The relationship of the prognostic nutrition index with short-term prognosis in adult patients with acute myocardial infarction: A population-based study

Abstract

This study aims to examine the association between the Prognostic Nutritional Index (PNI) and the short-term prognosis of adult patients experiencing acute myocardial infarction (AMI). A retrospective analysis was conducted using medical records from the MIMIC-IV database. The impact of PNI on survival outcomes was assessed using a restricted cubic spline (RCS) model. The predictive capacity of PNI was evaluated by calculating the receiver operating characteristic (ROC) curve. Kaplan–Meier (KM) analysis assessed survival rates across different PNI levels. Cox regression models were also utilized to investigate the relationship between PNI and 30-day mortality. Subgroup analyses were conducted to examine the effects of comorbidities and PNI on patient outcomes. 2347 adult patients experiencing AMI, 345 fatalities were recorded, yielding a mortality rate of 14.70%. After adjusting for potential confounders, the RCS analysis indicated a negative correlation between PNI < 43.15 and short-term mortality in these patients (P for nonlinearity < .001). The area under the ROC curve for 30-day all-cause mortality was 0.747. Concurrently, KM survival analysis demonstrated that patients with PNI < 43.15 had a poorer prognosis. Cox proportional hazards regression analysis identified PNI as an independent predictor of short-term prognosis in this patient population (HR = 0.484, 95% CI: 0.360–0.649, P < .001). Furthermore, subgroup analyses revealed significant interactions between PNI and other comorbidities regarding 30-day mortality risk (P for interaction < .05). PNI is associated with short-term mortality outcomes in patients with AMI and may serve as a valuable prognostic indicator for assessing mortality risk. retrospectively registered

1. Introduction

Acute myocardial infarction (AMI) is characterized by myocardial necrosis resulting from acute and sustained ischemia and hypoxia of the coronary arteries. Despite advancements in therapeutic interventions, including pharmacological thrombolysis, coronary artery bypass grafting, and percutaneous coronary intervention, the mortality rate associated with AMI remains high. This is primarily due to the irreversible death of myocardial cells and the consequent reduction in cardiac systolic function.^[1–3] The prognosis is inferior for patients with severe AMI admitted to the intensive care unit, where over 50% of these individuals are at an increased risk of complications and organ dysfunction.^[4] Research indicates a strong correlation among nutritional status, inflammatory processes, and the prognosis of AMI patients.^[5,6] The prognostic nutritional index (PNI) is a comprehensive biomarker that reflects nutritional, immune, and inflammatory status by assessing peripheral blood lymphocyte counts and serum albumin levels. It is frequently used to evaluate the prognosis of patients with various tumor types.^[7] Furthermore, clinical studies have demonstrated that PNI is also associated with the clinical outcomes of cardiovascular diseases, including heart failure, dilated cardiomyopathy, and coronary artery disease.^[8–10] However, the relationship between the PNI and the short-term prognosis of AMI patients remains ambiguous. Some studies suggest that PNI may not be an effective prognostic indicator for AMI.^[11]

This study investigates the correlation between PNI levels and 30-day all-cause mortality in AMI patients by analyzing data from the Medical Information Mart for Intensive Care (MIMIC) database. The aim is to enhance the identification of high-risk AMI patients, facilitate early clinical intervention, and improve nursing care measures.

2. Methods

2.1. Data collection and processing

A retrospective analysis was conducted of medical records from patients diagnosed with AMI, selected from the MIMIC-IV database. The inclusion criteria were as follows: a clinical diagnosis of primary acute myocardial infarction; consideration of only the first admission record in cases of multiple hospitalizations; age ≥ 18 years; and complete records of albumin and lymphocyte counts, along with comprehensive follow-up data. The exclusion criteria included incomplete admission or discharge records; age < 18 years; absence of albumin and lymphocyte count records; patients who succumbed within 24 hours of admission; and incomplete follow-up data. The patient screening process is depicted in Figure 1.

Figure 1.

Flow diagram for the process of study inclusion.

This study adheres to the principles outlined in the Declaration of Helsinki. As the MIMIC-IV database is publicly accessible and does not contain identifiable patient information, our institution’s ethical review board granted an exemption for this study.

