Prognostic Nutritional Index as a novel biomarker for predicting prognosis in sepsis-associated encephalopathy: A multicenter retrospective cohort study

LinaZhao1✉

ChaoQi1✉,2

QingheYan1

YuehaoShen1

DongxueHuang1

HaiyingLiu1

XuguangLi1

YunLi3✉Emailcfsyy_liyun@126.com

KeliangXie1✉,4Emailxiekeliang2009@hotmail.com

1Department of Critical Care MedicineTianjin Medical University General Hospital300052TianjinChina

2Department of AnesthesiologyNankai University Affiliated Beichen Hospital300400TianjinChina

3Department of AnesthesiologyThe Second Hospital of Tianjin Medical University300211TianjinChina

4Department of Anesthesiology, Tianjin Institute of AnesthesiologyTianjin Medical University General Hospital300052TianjinChina

+86 18847600734

Lina Zhao^1*, Chao Qi^1,2*_, Qinghe Yan¹_,Yuehao Shen¹, Dongxue Huang¹, Haiying Liu¹, Xuguang Li¹,, Yun Li^4**, Keliang Xie^1,5**

¹. Department of Critical Care Medicine, Tianjin Medical University General Hospital, Tianjin 300052, China

². Department of Anesthesiology, Nankai University Affiliated Beichen Hospital,Tianjin 300400, China

⁴ Department of Anesthesiology, The Second Hospital of Tianjin Medical University, Tianjin 300211, China

⁵.Department of Anesthesiology, Tianjin Institute of Anesthesiology, Tianjin Medical University General Hospital, Tianjin 300052, China

^**Corresponding author:

Keliang Xie: xiekeliang2009@hotmail.com; Department of Critical Care Medicine, Tianjin Medical University General Hospital, Tianjin 300052, China; +8615332112099.

Yun Li: cfsyy_liyun@126.com; Department of Anesthesiology, The Second Hospital of Tianjin Medical University, Tianjin, 300211, China; +86 18847600734

Lina Zhao, Chao Qi, Yun Li and Keliang Xie contributed equally to this work.

Abstract

Background Sepsis-associated encephalopathy (SAE) still has a high mortality rate, and there is a lack of effective biomarkers to assess the prognosis of SAE. This study aims to explore the relationship between prognostic nutritional index (PNI) and the prognosis of patients with SAE.

Methods This study is a multicenter cohort study, data from 2008–2019. The primary outcome was 28-day all-cause mortality in the SAE population. To explore the prognostic relationship between PNI and SAE patients, the multivariable Logistic regression, propensity score matching, inverse probability weighting were conducted to adjust confounders. In this study, the generalized additive model (GAM), Kaplan-Meier curve, receiver operating characteristic curve (ROC) curve and other methods were used to analyze the relationship between PNI and the 28-day mortality rate of SAE patients. The results of this study were validated by external data.

Results Among 3,202 SAE patients, multivariable analysis identified PNI as an independent predictor of 28-day mortality (OR: 0.85, 95% CI: 0.77–0.93) of original cohort. GAM of original cohort showed that a PNI of 34 was the optimal prognostic threshold for SAE patients. The Kaplan-Meier curves of both the original cohort and the external validation cohort showed that the 28-day mortality rate of SAE patients with PNI lower than 34 was significantly lower than that of patients with PIN higher than 34 (P < 0.001). ROC analysis showed superior predictive performance in original cohort (AUC: 0.879; sensitivity: 0.878; specificity: 0.880) versus external validation cohort (AUC: 0.724; sensitivity: 0.878; specificity: 0.569). Stratified analysis of the results of the study showed that elevated PNI correlated with higher Glasgow Coma Scale scores (P < 0.001).

Conclusions This large-scale multicenter study establishes the PNI as an independent predictor of 28-day mortality in patients with sepsis-associated encephalopathy (SAE). We identified that SAE patients with PNI < 34 exhibited significantly higher 28-day mortality rates and worse neurological function.

