Abstract
Objectives
There is currently a limited understanding regarding the effect of hepatic steatosis index (HSI) on stroke prognosis. Therefore, we aimed to investigate the effect of HSI on 3-month unfavorable outcomes in adults with acute ischemic stroke (AIS).
Methods
Between January 2010 and December 2016, a prospective cohort study was conducted involving 1906 AIS patients receiving treatment at a hospital in South Korea. The effect of HSI on 3-month unfavorable outcomes was evaluated using a binary logistic regression model. Smoothed curve fitting and a generalized additive model were also employed to investigate the nonlinear correlation between HSI and unfavorable outcomes.
Results
Multivariate analysis revealed a significant inverse relationship between HSI and the likelihood of 3-month unfavorable outcomes (OR = 0.97, 95% CI 0.94–1.00, P = 0.028). Particularly, individuals with HSI values of 30–36 had a significantly reduced risk of unfavorable outcomes (OR = 0.59, 95% CI 0.44–0.79, P < 0.001) compared with those with HSI < 30, while this association was not statistically significant when HSI > 36. A nonlinear association was identified, with an inflection point at an HSI of 31.78. For HSI values ≤ 31.78, a significant negative correlation with 3-month unfavorable outcomes was observed (OR = 0.92, 95% CI 0.87–0.96, P = 0.001). However, for HSI values exceeding 31.78, the association between HSI and unfavorable outcomes did not reach statistical significance (OR = 1.01, 95% CI 0.97–1.06, P = 0.595).
Conclusions
This study demonstrates a statistically significant inverse association between the HSI and 3-month unfavorable outcomes in patients with AIS, observed only when HSI values are below 31.78.
Supplementary Information
The online version contains supplementary material available at 10.1186/s40001-025-03456-9.
Keywords: Hepatic steatosis index, Acute ischemic stroke, Unfavorable outcomes, Nonlinear association, Non-alcoholic fatty liver disease
Introduction
Acute ischemic stroke (AIS) remains a leading cause of mortality and long-term disability globally, imposing a considerable burden on healthcare systems and societies [1–3]. Despite significant advancements in acute stroke management, a substantial proportion of patients continue to experience adverse outcomes [4]. The prognosis of stroke patients is influenced by a multitude of complex factors, underscoring the importance of identifying at-risk populations in both clinical practice and research.
Non-alcoholic fatty liver disease (NAFLD), marked by an abnormal buildup of fat within hepatocytes, has been implicated as a potential contributor to cardiovascular morbidity and stroke risk [5, 6]. To facilitate screening, the hepatic steatosis index (HSI) provides a non-invasive and readily computed tool that has been validated for estimating hepatic steatosis burden [7]. This index, derived from alanine aminotransferase (ALT), body mass index (BMI), and aspartate aminotransferase (AST), has proven effective in identifying individuals at risk for NAFLD [7, 8]. Elevated HSI was linked to insulin resistance, dyslipidemia, and systemic inflammation, conditions that are implicated in the pathogenesis of ischemic stroke [9, 10]. The current literature on the effect of HSI on neurological outcomes in AIS subjects is sparse and sometimes contradictory. For instance, a retrospective study involving 6950 patients reported an inverse relationship between HSI and 1-year combined outcomes post-stroke [11]. Conversely, other evidence has indicated that NAFLD may be related to more severe strokes and poorer outcomes [12–14]. Contrary to these findings, some perspectives argue that NAFLD does not correlate with stroke risk in matched cohorts [15]. Given these discrepancies, the precise relationship between NAFLD and stroke prognosis remains unclear. Our investigation aimed to clarify this association by examining the connection between HSI and 3-month unfavorable outcomes in the Korean population with AIS.
Methods
Study design and data source
This investigation utilized data from a comprehensive, single-center prospective registry, which encompassed AIS cases in South Korea between January 2010 and December 2016 [16]. This study is a secondary analysis based on a publicly available data set from PLoS One (10.1371/journal.pone.0228738). The original study was derived from a prospective registry database; however, no clinical trial registration number was reported in that publication. The primary aim of this investigation was to examine the 3-month unfavorable outcomes as the dependent variable, with the HSI serving as the independent variable.
We extend our gratitude to Kang MK and colleagues for their generous provision of the raw data used in this investigation at no cost [16]. This work falls under a Creative Commons Attribution License agreement, permitting unrestricted utilization, copying, and sharing provided appropriate credit is given to original contributors and proper citation of the source occurs.
Study population
The initial work was sanctioned by the Seoul National University Hospital Institutional Review Board approved the initial research, and patient informed consent was waived [16]. Conduct of the current investigation followed the ethical tenets of the Declaration of Helsinki, with all procedures listed in the Declaration’s section meeting relevant regulatory and professional standards.
In the initial research, Kang MK and colleagues extracted data from their registry, encompassing 2084 individuals diagnosed with AIS who were admitted within 7 days of symptom onset. Of these, 72 patients were excluded due to the absence of either a dysphagia assessment or essential laboratory data within the first 24 h of admission. In addition, 106 individuals were excluded because of the unavailability of a 3-month modified Rankin Scale (mRS) score [17] post-discharge. Consequently, the final study cohort comprised 1906 adults. The stepwise procedure used to select study participants is depicted in Fig. 1.
Fig. 1.
Flowchart of study participants
Covariates
The selection of covariates for this investigation was guided by established research literature and clinical expertise [11, 13, 16, 18, 19]. These covariates included: (1) categorical variables: diabetes mellitus (DM), sex, stroke etiology, history of transient ischemic attack (TIA) or stroke, coronary heart disease (CHD), age, and hypertension and (2) continuous variables: total cholesterol (TC), serum creatinine (Scr), C-reactive protein, hemoglobin (HGB), gamma-glutamyl transferase (GGT), low-density lipoprotein cholesterol (LDL-C), triglycerides (TG), National Institutes of Health Stroke Scale (NIHSS) score [20] at admission, serum albumin (ALB), and high-density lipoprotein cholesterol (HDL-C).