2.2. Study variables

Collected data included demographics (age, gender, BMI, race, nicotine, and alcohol use), comorbidities (e.g., hypertension, diabetes, renal failure, chronic obstructive pulmonary disease (COPD), sepsis, tumor, liver failure, hyperlipidemia, atrial fibrillation), lab parameters (e.g., lactate, albumin, troponin T, lymphocytes, calcium, creatinine, glucose, hematocrit, hemoglobin, PT, APTT, platelets, potassium, sodium, WBC count), systematic scores (SOFA, SAPS II), and primary treatment. All examination results pertain to the initial test conducted after the patient’s hospital admission. To minimize bias due to missing data, variables with more than 30% missing values were excluded from the study. Variables with <30% missing data were imputed using the MICE package in R with random forest. Five datasets were generated over 50 iterations to achieve convergence, and Rubin’s rules were used to obtain pooled estimates. All study variables were considered for multivariate Cox models due to their potential as confounders. The primary outcome was 30-day survival of AMI patients, categorizing them into Survivor and Non-survivor groups.

PNI calculation formula is as follows: PNI = 10 × Albumin(g/dL) + 5 × Lymphocyte(10^9/L).^[12]

3. Statistical analysis

Continuous variables following a normal distribution are expressed as the mean ± standard deviation (X ± S), whereas non-normally distributed continuous variables are represented by the median and inter-quartile range. Categorical variables are presented as frequencies and percentages. For continuous variables, either the independent-samples t-test or the Mann–Whitney U test was used, whereas the chi-square test was used for categorical variables. A restricted cubic spline (RCS) model with 4 knots at the 5th, 35th, 65th, and 95th percentiles of the PNI distribution was used to investigate the association between PNI and the risk of 30-day mortality, with the median PNI as the reference. The cutoff value of 43.15 was determined empirically as the point at which the hazard ratio equaled 1. To minimize the risk of overfitting, we limited the number of knots and incorporated a wide range of covariates into the multivariate models to adjust for. The 30-day survival rates of AMI patients with varying PNI levels were compared using Kaplan–Meier (KM) survival curves, and a receiver operating characteristic (ROC) curve was generated using bootstrapping with 1000 resamples to determine the area under the curve (AUC) for PNI in predicting all-cause mortality. Subsequently, univariate and multivariate Cox proportional hazards models were developed to investigate the association between PNI and clinical outcomes. In the multivariate model for 30-day all-cause mortality, Model 1 was unadjusted; Model 2 was adjusted for demographic factors such as age, gender, race, BMI, Nicotine, and Alcohol; and Model 3 included all variables. Simultaneously, we performed a subgroup analysis to investigate the impact of complications and the PNI on survival outcomes, presenting the findings as a forest plot. A 2-tailed P-value of <.05 was considered to indicate statistical significance. Multicollinearity among covariates was assessed using the variance inflation factor (VIF), with a threshold of VIF > 5 indicating significant multicollinearity; no variables exceeded this threshold. The proportional hazards assumption was tested for all Cox models using the Schoenfeld residuals test, and no significant violations were observed (all P > .05). Data processing and analysis were performed using R version 4.3.3 (2024-02-29), along with Zstats 1.0 (www.zstats.net).

4. Results

4.1. Baseline characteristics of the study population

Among a cohort of 2347 adult AMI patients, 345 individuals succumbed to the condition, resulting in a mortality rate of 14.70%. Compared to survivors, non-survivors were older and had a lower proportion of males and nicotine use. Non-survivors also had a higher prevalence of renal failure, COPD, malignancies, sepsis, tumors, liver failure, and atrial fibrillation. In contrast, hypertension and hyperlipidemia were less common in non-survivors. Furthermore, the non-survivor group exhibited significantly lower levels of PNI, calcium, hematocrit, hemoglobin, and PLT. Conversely, elevated levels of lactate, troponin T, creatinine, glucose, PT, potassium, and WBC, as well as higher scores on the SOFA and SAPS II, were observed, all of which were statistically significant (P < .05). No significant differences were noted in BMI, race, alcohol, diabetes, PTT, sodium, or the proportion of percutaneous coronary intervention (P > .05) (Table 1).

Table 1.

The characteristics of the population.

