Abstract
Introduction
The Neutrophil to Albumin Ratio (NAR) stands out as a novel integrative inflammatory biomarker that is associated with the occurrence of stroke-associated infection (SAI). Our objective is to comprehensively investigate the predictive capacity of NAR regarding the incidence of SAI and poor functional outcomes after spontaneous intracerebral hemorrhage (sICH).
Methods
We retrospectively enrolled 291 patients categorized into tertiles based on their NAR levels. Cox Proportional Hazards regressions, Kaplan-Meier curves with log-rank tests, and restricted cubic splines were utilized to elucidate the associations between NAR and SAI. Logistic regressions were used to investigate the associations between the poor functional outcome and NAR. To determine the role of SAI in the effect of NAR on the risk of poor functional outcome, mediation analyses were performed. The improvements in model discriminations and calibrations by incorporating NAR into ICH score and max-ICH score were comprehensively explored.
Results
With each standard deviation increase in NAR after full adjustments, the risks of SAI and poor functional outcome at 3 months rose by 43% and 76%, respectively. Furthermore, incorporating NAR into the ICH score and max-ICH score enhanced both the discrimination and calibration of model performance. Intriguingly, the proportion of the total effect mediated was 9.89% and 10.25% for poor functional outcome at 3 months (p = 0.010) and 1 year (p = 0.024).
Conclusion
Elevated NAR is a superior biomarker capable of predicting SAI and poor functional outcome after sICH. The association between NAR and the functional outcome is mildly yet significantly mediated by SAI.
Keywords: Spontaneous intracerebral hemorrhage, inflammation, infection, prognosis, neutrophil to albumin Ratio
Introduction
Spontaneous intracerebral hemorrhage (sICH) refers to the abrupt rupture of cerebral arteries, veins, or capillaries within the brain parenchyma without preceding trauma, resulting in the extravasation of blood into surrounding tissues [1]. sICH constitutes approximately 20% to 30% of all stroke subtypes in Asia [2], and is characterized by high mortality and disability rates [3,4]. In China, the disability-adjusted life years rate of sICH reached approximately 1500 per 100,000 population [5]. However, despite extensive efforts over the past decades, the therapeutic management aimed at advancing the prognosis of sICH patients remains limited. The significant disease burden caused by sICH underscores the urgency of identifying cost-efficient and applicable biomarkers, as well as establishing reliable prognostic models.
In particular, the occurrence of early complications has a substantial impact on patient outcome after stroke [6]. Stroke-associated infections (SAI), which encompass common subtypes such as stroke-associated pneumonia and urinary tract infections, are among the most prevalent and devastating complications of stroke, closely linked to short-term poor functional outcomes and mortality rates [7,8]. Indeed, prophylaxis or early treatment of these complications could substantially enhance outcomes by preventing life-threatening complications [9]. Predictions of these complications could enable physicians to promptly implement medical interventions and adjust treatment plans during the decision-making process.
Various grading scales currently exist for sICH, with the ICH score being the most widely used and validated [10]. This simple and accurate scale incorporates age, Glasgow Coma Scale (GCS) score, infratentorial ICH, hematoma volume (HV), and intraventricular hemorrhage (IVH), and has been validated to predict early functional outcomes in the Asian population [11,12]. However, there is potential for enhancement in this area. In 2017, Sembill et al. devised max-ICH score to provide severity assessment for functional outcomes based on the ICH score [13].
Neutrophils, as key mediators of systemic inflammation, have been closely associated with SAI and stroke prognosis [14]. Hypoalbuminemia, a classical biomarker of malnutrition and immune dysfunction, has also been reported to increase susceptibility to SAI, thereby promoting poor outcome in sICH patients [15]. The neutrophil to albumin ratio (NAR), a biomarker that integrates both systemic inflammation and nutritional status, has emerged as a novel prognostic indicator in various diseases, including cardiogenic shock [16], aneurysmal subarachnoid hemorrhage [17], and HBV-associated decompensated cirrhosis [18]. Recent studies have found that NAR or neutrophil percentage to albumin ratio (NPAR) serves as an effective predictor in sICH patients, showing associations with SAI, short-term poor functional outcomes, and mortality [19,20]. However, the association between NAR and long-term functional outcomes remains unclear, and the potential role of SAI in this association has not been explored. Moreover, it is yet to be determined whether incorporating NAR into established scoring systems, such as the ICH score and max-ICH score, provides incremental clinical utility.
Therefore, this study aims to (1) validate the prognostic significance of NAR for both SAI and short-term functional outcomes in patients with sICH; (2) assess the prognostic significance of NAR for long-term functional outcomes; (3) evaluate the additional predictive benefits of NAR when integrated into existing scoring systems; and (4) explore whether SAI mediates the effect of NAR on functional outcomes.
Materials and methods
Study population
We retrospectively enrolled sICH patients admitted in the Department of Neurology of the First Affiliated Hospital of Wenzhou Medical University from 2021.01.01, to 2021.12.31. This study was approved by the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University (KY2023-R123) and was conducted in compliance with the Declaration of Helsinki. Written informed consent was waived by our institutional ethics committee due to the retrospective nature of the study and the use of anonymized data.
The diagnosis was made in accordance with the diagnostic criteria established by the Cerebrovascular Disease Group of the Chinese Academy of Neurology and further confirmed via available computed tomography (CT) scans by clinicians.