The original study provided comprehensive methodologies for data collection. Anthropometric measurements were conducted by qualified nursing staff upon patient intake. A standard automated scale was employed to assess the height and weight of ambulatory individuals. A specialized bed scale was utilized for weight determination, while a flexible tape measure was used to ascertain height for patients with severe stroke-induced mobility limitations. Laboratory values, including TG, AST, HGB, TC, ALT, ALB, HDL-C, C-reactive protein, GGT, and LDL-C, were obtained from the patients’ electronic health records within the first 24 h of admission. Chronic conditions such as DM, CHD, and hypertension were also documented via the electronic medical record system.
Hepatic steatosis index
HSI was calculated using the following formulas: for females, HSI = 2 + (ALT/AST) × 8 + BMI (+ 2 if DM), and for males, HSI = (ALT/AST) × 8 + BMI (+ 2 if DM) [7].
Three-month unfavorable outcomes in AIS patients
Functional outcomes were assessed 3-month post-AIS onset using the mRS score [17]. Data were collected through structured interviews, either via telephone or during follow-up outpatient consultations [16]. Participants were stratified into two cohorts according to their functional outcomes: favorable and unfavorable. An mRS score of 3 or higher was designated as indicative of unfavorable outcomes [21].
Missing data handling
In this study, missing data were observed for TC in 1 participant (0.05%), LDL-C in 75 participants (3.93%), HDL-C in 99 participants (5.19%), TG in 107 participants (5.61%), C-reactive protein in 252 participants (13.22%), and GGT in 1570 participants (82.36%). To mitigate biases arising from missing data and ensure the sample’s statistical validity during modeling, we employed multiple imputation techniques [22]. Variables included in the imputation model were NIHSS score at admission, age, HDL-C, smoking status, history of stroke or TIA, sex, DM, TC, hypertension, CHD, TG, stroke etiology, LDL-C, GGT, Scr, ALB, HGB, and CRP. The analysis assumed that the data were missing at random [22, 23].
Statistical analysis
We categorized the patients into three groups based on previously established criteria for Asian populations: HSI > 36, 30 ≤ HSI ≤ 36, and HSI < 30 [24]. Participants’ demographics were analyzed employing the chi-square test for categorical data, the Kruskal–Wallis test for skewed continuous data, and one-way ANOVA for normally distributed continuous data. Categorical data were employed as population proportions and percentages, while continuous data were applied as means ± standard deviations or medians and interquartile ranges.
Multivariate logistic regression models were employed to investigate the effect of HSI on unfavorable outcomes across four different models. The models were as follows: (1) Model 1: without adjustment for covariates; (2) Model 2: controlling for age, smoking, and sex; (3) Model 3: controlling for DM, NIHSS score, stroke etiology, CHD, previous TIA or stroke, hypertension, and the variables in Model 2; and (4) Model 4: controlling for GGT, LDL-C, ALB, HDL-C, HGB, Scr, CRP, TG, and the variables in Model 3. Odds ratios (OR) and corresponding 95% confidence intervals (CI) were estimated. Adjustments for confounding variables were made based on clinical expertise, existing research, and findings from univariate analyses [11, 13, 16, 18, 19]. Due to the identified collinearity between TC and other variables (as displayed in Table S1), we excluded TC from the multivariate model.
To further analyze the non-linear connection between HSI and the incidence of unfavorable outcomes, we employed Generalized Additive Modeling (GAM) with smooth curve fitting via penalized splines. An iterative algorithm was applied to identify the inflection point after detecting a non-linear relationship. Subsequently, binary logistic regression models were developed on either side of this inflection point. The log-likelihood ratio test was utilized to determine the model that most accurately depicted the connection between HSI and the incidence of unfavorable outcomes.
To enhance the reliability of our findings, we implemented multiple sensitivity analyses. Recognizing the profound links between obesity, smoking, and hypertension with the prognosis of AIS patients [25–27], we excluded patients with smoking, hypertension, and BMI ≥ 25 kg/m2 in the sensitivity analyses to further test the effect of HSI on unfavorable outcomes. In addition, to address the likelihood of unaccounted confounding factors influencing the connection between HSI and unfavorable outcomes, we computed the E value [28].
Receiver operating characteristic (ROC) curve analyses were performed to assess the discriminative ability and predictive performance of HSI and conventional liver function biomarkers for the identification of 3-month unfavorable outcomes. All records pertaining to the study outcomes complied with the STROBE statement. Data processing employed EmpowerStats, and results were considered significant when two‑tailed P values were below 0.05.
Results
Characteristics of participants
Table 1 presents the clinical characteristics and demographics of the study cohort, comprising 1906 individuals, of whom 61.28% were male. Notably, 1470 individuals (77.12%) were over the age of 60. Individuals were stratified into three groups based on established Asian criteria for the Hepatic Steatosis Index (HSI): HSI < 30, 30 ≤ HSI ≤ 36, and HSI > 36. Compared to the moderate and high HSI groups (30 ≤ HSI ≤ 36 and HSI > 36), the low HSI group (HSI < 30) exhibited higher age, NIHSS score, C-reactive protein, and HDL-C, alongside lower TG, BMI, ALT, HGB, TC, and LDL-C. Furthermore, the high HSI group (HSI > 36) demonstrated the highest AST levels and a greater proportion of participants with DM compared to the low and moderate HSI groups (HSI < 30 and 30 ≤ HSI ≤ 36).
Table 1.
