-segment elevation myocardial infarction; T2DM, type 2 diabetes mellitus.

Clinical data collection, definitions and laboratory tests

Clinical data were collected from the medical records by trained clinicians who were blinded to the purpose of the study. The information gathered comprised demographic and clinical details of the patients, including their gender, age, height, weight, body mass index (BMI), duration of DM, history of hypertension, smoking status, alcohol intake, history of stroke, systolic and diastolic blood pressure (SBP and DBP), heart rate, left ventricular ejection fraction (LVEF) by echocardiography and medication at discharge.

Smoking status was classified as: non-smoker (no smoking history before admission), current smoker (smoked within 1 year before admission), ex-smoker (quit smoking over 1 year before admission) or current smoker. Alcohol intake was categorised into: non-drinker (no alcohol consumption before admission), current drinker (consumed alcohol within 1 year before admission) and ex-drinker (stopped drinking for more than 1 year before admission). Blood pressure was measured on the right upper arm using a standard sphygmomanometer in a quiet state, and the values of the blood pressure were averaged for two measurements taken at two different times.

In this study, all patients entered the hospital through the chest pain centre, ECG and blood sampling were performed at admission. Laboratory tests including complete blood cell counting, hs-TnT, creatine kinase MB, glucose, HbA1c, total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), serum creatinine, estimated glomerular filtration rate (eGFR) and C reactive protein were carried out within 1 hour of admission to ensure accuracy, timeliness and consistency of the data.

The Gensini score was calculated according to the degree of coronary artery stenosis and the number of lesions at the time of coronary angiography to assess the severity of coronary artery lesions.¹⁹ The Global Registry of Acute Coronary Events (GRACE) risk score was calculated for each patient to assess short-term risk based on eight variables on admission, including age, heart rate, SBP, serum creatinine, Killip class and the presence of elevated cardiac enzyme markers, whether the ST-segment of the ECG was altered and the presence of cardiac arrest on admission.²⁰ The PNI was calculated as serum albumin (g/L)+5×peripheral blood lymphocyte count (×10⁹/L).

Coronary angiography and PCI

All patients were given 300 mg chewable aspirin and 300–600 mg loading doses of clopidogrel on admission. Prior to the PCI procedure, patients also received standard intravenous heparin at 70 U/kg. All PCI procedures were performed by experienced interventional cardiologists using the transradial artery route and drug-eluting stents within 2 hours after admission and diagnosis.

Statistical analysis

Statistical analysis was performed using SPSS V.24.0 and R V.4.4.1. Subjects were classified according to the occurrence of MACCE during the follow-up period and the quartiles of PNI. Continuous variables in the case of normal distribution were presented as mean±SD, with the t-test employed for pairwise group comparisons, one-way analysis of variance was used for comparisons across multiple groups. Continuous variables in the case of non-normal distribution were expressed as the median (IQR), with the Mann-Whitney U test used for comparing groups. Categorical variables were described as frequencies and percentages, and group comparisons were made using the χ² test. The association between the PNI and traditional cardiovascular risk factors was assessed using stepwise multiple linear regression. The log-rank test was employed to compare PNI-associated Kaplan-Meier survival curves. Cox regression analysis was used to investigate the relationship between PNI and endpoint events. The PNI was analysed using two approaches: (1) as a categorical variable; (2) as a continuous variable. In the multivariate Cox regression analysis, the covariates included both demographic variables and other potential factors that could influence clinical outcomes. Three sequentially adjusted models evaluated the prognostic value of PNI: model 1 (demographic factors), model 2 (adding laboratory parameters) and model 3 (further incorporating clinical medications), progressively assessing PNI’s independent association with clinical outcomes. The results of the Cox regression analysis are presented as HR with 95% CIs.