Keywords:

Prognostic Nutritional Index

sepsis-associated encephalopathy

28-day mortality

Glasgow Coma Scale

1 Introduction

SAE is clinically defined as diffuse cerebral dysfunction secondary to systemic inflammatory responses in sepsis patients, in the absence of direct central nervous system (CNS) infection¹. Previous studies have shown that the incidence of SAE is as high as 50–70%^2,3. Critically, progression to SAE elevates sepsis mortality rates to approximately 50%, a marked increase compared to septic patients without encephalopathy^2,3. Currently, there is still a lack of effective indicators biomarkers to evaluate the prognosis of patients with SAE. Therefore, it is particularly important to explore the early evaluation of the prognosis of SAE patients and give effective interventions to reduce the mortality rate of SAE patients.

The PNI, calculated as serum albumin (g/dL) × 10 + total lymphocyte count (× 10⁹/L) × 5, integrates two pivotal pathophysiological pathways in SAE: hypoalbuminemia reflects persistent systemic inflammation and blood-brain barrier disruption, while lymphopenia indicates immunosuppression and impaired microbial clearance^4–6. Serum albumin, an endogenous neuroprotective agent, binds endotoxins and free radicals that contribute to neuroinflammation, lower serum albumin levels were independently associated with increased 90-day mortality in sepsis-associated encephalopathy patients⁷.In patients with delirium, higher albumin levels were associated with shorter hospital stays⁸. Concurrently, lymphocyte depletion correlates with secondary infections and failure to resolve systemic inflammation, both known drivers of SAE progression^9,10.

This dual biomarker synergy explains PNI's may be superior prognostic performance over isolated measures. We hypothesize that the PNI demonstrates superior prognostic accuracy compared to the conventional SOFA or SAPS II score in predicting prognosis of SAE patients. In this multicenter study, we evaluate the prognostic value of the PNI for 28-day mortality in patients with SAE, and independent validation cohorts validate the results of the exploration.

2 Materials and methods

2.1 Study Settings

This research leveraged the open-source medical information from the Medical Information Mart for Intensive Care IV (MIMIC-IV 2.0) database and multicenter database electronic Intensive Care Unit Collaborative Research Database (eICU-CRD v2.0). MIMIC-IV is a comprehensive repository that encompasses patient data from Beth Israel Deaconess Medical Center, spanning the years 2008 to 2019¹¹. The eICU-CRD aggregates ICU admission data from 208 U.S. hospitals during 2014–2015, representing a large-scale multicenter cohort of critically ill patients. The study protocol was reviewed and approved by the Institutional Review Board (IR No. 33690380), with all researchers completing mandatory human subjects protection training through the Collaborative Institutional Training Initiative program. The above databases have been approved by the Massachusetts Institute of Technology Review Committee. The raw data were extracted by employing structure query language (SQL) with Navicat and further processed using the R software.

2.2 Patients

All enrolled patients met the Sepsis-3 diagnostic criteria¹². SAE was defined by GCS < 15 or delirium at the presence of sepsis^13–16. For patients undergoing sedation or surgery, GCS scores before sedation or surgery were extracted. We excluded patients with: (1) primary neurological conditions that could independently impair consciousness (traumatic brain injury, acute ischemic/hemorrhagic stroke, active epilepsy, or intracranial infection); (2) pre-existing severe organ dysfunction (Child-Pugh class C liver cirrhosis or end-stage renal disease requiring dialysis); (3) acute life-threatening comorbidities (post-cardiac arrest status); (4) substance abuse disorders (chronic alcoholism or illicit drug use documented in medical records); (5) severe metabolic derangements (hyponatremia < 120 mmol/L, persistent hyperglycemia > 180 mg/dL, or hypoglycemia < 54 mg/dL despite correction); (6) insufficient observation time (ICU death/discharge within 24 hours of admission); (7) incomplete neurological assessments (missing GCS documentation); (8) unavailable nutritional-inflammatory data (missing albumin or lymphocyte counts for PNI calculation); and (9) age < 18 years.