Baseline characteristics of participants
| HSI | HSI (< 30) | HSI (30–36) | HSI (> 36) | P value |
|---|---|---|---|---|
| Participants | 717 | 834 | 355 | |
| Gender | 0.065 | |||
| Male | 463 (64.57%) | 498 (59.71%) | 207 (58.31%) | |
| Female | 254 (35.43%) | 336 (40.29%) | 148 (41.69%) | |
| Age (years) | < 0.001 | |||
| < 60 | 119 (16.60%) | 191 (22.90%) | 126 (35.49%) | |
| @60 to < 70 | 156 (21.76%) | 252 (30.22%) | 97 (27.32%) | |
| 70 to < 80 | 278 (38.77%) | 285 (34.17%) | 107 (30.14%) | |
| ≥ 80 | 164 (22.87%) | 106 (12.71%) | 25 (7.04%) | |
| BMI (kg/m2) | 20.91 ± 2.22 | 24.11 ± 2.00 | 27.29 ± 2.95 | < 0.001 |
| Smoking status | 0.456 | |||
| No | 423 (59.00%) | 518 (62.11%) | 215 (60.56%) | |
| Yes | 294 (41.00%) | 316 (37.89%) | 140 (39.44%) | |
| NIHSS score | 4 (2–9) | 3 (1–7) | 3 (1–5.5) | < 0.001 |
| Hypertension | ||||
| No | 317 (44.21%) | 276 (33.09%) | 102 (28.73%) | |
| Yes | 400 (55.79%) | 558 (66.91%) | 253 (71.27%) | |
| CHD | 0.244 | |||
| No | 640 (89.26%) | 741 (88.85%) | 305 (85.92%) | |
| Yes | 77 (10.74%) | 93 (11.15%) | 50 (14.08%) | |
| Previous stroke/TIA | 0.332 | |||
| No | 574 (80.06%) | 645 (77.34%) | 285 (80.28%) | |
| Yes | 143 (19.94%) | 189 (22.66%) | 70 (19.72%) | |
| DM | < 0.001 | |||
| No | 586 (81.73%) | 533 (63.91%) | 173 (48.73%) | |
| Yes | 131 (18.27%) | 301 (36.09%) | 182 (51.27%) | |
| Stroke etiology | 0.001 | |||
| SVO | 205 (28.59%) | 271 (32.49%) | 130 (36.62%) | |
| LAA | 113 (15.76%) | 175 (20.98%) | 77 (21.69%) | |
| CE | 211 (29.43%) | 211 (25.30%) | 71 (20.00%) | |
| Other determined | 74 (10.32%) | 66 (7.91%) | 31 (8.73%) | |
| Undetermined | 114 (15.90%) | 111 (13.31%) | 46 (12.96%) | |
| TC (mg/dL) | 176.19 ± 42.87 | 180.17 ± 41.90 | 183.99 ± 49.20 | 0.018 |
| TG (mg/dL) | 100.56 ± 47.87 | 114.63 ± 55.43 | 135.53 ± 70.70 | < 0.001 |
| HDL-C (mg/dL) | 46.74 ± 16.28 | 45.40 ± 13.79 | 42.56 ± 12.02 | < 0.001 |
| LDL-C (mg/dL) | 105.06 ± 36.61 | 108.17 ± 35.84 | 110.87 ± 43.85 | 0.048 |
| ALT (IU) | 14 (11–19) | 19 (15–25) | 30 (21–43.50) | < 0.001 |
| AST (IU) | 23 (19–29) | 22 (18–28) | 25 (19–32) | < 0.001 |
| GGT (IU) | 31.15 (10.98–69.44) | 32.95 (6.83–71.97) | 36.41 (9.92–73.50) | 0.697 |
| HSI | 26.96 ± 2.28 | 32.67 ± 1.64 | 39.48 ± 3.28 | < 0.001 |
| ALB (g/dL) | 3.93 ± 0.46 | 4.04 ± 0.41 | 4.14 ± 0.37 | < 0.001 |
| Scr (mg/dL) | 0.90 (0.73–1.11) | 0.88 (0.74–1.07) | 0.90 (0.74–1.06) | 0.548 |
| HGB (g/dL) | 12.98 ± 2.11 | 13.58 ± 1.88 | 14.23 ± 1.79 | < 0.001 |
| C-reactive protein (mg/dL) | 0.22 (0.07–1.10) | 0.15 (0.05–0.68) | 0.15 (0.06–0.59) | < 0.001 |
Values are n (%) or mean ± SD or median (quartile)
HSI, hepatic steatosis index; BMI, body mass index; NIHSS, national institute of health stroke scale; CHD, coronary heart disease; TIA, transient ischemia attack; DM, diabetes mellitus; SVO, small vessel occlusion; LAA, small vessel occlusion; CE, cardio embolism; TC, total cholesterol; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoproteins cholesterol; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT: gamma-glutamyl transferase; ALB, serum albumin; Scr, serum creatinine; HGB, hemoglobin concentration
The prevalence rate of 3-month unfavorable outcomes in AIS individuals
The occurrence of 3-month unfavorable outcomes is detailed in Table 2, revealing that 546 individuals experienced unfavorable outcomes, resulting in an overall incidence rate of 28.65%. More specifically, the incidence rates of 3-month unfavorable outcomes were 36.26%, 24.10%, and 23.94% in the HSI < 30, 30 ≤ HSI ≤ 36, and HSI > 36 groups, respectively. Notably, patients in the HSI > 36 group exhibited the lowest incidence of 3-month unfavorable outcomes.
Table 2.
Incidence rate of 3-month unfavorable outcomes after acute ischemic stroke
| HSI | Participants(n) | Events(n) | Incidence (95%CI) (%) |
|---|---|---|---|
| All participants | 1906 | 546 | 28.65 (26.61–30.68) |
| HSI (< 30) | 717 | 260 | 36.26 (32.73–39.79) |
| HSI (30–36) | 834 | 201 | 24.10 (21.19–27.01) |
| HSI (> 36) | 355 | 85 | 23.94 (19.48–28.40) |
HSI, hepatic steatosis index; CI, confidence intervals
The effect of HSI on 3-month unfavorable outcomes
Table 3 illustrates the effect of HSI on unfavorable outcomes. In Model 1, each additional unit of HSI was linked to a 6% reduction in the risk of unfavorable outcomes (OR: 0.94, 95% CI 0.92–0.96, P < 0.001), demonstrating a strong inverse relationship. Model 2, which accounted solely for demographic factors (age, smoking status, gender), showed that each one-unit increase in HSI correlated with a 4% decrease in the incidence of poor outcomes 3-month post-AIS (OR: 0.96, 95% CI 0.94–0.98, P < 0.001). Further adjustments in Model 3, which included age, sex, smoking, previous stroke or TIA, hypertension, DM, CHD, NIHSS score, and stroke etiology, revealed that a one-unit rise in HSI was related to a 4% reduction in the incidence of adverse outcomes (OR: 0.96, 95% CI 0.93–0.99; P = 0.003). Model 4 expanded these adjustments to include TG, HDL-C, LDL-C, GGT, albumin, serum creatinine, hemoglobin, and C-reactive protein, revealing that each unit increase in HSI corresponded to a 3% decrease in the incidence of unfavorable outcomes (OR: 0.97, 95% CI 0.94–1.00; P = 0.028).