The inclusion criteria were as follows: (1) patients diagnosed with sICH who did not undergo surgical intervention; (2) patients who had an initial CT scan performed within 72 h post-stroke; and 3) patients with well-documented electronic medical records. A total of 406 consecutive sICH patients were initially included in the study. The exclusion criteria were as follows: (1) patients with primary ventricular hemorrhage (PVH); (2) patients with hemorrhage resulting from secondary causes, including traumatic brain injury, cerebral aneurysm, arteriovenous malformation, brain tumors, oral anticoagulant or coagulopathy, or hemorrhagic infarction; 3) patients with severe hepatic or renal dysfunction; (4) patients with malignancy; (5) patients with autoimmune disease; (6) patients with an active infection or suggestive symptoms of infection before hospitalization [21]; (7) patients with a premorbid modified Rankin scale (mRS) ≥ 3; 8) patients missing 3-month mRS scores. Eventually, a total of 291 eligible sICH patients were ultimately included.
Patients were treated according to the Chinese Stroke Association guideline [22]. To be specific, for patients presenting SBP >150 mmHg and with no contraindications to acute anti-hypertensive therapy, the target is to reduce the SBP to less than 140 mmHg. Aggressive reduction of BP with continuous BP monitoring is considered if SBP >220 mmHg.
Data collection
Demographic and clinical data of patients about age, sex, medications (antiplatelet and anticoagulation), clinical characteristics including hypertension, diabetes, previous stroke, alcohol abuse, smoking, and blood pressure, were retrieved from the electronic medical records.
The severity of ICH on admission was assessed by experienced clinicians using National Institutes of Health Stroke Scale (NIHSS) scores and Glasgow Coma Scale (GCS). Head CT scan using 64-slice CT scanner was performed as a part of routine clinical examinations. Briefly, the hyperdense region on CT scan represents the hematoma, and the hypodense region surrounding the hematoma indicates perihematomal edema (PHE). Hematoma and total lesion (the hyperdense region + the hypodense region) volumes were calculated by ABC/2 method, where A represents the greatest hemorrhage or total diameter, B represents the diameter 90° to A, and C is the approximate number of CT slices multiplied by the slice thickness. PHE was then measured by subtracting the hyperdense volume (hematoma) from the total lesion area. ICH location was divided into lobar, deep, and infratentorial.
Blood samples were collected within the first 24 h after admission in EDTA tubes (for plasma) or vacutainer tubes (for serum) for each patient. The cell counts were measured by Automated Hematology Analyzer Sysmex XE-2100 (Sysmex, Kobe, Japan) while biochemical parameters including albumin were measured using Clinical Analyzer Beckman Coulter AU-5831 (Beckman Coulter, CA, USA).
The 3-month and 1-year mRS were assessed through phone interviews and poor functional outcome was defined as mRS > 2. The status of SAI, defined as an infection diagnosed by trained and experienced clinicians during the hospitalization period [7], was extracted from clinical records and reviewed by ZT and HJQ to confirm the infection diagnosis.
Statistical analysis
Statistics were performed with the SPSS software (version 25.0), R software (version 4.1.3), and Medcalc (version 20.0). Kolmogorov–Smirnov test was used to assess the distribution normality, and continuous variables with normal distribution were described as mean ± standard deviation (SD) while variables with non-normal distribution were described as median (interquartile range [IQR]). Two-sample t tests and one-way analyses of variance were performed to analyze the intergroup difference of normally distributed continuous variables. Mann–Whitney U-tests and Kruskal–Wallis tests were performed to analyze non-normally distributed continuous variables. Categorical variables were described as percentage numbers and analyzed by χ2 test. Subjects were grouped into tertiles according to their values of NAR. We used Mantel-Haenszel test to confirm the linearity between mRS as an ordinal variable and NAR tertiles. The relationship between NAR and other parameters was evaluated by Spearman’s correlation test.
Cox proportional hazards regressions were used to investigate the associations between the time from stroke onset to SAI and NAR. Patients who did not experience SAI were censored using their lengths of hospitalization. Schoenfeld residuals tests confirmed the proportional hazards assumption, and no relevant violations were discovered. Kaplan–Meier curves were used to describe the cumulative incidence with log-rank tests to compare the difference of the curves.
Logistic regressions were used to investigate the associations between the poor functional outcome and NAR. To better explain the results of regressions and avoid the limitations of the unit, NAR has been transformed into standardized continuous parameter to explore the change of OR and HR with per SD increase in NAR. To investigate the independent association between NAR and clinical outcomes, multivariable logistic regression models were constructed following a pre-specified hierarchical adjustment strategy, based on clinical relevance and established prognostic framework. In model 1, we adjusted for basic demographic confounders, including age and sex. In model 2, adjustments were further made for constituent variables of the ICH score (age, GCS score, infratentorial ICH, HV, and IVH). In model 3, we further adjusted for max-ICH score variables (age, NIHSS, HV, IVH and oral anticoagulation). Absolute values of each variable were used in the model. p trends were computed for categorical NAR in regressions.
Subsequently, restricted cubic splines (RCS) with 3 knots (at 10th, 50th, 90th percentiles) were plotted to examine the potential nonlinear associations between NAR and SAI. The median NAR was set as a reference (HR = 1.00). To determine the role of SAI in the effect of NAR on the risk of poor functional outcome, we carried out a mediation analysis using R mediation package to calculate the mediated proportion.
The improvements in model performance by adding NAR into ICH score and max-ICH score to predict poor functional outcome were assessed through a series of statistical indexes. Regarding discrimination performance, C-index, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated while the results of Hosmer-Lemeshow (H-L) test, Akaike information criterion (AIC) and Bayesian information (BIC) were applied to estimate model calibrations. Delong test was used to compare the differences between C-indices. A p value < 0.05 was regarded as statistically significant.