2.3 Data Collection

We collected comprehensive clinical data, including: (1) demographic characteristics (age and sex); (2) 28-day mortality outcomes; (3) comorbidities classified using the International Classification of Diseases, Ninth Revision (ICD-9) criteria; (4) mean vital sign measurements prior to SAE diagnosis; (5) mean laboratory measurements prior to SAE diagnosis, first available laboratory results following ICU admission; and (6) infection sites and causative microorganisms. Disease severity was assessed using standardized scoring systems recorded within 24 hours of ICU admission: the Simplified Acute Physiology Score II (SAPS II), Logistic Organ Dysfunction System (LODS), Systemic Inflammatory Response Syndrome (SIRS) criteria, Sequential Organ Failure Assessment (SOFA) score, Charlson Comorbidity Index (CCI), and Glasgow Coma Scale (GCS).

2.4 Statistical Analysis

Continuous variables were expressed as mean ± standard deviation or median (interquartile range), and categorical variables as frequencies (percentages). Multivariable Logistic regression analysis was employed to evaluate independent predictors of 28-day mortality in SAE patients, adjusting for potential confounding variables. A GAM was applied to explore linear associations and identify optimal thresholds. KM survival curves were generated to compare 28-day survival probabilities between PNI-stratified groups in both MIMIC and eICU cohorts. ROC curves were constructed to assess the predictive performance of PNI for mortality. Boxplot analysis was conducted to examine the relationship between PNI values and GCS scores. All statistical analyses were performed using R software, with P < 0.05 considered statistically significant.

3 Results

3.1 Baseline characteristics

A total of 3,202 patients with SAE were included based on predefined criteria (Supplementary Fig. 1). Participants were stratified into non-survivors (n = 1,062) and survivors (n = 2,140) according to 28-day mortality. Non-survivors were significantly older and demonstrated profoundly impaired nutritional-inmunological status, characterized by substantially lower serum albumin, reduced lymphocyte counts, and critically depressed PNI levels. Furthermore, non-survivors exhibited exacerbated organ dysfunction and higher disease severity, evidenced by markedly elevated SOFA, SAPS II, and LODS scores compared to survivors. All baseline variables are presented in Table 1.