Table 3.
Relationship between HSI and 3-month unfavorable outcomes after acute ischemic stroke in different models
| Exposure | Model 1 (OR,95%CI) P | Model 2 (OR,95%CI) P | Model 3 (OR,95%CI) P | Model 4 (OR,95%CI) P |
|---|---|---|---|---|
| HSI | 0.94 (0.92, 0.96) < 0.001 | 0.96 (0.94, 0.98) < 0.001 | 0.96 (0.93, 0.99) 0.003 | 0.97 (0.94, 1.00) 0.028 |
| HSI group | ||||
| HSI (< 30) | Ref. | Ref. | Ref. | Ref. |
| HSI (30–36) | 0.62 (0.49, 0.79) < 0.001 | 0.61 (0.48, 0.76) < 0.001 | 0.56 (0.42, 0.74) < 0.001 | 0.59 (0.44, 0.79) < 0.001 |
| HSI (> 36) | 0.50 (0.39, 0.64) < 0.001 | 0.66 (0.49, 0.90) 0.007 | 0.66 (0.46, 0.96) 0.028 | 0.73 (0.50, 1.07) 0.111 |
| P for trend | < 0.001 | < 0.001 | 0.004 | 0.025 |
Model 1: we did not adjust any covariates
Model 2: we adjusted age, sex, and smoking
Model 3: we adjusted age, sex, smoking, hypertension, DM, CHD, previous stroke or TIA, stroke etiology, and NIHSS score
Model 4: we adjusted age, sex, smoking, hypertension, DM, CHD, previous stroke or TIA, stroke etiology, NIHSS score, TG, HDL-C, LDL-C, GGT, ALB, Scr, HGB, and C-reactive protein
HSI, hepatic steatosis index; OR, odds ratios; CI, confidence intervals; Ref: reference
In addition, the HSI, initially a continuous variable, was transformed into a categorical variable and subsequently reintegrated into the model in its categorical form. After accounting for confounding variables, the OR in the 30 ≤ HSI ≤ 36 group and HSI > 36 group were 0.59 (95% CI 0.44–0.79) and 0.73 (95% CI 0.50–1.07) relative to the HSI < 30 group (Table 3, Model 4). Analysis of the confidence intervals revealed that individuals in the HSI > 36 group did not demonstrate a statistically significant variation in the risk of unfavorable outcomes compared to those in the HSI < 30 group. Conversely, participants in the 30 ≤ HSI ≤ 36 group exhibited a notable reduction in the risk of unfavorable outcomes. Furthermore, a statistically significant trend in effect size was observed (trend P = 0.025).
Sensitivity analysis
To uphold the robustness of the study findings, a comprehensive set of sensitivity analyses was conducted. An E value was undertaken to evaluate the possible effect of unmeasured confounders on the research outcomes. The calculated E value of 1.17 surpassed the relative risk attributed to unmeasured confounding factors and the HSI (1.15), indicating minimal influence of unidentified confounders on the effect of HSI on unfavorable outcomes. In addition, to further validate the study’s findings, supplementary sensitivity analyses were conducted by excluding specific subgroups. When individuals with a BMI ≥ 25 kg/m2 were removed from the analysis, a negative correlation between HSI and the risk of unfavorable outcomes in AIS individuals remained significant after adjusting for confounding variables (OR: 0.94, 95% CI 0.90–0.98, P = 0.004) (Table 4, Model 5). Similarly, excluding participants who were smokers yielded consistent results (OR: 0.94, 95% CI 0.89–0.99, P = 0.029) (Table 4, Model 6). Furthermore, when restricting the analysis to individuals without hypertension, the inverse association between HSI and unfavorable outcomes persisted (OR: 0.96, 95% CI 0.92–0.99, P = 0.017) (Table 4, Model 7). Collectively, the outcomes from all sensitivity analyses reinforce the reliability of the study conclusions.
Table 4.
Relationship between HSI and 3-month unfavorable outcomes after acute ischemic stroke in different sensitivity analyses
| Exposure | Model 5 (OR,95%CI) P | Model 6 (OR,95%CI) P | Model 7 (OR,95%CI) P |
|---|---|---|---|
| HSI | 0.94 (0.90, 0.98) 0.004 | 0.94 (0.89, 0.99) 0.029 | 0.96 (0.92, 0.99) 0.017 |
| HSI group | |||
| HSI (< 30) | Ref. | Ref. | Ref. |
| HSI (30–36) | 0.60 (0.43, 0.83) 0.002 | 0.47 (0.28, 0.79) 0.004 | 0.46 (0.31, 0.66) < 0.001 |
| HSI (> 36) | 0.50 (0.24, 1.01) 0.054 | 0.73 (0.36, 1.51) 0.397 | 0.66 (0.40, 1.07) 0.094 |
| P for trend | 0.002 | 0.073 | 0.017 |
Model 5 was a sensitivity analysis in participants without BMI ≥ 25 kg/m2. We adjusted age, sex, smoking, hypertension, DM, CHD, previous stroke or TIA, stroke etiology, NIHSS score, TG, HDL-C, LDL-C, GGT, ALB, Scr, HGB, and C-reactive protein
Model 6 was a sensitivity analysis in participants without hypertension. We adjusted age, sex, smoking, DM, CHD, previous stroke or TIA, stroke etiology, NIHSS score, TG, HDL-C, LDL-C, GGT, ALB, Scr, HGB, and C-reactive protein
Model 7 was a sensitivity analysis in participants without smoking. We adjusted age, sex, hypertension, DM, CHD, previous stroke or TIA, stroke etiology, NIHSS score, TG, HDL-C, LDL-C, GGT, ALB, Scr, HGB, and C-reactive protein
HSI, hepatic steatosis index; OR, odds ratios; CI, confidence intervals; Ref: reference
Non-linear relationship
Figure 2 and Table 5 reveal a nonlinear association between HSI and the risk of unfavorable outcomes in AIS patients. Using the recursive partitioning method, an inflection point was identified at 31.78. Below this inflection point, HSI demonstrated a significant negative association with the risk of unfavorable outcomes (OR: 0.92, 95%CI 0.87–0.96, P = 0.001). However, when HSI levels exceeded 31.78, the association was not statistically significant (OR: 1.01, 95%CI 0.97–1.06, P = 0.595).
