From January 2006 to December 2012, 720 patients aged ≥80 admitted to the cardiology department of the Chinese People’s Liberation Army (PLA) General Hospital for coronary angiography due to ACS symptoms, were enrolled (Figure 1). A total of 699 patients signed their informed consent. The exclusion criteria were: (1) patients with severe valvular heart disease, severe pulmonary hypertension, severe liver or renal insufficiency, rheumatoid arthritis, infectious diseases, and malignant tumors; (2) patients with familial hypertriglyceridemia (triglyceride ≥5.65 mmol/L); and (3) patients with neuropsychiatric problems that prevent them cooperating with the researchers. Most critically, our study followed the Declaration of Helsinki and was certified by the Chinese PLA General Hospital’s Ethics Service Center.

To confirm the diagnosis of coronary heart disease, the cardiac intervention center of the PLA General Hospital performed coronary intervention and perioperative treatment according to current guidelines. All of the angiography results were analyzed using the same image analysis tool. Loading doses of aspirin (300 mg) and clopidogrel (300 mg) were administered before the intervention. The degree of coronary stenosis was determined using the Gensini score (26), and two experienced experts trained the recorders. Individualized interventions, such as intensive medication therapy, percutaneous coronary intervention (PCI), or coronary artery bypass grafting (CABG), were administered based on coronary angiography results, and long-term follow-up was conducted after being discharged.

We collected demographic data (age and gender), anthropometric data [height, body weight, BMI, heart rate, systolic blood pressure (SBP), and diastolic blood pressure (DBP)], laboratory data [total cholesterol (TC), triglycerides (TG), low-density lipid-cholesterol (LDL-C), high-density lipid-cholesterol (HDL-C), estimated glomerular filtration rate (eGFR), fasting plasma glucose (FPG), uric acid (UA), albumin], left ventricular ejection fraction (LVEF), comorbidity data [diabetes mellitus type 2 (T2DM), hypertension, stroke, prior myocardial infarction (MI), hyperlipidemia and chronic kidney disease (CKD)], smoking, drug-usage data [aspirin, clopidogrel, statins, β-blocker and angiotensin-converting enzyme inhibitor (ACEI)/angiotensin receptor blocker (ARB)], coronary lesions data[left anterior descending branch (LAD), left circumflex branch (LCX), right coronary artery (RCA), left main coronary artery (LM), multivessel lesions and Gensini score] and treatment data (intensive medications, PCI and CABG).

The body mass index (BMI) was calculated as follows:

According to WHO criteria in the Asian population, patients could be classified as overweight (BMI > 24.9 kg/m²), normal-weight (BMI 18.5–24.9 kg/m²), and under-weight (BMI < 18.5 kg/m²) (27).

The eGFR was calculated by the Chinese modified Modification of Diet in Renal Disease equation:

Standardized creatinine (Scr) was calculated by the calibration equation:

Chronic kidney disease was defined as eGFR <60 ml/min/1.73 m².

The diagnostic criteria for diabetes mellitus type 2 (T2DM) were: FPG ≥ 7.0 mmol/L; and (or) random blood glucose (RBG) ≥ 11.1 mmol/L; blood glucose ≥11.1 mmol/L 2 h after oral glucose tolerance test (OGTT).

Hyperlipidemia was defined as the use of lipid-lowering drugs or total serum cholesterol ≥240 mg/dl.

Based on coronary angiography results, the multivessel lesion was defined as having more than 2 vessels with significant diameter stenosis of 50%.

The geriatric nutritional risk index (GNRI) was used in this study to assess the nutritional status of the oldest-old patients with ACS. GNRI is calculated as follow (17):

The ideal weight was calculated as follows: 22 × square of height (m²) (18). It is worth noting that GNRI was created to identify and predict nutritional-related complications (17). The original GNRI cut-off values and grades of nutrition-related risk were: severe risk (GNRI <82), moderate risk (GNRI 82–92), low risk (GNRI 92–98), and no risk (GNRI >98). Patients were often considered as having a normal nutritional status if their GNRI >98 (19). Hence, we used 98 as the cut-off value in the present study. GNRI >98 was defined as well-nourished, while GNRI≤98 was defined as at malnutrition risk.

The follow-up period lasted up to 10 years performed every 12 months after discharge via outpatient visits, telephone records, or medical records of outcomes. During the follow-up period, 37 patients were lost follow-up, leaving 662 (94.7%) patients enrolled in the final statistical analysis. All-cause mortality (cardiac and non-cardiac) was the ultimate endpoint of our research.