Table 1
Baseline characteristics and outcomes of sepsis-associated encephalopathy patients
	Original cohort			Match cohort
	Survival group (n = 2140)	Non-survival group (n = 1062)	P	Survival group (n = 965)	Non-survival group (n = 965)	P
Baseline variables
Age (years) (median[IQR])	71.00 [59.00,81.00]	68.00 [59.00, 77.00]	< 0.001	69.00 [58.00, 78.00]	68.00 [60.00, 78.00]	0.885
Gender,M (%)	1205 (56.3)	681 ( 64.1)	< 0.001	594 (61.6)	611 (63.3)	0.452
Laboratory parameters (median [IQR])
Albumin (g/dL)	3.50 [2.90,3.90]	1.60 [1.10,2.10]	< 0.001	3.50 [2.90,4.00]	1.60 [1.10, 2.10]	< 0.001
Lymphocyte (× 10⁹/L)	1.85 [1.42, 2.42]	1.13 [0.73, 1.66]	< 0.001	1.79 [1.38, 2.42]	1.13 [0.73, 1.64]	< 0.001
PNI	44.15 [38.60,49.56]	22.43 [17.85, 28.89]	< 0.001	44.75 [38.70, 50.42]	22.40 [17.85,29.40]	< 0.001
Critical illness score(median [IQR])
CCI	5.00 [3.00, 7.00]	5.00 [3.00, 6.00]	0.055	5.00 [3.00,7.00]	5.00 [3.00, 6.00]	0.055
GCS	8.00 [6.00, 9.00]	3.00 [3.00,3.00]	< 0.001	8.00 [6.00, 9.00]	3.00 [3.00, 3.00]	< 0.001
SOFA	6.00 [4.00, 8.00]	7.00 [6.00, 9.00]	< 0.001	6.00 [4.00, 8.00]	7.00 [6.00, 9.00]	< 0.001
SAPS II	41.00 [32.75, 51.00]	54.00 [40.00, 63.00]	< 0.001	42.00 [33.00, 51.00]	54.00 [40.00, 63.00]	< 0.001
LODS	5.00 [3.00, 8.00]	8.00 [6.00, 9.00]	< 0.001	6.00 [4.00, 8.00]	7.00 [6.00, 9.00]	< 0.001
SIRS	3.00 [2.00, 3.00]	3.00 [2.00, 3.00]	0.005	3.00 [2.00, 3.00]	3.00 [2.00, 3.00]	0.121
Clinical outcome
Ventdurations (n(%))	1379 (64.4)	924 ( 87.0)	< 0.001	839 (86.9)	829 ( 85.9)	0.550
Vasopressin (n(%))	214 (10.0)	172 ( 16.2)	< 0.001	153 (15.9)	151 ( 15.6)	0.950
Length of stay (median [IQR])	3.90 [1.70, 9.20]	5.30 [2.30, 10.10]	< 0.001	3.80 [1.50, 10.30]	5.20 [2.30, 10.20]	< 0.001
PNI = 10 × Albumin (g/dL) + 5 × Lymphocyte (× 10⁹/L); CCI: Charlson Comorbidity Index; GCS: Glasgow Coma Scale; SOFA: Sequential Organ Failure Assessment; SAPS II: Simplified Acute Physiology Score II; LODS: Logistic Organ Dysfunction System; SIRS: Systemic Inflammatory Response Syndrome

3.2 Association between PNI and 28-day mortality in SAE

The GAM demonstrated a significant linear relationship between PNI and 28-day mortality in SAE (Fig. 1). A knot value of about 34 was identified, with mortality risk decreasing significantly when PNI exceeded this threshold (Fig. 1). Multivariate Logistic regression analysis demonstrated that higher PNI was significantly associated with reduced 28-day mortality in SAE patients (adjusted OR: 0.85, 95% CI: 0.77–0.93, P < 0.001). Consistent findings were observed across alternative analytical approaches: propensity score-matched analysis (OR: 0.87, 95% CI: 0.81–0.93, P < 0.001), IPW (OR: 0.82, 95% CI: 0.78–0.86, P < 0.001), and doubly robust estimation (OR: 0.81, 95% CI: 0.76–0.86, P < 0.001) (Table 2).

Fig. 1

Linear association between PNI and 28-day mortality in SAE. SAE: sepsis-associated encephalopathy

Table 2
Association between Prognostic Nutritional Index and 28-day mortality in sepsis-associated encephalopathy
Models	OR	CI		P
Models	OR	2.5%	97.5%	P
Multivariate Logistic analysis*	0.85	0.77	0.80	< 0.001
Propensity score matching*	0.77	0.81	0.88	< 0.001
Inverse probability weighting*	0.80	0.78	0.81	< 0.001
Doubly robust with all coariates#	0.10 exp(coef) [confint] p	0.09 exp(coef) [confint] p	0.12 exp(coef) [confint] p	< 0.001 exp(coef) [confint] p
*Analysis was conducted using the continuous variable of PNI# The doubly robust method requires converting the continuous PNI into a binary categorical variable using a predefined cut-off value for analysis; OR: odds ratio; CI: confidence interval; P < 0.05, statistically significant.