Fig. 2.
Nonlinear relationship between HSI and 3-month adverse clinical outcomes in AIS patients. A nonlinear relationship between them was detected in AIS patients after adjusting for age, sex, smoking, hypertension, DM, CHD, previous stroke or TIA, stroke etiology, NIHSS score, TG, HDL-C, LDL-C, GGT, ALB, Scr, HGB, and C-reactive protein
Table 5.
Relationship between HSI and 3-month unfavorable outcomes after acute ischemic stroke using a two-piecewise binary logistic regression model
| 3-month adverse clinical outcomes | OR (95%CI) | P |
|---|---|---|
| Fitting Model by standard linear regression | 0.97 (0.94, 1.00) | 0.028 |
| Fitting Model by two-piecewise binary logistic regression | ||
| Inflection point of HSI | 31.78 | |
| ≤ 31.78 | 0.92 (0.87, 0.96) | 0.001 |
| > 31.78 | 1.01 (0.97, 1.06) | 0.595 |
| P for log-likelihood ratio test | 0.012 | |
We adjusted age, sex, smoking, hypertension, DM, CHD, previous stroke or TIA, stroke etiology, NIHSS score, TG, HDL-C, LDL-C, GGT, ALB, Scr, HGB, and C-reactive protein
HSI, hepatic steatosis index; OR, odds ratios; CI, confidence intervals; Ref: reference
Predictive performance of HSI and conventional liver function biomarkers
To evaluate the relative prognostic utility of HSI compared with traditional liver function parameters, we conducted ROC analyses for predicting 3-month unfavorable outcomes. As presented in Table S2 and Figure S1, HSI demonstrated moderate discriminative ability (AUC = 0.5877, 95% CI 0.5588–0.6166), which was superior to traditional liver enzymes, including AST (AUC = 0.5259, 95% CI 0.4967–0.5551), ALT (AUC = 0.5567, 95% CI 0.5271–0.5863), and GGT (AUC = 0.5294, 95% CI 0.5007–0.5581). However, albumin showed the highest prognostic performance among all parameters examined (AUC = 0.6479, 95% CI 0.6205–0.6753), significantly outperforming HSI. The optimal threshold for HSI in predicting unfavorable outcomes was 28.76, with a specificity of 0.7647 and sensitivity of 0.3864. These findings suggest that while HSI offers improved prognostic value over individual liver enzymes, albumin remains a more powerful predictor of stroke outcomes.
Discussion
This prospective investigation demonstrated an inverse correlation between the HSI and the risk of unfavorable outcomes in AIS patients after adjusting for confounders. In addition, our results revealed a non-linear connection between HSI and the likelihood of unfavorable outcomes, with an inflection point identified at 31.78. The reliability of these findings was further corroborated through comprehensive sensitivity analyses.
The HSI, an essential measure of hepatic fat accumulation severity, serves as an efficient tool for screening NAFLD [7]. Recent studies have demonstrated that HSI is associated not only with hepatic steatosis but also with a range of systemic conditions, including diabetes, cardiovascular disease, chronic kidney disease, and carotid atherosclerosis [29–32]. However, the relationship between NAFLD and post-stroke outcomes remains contentious, with available evidence being limited and conflicting. In a retrospective analysis involving 306 patients with brainstem infarction, multivariate adjustments for age, sex, diabetes mellitus, fibrinogen levels, and C-reactive protein indicated that NAFLD is an independent risk factor for worsening brainstem conditions [12]. Furthermore, a prospective study of 200 AIS patients suggested that those with NAFLD experienced more severe strokes and poorer prognoses compared to those without NAFLD [13]. Conversely, a retrospective cohort study of 1601 patients found no significant impact of NAFLD on stroke-related disability or mortality [33]. Another recent prospective investigation involving 6950 participants demonstrated an inverse relationship between HSI and poor stroke prognosis after accounting for variables, such as age, stroke subtype, initial NIHSS score, sex, diabetes, coronary heart disease, smoking, dyslipidemia, and hypertension [11]. Moreover, a study involving 321 AIS or TIA patients revealed that after adjusting for age, metabolic syndrome, sex, hypercholesterolemia, BMI, hypertension, and diabetes, NAFLD was associated with less severe strokes and improved functional outcomes post-stroke [34]. The discrepancies in these findings may stem from several factors, including differences in study populations, sample sizes, and the covariate adjustments utilized in the analyses. In addition, nonlinear associations could contribute to these variations.
This study reinforced the hypothesis that a higher HSI was linked to a lower incidence of 3-month unfavorable outcomes in AIS individuals. By examining HSI as both categorical and continuous data, our research minimizes data loss and provides a comprehensive quantification of the effect of HSI on unfavorable outcomes. The stability of these findings was confirmed through sensitivity analyses in prespecified subgroups (BMI < 25 kg/m2, non-hypertensive individuals, and smokers). These insights into the HSI–stroke outcome link provide a new perspective to guide improvements in post-stroke rehabilitation and overall management, with potential gains in patient well-being.