Results
Demographics, clinical characteristics, and radiological characteristics
The study retrospectively enrolled patients with sICH confirmed by CT scans from 2021.01.01 to 2021.12.31. A total of 406 eligible ICH patients were included, of whom 76 were excluded to ensure the validity and reliability of the study design. Among the remaining 330 patients, 39 were excluded due to lack of 3-month mRS scores. Ultimately, 291 sICH patients were included for analysis (Figure 1).
Figure 1.
A flowchart of patients’ selection.
The mean age of the participants was 62.45 years (SD = 12.29), with 71.13% being male. Among the included patients, 90 had poor functional outcomes at 3 months, 55 had poor functional outcomes at 1 year, and 71 were diagnosed with SAI. The overall mean time from stroke onset to SAI diagnosis was 7.58 days (SD = 3.95). The median level of NAR was 1.41 (IQR = 1.12–2.03) (Supplementary Table 1).
Subsequently, the subjects were divided into 3 tertiles based on their NAR levels (< 1.20, 1.20–1.76, and > 1.76). It was observed that baseline NIHSS, baseline GCS, ICH score, and max-ICH score differed significantly among the tertiles (p < 0.001 for all). Patients with higher tertiles of NAR were prone to have higher frequencies of SAI, poor functional outcome at 3 months, and poor functional outcome at 1 year (p < 0.001 for all) (Table 1).
Table 1.
Baseline demographics, clinical characteristics, and radiological characteristics stratified by NAR tertiles.
| T1 (< 1.20) | T2 (1.20–1.76) | T3 (> 1.76) | p-value | |
|---|---|---|---|---|
| n | 96 | 98 | 97 | |
| Age (years) | 60.54 ± 11.89 | 63.47 ± 13.12 | 63.31 ± 11.69 | 0.177 |
| Male (n%) | 67 (69.79) | 68 (69.39) | 72 (74.23) | 0.711 |
| Medications (n%) | ||||
| Antiplatelet | 4 (4.17) | 8 (8.16) | 5 (5.15) | 0.465 |
| Anticoagulation | 2 (2.08) | 4 (4.08) | 6 (6.19) | 0.358 |
| Clinical characteristics | ||||
| Hypertension (n%) | 76 (79.17) | 70 (71.43) | 75 (77.32) | 0.419 |
| Diabetes (n%) | 14 (14.58) | 20 (20.41) | 11 (11.34) | 0.207 |
| Previous stroke (n%) | 12 (12.50) | 17 (17.35) | 23 (23.71) | 0.125 |
| Alcohol abuse (n%) | 44 (45.83) | 35 (35.71) | 43 (44.33) | 0.303 |
| Smoking (n%) | 37 (38.54) | 35 (35.71) | 45 (46.39) | 0.290 |
| SBP (mmHg) | 154.88 ± 22.41 | 155 ± 23.65 | 159.94 ± 22.43 | 0.210 |
| DBP (mmHg) | 93 (78–102) | 87 (78–97) | 86 (79–100) | 0.224 |
| Baseline NIHSS | 4 (2–6) | 6 (3–11) | 9 (5–14) | < 0.001 |
| Baseline GCS | 15 (15-15) | 15 (15-15) | 15 (14–15) | < 0.001 |
| ICH score | 0 (0-0) | 0 (0–1) | 1 (0–1) | < 0.001 |
| max-ICH score | 1 (0–2) | 2 (0–3) | 2 (1–4) | < 0.001 |
| Radiological characteristics | ||||
| Time from onset to first imaging (h) | 9.5 (3–21) | 6 (3–14) | 5 (2–11) | 0.003 |
| HV (mL) | 4.59 (1.85–9.68) | 6.01 (2.31–15.10) | 12.79 (7.46–21.78) | < 0.001 |
| PHE (mL) | 2.25 (0.43–4.09) | 2.26 (0.62–5.74) | 4.32 (1.48–8.73) | < 0.001 |
| ICH location (n%) | ||||
| Lobar | 10 (10.42) | 12 (12.24) | 18 (18.56) | 0.226 |
| Deep | 74 (77.08) | 76 (77.55) | 72 (74.23) | 0.840 |
| Infratentorial | 12 (12.50) | 10 (10.21) | 7 (7.21) | 0.470 |
| IVH (n%) | 11 (11.46) | 24 (24.49) | 40 (41.24) | < 0.001 |
| SAI (n%) | 7 (7.29) | 22 (22.45) | 42 (43.30) | < 0.001 |
| Time from onset to SAI (days) | 8.28 ± 3.50 | 7.82 ± 3.3 | 6.65 ± 4.74 | 0.012 |
| Poor functional outcome at 3 months (n%) | 10 (10.42) | 28 (28.57) | 52 (53.61) | < 0.001 |
| Poor functional outcome at 1 year (n%) | 2 (2.08) | 17 (17.35) | 36 (37.11) | < 0.001 |
Abbreviations: ICH, intracerebral hemorrhage; mRS, modified rankin scale; SBP, systolic blood pressure; DBP, diastolic blood pressure; NIHSS, National Institutes of Health Stroke Scale; GCS, Glasgow Coma Scale; HV, hematoma volume; PHE, perihematomal edema; IVH, intraventricular hemorrhage; SAI, stroke-associated infection; NAR, neutrophil to albumin ratio.
Correlation analyses
Spearman’s correlation test revealed a positive correlation between NAR and SAI (rs = 0.36, p < 0.001), as well as with poor functional outcomes at 3 months (rs = 0.37, p < 0.001) (Figure 2A). Additionally, a linear trend, as assessed by the Mantel-Haenszel test, between NAR tertiles (1–3) and ordinal 3-month mRS levels (0–6) was observed (χ2 = 34.281, p < 0.001). The correlation demonstrated a moderate and positive association (rs = 0.30, p < 0.001) (Figure 2B).
Figure 2.