3.3 Predictive performance of prognostic models for 28-day mortality in

SAE

Kaplan-Meier survival analysis demonstrated significantly prolonged 28-day survival in SAE patients with PNI > 34 compared to those with PNI ≤ 34 in both MIMIC (P < 0.001) and EICU (P < 0.001) cohorts (Fig. 2). Receiver operating characteristic (ROC) curve analysis demonstrated that the PNI exhibited excellent discriminative ability for predicting 28-day mortality in SAE patients, with area under the curve (AUC) values of 0.879 (95% CI: 0.867–0.891) in the MIMIC IV database and 0.724 (95% CI: 0.707–0.741) in the eICU database (Fig. 3). Model 4 (PNI + SOFA + SAPS II) showed superior predictive performance, with AUCs of 0.907 (MIMIC) and 0.766 (EICU).

Fig. 2

KM survival curves of 28-day mortality in SAE patients and PNI

levels. Patients were stratified into two groups based on the PNI, GROUP1:

PNI ≤ 34; GROUP2: PNI > 34. Figure 2A: Analysis of the MIMIC-IV

cohort; Fig. 2B: Analysis of the eICU-CRD cohort. KM: Kaplan-Meier; SAE: sepsis-associated encephalopathy; PNI: Prognostic Nutritional Index; MIMIC-IV: Medical Information Mart for Intensive Care IV; eICU-CRD: electronic

Intensive Care Unit Collaborative Research Database

Fig. 3

Predictive performance of prognostic models for 28-day mortality in SAE. Model 1 = PNI; Model 2 = SOFA; Model 3 = SAPS II; Model 4 =

PNI + SOFA + SAPS II. Figure 3A: ROC curves for predicting 28-day mortality in SAE of MIMIC-IV cohort; Fig. 3B: ROC curves for predicting 28-day mortality in SAE of eICU-CRD cohort; Fig. 3C: Comparison of predictive

performance for 28-day mortality in SAE of MIMIC-IV cohort; Fig. 3D:

Comparison of predictive performance for 28-day mortality in SAE of

eICU-CRD cohort. SAE: sepsis-associated encephalopathy; PNI: Prognostic

Nutritional Index; SOFA: Sequential Organ Failure Assessment; SAPS II:

Simplified Acute Physiology Score; ROC: receiver operating characteristic;

MIMIC-IV: Medical Information Mart for Intensive Care IV; eICU-CRD:

electronic Intensive Care Unit Collaborative Research Database.

3.4 The relationship between PNI and GCS score

Boxplot analysis revealed a significant inverse correlation between PNI levels and GCS severity categories ( GCS grade 1: GCS score 3–8; GCS grade 2༚ GCS score 9–12; GCS grade 3༚GCS score 13–14) in both the MIMIC-IV database (P < 0.001) and eICU database (P < 0.001) cohorts (Fig. 4). The PNI values progressively decreased with increasing GCS severity: 38.2 (34.5–42.1) in grade 1 versus 45.6 (41.2–49.8) in grade 3 (P < 0.001).

Fig. 4

The relationship between PNI and GCS grade. GCS grade 1: GCS score 3–8; GCS grade 2༚ GCS score 9–12; GCS grade 3༚ GCS score 13–14. Figure 4A: Analysis of the MIMIC-IV cohort; Fig. 4B: Analysis of the

eICU-CRD cohort.

PNI: Prognostic Nutritional Index; GCS: Glasgow Coma Scale; MIMIC-IV: Medical Information Mart for Intensive Care IV; eICU-CRD: electronic Intensive Care Unit Collaborative Research Database.

4 Discussion

Our findings demonstrate that SAE patients had a 28-day mortality rate of 33.2%. The PNI as a significant predictor of 28-day mortality of SAE superior than SOFA score and SAPS II score, with PNI levels below 34 showing a linear correlation with increased mortality risk. Stratified analysis show that progressively lower PNI levels were associated with worsening degrees of consciousness impairment.

The multicenter analysis demonstrates unabated high 28-day mortality rates among SAE patients, consistent with established clinical evidence^2,3. Despite substantial research efforts, the current lack of sensitive prognostic biomarkers impedes early risk stratification and targeted interventions for SAE. Our findings reinforce the critical need to identify reliable indicators that can dynamically predict outcomes and guide therapeutic escalation in this vulnerable population.