This investigation represented a pioneering examination of the non-linear connection between HSI and unfavorable outcomes, with an inflection point identified at 31.78. Below this threshold, each incremental unit rise in HSI was linked to a 6% decrease in the risk of unfavorable outcomes. Conversely, when HSI exceeded 31.78, there was no statistically significant relationship with the risk of unfavorable outcomes. Essentially, the incidence rate of unfavorable outcomes in AIS patients diminishes with increasing HSI values up to the threshold of 31.78; beyond this point, further elevations in HSI do not correspondingly raise the risk of unfavorable outcomes. Clinically, this cutoff provides an accessible and non-invasive metric to help identify AIS patients who may derive the most benefit from metabolic and nutritional optimization post-stroke. It supports the hypothesis that moderate hepatic steatosis, as estimated by HSI, may reflect a metabolic reserve or more favorable nutritional status, conferring resilience during the catabolic stress of acute ischemic stroke. This is particularly relevant as malnutrition and low metabolic reserve have been linked to poorer functional recovery in stroke survivors. Furthermore, the use of 31.78 as a clinically relevant cutoff offers practical guidance for both acute care and rehabilitation teams by distinguishing patients who may require closer monitoring for malnutrition or more proactive nutritional management. Future research could further validate this threshold in diverse populations and evaluate its integration into routine post-stroke prognostic assessment, potentially paving the way toward more individualized and effective rehabilitation strategies.
Although higher HSI values commonly indicate greater metabolic dysfunction, our findings reveal a nonlinear, threshold‑dependent association with stroke outcomes, defined by an inflection point at 31.78. Within the lower to moderate range, higher HSI may reflect superior metabolic and nutritional reserves—consistent with the “obesity paradox” frequently reported in stroke research—thereby contributing to improved outcomes [35, 36]. However, once HSI exceeds 31.78, the detrimental effects of progressive hepatic steatosis—including insulin resistance, oxidative stress, and endothelial dysfunction—may outweigh any protective influence. This threshold-dependent pattern provides a plausible explanation for the apparent inconsistency observed in Table 3 and underscores the complex interplay between hepatic steatosis, systemic metabolism, and functional recovery after stroke.
The HSI (HSI = 8 × ALT/AST ratio + BMI + sex/diabetes modifiers) integrates hepatocellular enzyme balance, overall adiposity and nutritional status, and key metabolic modifiers (sex and diabetes). In this study, HSI behaved as a protective marker rather than a risk factor within the lower-to-moderate range. Specifically, when HSI ≤ 31.78, each unit increase was associated with lower odds of 3-month unfavorable outcomes. Several mechanisms may underlie this protective association. First, higher HSI within this range likely reflects more robust metabolic and nutritional reserves that buffer the profound catabolic stress of AIS, providing substrates necessary for neuroplasticity, immune competence, and tissue repair [36, 37]. Second, moderately elevated HSI often co-occurs with obesity phenotypes characterized by chronic low-grade inflammation and features of immune tolerance; this state may attenuate excessive post-stroke inflammatory cascades and limit collateral tissue injury [36, 38]. Third, preserved hepatic function within this range may support balanced acute-phase responses and the synthesis of albumin, clotting factors, and carrier proteins, thereby maintaining vascular integrity and optimizing pharmacokinetics during recovery. The relationship between HSI and outcomes was non-linear, with attenuation of benefit beyond a threshold. The protective association weakened between HSI 31.78, and was no longer statistically significant at higher levels. Several complementary mechanisms may explain this plateau. A ceiling (sufficiency) effect likely operates once metabolic and nutritional requirements for neural repair are met, such that additional energy reserves confer diminishing returns [39]. At higher HSI values, counterbalancing metabolic dysregulation—insulin resistance, altered adipokine signaling, microvascular dysfunction, and dysregulated inflammation—may offset the benefits of reserve, yielding a net effect near unity [40–42]. These findings support a model in which moderate HSI elevation marks a biologically favorable state—adequate reserves and adaptive inflammatory tuning—associated with improved early recovery, whereas further increases do not translate into additional benefit. Clinically, HSI should not be interpreted as a linear risk metric across its spectrum; rather, its prognostic value appears confined to a lower-to-moderate range, with no evidence of harm at higher levels in our data, but with loss of incremental protection beyond the identified thresholds.
This investigation demonstrates several methodological strengths that bolster the robustness of its results. First, the study adeptly employed both categorical and continuous analyses of the HSI to examine its connection with unfavorable outcomes in AIS individuals. This nuanced approach minimized information loss and effectively quantified the connection between HSI and unfavorable outcomes. Second, by addressing the non-linear nature of this relationship, the study represents a noteworthy advancement over prior research endeavors. Third, the use of multiple imputation techniques to manage missing data enhanced the statistical power of the analysis and mitigated potential biases arising from incomplete covariate information. Furthermore, the implementation of a series of sensitivity analyses, including variable transformations, E value calculations to probe unmeasured confounding factors, and re-evaluation of the HSI-adverse outcomes association after excluding specific participant subgroups, at tests to the diligence in ensuring the robustness and reliability of the study results.
Nonetheless, certain limitations warrant consideration. First, the study cohort was comprised solely of Korean individuals, necessitating caution in extrapolating these findings to other ethnic groups without validation in diverse populations. Second, specific variable details, such as the precise ages of patients, were lacking and provided only in broad stratifications, potentially introducing incomplete variable information. Future research might benefit from designing studies to obtain more granular data on variables of interest. Third, the calculation of HSI only upon admission raises concerns regarding the temporal dynamics of HSI, as changes in HSI within the initial 24 h post-admission remain unexplored, highlighting the need for further longitudinal assessments. Fourth, inherent to observational studies, residual confounding remains possible despite efforts to control for acknowledged potential confounders. Fifth, our analysis was based on a publicly available data set from PLoS One, and the original database did not include detailed information on major adverse cardiovascular events, such as ischemic stroke recurrence, myocardial infarction, or other vascular complications. Therefore, we were unable to perform additional sensitivity analyses using these hard clinical endpoints. Future prospective studies with comprehensive follow-up data are warranted to further validate the association between HSI and cardiovascular outcomes. Sixth, the source database did not include granular information on acute reperfusion therapies (intravenous thrombolysis and endovascular thrombectomy) or post-stroke pharmacotherapy (antiplatelet agents, statins, antihypertensives, and glucose-lowering medications). Because these treatments can substantially influence neurological recovery and vascular risk, their absence precluded direct adjustment for treatment-related confounding. To assess the robustness of our findings to unmeasured confounding, we conducted a quantitative bias analysis using the E value; the estimate suggests that unmeasured factors of plausible magnitude are unlikely to fully explain the observed associations. Future work will incorporate detailed reperfusion metrics (e.g., timing, dosing, and procedural characteristics) and comprehensive medication data to more precisely delineate the independent relationship between HSI and post-stroke outcomes. Seventh, it is important to acknowledge that HSI, while widely used as a non-invasive surrogate marker, has several well-documented limitations in diagnosing hepatic steatosis. Its diagnostic accuracy varies considerably across different populations and is particularly compromised in the context of obesity, metabolic syndrome, liver fibrosis, and inflammatory states. Furthermore, HSI cannot reliably quantify steatosis severity and shows inconsistent performance when validated in external cohorts. As emphasized in the guidelines from the American Association for the Study of Liver Diseases, HSI should be regarded primarily as a screening tool rather than a definitive diagnostic or monitoring index. Consequently, the associations observed between HSI and stroke outcomes in this study should be interpreted cautiously, as reflecting potential relationships between metabolic–hepatic dysfunction and stroke prognosis rather than definitive evidence of direct causation by hepatic steatosis itself. Future studies incorporating more precise imaging-based assessments of hepatic steatosis would be valuable to confirm these findings.