Correlation between NAR and other variables. (A) Correlation matrix of NAR and other variables ; (B) A graphical representation of the distribution of cases across NAR tertiles (1–3) and 3-month ordinal mRS levels (0–6). The size and color of bubbles indicates number of cases.
Predictive value of NAR for SAI and poor functional outcome
Univariate Cox regressions and logistic regressions indicated that high NAR (per SD increase) was associated with increased risk of SAI (crude HR: 1.69, 95% CI = 1.44–1.98, p < 0.001), poor functional outcome at 3 months (crude OR: 2.29, 95% CI = 1.68–3.11, p < 0.001), and poor functional outcome at 1 year (crude OR: 2.01, 95% CI = 1.48–2.73, p < 0.001) (Supplementary Table 2).
It was observed that individuals with higher NAR tertile exhibited an increased SAI risk during hospitalization (p < 0.001) (Figure 3A). Moreover, RCS after adjusting max-ICH score variables demonstrated a linear association of NAR with the risk of SAI during hospitalization (p overall association = 0.002, p nonlinearity = 0.127) (Figure 3B). In multivariable analyses, in the fully adjusted model 3 (adjusting for age, sex, and max-ICH score variables), the risk of SAI occurrence increased by 43% for per SD increase in NAR (HR: 1.43, 95% CI = 1.16–1.76). HR for tertile 3 was as high as 3.47 (95% CI = 1.43–8.46) compared to tertile 1, and a significant trend between categorical NAR and HR was found (p trend= 0.003) (Table 2).
Figure 3.
(A) Kaplan-Meier curves for SAI based on categorical NAR as tertiles; (B) Restricted cubic splines (RCS) adjusting for max-ICH score variables, the reference was set as NAR 1.41 (median).
Table 2.
Multivariable cox proportional hazards for SAI and multivariable logistic regressions for poor functional outcome.
| Categorical NAR |
Continuous NAR (per SD increase) | Continuous NAR (per unit increase) |
||||
|---|---|---|---|---|---|---|
| T1 | T2 | T3 | P trend | |||
| SAI | ||||||
| Unadjusted HR | Reference | 3.15 (1.35–7.38) | 6.89 (3.09–15.36) | < 0.001 | 1.69 (1.44–1.98) | 2.05 (1.64–2.55) |
| Model 1 HR | Reference | 2.76 (1.17–6.49) | 6.43 (2.88–14.33) | < 0.001 | 1.58(1.34–1.86) | 1.86 (1.49–2.33) |
| Model 2 HR | Reference | 2.41 (1.02–5.71) | 4.03 (1.70–9.55) | 0.005 | 1.46 (1.19–1.79) | 1.67 (1.27–2.21) |
| Model 3 HR | Reference | 2.07 (0.86–4.95) | 3.47 (1.43–8.46) | 0.003 | 1.43 (1.16–1.76) | 1.64 (1.23–2.17) |
| 3-month poor functional outcome | ||||||
| Unadjusted OR | Reference | 3.44 (1.56–7.56) | 9.94 (4.62–21.40) | < 0.001 | 2.29 (1.68–3.11) | 3.10 (2.04–4.72) |
| Model 1 OR | Reference | 3.20 (1.44– 7.10) | 9.88 (4.54–21.50) | < 0.001 | 2.33 (1.69–3.21) | 3.18 (2.06–4.93) |
| Model 2 OR | Reference | 2.56 (1.12– 5.82) | 5.82 (2.46–13.77) | < 0.001 | 1.82 (1.31–2.54) | 2.27 (1.45–3.57) |
| Model 3 OR | Reference | 1.54 (0.59–4.03) | 4.16 (1.54–11.25) | 0.003 | 1.76 (1.23–2.51) | 2.16 (1.32–3.53) |
| 1-year poor functional outcome | ||||||
| Unadjusted OR | Reference | 9.86 (2.21–43.98) | 27.74 (6.44–119.42) | < 0.001 | 2.01 (1.48–2.73) | 2.60 (1.71–3.96) |
| Model 1 OR | Reference | 8.44 (1.85–38.51) | 29.92 (6.75–132.57) | <0.001 | 2.00 (1.42–2.84) | 2.59 (1.61–4.17) |
| Model 2 OR | Reference | 6.40 (1.38–29.74) | 15.09 (3.18–71.59) | < 0.001 | 1.46 (1.03–2.08) | 1.68 (1.04–2.73) |
| Model 3 OR | Reference | 4.06 (0.80–20.56) | 9.90 (1.96–50.12) | 0.004 | 1.47 (1.00–2.18) | 1.70 (0.99–2.91) |
Model 1: NAR + age + sex.
Model 2: NAR + sex + ICH score variables (age, GCS score, infratentorial ICH, HV, and IVH, absolute values of each variable were used in the model).
Model 3: NAR + sex + max-ICH score variables (age, NIHSS on admission, HV, IVH and oral anticoagulation, absolute values of each variable were used in the model).
Abbreviations: SAI, stroke-associated infection; NAR, neutrophil to albumin ratio; SBP, systolic blood pressure; NIHSS, National Institutes of Health Stroke Scale; GCS, Glasgow Coma Scale; HV, hematoma volume; PHE, Perihematomal edema; IVH, intraventricular hemorrhage.
NAR demonstrated a significant predictive value for poor functional outcome. In model 3 adjusting for age, sex, and max-ICH score variables, the risk of poor functional outcome at 3 months increased by 76% for per SD increase in NAR (OR: 1.76, 95% CI = 1.23–2.51). In model 3, patients in the highest tertile had 9.90 times the risk of poor functional outcome at 1 year compared to those in tertile 1, and a significant trend between categorical NAR and OR was observed (p trend= 0.004).