The linear and non-linear relationship between the PNI and MACCE was analysed using restricted cubic splines with four knots. Time-dependent receiver operating characteristic (Time-ROC) curves, along with the area under the curve (AUC), were used to estimate the predictive value of the PNI for the occurrence of new-onset MACCE. Furthermore, plotting the AUC values at different time points yields the time-dependent AUC (Time-AUC), which illustrates the variation in the model’s predictive performance over time. The C-statistic quantifies the predictive accuracy of a model, with values ranging from 0 to 1. A value of 0.5 indicates that the model’s predictive performance is no better than random chance, while values closer to 1 signify superior predictive performance. The incremental predictive value of PNI for MACCE was assessed by comparing the Time-AUC and C-statistics of the fully adjusted model (model 3) with and without PNI. In the end, subgroup analysis was conducted to investigate the consistency of the prognostic impact of the PNI on the primary endpoint across different subgroups defined by age (≤65 years and >65 years), gender, ethnicity, BMI (≤28 and >28 kg/m²), history of hypertension, HbA1c (≤7% and >7%), low-density lipoprotein cholesterol (≤70 and >70 mg/dL) and pre-admission medication use (including insulin and sodium-glucose transport protein 2 (SGLT-2) inhibitors). The models used in the subgroup analysis included all covariates from model 3. A two-sided analysis with a p value <0.05 was considered significant.

Patient and public involvement

Patients and the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Results

Occurrence of MACCE during the follow-up period

A total of 1582 STEMI patients with T2DM were included in this study, with a follow-up period ranging from 5 to 108 months (median: 48 months; IQR: 24–84 months). Of them, a total of 282 patients (17.8%) experienced MACCE including 138 cases (8.7%) of all-cause death, 84 cases (5.3%) of non-fatal MI and 60 cases (3.8%) of non-fatal stroke.

Baseline characteristics of study population

The baseline characteristics of the study population with and without MACCE during the follow-up period are presented in table 1. The information on medication of the study population according to the occurrence of MACCE is presented in online supplemental table S1. Compared with the patients without MACCE, the patients developed MACCE had a higher proportion of female gender and history of cerebral infarction with significantly higher age, lower DBP, LVEF, PNI, eGFR and TG and significantly higher heart rate, glucose levels, GRACE score and Gensini score (all p<0.05).

Table 1. Baseline characteristics of the study population according to the occurrence of MACCE.

Variable	Overall (n=1582)	Non-MACCE (n=1300)	MACCE (n=282)	P value
Gender				0.002
Male, n (%)	1206 (76.2%)	1011 (77.8%)	195 (69.1%)
Female, n (%)	376 (23.8%)	289 (22.2%)	87 (30.9%)
Age, years	64.3±6.0	63.9±6.1	66.4±5.1	<0.001
Ethnicity				0.64
Han, n (%)	1005 (63.5%)	824 (63.4%)	181 (64.2%)
Uygur, n (%)	395 (25.0%)	328 (25.2%)	67 (23.8%)
Kazakh, n (%)	34 (2.2%)	26 (2.00%)	8 (2.8%)
Hui, n (%)	111 (7.0%)	94 (7.2%)	17 (6.0%)
Others, n (%)	37 (2.3%)	28 (2.2%)	9 (3.2%)
Length of hospital stay, day	8 (5.0, 11.0)	8 (6.0, 11.0)	7 (4.0, 12.0)	0.19
Hypertension, n (%)	975 (61.6%)	792 (60.9%)	183 (64.9%)	0.214
Smoking status				0.007
Non-smoker, n (%)	793 (50.1%)	637 (49.00%)	156 (55.3%)
Current smoker, n (%)	553 (35.0%)	477 (36.7%)	76 (27.0%)
Ex-smoker, n (%)	236 (14.9%)	186 (14.3%)	50 (17.7%)
Alcohol intake				0.258
Non-drinker, n (%)	1026 (68.9%)	837 (68.5%)	189 (70.8%)
Current drinker, n (%)	335 (22.5%)	284 (23.2%)	51 (19.1%)
Ex-drinker, n (%)	128 (8.6%)	101 (8.3%)	27 (10.1%)
Cerebral haemorrhage, n (%)	19 (1.2%)	14 (1.1%)	5 (1.8%)	0.361
Cerebral infarction, n (%)	123 (7.8%)	84 (6.5%)	39 (13.8%)	<0.001
Heart rate, beats per minute	84±17	83±16	88±22	<0.001
DBP, mm Hg	74±13	75±12	72±15	0.03
SBP, mm Hg	123±21	123±20	121±23	0.196
BMI, kg/m2	26.8±22.9	26.9±24.6	26.3±12.2	0.672
LVEF, %	54.9±8.2	55.4±7.9	52.6±9.3	<0.001
PNI	45.0±7.5	45.9±7.3	41.1±7.4	<0.001
eGFR, mL/min/1.73m2	95.13±39.05	97.83±37.18	82.68±44.73	<0.001
TC, mmol/L	3.68±1.03	3.69±1.02	3.63±1.08	0.321
TG, mmol/L	1.82±1.22	1.87±1.27	1.62±0.97	<0.001
HDL-C, mmol/L	0.87±0.24	0.87±0.23	0.87±0.26	0.954
LDL-C, mmol/L	2.35±0.84	2.36±0.83	2.32±0.89	0.515
Glucose, mmol/L	11.16±4.72	10.88±4.48	12.44±5.55	<0.001
HbA1c, %	8.47±1.67	8.46±1.66	8.47±1.70	0.956
C reactive protein, mg/L	22.57 (11.90, 42.13)	21.95 (11.90, 41.88)	26.12 (12.03, 44.91)	0.184
CK-MB, U/L	33.16±47.89	33.20±48.52	33.01±44.91	0.951
hs-TnT, ng/mL	15.80±19.22	15.42±18.82	17.56±20.90	0.115
Gensini score	56.3±15.4	54.9±15.1	62.6±15.1	<0.001
GRACE score	131±27	128±27	143±27	<0.001