Conclusion
This study elucidates a negative and nonlinear relationship between HSI and 3-month unfavorable outcomes in AIS patients. A significant negative relationship between HSI and 3-month unfavorable outcomes was observed only when HSI values were below 31.78. These findings offer valuable insights for the clinical optimization of rehabilitation strategies in AIS patients.
Supplementary Information
Acknowledgements
As this is a secondary study, much of the data and methodological description are derived from previous works, specifically: Kang MK, Kim TJ, Kim Y, Nam KW, Jeong HY, Kim SK, et al. (2020) Automated undernutrition screen tool: Geriatric nutritional risk score predicts poor outcomes in individuals with acute ischemic stroke. PLoS ONE 15(2): e0228738. 10.1371/journal.pone.0228738. We extend our gratitude to the entire study authorship team.
Abbreviations
- AIS
Acute ischemic stroke
- ALB
Serum albumin
- ALT
Alanine aminotransferase
- AST
Aspartate aminotransferase
- BMI
Body mass index
- CHD
Coronary heart disease
- CI
Confidence intervals
- DM
Diabetes mellitus
- GAM
Generalized additive models
- GGT
Gamma-glutamyl transferase
- HGB
Hemoglobin concentration
- HDL-C
High-density lipoprotein cholesterol
- HSI
Hepatic steatosis index
- LDL-C
Low-density lipoproteins cholesterol
- mRS
Modified Rankin Scale
- NAFLD
Non-alcoholic fatty liver disease
- NIHSS
National Institutes of Health Stroke Scale
- OR
Odds ratios
- Scr
Serum creatinine
- TC
Total cholesterol
- TG
Triglyceride
- TIA
Transient ischemic attack
Author contributions
Changchun Cao and Mingxing Lou were responsible for conceiving the research, drafting the manuscript, and conducting the statistical analysis. Jie Jia took part in revising the manuscript and designing the study. All authors read and approved the final manuscript.
Funding
This study was supported by the Joint Funds for the innovation of science and Technology, Fujian province (2021Y9130), and the National Key Research and Development Program of Ministry of Science and Technology of the People’s Republic of China (2018YFC2002300).
Data availability
Data can be accessed and downloaded from the “PLoS One” database (10.1371/journal.pone.0228738).
Declarations
Ethics approval and consent to participate
The original research was approved by the Seoul National University Hospital Institutional Review Board, and the Seoul National University Hospital Institutional Review Board waived the requirement for patient-informed consent (No.1009-062-332). This study was carried out ethically and in compliance with the Helsinki Declaration.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Changchun Cao and Mingxing Lou have contributed equally to this work.
References
- 1.Ding Q, Liu S, Yao Y, Liu H, Cai T, Han L. Global, regional, and national burden of ischemic stroke, 1990–2019. Neurology. 2022;98(3):e279–90. [DOI] [PubMed] [Google Scholar]
- 2.Global, regional, and national burden of stroke and its risk factors, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Neurol 2021;20(10):795–820. [DOI] [PMC free article] [PubMed]
- 3.Prabhakaran S, Ruff I, Bernstein RA. Acute stroke intervention: a systematic review. JAMA J Am Med Assoc. 2015;313(14):1451–62. [DOI] [PubMed] [Google Scholar]
- 4.Luengo-Fernandez R, Paul NL, Gray AM, Pendlebury ST, Bull LM, Welch SJ, et al. Population-based study of disability and institutionalization after transient ischemic attack and stroke: 10-year results of the Oxford Vascular Study. Stroke. 2013;44(10):2854–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Patil R, Sood GK. Non-alcoholic fatty liver disease and cardiovascular risk. World J Gastrointest Pathophys. 2017;8(2):51–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Alkagiet S, Papagiannis A, Tziomalos K. Associations between nonalcoholic fatty liver disease and ischemic stroke. World J Hepatol. 2018;10(7):474–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lee JH, Kim D, Kim HJ, Lee CH, Yang JI, Kim W, et al. Hepatic steatosis index: a simple screening tool reflecting nonalcoholic fatty liver disease. Dig Liver Dis. 2010;42(7):503–8. [DOI] [PubMed] [Google Scholar]
- 8.Meffert PJ, Baumeister SE, Lerch MM, Mayerle J, Kratzer W, Volzke H. Development, external validation, and comparative assessment of a new diagnostic score for hepatic steatosis. Am J Gastroenterol. 2014;109(9):1404–14. [DOI] [PubMed] [Google Scholar]
- 9.Francque SM, van der Graaff D, Kwanten WJ. Non-alcoholic fatty liver disease and cardiovascular risk: pathophysiological mechanisms and implications. J Hepatol. 2016;65(2):425–43. [DOI] [PubMed] [Google Scholar]
- 10.Sookoian S, Pirola CJ. Non-alcoholic fatty liver disease is strongly associated with carotid atherosclerosis: a systematic review. J Hepatol. 2008;49(4):600–7. [DOI] [PubMed] [Google Scholar]
- 11.Chen K, Pan Y, Xiang X, Meng X, Yao D, Lin L, et al. The nonalcoholic fatty liver risk in prediction of unfavorable outcome after stroke: a nationwide registry analysis. Comput Biol Med. 2023;157:106692. [DOI] [PubMed] [Google Scholar]
- 12.Li H, Hu B, Wei L, Zhou L, Zhang L, Lin Y, et al. Non-alcoholic fatty liver disease is associated with stroke severity and progression of brainstem infarctions. Eur J Neurol. 2018;25(3):534–77. [DOI] [PubMed] [Google Scholar]