Variables	All patients	Survivor	Non-survivor	P
Patients	n = 2347	n = 2002	n = 345
Demographics
Age (years)	68.00 (59.00, 76.00)	68.00 (59.00, 75.00)	73.00 (64.00, 81.00)	<.001
BMI (kg/m2)	27.62 (24.38, 31.88)	27.70 (24.54, 31.78)	27.08 (23.72, 32.22)	.186
Gender (Male, n[%])	1571 (66.94)	1359 (67.88)	212 (61.45)	0.019
Race (White, n[%])	1288 (54.88)	1107 (55.29)	181 (52.46)	.506
Nicotine, n(%)	368 (15.68)	331 (16.53)	37 (10.72)	.006
Alcoholic, n(%)	134 (5.71)	113 (5.64)	21 (6.09)	.743
Comorbidities
Hypertension, n(%)	783 (33.36)	714 (35.66)	69 (20.00)	<.001
Diabetes, n(%)	1084 (46.19)	932 (46.55)	152 (44.06)	.391
Renal failure, n(%)	1293 (55.09)	1005 (50.20)	288 (83.48)	<.001
COPD, n(%)	312 (13.29)	244 (12.19)	68 (19.71)	<.001
Sepsis, n(%)	358 (15.25)	214 (10.69)	144 (41.74)	<.001
Tumor, n(%)	245 (10.44)	183 (9.14)	62 (17.97)	<.001
Liver failure, n(%)	201 (8.56)	107 (5.34)	94 (27.25)	<.001
Hyperlipidemia, n(%)	1535 (65.40)	1342 (67.03)	193 (55.94)	<.001
Atrial fibrillation, n(%)	894 (38.09)	743 (37.11)	151 (43.77)	.019
Laboratory variables
PNI	43.15 (37.15, 49.20)	44.35 (38.70, 50.10)	36.55 (30.15, 41.30)	<.001
Lactate(mmol/L)	1.50 (1.20, 2.10)	1.50 (1.10, 2.00)	2.20 (1.50, 4.10)	<.001
Troponin T(ng/mL)	0.35 (0.12, 1.21)	0.34 (0.12, 1.13)	0.51 (0.15, 1.57)	<.001
Calcium (mg/dL)	8.70 (8.20, 9.20)	8.80 (8.30, 9.20)	8.50 (7.90, 9.00)	<.001
Creatinine (mg/dL)	1.10 (0.90, 1.60)	1.10 (0.80, 1.50)	1.50 (1.10, 2.30)	<.001
Glucose (mg/dL)	133.00 (106.00, 187.00)	130.50 (105.00, 178.00)	158.00 (116.00, 219.00)	<.001
Hematocrit (%)	36.20 (30.60, 41.00)	36.70 (31.50, 41.27)	32.40 (28.40, 38.20)	<.001
Hemoglobin (g/dL)	11.80 (9.80, 13.60)	12.00 (10.10, 13.70)	10.40 (8.90, 12.30)	<.001
PT (s)	12.80 (11.80, 14.45)	12.70 (11.70, 14.00)	14.10 (12.60, 17.90)	<.001
PTT (s)	35.20 (29.40, 56.50)	35.45 (29.50, 56.80)	34.30 (28.60, 55.30)	.356
PLT (K/uL)	205.00 (160.00, 260.00)	208.00 (165.00, 259.00)	186.00 (140.00, 265.00)	.002
Potassium (mmol/L)	4.20 (3.90, 4.60)	4.20 (3.90, 4.50)	4.30 (3.90, 5.00)	<.001
Sodium (mmol/L)	138.00 (136.00, 141.00)	139.00 (136.00, 141.00)	138.00 (135.00, 141.00)	.067
WBC (K/uL)	9.70 (7.50, 13.20)	9.40 (7.30, 12.60)	12.30 (8.20, 17.80)	<.001
Clinical variables
ST elevated (yes, n[%])	464 (19.77)	372 (18.58)	92 (26.67)	<.001
PCI (yes, n[%])	285 (12.14)	248 (12.39)	37 (10.72)	.382
Scoring systems
SOFA	5.00 (3.00, 8.00)	5.00 (3.00, 7.00)	8.00 (5.00, 11.00)	<.001
SAPSII	37.00 (31.00, 47.00)	36.00 (30.00, 44.00)	49.00 (39.00, 60.00)	<.001

Note: p-values less than.05 are indicated in bold.