Continuous variables were presented as mean±SD or the median (IQR) in the case of normal or non-normal distribution and categorical variables were described as frequencies and percentages. P values were based on χ2 tests or t-test.

BMI, body mass index; CK-MB, creatine kinase-muscle/brain type isoenzyme; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; GRACE, Global Registry of Acute Coronary Events; HbA1c, glycosylated haemoglobin; HDL-C, high-density lipoprotein cholesterol; hs-TnT, cardiac high sensitive-troponin T; LDL-C, low-density lipoprotein cholesterol; LVEF, left ventricular ejection fraction; MACCE, major adverse cardiac and cerebrovascular events; PNI, prognostic nutritional index; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides.

The baseline characteristics of the study population according to the quartiles of the PNI are presented in online supplemental table S2. Patients with lower PNI had a higher proportion of females, longer hospital stays, more history of cerebral infarction and higher heart rate, glucose, C reactive protein, hs-TnT, GRACE score and Gensini score compared with those with higher PNI. Patients with lower PNI also exhibited lower eGFR, DBP, SBP, LVEF, TC and TG than those with higher PNI (p<0.05).

Correlations between PNI and traditional cardiovascular risk factors

Correlations between PNI and traditional cardiovascular risk factors were assessed using stepwise multiple linear regression (online supplemental table S3). Initially, univariate linear regression analysis was conducted to examine the relationship between PNI and traditional cardiovascular risk factors, followed by multivariate linear regression analysis for variables that showed statistical significance in the univariate analysis. The results indicated that PNI levels were positively correlated with DBP, LVEF, eGFR, TG and HDL-C, and negatively correlated with age, heart rate, SBP, glucose, GRACE score and Gensini score (p<0.05).

Relationship between PNI and MACCE

To further elucidate the relationship between PNI and MACCE, the distribution of various MACCE according to the quartiles of PNI was examined. Compared with patients with higher PNI, those with lower PNI had significantly higher incidence of MACCE and its components, that is, all-cause death, non-fatal MI and non-fatal stroke (all p<0.01, table 2).

Table 2. The relationship between PNI and MACCE.

	Quartile 1(21.1–40.0)n=396	Quartile 2(40.0–45.1)n=396	Quartile 3(45.1–49.9)n=395	Quartile 4(49.9–72.5)n=395	P value
MACCE, n (%)	127 (32.1%)	73 (18.4%)	56 (14.2%)	26 (6.6%)	<0.001
All-cause death, n (%)	65 (16.4%)	33 (8.3%)	26 (6.6%)	14 (3.5%)	<0.001
Non-fatal MI, n (%)	38 (9.6%)	22 (5.6%)	16 (4.1%)	8 (2.0%)	<0.001
Non-fatal stroke, n (%)	24 (6.1%)	18 (4.6%)	14 (3.5%)	4 (1.0%)	0.002

P values were based on χ2 tests. Results were analysed by χ2.