The baseline characteristics of the participants were shown according to GNRI >98 and GNRI ≤98. The measurement data of normal distribution were expressed as mean ± SD, and the T-test was used for homogeneity of variance. If the variance is not uniform, the rank-sum test was used. Non-normally distributed measurements were represented by the median and interquartile range (IQR). Statistical data were expressed by quantity, the chi-square test was used to evaluate differences between groups, and an analysis of variance was used to compare data between groups. Pearson correlation test was used to evaluate the correlation between GNRI and clinical parameters. Unadjusted survival curves were generated using log-rank tests in Kaplan–Meier plots. Univariate Cox regression analysis (HR, 95% CI) was used to identify the factors associated with all-cause mortality. p < 0.05 was considered statistically significant. The Cox proportional hazards model was used to estimate the association between GNRI and all-cause mortality. We built three regression models: model 1 is the unadjusted model, model 2 is the partially adjusted model (age and gender), and model 3 is the completely adjusted model (age, gender, diabetes, stroke, CKD, aspirin, eGFR, FPG, UA, EF, Gensini score, LM lesions, multivessel lesions, and HDL-C). The GNRI was transformed into three classification variables for the primary analysis. For the trend test, the new categorical variables were recorded as continuous variables and entered into the regression model. We also standardized the GNRI and then put it into a regression model to determine the relationship between the change in GNRI per SD and all-cause mortality. In addition, we performed a subgroup analysis to explore whether the relationship between GNRI and all-cause mortality could be modified by the following variables: gender, diabetes, hypertension, prior MI, hyperlipidemia, CKD, smoking, LM lesions, and multivessel lesions. Interactions between GNRI and the above variables were tested. Results were reported as HR and 95% CI. Two-sided p < 0.05 was considered statistically significant. All analyses were performed using the statistical software packages R (28) and Empower Stats (29).

There were 662 oldest-old patients with ACS enrolled in our study, 71.9% (n = 476) of whom were males (Table 1). The mean age of the participants was 81.87 ± 2.14 years old (IQR 80–89), and the average GNRI at admission was 102.47 ± 11.06. According to GNRI, the patients were classified as well-nourished (GNRI >98, n = 456, 68.89%), and at malnutrition risk (GNRI ≤98, n = 206, 31.11%). Patients with low GNRI had lower body weight, BMI, SBP, LVEF, albumin, and eGFR, while they have higher height and heart rate. There were no statistical differences in TC, TG, LDL-C, HDL-C, FPG, and UA between the two groups. Patients at malnutrition risk had a higher proportion of CKD but a lower proportion of hypertension. There was no significant difference in the prevalence of other common complications such as diabetes, hyperlipidemia, or stroke. What is more, we noticed that patients with low GNRI were more likely to have RCA lesions and multivessel lesions, and got a higher Gensini score. In terms of smoking behavior, medication usage, and treatment manners, there was no significant difference between the two groups.

We further analyzed GNRI in patients with different BMI and albumin levels (Figure 2). The high prevalence of malnutrition was found in patients with BMI <18.5 kg/m² (94.12%), and patients with albumin<35 g/L (74.68%). What is more, there were substantial malnutritional patients in the normal-to over-weight groups.

The participants were followed for a median of 63 months (IQR 51–74). There were 201 endpoint-events during the follow-up, with 82 occurring in the low GNRI group and 119 in the other. As a nutritional screening index, GNRI was associated with many traditional cardiovascular risk factors (Supplementary Table 1). In univariate analysis (Table 2), a strong positive association was found between all-cause mortality and age, diabetes, stroke, CKD, FPG, Gensini scores, LM lesions, and multivessel lesions. Meanwhile, aspirin, HDL-C, eGFR, LVEF, albumin, and GNRI (HR = 0.97, 95% CI [0.96, 0.99]) were closely associated with a reduction in all-cause mortality. However, male, hypertension, BMI, TC, and LDL-C did not show a significant association with the outcome.

We further assessed the association between GNRI and all-cause mortality by multivariable Cox regression analysis (Table 3 and Figure 3). For each SD increase in GNRI, the risk of all-cause mortality was lowered by 23%. And when compared with the high GNRI group, the HR of all-cause mortality in the low GNRI group was 1.39 (95% CI [1.04, 1.86], p < 0.05). Then we divided the patients into three groups according to the tertiles of GNRI. Compared with tertile 1, the HRs of tertile 2 and tertile 3 were 1.49 (95% CI [1.02, 2.19], p < 0.05) and 1.74 (95% CI [1.22, 2.50], p < 0.01), respectively. Subsequently, the trend test revealed a dose–response between GNRI and all-cause mortality.

According to the Kaplan–Meier survival analysis (Figure 4A), we found that patients with high GNRI had a longer survival time (p < 0.001). Then we further divided GNRI into the tertiles and analyzed them. As shown in Figure 4B, all-cause mortality in tertile 3 was significantly lower than in tertile 1 and tertile 2 (p < 0.001).

In the subgroup analysis (Figure 5), we adjusted sex, age, diabetes, stroke, CKD, aspirin, eGFR, FPG, UA, LVEF, Gensini score, LM lesions, multivessel lesions, and HDL-C except for the stratified variables. The results showed that low GNRI was associated with increased all-cause mortality in those males, with diabetes mellitus, with previous MI, without multivessel lesions, and without LM lesions. But the values of p for interaction were all >0.05, suggesting the inverse association between low GNRI with all-cause mortality across all those subgroups.

This was the first time enrolled oldest-old patients with ACS, assessed their nutritional status with GNRI on admission, and followed for 10 years. Then we evaluated the association between GNRI and long-term all-cause mortality. In the end, we demonstrated that GNRI is a nutrition assessing tool with prominent advantages, and it may be a hopeful tool to quickly and accurately evaluate nutritional status and predict their prognosis. The results of our study were reliable and practical for clinical use.

However, there were some deficits inevitably. First, this was a single-center cohort study and all participants were Chinese and our was something old, so some selection bias existed. Second, we did not record dietary intake, physical activity, and other factors during the follow-up that may affect nutritional status. Finally, we took 98 as the cut-off value of GNRI, it may need further adjustment according to race or age, etc.