Abbreviations: BMI = body mass index, COPD = chronic obstructive pulmonary disease, PCI = percutaneous coronary intervention, PLT = platelets, PNI = prognostic nutrition index, PT = prothrombin time, PTT = partial thromboplastin time, SAPASII = simplified acute physiology score II, SOFA = sequential organ failure assessment, WBC = white blood cell.

4.2. The capacity of PNI to forecast the short-term mortality

The prognostic value of the PNI for short-term mortality was assessed using the RCS analysis. In the unadjusted model, a PNI value of 43.15 corresponded to a hazard ratio (HR) of 1. For PNI values below 43.15, there was a significant negative correlation with short-term mortality in adult AMI patients (P nonlinear < .001) (Fig. 2A). After adjusting for all confounding variables, this nonlinear negative correlation persisted for PNI values below 43.15, concerning the 30-day mortality risk (P nonlinear = .002) (Fig. 2B). ROC analysis demonstrated that the area under the curve (AUC) for PNI in predicting the 30-day mortality risk of AMI patients was 0.747 (Fig. 3A). Furthermore, the KM survival curve analysis indicated a statistically significant difference in 30-day mortality across different PNI levels. Specifically, patients in the low PNI group (PNI < 43.15) exhibited a significantly higher risk of 30-day all-cause mortality compared to those in the high PNI group (PNI ≥ 43.15) (P < .001) (Fig. 3B).

Figure 2.

The RCS plot between PNI and the short-term mortality rate of adult AMI patients. (A) In the unadjusted model, (B) The model was adjusted for all factors. RCS, restricted cubic spline. PNI, prognostic nutrition index. AMI, acute myocardial infarction.

Figure 3.

The ROC curve analysis of PNI predicting the 30-day death risk of AMI patients (A) and the KM survival plot of AMI patients with different PNI levels (B). ROC, receiver operating characteristic. PNI, prognostic nutrition index. AMI, acute myocardial infarction.

4.3. Cox proportional hazards model

Table 2 presents the findings from the Cox proportional hazards model assessing the 30-day all-cause mortality risk among PNI and AMI patients. The analysis indicates that PNI, whether treated as a continuous variable or categorized according to the RCS results, is an independent predictor of 30-day survival in AMI patients (P < .001). In the unadjusted Cox model (Model I), a higher PNI is identified as a protective factor against short-term mortality risk (HR = 0.226, 95%CI: 0.174–0.295, P < .001). Upon adjusting for demographic variables such as Age, Gender, and Race in Model II, a high PNI remains an independent protective factor for short-term mortality risk (HR = 0.247, 95% CI: 0.189–0.323, P < .001). In Model III, after adjusting for all variables except PNI, a high PNI remains an independent protective factor for short-term mortality in AMI patients (HR = 0.484, 95% CI: 0.360–0.649, P < .001).

Table 2.

The association between PNI and outcomes.

Variables	Model 1		Model 2		Model 3
	HR (95%CI)	P	HR (95%CI)	P	HR (95%CI)	P
PNI (continuous)	0.916 (0.905–0.927)	<.001	0.916 (0.905–0.928)	<.001	0.955 (0.941–0.970)	<.001
PNI
Low	1.000 (Reference)		1.000 (Reference)		1.000 (Reference)
High	0.226 (0.174–0.295)	<.001	0.247 (0.189–0.323)	<.001	0.484 (0.360–0.649)	<.001

CI = confidence interval, HR = hazard ratio, PNI = prognostic nutritional index.

Model 1: Crude.

Model 2: Adjust: Age, BMI, Gender, Race, Nicotine, Alcoholic.

Model 3: Adjust: Age, BMI, Gender, Race, Nicotine, Alcoholic, Hypertension, Diabetes, Renal Failure, COPD, Sepsis, Tumor, Liver Failure, Hyperlipidemia, Atrial Fibrillation, Lactate, Troponin T, Calcium, Creatinine, Glucose, Hematocrit, Hemoglobin, PT, PTT, PLT, Potassium, Sodium, WBC, ST elevated, PCI, SOFA, SAPSII.