MACCE, major adverse cardiac and cerebrovascular events; MI, myocardial infarction; PNI, prognostic nutritional index.

The Kaplan-Meier curves for the incidence of MACCE and its individual components are depicted in figure 2. The Kaplan-Meier analysis revealed statistically significant differences in the incidence of MACCE, all-cause death, non-fatal MI and non-fatal stroke among different PNI levels (log-rank p<0.01). The incidence of MACCE, all-cause death and non-fatal MI in quartile 1 was significantly higher compared with quartiles 2, 3 and 4. Quartile 2 exhibited higher rates of MACCE, all-cause death, non-fatal MI and non-fatal stroke than quartile 4. Additionally, quartile 3 had higher rates of MACCE and non-fatal stroke than quartile 4 (figure 2A–D).

Figure 2. Kaplan-Meier curve showed the incidence of MACCE, all-cause death, non-fatal MI and non-fatal stroke at different PNI levels. (A) Kaplan-Meier curve of MACCE; (B) Kaplan-Meier curve of all-cause death; (C) Kaplan-Meier curve of non-fatal MI; (D) Kaplan-Meier curve for non-fatal stroke. MI, myocardial infarction; MACCE, major adverse cardiac and cerebrovascular events.

In the multivariate Cox regression analysis, three models were constructed to assess the predictive potential of the PNI for MACCE. The results indicated that, after adjusting for multiple confounders, a lower PNI was associated with an increased risk of MACCE and all-cause death in patients with STEMI complicated by T2DM, regardless of whether PNI was included as a continuous or categorical variable, making it an independent risk predictor for both outcomes. Additionally, PNI was not identified as an independent risk predictor for non-fatal MI and non-fatal stroke (table 3).

Table 3. Cox proportional HRs for MACCE, all-cause death, non-fatal MI and non-fatal stroke.

	Model 1			Model 2			Model 3
	HR (95% CI)	P value	P for trend	HR (95% CI)	P value	P for trend	HR (95% CI)	P value	P for trend
MACCE
Continuous	0.94 (0.92 to 0.96)	<0.001		0.95 (0.93 to 0.98)	<0.001		0.95 (0.93 to 0.97)	<0.001
Quartile 1	Ref.		<0.001	Ref.		0.001	Ref.		0.001
Quartile 2	0.60 (0.44 to 0.82)	0.001		0.62 (0.41 to 0.95)	0.027		0.63 (0.41 to 0.96)	0.031
Quartile 3	0.48 (0.34 to 0.67)	<0.001		0.53 (0.33 to 0.83)	0.006		0.53 (0.33 to 0.84)	0.007
Quartile 4	0.24 (0.15 to 0.39)	<0.001		0.29 (0.14 to 0.57)	<0.001		0.29 (0.15 to 0.60)	0.001
All-cause death
Continuous	0.93 (0.91 to 0.96)	<0.001		0.93 (0.90 to 0.97)	<0.001		0.93 (0.90 to 0.97)	<0.001
Quartile 1	Ref.		<0.001	Ref.		0.001	Ref.		0.001
Quartile 2	0.50 (0.32 to 0.79)	0.003		0.38 (0.21 to 0.71)	0.002		0.40 (0.22 to 0.73)	0.003
Quartile 3	0.41 (0.25 to 0.67)	<0.001		0.33 (0.17 to 0.63)	0.001		0.34 (0.17 to 0.66)	0.001
Quartile 4	0.23 (0.12 to 0.45)	<0.001		0.16 (0.06 to 0.43)	<0.001		0.17 (0.06 to 0.46)	<0.001
Non-fatal MI
Continuous	0.96 (0.93 to 0.99)	0.021		0.98 (0.94 to 1.03)	0.390		0.98 (0.93 to 1.03)	0.371
Quartile 1	Ref.		0.012	Ref.		0.238	Ref.		0.211
Quartile 2	0.72 (0.40 to 1.28)	0.264		0.80 (0.37 to 1.73)	0.574		0.82 (0.37 to 1.80)	0.621
Quartile 3	0.58 (0.31 to 1.07)	0.079		0.67 (0.29 to 1.54)	0.342		0.65 (0.28 to 1.50)	0.31
Quartile 4	0.37 (0.16 to 0.87)	0.023		0.50 (0.14 to 1.76)	0.281		0.50 (0.14 to 1.81)	0.293
Non-fatal stroke
Continuous	0.92 (0.89 to 0.96)	<0.001		0.96 (0.90 to 1.01)	0.117		0.95 (0.90 to 1.01)	0.091
Quartile 1	Ref.		0.001	Ref.		0.467	Ref.		0.475
Quartile 2	0.67 (0.35 to 1.29)	0.232		1.32 (0.53 to 3.27)	0.552		1.35 (0.55 to 3.41)	0.521
Quartile 3	0.50 (0.24 to 1.04)	0.064		1.16 (0.41 to 3.27)	0.782		1.17 (0.41 to 3.32)	0.766
Quartile 4	0.13 (0.04 to 0.43)	0.001		0.49 (0.09 to 2.55)	0.395		0.49 (0.09 to 2.63)	0.408