We established the PNI as a novel independent predictor of SAE mortality, outperforming conventional severity scores (SOFA/SAPS II) in both discrimination and clinical utility. The optimal threshold (PNI = 34) robustly stratified mortality risk across cohorts. The mechanisms underlying PNI's predictive value in SAE likely involve dual pathways: (1) Albumin were associated with SAE and were supported by medium- to high-quality evidence⁶. Use of albumin decreased the risk of sepsis-associated delirium⁵. Hypoalbuminemia exacerbates blood-brain barrier disruption via oxidative stress and endothelial dysfunction (2) Lymphocytopenia reflects impaired immunomodulation, worsening neuroinflammation. The lymphocyte population comprises three principal immunophenotypically distinct subsets: T lymphocytes (T cells), B lymphocytes (B cells), and natural killer cells (NK cells). T cells constitute a major subset of lymphocytes, representing one of the principal cellular components within the lymphoid lineage. V-domain immunoglobulin suppressor of T cell activation (VISTA) has emerged as a crucial player in the pathogenesis of neurological disorders¹⁷. CD86 in CD3 + CD56 + natural killer T (NKT) cells is an independent risk factor of SAE¹⁸. Mechanistically, PNI integrates hypoalbuminemia-driven blood-brain barrier disruption and lymphopenia-mediated neuroinflammation – core pathways in SAE pathogenesis^19–23. We therefore recommend serial PNI monitoring in SAE management, with prompt nutritional-immunomodulatory therapy when PNI falls below 34.

Stratified analysis demonstrated a significant correlation between higher PNI and improved GCS scores indicating PNI's capacity to reflect SAE-induced consciousness impairment severity. This association may be attributed to the potential roles of albumin in attenuating oxidative neuronal injury and lymphocytes in controlling CNS infections, with both processes possibly influencing neurological recovery pathways. Clinicians should utilize PNI trends to anticipate neurological trajectory and personalize neuroprotective strategies (e.g., antioxidant supplementation, infection source control) in comatose SAE patients.

6. Limitations

Several study limitations warrant acknowledgment. First, the retrospective design inherently carries potential selection bias, notwithstanding comprehensive statistical adjustment. Second, while validated across two independent databases, the generalizability of the established PNI threshold (34) requires prospective verification. Third, residual confounding from unmeasured variables (e.g., nutritional supplementation regimens, underlying comorbidities) may potentially influence clinical outcomes. Finally, the pathophysiological mechanisms mediating the observed PNI-GCS association remain hypothetical and necessitate further mechanistic investigation.

7. Conclusions

This multicenter study identified the PNI as a robust and independent predictor of 28-day mortality in patients with SAE, demonstrating an optimal prognostic threshold of 34. The consistent predictive performance across diverse analytical methodologies and independent validation cohorts substantiates its clinical utility for risk stratification in SAE management.

List of abbreviations:

CCI

Charlson Comorbidity Index

confidence interval

eICU-CRD

electronic Intensive Care Unit Collaborative Research Database

GAM

generalized additive model

GCS

Glasgow Coma Scale

INR

international normalized ratio

IPW

inverse probability weighting

Kaplan-Meier

LODS

Logistic Organ Dysfunction System

MIMIC-IV

Medical Information Mart for Intensive Care IV

odds ratio

PNI

Prognostic Nutritional Index

PSM

propensity score matching

prothrombin time

PTT

partial thromboplastin time

ROC

receiver operating characteristic

SAE

sepsis-associated encephalopathy

SAPS II

Simplified Acute Physiology Score II

SIRS

Systemic Inflammatory Response Syndrome

SMD

standardized mean differences

SOFA

Sequential Organ Failure Assessment

SQL

Structured Query Language

Declarations

Ethics approval and consent to participate:

The use of the MIMIC-IV and eICU-CRD databases was approved by the institutional review boards (IRBs) of the Massachusetts Institute of Technology (MIT) and Beth Israel Deaconess Medical Center (BIDMC).