4.4. Subgroup analyses

The RCS, KM survival analysis, and Cox regression analyses indicate a poor short-term prognosis when the PNI is below 43.15. A subgroup analysis was performed to investigate whether complications influence this association, incorporating conditions such as Hypertension, Diabetes, Renal failure, COPD, Sepsis, Tumor, Liver failure, hyperlipidemia, and Atrial fibrillation. The 30-day all-cause mortality rate was markedly elevated in the low prognostic nutritional index (PNI) group (P < .001). As illustrated in Figure 4A, significant interactions were observed between PNI and comorbid conditions such as hypertension, renal failure, sepsis, and liver failure (P for interaction < .05), suggesting that the relationship between PNI and 30-day mortality may differ across these comorbidities. Conversely, for conditions such as diabetes, COPD, tumors, hyperlipidemia, and atrial fibrillation, the interactions were not statistically significant (P for interaction > .05), indicating that the prognostic value of PNI remained stable across these subgroups. This pattern was consistent even after controlling for demographic variables, as depicted in Figure 4B.

Figure 4.

Subgroup analyses of the association between PNI and the 30-day death risk of AMI patients with different comorbidities. (A) Unadjusted variables. (B) Adjust: Age, BMI, Gender, Race, Nicotine, Alcoholic. PNI, prognostic nutrition index. AMI, acute myocardial infarction.

5. Discussion

AMI is a prevalent cardiovascular condition observed in clinical settings, with hypertension, obesity, and diabetes being significantly represented among patients diagnosed with AMI.^[13] Risk factors for AMI include advanced age, hyperlipidemia, and diabetes. The rupture of atherosclerotic plaques and subsequent thrombosis can exacerbate coronary artery obstruction, thereby restricting myocardial blood flow and elevating the risk of mortality.^[14] Given the recent trend of an aging population, the incidence of AMI is projected to rise annually, posing a substantial threat to public health. Consequently, identifying highly sensitive biomarkers is crucial for accurately predicting mortality risk in AMI patients. Such biomarkers would facilitate improved targeted monitoring, management, and therapeutic decision-making, ultimately enhancing patient prognosis.

The PNI serves as a comprehensive biomarker reflecting nutritional status, inflammation, and immune function, calculated based on serum albumin levels and peripheral blood lymphocyte counts. In critically ill patients, malnutrition can exacerbate adverse outcomes, including significantly increased mortality rates, clinical complications, and prolonged hospital stays.^[15] Recently, there has been growing interest in using PNI for prognostic evaluation in cardiovascular diseases.^[16] Notably, the application of PNI in patients with coronary heart disease highlights its potential as a predictive indicator. A study investigating the association between PNI and mortality risk in coronary heart disease patients demonstrated that individuals with low PNI exhibited a significantly increased risk of all-cause mortality during the follow-up period, particularly among diabetic patients, where the negative impact of low PNI was more pronounced.^[17] Furthermore, another study examining the role of PNI in patients undergoing coronary artery bypass grafting revealed that those with higher preoperative PNI had lower in-hospital mortality rates, indicating that PNI is an independent predictor of short-term prognosis after surgery.^[18] Secondly, the PNI has been established as a significant prognostic marker in patients with heart failure. A meta-analysis that synthesized data from multiple studies demonstrated that heart failure patients with a low PNI are at a substantially elevated risk of all-cause mortality and major adverse cardiac events.^[19] This evidence suggests that PNI can serve as a bedside biomarker for assessing malnutrition and inflammation in heart failure patients. Serum albumin, a key component of PNI, reflects nutritional status and acts as a potent circulating antioxidant. It inhibits platelet activation and aggregation and prevents vascular endothelial cell apoptosis.^[20,21] A retrospective study has identified serum albumin levels at admission as an independent predictor of heart failure and cardiovascular mortality in AMI patients.^[22] Furthermore, a recent study involving 1424 participants found that a persistently low albumin status following AMI is associated with adverse long-term outcomes, irrespective of acute albumin levels (HR = 4.02, 95% CI = 2.36–6.87, P < .001).^[23] Lymphocytes are another immunological parameter that influences the PNI and are integral to the inflammatory process. Patients experiencing AMI who exhibit lymphopenia are predisposed to endothelial dysfunction, platelet activation, and thrombosis.^[24,25] Consequently, the PNI offers a distinct advantage in prognostic assessment of AMI patients.