P values were based on Cox regression analysis.

Model 1, adjusted for age, gender, ethnicity, DBP, SBP, heart rate, smoking status and history of cerebral infarction;.

Model 2, adjusted for Model 1, TC, TG, HDL-C, eGFR, glucose, C reactive protein, CK-MB, hs-TnT and Gensini score.

Model 3, adjusted for model 2, post-discharge use of dual antiplatelet therapy, interruption of dual antiplatelet or statin therapy within 12 months post-discharge and post-discharge use of statins, β-blockers, metformin, SGLT-2 inhibitors, ACEI/ARB/ARNI, insulin and GLP-1 receptor agonists.

ACEI, ACE inhibitors; ARB, angiotensin receptor blockers; ARNI, angiotensin receptor-neprilysin inhibitors; CCB, calcium channel blockers; CK-MB, creatine kinase MB; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; GLP-1 receptor agonists, glucagon-like peptide-1 agonists; HDL-C, high-density lipoprotein cholesterol; hs-TnT, cardiac high sensitive-troponin T; MACCE, major adverse cardiac and cerebrovascular events; MI, myocardial infarction; SBP, systolic blood pressure; SGLT-2 inhibitors, sodium-glucose transport protein 2; TC, total cholesterol; TG, triglycerides.

When PNI was a continuous variable, in model 1 (HR, 0.94 (0.92 to 0.96), p<0.001), model 2 (HR, 0.95 (0.93 to 0.98), p<0.001) and the fully adjusted model 3 (HR, 0.95 (0.93 to 0.97), p<0.001), a lower PNI was associated with an increased risk of MACCE in patients with STEMI and T2DM. Furthermore, in model 1 (HR, 0.93 (0.91 to 0.96), p<0.001), model 2 (HR, 0.93 (0.90 to 0.97), p<0.001) and the fully adjusted model 3 (HR, 0.93 (0.90 to 0.97), p<0.001), a lower PNI was also a significant risk factor for all-cause death. When PNI was a categorical variable, in these Cox proportional hazards models established, compared with patients in the lowest quartile (quartile 1), those with higher PNI exhibited significantly lower rates of MACCE and all-cause death, with both risks decreasing progressively as PNI increased.

Furthermore, figure 3 displays the restricted cubic spline curves for the relationship between PNI and the occurrence of MACCE events and all-cause death after adjusting the variables of model 3. The results indicate a positive linear correlation between PNI and the incidence of both MACCE events and all-cause death (p for both <0.001, non-linear p for MACCE=0.282, non-linear p for all-cause death=0.069), with a critical value of 45.10 for MACCE and 45.09 for all-cause death.

Figure 3. Restricted cubic spline curves of PNI and MACCE events and all-cause deaths after adjusting the variables of model 3. (A) Restricted cubic spline curve of MACCE; (B) restricted cubic spline curves for all-cause death. MACCE, major adverse cardiac and cerebrovascular events; PNI, prognostic nutritional index.