Improvements of model performance with NAR
Regarding the discrimination of the predictive models, incorporating categorical NAR and continuous NAR (per SD increase) to the ICH score resulted in significant enhancements of the C-index for the prediction of poor functional outcomes at 3 months and 1 year. Moreover, both categorical NAR and continuous NAR demonstrated improvements in the NRI and IDI (Table 3).
Table 3.
Enhancements of model discrimination and calibration for poor functional outcome at 3-month and 1-year through integration of NAR.
| Discrimination |
Calibration |
||||||||
|---|---|---|---|---|---|---|---|---|---|
| C-index | p | NRI | p | IDI | p | H-L test, p | AIC | BIC | |
| 3 months poor functional outcome | |||||||||
| ICH score | 0.612 | Reference | Reference | Reference | Reference | Reference | 0.806 | 350.225 | 357.571 |
| ICH score + Categorical NAR | 0.739 | < 0.001 | 0.619 (0.389–0.848) | < 0.001 | 0.109 (0.075–0.143) | < 0.001 | 0.342 | 318.756 | 329.776 |
| ICH score + Continuous NAR (per SD increase) | 0.730 | < 0.001 | 0.663 (0.426–0.899) | < 0.001 | 0.090 (0.055–0.126) | < 0.001 | 0.823 | 327.288 | 338.308 |
| Max-ICH score | 0.800 | Reference | Reference | Reference | Reference | Reference | 0.293 | 286.004 | 293.351 |
| Max-ICH score + Categorical NAR | 0.823 | 0.175 | 0.491 (0.249–0.733) | < 0.001 | 0.050 (0.025–0.074) | < 0.001 | 0.120 | 272.293 | 283.313 |
| Max-ICH score + Continuous NAR (per SD increase) | 0.821 | 0.051 | 0.500 (0.259–0.740) | < 0.001 | 0.032 (0.011–0.053) | 0.003 | 0.739 | 278.856 | 289.876 |
| 1 year poor functional outcome | |||||||||
| ICH score | 0.699 | Reference | Reference | Reference | Reference | Reference | 0.730 | 255.63 | 262.97 |
| ICH score + Categorical NAR | 0.788 | < 0.001 | 0.719 (0.439–1.000) | < 0.001 | 0.079 (0.047–0.110) | < 0.001 | 0.665 | 231.99 | 243.01 |
| ICH score + Continuous NAR (per SD increase) | 0.774 | 0.004 | 0.615 (0.332–0.900) | < 0.001 | 0.072 (0.028–0.115) | 0.001 | 0.445 | 246.96 | 257.98 |
| Max-ICH score | 0.823 | Reference | Reference | Reference | Reference | Reference | 0.509 | 218.53 | 225.88 |
| Max-ICH score + Categorical NAR | 0.856 | 0.012 | 0.792 (0.517–1.067) | < 0.001 | 0.053 (0.023–0.082) | < 0.001 | 0.436 | 204.41 | 215.43 |
| Max-ICH score + Continuous NAR (per SD increase) | 0.840 | 0.259 | 0.408 (0.119–0.697) | 0.005 | 0.015 (0.005–0.034) | 0.140 | 0.232 | 216.25 | 227.27 |
Abbreviations: ICH, intracerebral hemorrhage; NAR, neutrophil to albumin ratio; mRS, modified Rankin Scale; NRI, net reclassification improvement; IDI, integrated discrimination improvement; H-L test, Hosmer-Lemeshow test; AIC, Akaike information criterion; BIC, Bayesian information criterion.
The C-index of the max-ICH score reached notable values of 0.800 at 3 months and 0.823 at 1 year. Augmenting this score with categorical NAR led to significant enhancements in the C-index for predicting poor functional outcomes at 1 year (0.823 vs. 0.856, p = 0.012).
In terms of predictive model calibration, the H-L tests produced non-significant statistics for all models listed in Table 3, indicating that these models were well calibrated and exhibited no departure from perfect fitting. Additionally, the calibration curves for all models closely followed the diagonal line, further confirming their good calibration (Figure 4). Furthermore, the AIC and BIC values almost all decreased after incorporating NAR into the ICH score and max-ICH score.
Figure 4.
(A and C) Calibration curves of ICH score, ICH score combined with categorical NAR, and ICH score combined with continuous NAR; (B and D) Calibration curves of max-ICH score, max-ICH score combined with categorical NAR, and max-ICH score combined with continuous NAR.
Mediation analyses
We hypothesized that NAR influenced functional outcomes through the mediation of SAI. Hence, a single-mediator mediation analysis was conducted. Upon including SAI as a mediator, we observed significant mediation effects on the relationship between NAR and poor functional outcomes. The proportion of the total effect mediated was 9.89% and 10.25% for poor functional outcome at 3 months (p = 0.010) and 1 year (p = 0.024), respectively (Figure 5).
Figure 5.
Mediation analyses of NAR for poor functional outcome through SAI. (A) Pathways of the single mediator model at 3 months. (B) Graphical summary of indirect, direct, and total effects along with their confidence intervals from the mediate function at 3 months. (C) Pathways of the single mediator model at 1 year. (D) Graphical summary of indirect, direct, and total effects along with their confidence intervals from the mediate function at 1 year.
Discussion
The estimated prevalence, incidence, and mortality rate of stroke in China in 2020 were 2.6%, 505.2 per 100 000 person-years, and 343.4 per 100 000 person-years, respectively, indicating the desperate need for an improved stroke prevention strategy in the general Chinese population [23]. Among stroke subtypes, ICH is the most dangerous with 1 year mortality rate reaching 17.9% [24]. Existing scoring systems including ICH score [12] and max-ICH score [25] have been widely used in the clinical practice.