All methods were carried out in accordance with the Declaration of Helsinki and relevant institutional guidelines and regulations.

Informed consent was waived by the IRBs of MIT and BIDMC because the study did not impact clinical care, and all protected health information was de-identified.

Clinical Trial

Not applicable

Consent for Publication:

The authors confirm that they have reviewed and approved the final version of the manuscript and consent to its publication in BMC Neurolpgy.

Availability of data and materials:

If the reason is reasonable, the original data can be requested from the corresponding author.

Conflict of Interest:

The authors declare no competing interests.

Funding:

This work was supported by Joint Funds of the Natural Science Foundation of Tianjin (No. 25JCLMJC00350).

Author Contribution

Dr LNZ had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.Concept and design: LNZ, CQ, KLX.Acquisition, analysis, or interpretation of data: All authors.Drafting of the manuscript: LNZ, CQ, KLX.Critical review of the manuscript for important intellectual content: All authors.Data collection and statistical analysis: CQ, YHS, HYL, XGL, DXH, QHY.Obtained funding: LNZ.Administrative, technical, or material support: YL, YHS, HYL, XGL, DXH, QHY.Supervision: YL, KLX.

Concept and design: LNZ, CQ, KLX.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: LNZ, CQ, KLX.

Critical review of the manuscript for important intellectual content: All authors.

Data collection and statistical analysis: CQ, YHS, HYL, XGL, DXH, QHY.

Obtained funding: LNZ.

Administrative, technical, or material support: YL, YHS, HYL, XGL, DXH, QHY.

Supervision: YL, KLX.

Acknowledgments

Not applicable

Electronic Supplementary Material

Below is the link to the electronic supplementary material

Supplementary Material 1

Reference

Song Z, Chen H, Xu W, et al. The hexapeptide functionalized gold nanoparticles protect against sepsis-associated encephalopathy by forming specific protein corona and regulating macrophage activation. Mater Today Bio. 2025;32:101704.

Chen L, Luo S, Liu T, et al. Growth differentiation factor 15 aggravates sepsis-induced cognitive and memory impairments by promoting microglial inflammatory responses and phagocytosis. J Neuroinflammation. 2025;22(1):44.

Zeng QQ, Wang J, Yue RC, et al. Gelsevirine ameliorates sepsis-associated encephalopathy by inhibiting the STING signalling-mediated pyroptosis pathway in microglia. Phytomedicine. 2024;135:156071.

Hotchkiss RS, Monneret G, Payen D. Sepsis-induced immunosuppression: from cellular dysfunctions to immunotherapy. Nat Rev Immunol. 2013;13(12):862–74.

Kim Y, Jin Y, Jin T, Lee SM, et al. Risk factors and outcomes of sepsis-associated delirium in intensive care unit patients: A secondary data analysis. Intensive Crit Care Nurs. 2020;59:102844.

Zhang Z, Guo L, Jia L, et al. Factors contributing to sepsis-associated encephalopathy: a comprehensive systematic review and meta-analysis. Front Med (Lausanne). 2024;11:1379019.

Wang G, Gao Y, Fu Y, et al. A simple nomogram for predicting the mortality of PICU patients with sepsis-associated encephalopathy: a multicenter retrospective study. Front Neurol. 2024;15:1418405.

Chi R, Perkins AJ, Khalifeh Y, et al. Serum Albumin Level at Intensive Care Unit Admission and Delirium Duration and Severity in Critically Ill Adults. Am J Crit Care. 2024;33(6):412–20.

Sun Y, Lu Y, Pan X, et al. Early B lymphocyte subsets in blood predict prognosis in sepsis. Front Immunol. 2024;15:1437864.

10.

Tomino A, Tsuda M, Aoki R, et al. Increased PD-1 Expression and Altered T Cell Repertoire Diversity Predict Mortality in Patients with Septic Shock: A Preliminary Study. PLoS ONE. 2017;12(1):e0169653.