Evaluation of PNI for predicting MACCE

To further assess the prognostic value and predictive performance of the PNI, Time-ROC curves and Time-AUC curves for PNI in predicting the occurrence of MACCE in STEMI patients with T2DM were constructed, as shown in figure 4A,B. AUC for PNI predicting MACCE at 2 years was 0.665 (0.615 to 0.715), at 4 years was 0.665 (0.625 to 0.706), at 6 years was 0.671 (0.630 to 0.712) and at 8 years was 0.664 (0.607 to 0.720). The Time-AUC curves for predicting MACCE with and without PNI in the model are depicted in figure 4C. Compared with model 3 without PNI, the risk prediction of model 3 with PNI showed improvement (C-statistic increased from 0.583 (0.516 to 0.650) to 0.611 (0.544 to 0.678), p<0.001). In addition, to further elucidate the potential value of PNI in clinical practice, a C-statistic comparison of PNI with the established risk score, GRACE score, was made. The results showed a C-statistic of 0.662 (0.628 to 0.697) for the GRACE score, which was not significantly different from the unadjusted PNI (0.631 (0.615 to 0.648)) in its predictive efficacy for MACCE events in STEMI patients with T2DM (p=0.285).

Figure 4. (A) Time-ROC curve of PNI prediction of MACCE occurrence. (B) Time-AUC curve of PNI predicting MACCE occurrence. (C) Time-AUC curve with and without PNI predicted the time correlation of MACCE occurrence. The dotted line represents the respective 95% CI. AUC, area under the curve; MACCE, major adverse cardiac and cerebrovascular events; PNI, prognostic nutritional index; ROC, receiver operating characteristic.

Subgroup analysis

Stratified subgroup analyses were conducted based on age (≤65 and >65 years), gender, ethnicity, BMI (≤28 and >28 kg/m²), history of hypertension, smoking status, HbA1c (≤7% and >7%), LDL-C (≤70 and >70 mg/dL) and pre-admission medication use (including insulin and SGLT-2 inhibitors). After adjusting for multiple factors, there was no interaction between PNI and the variables of ethnicity, history of hypertension, BMI, smoking status, pre-admission use of insulin and SGLT-2 inhibitors (All p values for interaction ≥0.05). However, there was a significant interaction between PNI and the variables of gender, age, HbA1c and LDL-C (p values for interaction <0.05). Specifically, there is a significant association between PNI and the occurrence of MACCE among patients those are male and over the age of 65 with HbA1c >7.0% or LDL-C >1.81 mmol/L (figure 5).

Figure 5. Cox regression analysis evaluated the prognostic value of PNI in different stratifications. PNI was incorporated into model 3 as a continuous variable. BMI, body mass index; HbA1c, glycosylated haemoglobin; LDL-C, low-density lipoprotein cholesterol; PNI, prognostic nutritional index; SGLT-2, sodium-glucose transport protein 2.

Discussion

This study thoroughly investigated the prognostic value of PNI in predicting MACCE in STEMI patients with T2DM who underwent PCI. Through the retrospective cohort study of 1582 patients, we found that (1) low PNI is associated with an increased risk of long-term MACCE in STEMI patients with T2DM, independent of traditional cardiovascular risk factors; (2) the significant correlation between PNI and long-term MACCE is mainly observed in male, patients over the age of 65, with HbA1c >7.0% and LDL-C >1.81 mmol/L and (3) incorporating PNI into existing risk prediction models can improve the predictive ability for outcomes in STEMI patients with T2DM after PCI, optimising the risk stratification of recurrent cardiac and cerebrovascular risk. Overall, our findings suggest that lower PNI is independently associated with an increased risk of long-term MACCE in this study population.