In our current study, we found a significant association between NAR and short-term poor functional outcomes in sICH, which is consistent with previous investigations that have highlighted the prognostic value of NAR/NPAR [19,20]. Importantly, our study extends these findings by demonstrating that NAR is also associated with long-term functional outcomes, expanding its prognostic implication beyond prior studies. Moreover, our study provides novel insights by proving that SAI partially mediates the association between NAR and functional outcomes. In addition, we systematically assessed the incremental predictive value of incorporating NAR into established prognostic models, further supporting its clinical utility for risk stratification in sICH. With its cost-effectiveness and ease of acquisition, NAR may serve as a superior biomarker for physicians to promptly identify and stratify individuals at risk of both short-term and long-term poor functional outcomes, thus facilitating prevention and ultimately reducing the burden of disease.
Recent advancements suggests that early minimally invasive removal has significant benefits, with patients undergoing this procedure exhibiting better functional outcomes at 180 days compared to those receiving guideline-based medical management [26]. Additionally, the care bundle protocol, including early intensive lowering of systolic blood pressure, strict glucose control, and rapid reversal of warfarin-related anticoagulation within 1 h, has demonstrated considerable advantages [27]. Currently, not enough patients receive these interventions in the acute phase. Therefore, advocating for the widespread adoption of this early care bundle appears beneficial for ICH and should be implemented more broadly [28,29], and the elevated NAR may serve as an early indicator for timely adjustments in treatment strategies.
Our current study offers further support for the association between SAI and poor functional outcome, aligning with prior research findings. The negative impact of SAI on functional outcomes may be attributed in part to prolonged hospital stays, leading to immobility and general frailty that hinder rehabilitation efforts [7]. Additionally, SAI exacerbates the inflammatory process, directly impeding recovery. Furthermore, in clinical settings, there’s a risk of excessive antibiotic use in SAI treatment, a factor known to worsen prognosis for sICH patients [30]. Therefore, monitoring elevated SAI levels to identify the risk of SAI is crucial for early clinical intervention and preventing poor prognostic outcomes.
The underlying mechanism between NAR and poor outcome could be partially explained by the central nervous system (CNS)-peripheral immune interactions. Emerging evidence now suggests that ICH-induced brain injury triggers peripheral immune responses via the regulations of autonomic nervous system, neuroendocrine system, meningeal lymphatic vessels, and stroke-induced immune depression syndrome [31]. The peripheral inflammatory response, in turn, exacerbates neuroinflammation through a variety of pathways, contributing to the secondary brain injury [31]. This CNS-peripheral immune crosstalk not only contributes to infectious complications like SAI but also significantly impairs functional outcomes of stroke patients [32].
Neutrophils, as the initial leukocytes recruited from the peripheral blood to the brain, initiate peripheral inflammatory infiltration triggered by hematoma components [33]. This neutrophil infiltration exacerbates brain damage by releasing various pro-inflammatory mediators and reactive oxygen species, thereby contributing to the destruction of the blood-brain barrier [34]. Neutrophil-derived cytokines, such as IL-1β, can further exacerbate brain edema [35]. Additionally, certain inflammatory mediators exacerbate inflammation by recruiting monocytes or macrophages [36]. The association between serum albumin levels and poor functional outcomes may be explained by subclinical inflammation, as studies have demonstrated the correlation between low albumin levels and inflammation-associated cytokines like IL-6. The production of cytokines during inflammatory activation leads to a shift in protein synthesis in the liver from albumin to other acute-phase proteins [37,38]. Furthermore, the subsequent repression of serum albumin levels may impair the neuroprotective function of albumin, which typically helps reduce oxidative stress and neuronal apoptosis [39]. Moreover, in clinical practice, serum albumin is widely used as a diagnostic marker for malnutrition [40], which can lead to impaired immune responses and has been identified as an independent predictor of poor outcomes in patients with sICH [41,42].
There are several limitations in our study. First, the retrospective nature of our research precludes the establishment of any causal relationship. Second, the relatively small hematoma volumes observed in our study, which commonly results in mild neurological impairment, may confine the applications of our findings to individuals with small to medium-sized ICH. Thirdly, the limited number of cases prevented us from exploring subtypes of SAI, such as urinary tract infections. Additionally, although previous studies have indicated that the severity of infections could significantly influence long-term outcomes in sICH patients [43], we were unable to further investigate the impact of infection severity due to the lack of detailed records in our retrospective dataset. Prospective studies with a larger cohort of sICH patients, incorporating the severity and types of infections, are warranted to validate the relationship and dynamic changes between NAR, SAI, and clinical outcomes.
Conclusion
Elevated NAR is a superior biomarker capable of predicting SAI and poor functional outcome after sICH. NAR’s inclusion enhances the predictive performance of established compound scores such as the ICH score and max-ICH score. The association between NAR and the functional outcome is mildly yet significantly mediated by SAI.
Supplementary Material
Funding Statement
This study was funded by Zhejiang Provincial Science and Technology Innovation Program (New Young Talent Program) for College Students (No.2023R413015).
Ethics statement
This study was approved by the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University (KY2023-R123) and was conducted in compliance with the Declaration of Helsinki. Written informed consent was waived by our institutional ethics committee due to the retrospective nature of the study and the use of anonymized data.
Disclosure statement
The authors declare that they have no conflicts of interest with respect to the research, authorship, and/or publication of this manuscript. No financial or personal relationships with individuals or organizations have influenced or could be perceived to have influenced the work presented in this paper.
Data availability statement
The raw data supporting the conclusions of this study is available from the corresponding author on reasonable request.