This study investigated the relationship between PNI levels and the short-term prognosis of adult patients with AMI by analyzing clinical data from the MIMIC-IV database. The findings revealed that patients in the non-survivor group had lower PNI levels than those in the survivor group. A nonlinear negative correlation was observed between PNI levels and the risk of all-cause mortality within 30 days, with a cutoff of 43.15. Patients in the low PNI group demonstrated a higher risk of 30-day mortality compared to those in the high PNI group. These results align with the study by Huang et al.,^[26] suggesting that a low PNI may indicate a poorer prognosis in patients. Multivariate Cox regression analysis further established that a low PNI is an independent predictor of all-cause mortality in adult AMI patients within 30 days. The area under the ROC curve for predicting 30-day mortality risk in adult AMI patients was 0.747, indicating that PNI possesses significant clinical predictive value.

Subgroup analysis revealed an interaction between PNI and factors such as hypertension, renal failure, sepsis, and liver failure regarding 30-day mortality risk (P for interaction < .05). This suggests that the presence of conditions such as hypertension, renal failure, sepsis, and liver failure may alter the influence of the PNI on the short-term prognosis of AMI patients. In patients with hypertension, the combined effect of hypertension and AMI substantially elevates the risk of mortality.^[27] Additionally, hypertensive patients exhibit significantly lower peripheral blood lymphocyte counts compared to non-hypertensive individuals, with the reduction in lymphocyte proportion being positively correlated with the duration of hypertension.^[28] Consequently, in hypertensive patients with AMI, the impact of elevated blood pressure may outweigh that of PNI. Patients with renal failure typically present with markedly low albumin levels and lymphocyte counts,^[29] and renal failure is recognized as a negative prognostic factor for survival following AMI. It acts synergistically with albumin depletion and lymphocyte imbalance, thereby increasing the risk of cardiovascular mortality and reinfarction.^[30] Furthermore, the overall mortality rate in patients with Sepsis is notably high, reported at 36.3%, with the mortality rate for septic shock reaching 50.8%[.^[31] Moreover, a decrease in albumin levels in septic patients correlates with a more severe condition, and a significant reduction in peripheral blood T lymphocytes further exacerbates the risk of adverse AMI outcomes.^[32,33] As the liver is the primary organ responsible for albumin synthesis, liver failure compromises albumin function and reduces lymphocyte levels, exacerbating the systemic inflammatory state, promoting oxidative stress, and facilitating toxin accumulation. These processes damage cardiomyocytes, increase cardiac afterload, and increase the risk of renal impairment.^[34,35] In patients without hypertension, renal failure, sepsis, or liver failure, PNI levels were significantly associated with the risk of mortality in AMI patients at 30 days. Consequently, we posit that a low PNI is linked to an elevated risk of mortality in AMI patients within 30 days.

Nevertheless, this study is subject to several limitations. Firstly, it employs a retrospective design, which may introduce bias in subject selection. Secondly, the sample size is insufficient, particularly within the mortality group. Following subgroup analysis, disparities in subgroup sizes were observed, and for specific analyses involving concomitant diseases, the sample size was notably small, potentially leading to deviations. Other unaccounted variables may still influence our findings despite adjustments for particular factors. Thirdly, our findings are based on data obtained from the MIMIC-IV database, which primarily reflects a North American population. As a result, racial and regional factors may influence the outcomes. Furthermore, the PNI cutoff value was established using a data-driven method without internal validation techniques such as bootstrapping or cross-validation, potentially leading to overfitting. Fourthly, the subgroup analyses involved multiple comparisons, thereby increasing the likelihood of false-positive results. Consequently, the observed interactions should be interpreted with caution, and our findings require validation in independent prospective cohorts.

In conclusion, a low PNI (<43.15) significantly impacts the short-term prognosis of AMI patients and holds potential as a critical tool for risk stratification and nutritional intervention in this patient population. Further research is warranted to investigate the integration of PNI with other clinical indicators and its applicability across various clinical stages and treatment modalities. Such studies could offer more comprehensive and individualized guidance for managing AMI patients.

Acknowledgments

We appreciate the MIMIC-IV database for providing clinical data. Furthermore, we sincerely thank our international colleagues for their invaluable assistance with language-related matters.

Author contributions

Conceptualization: Qin Shen, Lingling Song.

Data curation: Qin Shen, Lingling Song.

Methodology: Qin Shen.

Project administration: Lingling Song.

Resources: Lingling Song.

Software: Qin Shen.

Supervision: Lingling Song.

Validation: Lingling Song.

Visualization: Lingling Song.

Writing – original draft: Qin Shen.

Writing – review & editing: Lingling Song.