PNI is a composite marker of nutritional and immune status, widely recognised for its prognostic significance in a variety of diseases. It has been used to predict the prognosis of patients with different types of cancers and mortality in patients with various heart diseases. A meta-analysis confirmed the association between PNI and the prognosis of patients with coronary artery disease, indicating that a lower PNI is associated with a higher risk of major adverse cardiovascular events.²¹ It is well-known that the pathological state of hyperglycaemia in diabetic patients not only affects their daily quality of life but also leads to a range of serious complications. Studies have confirmed that PNI can serve as a novel marker for diabetic retinopathy in patients with T2DM,²² as well as an independent predictor of all-cause death risk in patients with diabetic nephropathy.²³ Diabetes and its complications constitute a complex pathological network, ranging from microvascular damage to microvascular complications to macrovascular complications and even cardiovascular diseases (CVDs), which are interconnected and collectively impact the disease progression and prognosis. It is interesting to note that the presence of diabetes amplifies the negative impact of low PNI on all-cause death in patients with coronary heart disease, with poor nutritional immune status exceeding diabetes itself.⁶ Our study adds new evidence that PNI may serve as an independent risk predictor for long-term MACCE in STEMI patients with T2DM who underwent PCI. In the correlation analysis, we found that PNI was significantly associated with traditional cardiovascular risk factors (positively correlated with DBP, LVEF, eGFR, TG, HDL-C, and negatively correlated with AGE, HR, SBP, glucose, GRACE score, Gensini score). This finding suggested that PNI reflects multidimensional information on nutritional, metabolic and inflammatory status and organ function, providing a new supplementary indicator for the traditional risk assessment system. The multivariate Cox regression analysis highlighted the independent predictive power of PNI for clinical outcomes, and even after adjusting for traditional cardiovascular risk factors and medications. It confirmed that PNI could provide additional predictive information beyond traditional risk models and has promising clinical applications. Time-AUC curve analysis confirmed that the predictive value of admission PNI remained stable with increasing follow-up time, demonstrating the reliability of admission PNI in long-term prediction, which is a key finding of our study. The GRACE score is currently a globally recognised predictive index with a complex and comprehensive composition that incorporates multiple clinical indicators (including haemodynamics, electrocardiographic features and biochemical parameters). In contrast, the admission PNI combined two key factors, reflecting nutritional and inflammatory status. Its predictive capacity for MACCE in STEMI patients with T2DM after PCI was comparable to that of GRACE score. This finding suggests that PNI may offer a complementary perspective for assessing patient risk. Furthermore, our study explored the predictive value of PNI in different patient subgroups and found that its significant correlation with long-term MACCE was mainly seen in patients who were male, aged >65 years, or with HbA1c >7.0, and LDL-C>1.81 mmol/L. It may help clinicians identify patient groups that benefit more from PNI monitoring, thereby improving the targeting and efficiency of risk assessment. For example, in older patients or those with poorly controlled diabetes, PNI may become a more sensitive prognostic index to guide clinical management.

The association between PNI and the prognosis of CVDs is not yet fully understood. However, the underlying mechanism may lie in its reflection of systemic inflammation and nutritional status, both of which play a key role in the pathogenesis of CVDs and subsequent adverse events.^{24 25} As an important component of the immune system, lymphocytes have multifaceted functions. It can inhibit neutrophil-mediated excessive inflammatory responses and maintain immune homeostasis by secreting anti-inflammatory factors (eg, interleukin (IL) 10, transforming growth factor beta).²⁶ In general, despite revascularisation after PCI, a persistent inflammatory response may lead to a ‘residual inflammatory risk’ that is an important driver of poor long-term prognosis. Sustained inflammatory status causes further myocardial damage and impaired repair process through neutrophil- or macrophage-derived proinflammatory factors, arachidonic acid metabolites and platelet aggregation.^{27 28} Low lymphocytes after PCI may impair its anti-inflammatory capacity that exacerbates vascular endothelial injury and plaque instability, even poor prognosis.²⁹ Several studies have shown that a low lymphocyte count is associated with an increased risk of death in patients with CVDs,³⁰ and is also significantly associated with a poorer prognosis in patients with heart failure, chronic IHD and ACS.³¹ Interestingly, ACS patients with low lymphocyte counts after PCI had higher plaque vulnerability and a significantly increased risk of reinfarction.³² This may be due to elevated serum cortisol and catecholamine levels during systemic stress response and activation of the local immune system, accompanied by increased lymphocyte apoptosis, downregulation of lymphocyte proliferation and differentiation as well as redistribution of lymphocytes leading to diminished anti-inflammatory and atherosclerosis protective effects.^{33 34} Mechanistically, B-lymphocytes specifically recognise oxidative epitopes of oxidised low-density lipoprotein by secreting natural IgM antibodies and block macrophage CD36/scavenger receptors-B1-mediated foam cell formation,³⁵ while inhibiting inflammatory mediator production by macrophages and antigen presentation via IL-10.³⁶ It is reported that B-lymphocytes derived IL-10 and IgM also improves insulin resistance, synergistically regulates the metabolic-inflammatory axis, and attenuates T2DM.³⁷ Furthermore, diabetes is closely associated with a chronic low-grade inflammatory status. A hyperglycaemic environment exacerbates the inflammatory response, which in turn further exacerbates insulin resistance, creating a vicious cycle and exacerbating the risk and progression of CVDs.³⁸ Based on this, PNI can reflect the long-term status of inflammatory-immune balance and help identify high-risk populations associated with chronic inflammatory status.