References
- 1.Sheth KN. Spontaneous intracerebral hemorrhage. N Engl J Med. 2022;. 387(17):1589–1596. doi: 10.1056/NEJMra2201449. [DOI] [PubMed] [Google Scholar]
- 2.Tschoe C, Bushnell CD, Duncan PW, et al. Neuroinflammation after intracerebral hemorrhage and potential therapeutic targets. J Stroke. 2020;22(1):29–46. doi: 10.5853/jos.2019.02236. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Qureshi A, Mendelow A, Hanley DJL.. Intracerebral Haemorrhage. Lancet. 2009;373(9675):1632–1644. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Schrag M, Kirshner HJ.. Management of intracerebral hemorrhage: JACC focus seminar. J Am Coll Cardiol. 2020;75(15):1819–1831. doi: 10.1016/j.jacc.2019.10.066. [DOI] [PubMed] [Google Scholar]
- 5.Ma Q, et al. Temporal trend and attributable risk factors of stroke burden in China, 1990-2019: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2021;6(12):e897–e906. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Grube MM, Koennecke H-C, Walter G, et al. Influence of acute complications on outcome 3 months after ischemic stroke. PLoS One. 2013;8(9):e75719. doi: 10.1371/journal.pone.0075719. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Suda S, Aoki J, Shimoyama T, et al. Stroke-associated infection independently predicts 3-month poor functional outcome and mortality. J Neurol. 2018;265(2):370–375. -doi: 10.1007/s00415-017-8714-6. [DOI] [PubMed] [Google Scholar]
- 8.Aslanyan S, Weir CJ, Diener H-C, et al. Pneumonia and urinary tract infection after acute ischaemic stroke: a tertiary analysis of the GAIN International trial. Eur J Neurol. 2004;11(1):49–53. -doi: 10.1046/j.1468-1331.2003.00749.x. [DOI] [PubMed] [Google Scholar]
- 9.Bustamante A, García-Berrocoso T, Rodriguez N, et al. Ischemic stroke outcome: a review of the influence of post-stroke complications within the different scenarios of stroke care. Eur J Intern Med. 2016;29:9–21. doi: 10.1016/j.ejim.2015.11.030. [DOI] [PubMed] [Google Scholar]
- 10.Hemphill JC, Greenberg SM, Anderson CS, et al. Guidelines for the management of spontaneous intracerebral hemorrhage: a guideline for healthcare professionals from the American heart Association/American Stroke Association. Stroke. 2015;46(7):2032–2060. -doi: 10.1161/STR.0000000000000069. [DOI] [PubMed] [Google Scholar]
- 11.Jamora RDG, Kishi-Generao EM, Bitanga ES, et al. The ICH score: predicting mortality and functional outcome in an Asian population. Stroke. 2003;34(1):6–7. doi: 10.1161/01.str.0000047847.18178.d3. [DOI] [PubMed] [Google Scholar]
- 12.Hemphill JC, Bonovich DC, Besmertis L, et al. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke. 2001;32(4):891–897.): -doi: 10.1161/01.str.32.4.891. [DOI] [PubMed] [Google Scholar]
- 13.Sembill JA, Gerner ST, Volbers B, et al. Severity assessment in maximally treated ICH patients: the max-ICH score. Neurology. 2017;89(5):423–431. doi: 10.1212/WNL.0000000000004174. [DOI] [PubMed] [Google Scholar]
- 14.Deng Q-W, Gong P-Y, Chen X-L, et al. Admission blood cell counts are predictive of stroke-associated infection in acute ischemic stroke patients treated with endovascular therapy. Neurol Sci. 2021;42(6):2397–2409. doi: 10.1007/s10072-020-04827-2. [DOI] [PubMed] [Google Scholar]
- 15.Morotti A, Marini S, Lena UK, et al. Significance of admission hypoalbuminemia in acute intracerebral hemorrhage. J Neurol. 2017;264(5):905–911. doi: 10.1007/s00415-017-8451-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Peng Y, Xue Y, Wang J, et al. Association between neutrophil-to-albumin ratio and mortality in patients with cardiogenic shock: a retrospective cohort study. BMJ Open. 2020;10(10):e039860. doi: 10.1136/bmjopen-2020-039860. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zhang X, Liu Y, Zhang S, et al. Neutrophil-to-albumin ratio as a biomarker of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. World Neurosurg. 2021;147:e453–e458. doi: 10.1016/j.wneu.2020.12.084. [DOI] [PubMed] [Google Scholar]
- 18.Han Z, He X, Peng S.. Neutrophil count to albumin ratio as a prognostic indicator for HBV-associated decompensated cirrhosis. J Clin Lab Anal. 2021;35(4):e23730. doi: 10.1002/jcla.23730. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lv X-N, Shen Y-Q, Li Z-Q, et al. Neutrophil percentage to albumin ratio is associated with stroke-associated pneumonia and poor outcome in patients with spontaneous intracerebral hemorrhage. Front Immunol. 2023;14:1173718. doi: 10.3389/fimmu.2023.1173718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wan J, Zhang S, Yuan X, et al. Neutrophil-to-albumin ratio as a predictor of mortality in patients with intracerebral hemorrhage: a multicenter retrospective cohort study. BMC Neurol. 2025;25(1):209.): doi: 10.1186/s12883-025-04220-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Schweizer J, Bustamante A, Lapierre-Fétaud V, et al. SAA (serum amyloid A): a novel predictor of stroke-associated infections. Stroke. 2020;51(12):3523–3530. -doi: 10.1161/STROKEAHA.120.030064. [DOI] [PubMed] [Google Scholar]