11.

Shi X, Xu L, Ren J, et al. Triglyceride-glucose index: a novel prognostic marker for sepsis-associated encephalopathy severity and outcomes. Front Neurol. 2025;16:1468419.

12.

Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801–10.

13.

Sonneville R, de Montmollin E, Poujade J, et al. Potentially modifiable factors contributing to sepsis-associated encephalopathy. Intensive Care Med. 2017;43(8):1075–84.

14.

Iwashyna TJ, Ely EW, Smith DM, et al. Long-term cognitive impairment and functional disability among survivors of severe sepsis. JAMA. 2010;304(16):1787–94.

15.

Yang Y, Liang S, Geng J, et al. Development of a nomogram to predict 30-day mortality of patients with sepsis-associated encephalopathy: a retrospective cohort study. J Intensive Care. 2020;8:45.

16.

Zhao L, Li Y, Wang Y, et al. Development and Validation of a Nomogram for the Prediction of Hospital Mortality of Patients With Encephalopathy Caused by Microbial Infection: A Retrospective Cohort Study. Front Microbiol. 2021;12:737066.

17.

Xu Y, Zhu Y, Shi Y, et al. Immune Checkpoint VISTA Negatively Regulates Microglia Glycolysis and Activation via TRIM28-Mediated Ubiquitination of HK2 in Sepsis-Associated Encephalopathy. Mol Neurobiol. 2025;62(4):4452–65.

18.

Chen SL, Liu XY, Huang JH, et al. The expression of CD86 in CD3(+)CD56(+) NKT cells is associated with the occurrence and prognosis of sepsis-associated encephalopathy in sepsis patients: a prospective observational cohort study. Immunol Res. 2023;71(6):929–40.

19.

Saito M, Fujinami Y, Ono Y, et al. Infiltrated regulatory T cells and Th2 cells in the brain contribute to attenuation of sepsis-associated encephalopathy and alleviation of mental impairments in mice with polymicrobial sepsis. Brain Behav Immun. 2021;92:25–38.

20.

Zhang Y, Xin J, Zhao D, et al. Magnesium hexacyanoferrate mitigates sepsis-associated encephalopathy through inhibiting microglial activation and neuronal cuproptosis. Biomaterials. 2025;321:123279.

21.

Ren C, Yao RQ, Zhang H, et al. Sepsis-associated encephalopathy: a vicious cycle of immunosuppression. J Neuroinflammation. 2020;17(1):14.

22.

Heming N, Mazeraud A, Verdonk F, et al. Neuroanatomy of sepsis-associated encephalopathy. Crit Care. 2017;21(1):65.

23.

Luo RY, Luo C, Zhong F, et al. ProBDNF promotes sepsis-associated encephalopathy in mice by dampening the immune activity of meningeal CD4(+) T cells. J Neuroinflammation. 2020;17(1):169.

Captions.

Table 1 Baseline characteristics and outcomes of sepsis-associated encephalopathy patients

Table 2 Association between PNI and 28-day mortality in SAE. SAE: sepsis-associated encephalopathy; PNI: Prognostic Nutritional Index

Supplementary Material

eTable 1 Baseline characteristics of sepsis-associated encephalopathy patients

eTable 2 Multivariable Logistic analysis of factors associated with 28-day mortality in patients with sepsis-associated encephalopathy

eFigure 1 Flow chart for patient selection. MIMIC-IV: Medical Information Mart for Intensive Care IV; eICU-CRD: electronic Intensive Care Unit

Collaborative Research Database

eFigure 2 The SMD of propensity-matched. SMD: standardized mean differences

eFigure 3 Forest plot of multivariable Logistic regression analysis influencing the primary outcomes of SAE patients. The OR point estimates for each predictor variable are represented by red dots. Horizontal lines denote the corresponding 95% CI. SAE: sepsis-associated encephalopathy; OR: odds ratio; CI: confidence interval.

Yes