It is well known that albumin, a plasma protein synthesised by the liver, is the core functional protein in human plasma. It not only maintains colloid osmotic pressure, but also participates in systemic homeostatic regulation through multiple mechanisms. On the one hand, the sulfhydryl group of albumins can directly scavenge oxygen free radicals and reduce oxidative stress damage.³⁹ On the other hand, albumin can reduce microvascular leakage and improve tissue oxygen supply by maintaining the function of the vascular endothelial barrier.40,42 It also inhibits the inflammatory cascade by binding endogenous toxins (eg, lipopolysaccharides) and exogenous inflammatory substances (eg, cytokines).⁴³ Albumin also plays an important role in the regulation of immune repair function. First, albumin can promote the secretion of IL-2 by antigen-presenting cells to activate the immune response of lymphocytes⁴⁴; second, it can unblock the Fcγ receptor, inhibit the apoptosis of immune cells and improve the immune function^{44 45}; and finally, it can neutralise microorganisms and toxins by its broad-spectrum antigen-binding ability and inhibit excessive inflammatory reactions triggered by infections.^{45 46} Hypoalbuminaemia has been found to have a prevalence of 13% in stable coronary artery disease and 20%–30% in ACS.47,49 Reduced albumin levels are typically associated with a worse inflammatory state and can serve as an independent predictor of MI.⁵⁰ It is worth mentioning that this unfavourable condition is even more concerning in patients with a background of diabetes. This is because serum albumin levels are reduced in patients with T2DM, mainly involving impaired albumin synthesis, clearance mediated by immune responses due to glycation, and albuminuria.⁵¹ Elevated glycosylated albumin due to chronic hyperglycaemia may imply worsened oxidative stress and inflammatory response, endothelial damage and microcirculatory impairment, as well as disturbances in cholesterol and HDL metabolism.⁵² On the other hand, in a large cohort study from the UK Biobank, results found that serum albumin levels were inversely correlated with the incidence of diabetes and diabetic microvascular complications,⁵³ suggesting that low albumin levels are a risk factor for the occurrence and progression of T2DM.⁵¹ These observations demonstrate an association between diabetes and serum albumin levels. Changes in serum albumin levels in the context of diabetes and CVDs are linked to alterations in homeostasis and metabolism. Consequently, monitoring albumin levels has potential clinical value for assessing disease status and prognosis.

Overall, lymphocyte counts, albumin levels and diabetic status are closely related to the pathogenesis and clinical prognosis of AMI. However, the universal applicability of PNI across different populations and disease states is questioned, as the optimal cut-off value of PNI may vary depending on the specific clinical context, which could limit its standardisation in clinical practice. The findings from our single-centre and retrospective study may limit its generalisability. Meanwhile, the hyperinflammatory status of patients with STEMI in the acute phase and the hyperglycaemic status and chronic inflammatory context of patients with T2DM may largely influence albumin and lymphocyte levels, which may in turn affect the PNI.54,56 Therefore, dynamic monitoring of the PNI during the acute phase and follow-up period may more precisely reflect patient’s nutritional and immune status. Future multicentre, larger-scale, prospective studies may provide better evidence for the usefulness of PNI in predicting clinical outcomes.

Conclusion

In this retrospective study, a lower PNI was independently associated with an increased risk of MACCE, especially in male, older patients and those with higher HbA1c and LDL-C levels, as well as with an increased risk of all-cause death. Our findings emphasise the potential value of PNI in clinical practice for evaluating the nutritional and inflammatory status in this study population. This can help in the early identification of high-risk patients who may require closer monitoring or more aggressive treatment strategies.

Data availability statement

The datasets used and/or analysed for the current study are available upon request.