- 22.Cao Y, Yu S, Zhang Q, et al. Chinese Stroke Association guidelines for clinical management of cerebrovascular disorders: executive summary and 2019 update of clinical management of intracerebral haemorrhage. Stroke Vasc Neurol. 2020;5(4):396–402.): -doi: 10.1136/svn-2020-000433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tu W-J, Zhao Z, Yin P, et al. Estimated burden of stroke in China in 2020. JAMA Netw Open. 2023;6(3):e231455. doi: 10.1001/jamanetworkopen.2023.1455. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tu W, Wang L. China stroke surveillance report 2021. Mil Med Res. 2023;10(1):33. doi: 10.1186/s40779-023-00463-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Sembill JA, Castello JP, Sprügel MI, et al. Multicenter validation of the max-ICH score in intracerebral hemorrhage. Ann Neurol. 2021;89(3):474–484. doi: 10.1002/ana.25969. [DOI] [PubMed] [Google Scholar]
- 26.Pradilla G, Ratcliff JJ, Hall AJ, et al. Trial of early minimally invasive removal of intracerebral hemorrhage. N Engl J Med. 2024;390(14):1277–1289. doi: 10.1056/NEJMoa2308440. [DOI] [PubMed] [Google Scholar]
- 27.Ma L, Hu X, Song L, et al. The third intensive care bundle with blood pressure reduction in acute cerebral haemorrhage trial (INTERACT3): an international, stepped wedge cluster randomised controlled trial. Lancet. 2023;402(10395):27–40. doi: 10.1016/S0140-6736(23)00806-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Parry-Jones AR, Järhult SJ, Kreitzer N, et al. Acute care bundles should be used for patients with intracerebral haemorrhage: An expert consensus statement. Eur Stroke J. 2024;9(2):295–302. -doi: 10.1177/23969873231220235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Li Q, Yakhkind A, Alexandrov AW, et al. Code ICH: a call to action. Stroke. 2024;55(2):494–505. doi: 10.1161/STROKEAHA.123.043033. [DOI] [PubMed] [Google Scholar]
- 30.Wang F, Wang Y, Shi J, et al. Excessive antibiotics use increased in-hospital mortality in intracerebral hemorrhage patients with stroke-associated infection. Sci Rep. 2020;10(1):1950. doi: 10.1038/s41598-020-58772-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Duan M, Xu Y, Li Y, et al. Targeting brain-peripheral immune responses for secondary brain injury after ischemic and hemorrhagic stroke. J Neuroinflammation. 2024;21(1):102. doi: 10.1186/s12974-024-03101-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Li X, Chen G.. CNS-peripheral immune interactions in hemorrhagic stroke. J Cereb Blood Flow Metab. 2023;. 43(2):185–197. doi: 10.1177/0271678X221145089. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Lattanzi S, Brigo F, Trinka E, et al. Neutrophil-to-lymphocyte ratio in acute cerebral hemorrhage: a system review. Transl Stroke Res. 2019;10(2):137–145. doi: 10.1007/s12975-018-0649-4. [DOI] [PubMed] [Google Scholar]
- 34.Moxon-Emre I, Schlichter LJ.. Neutrophil depletion reduces blood-brain barrier breakdown, axon injury, and inflammation after intracerebral hemorrhage. J Neuropathol Exp Neurol. 2011;70(3):218–235. doi: 10.1097/NEN.0b013e31820d94a5. [DOI] [PubMed] [Google Scholar]
- 35.Chen S, Yang Q, Chen G, et al. An update on inflammation in the acute phase of intracerebral hemorrhage. Transl Stroke Res. 2015;6(1):4–8. doi: 10.1007/s12975-014-0384-4. [DOI] [PubMed] [Google Scholar]
- 36.Soehnlein O, Lindbom LJ.. Phagocyte partnership during the onset and resolution of inflammation. Nat Rev Immunol. 2010;10(6):427–439. doi: 10.1038/nri2779. [DOI] [PubMed] [Google Scholar]
- 37.Sullivan DH, Roberson PK, Johnson LE, et al. Association between inflammation-associated cytokines, serum albumins, and mortality in the elderly. J Am Med Dir Assoc. 2007;8(7):458–463. doi: 10.1016/j.jamda.2007.04.004. [DOI] [PubMed] [Google Scholar]
- 38.Gabay C, Kushner I.. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med. 1999;. 340(6):448–454. doi: 10.1056/NEJM199902113400607. [DOI] [PubMed] [Google Scholar]
- 39.Deng S, Liu S, Jin P, et al. Albumin reduces oxidative stress and neuronal apoptosis via the ERK/Nrf2/HO-1 pathway after intracerebral hemorrhage in rats. Oxid Med Cell Longev. 2021;2021:8891373. doi: 10.1155/2021/8891373. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Eckart A, Struja T, Kutz A, et al. Relationship of nutritional status, inflammation, and serum albumin levels during acute illness: a prospective study. Am J Med. 2020;133(6):713–722.e7. doi: 10.1016/j.amjmed.2019.10.031. [DOI] [PubMed] [Google Scholar]
- 41.Shiga Y, Nezu T, Shimomura R, et al. Various effects of nutritional status on clinical outcomes after intracerebral hemorrhage. Intern Emerg Med. 2022;17(4):1043–1052. doi: 10.1007/s11739-021-02901-x. [DOI] [PubMed] [Google Scholar]
- 42.Schaible UE, Kaufmann SH.. Malnutrition and infection: complex mechanisms and global impacts. PLoS Med. 2007;4(5):e115. doi: 10.1371/journal.pmed.0040115. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Lin J, He M, Tan B, et al. Impact and risk factors of sepsis on long-term outcomes after spontaneous intracerebral hemorrhage. Chin Med J (Engl). 2022;135(8):1006–1008. doi: 10.1097/CM9.0000000000001954. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The raw data supporting the conclusions of this study is available from the corresponding author on reasonable request.