- 13.Abdeldyem SM, Goda T, Khodeir SA, Abou SS, Abd-Elsalam S. Nonalcoholic fatty liver disease in patients with acute ischemic stroke is associated with more severe stroke and worse outcome. J CLIN LIPIDOL. 2017;11(4):915–9. [DOI] [PubMed] [Google Scholar]
- 14.Parikh NS, Merkler AE, Schneider Y, Navi BB, Kamel H. Discharge disposition after stroke in patients with liver disease. Stroke. 2017;48(2):476–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Alexander M, Loomis AK, van der Lei J, Duarte-Salles T, Prieto-Alhambra D, Ansell D, et al. Non-alcoholic fatty liver disease and risk of incident acute myocardial infarction and stroke: findings from matched cohort study of 18 million European adults. BMJ. 2019;367:l5367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kang MK, Kim TJ, Kim Y, Nam KW, Jeong HY, Kim SK, et al. Geriatric nutritional risk index predicts poor outcomes in patients with acute ischemic stroke-automated undernutrition screen tool. PLoS ONE. 2020;15(2):e228738. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988;19(5):604–7. [DOI] [PubMed] [Google Scholar]
- 18.Cao C, Wang TB, Hu H, Han Y, Zhang X, Wang Y. Relationship between glycated hemoglobin levels and three-month outcomes in acute ischemic stroke patients with or without diabetes: a prospective Korean cohort study. BMC Neurol. 2024;24(1):85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zan Y, Xiong W, Zhang X, Han Y, Cao C, Hu H, et al. Body mass index has a non-linear association with three-month outcomes in men with acute ischemic stroke: an analysis based on data from a prospective cohort study. Front Endocrinol. 2022;13:1041379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Brott T, Adams HJ, Olinger CP, Marler JR, Barsan WG, Biller J, et al. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989;20(7):864–70. [DOI] [PubMed] [Google Scholar]
- 21.Haggag H, Hodgson C. Clinimetrics: modified rankin scale (mRS). J PHYSIOTHER. 2022;68(4):281. [DOI] [PubMed] [Google Scholar]
- 22.Groenwold RH, White IR, Donders AR, Carpenter JR, Altman DG, Moons KG. Missing covariate data in clinical research: when and when not to use the missing-indicator method for analysis. Can Med Assoc J. 2012;184(11):1265–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377–99. [DOI] [PubMed] [Google Scholar]
- 24.Yang BL, Wu WC, Fang KC, Wang YC, Huo TI, Huang YH, et al. External validation of fatty liver index for identifying ultrasonographic fatty liver in a large-scale cross-sectional study in Taiwan. PLoS ONE. 2015;10(3):e120443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Sun W, Huang Y, Xian Y, Zhu S, Jia Z, Liu R, et al. Association of body mass index with mortality and functional outcome after acute ischemic stroke. Sci Rep. 2017;7(1):2507. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Matsuo R, Ago T, Kiyuna F, Sato N, Nakamura K, Kuroda J, et al. Smoking status and functional outcomes after acute ischemic stroke. Stroke. 2020;51(3):846–52. [DOI] [PubMed] [Google Scholar]
- 27.Giantin V, Franchin A, Seresin C, Toffanello ED, Simonato M, Panzuto G, et al. Hypertension after acute ischemic stroke in older patients: effect on neurological outcome. J Am Geriatr Soc. 2008;56(10):1971–3. [DOI] [PubMed] [Google Scholar]
- 28.Haneuse S, VanderWeele TJ, Arterburn D. Using the e-value to assess the potential effect of unmeasured confounding in observational studies. JAMA-J Am Med Assoc. 2019;321(6):602–3. [DOI] [PubMed] [Google Scholar]
- 29.Si R, Xiao J, Zheng K, Yin Y, Li Y. Association between the hepatic steatosis index and risk of incident type 2 diabetes mellitus in the normoglycemic population: a longitudinal prospective study in Japan. Diabet Metab Synd Ob. 2024;17:2317–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wang C, Cai Z, Deng X, Li H, Zhao Z, Guo C, et al. Association of hepatic steatosis index and fatty liver index with carotid atherosclerosis in type 2 diabetes. Int J Med Sci. 2021;18(14):3280–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Ochiai H, Shirasawa T, Yoshimoto T, Nagahama S, Sakamoto K, Azuma M, et al. Hepatic steatosis index and chronic kidney disease among middle-aged individuals: a large-scale study in Japan. Dis Markers. 2021;2021:9941834. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Kunutsor SK, Bakker S, Blokzijl H, Dullaart R. Associations of the fatty liver and hepatic steatosis indices with risk of cardiovascular disease: interrelationship with age. Clin Chim Acta. 2017;466:54–60. [DOI] [PubMed] [Google Scholar]
- 33.Canillas L, Soriano-Varela A, Rodriguez-Campello A, Giralt-Steinhauer E, Cuadrado-Godia E, Broquetas T. High prevalence of non-alcoholic fatty liver disease in patients with a first episode of acute ischemic stroke. Impact on disability and death. Front Endocrinol. 2022;13:1003878. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Baik M, Kim SU, Nam HS, Heo JH, Kim YD. The paradoxical protective effect of liver steatosis on severity and functional outcome of ischemic stroke. Front Neurol. 2019;10:375. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Qin J, Zhang T, Chen Y, Wei X, Yang Y, Yuan Y, et al. The effect of body mass index on stroke prognosis: a systematic review and meta-analysis of 32 cohort studies with 330,353 patients. Int J Stroke. 2024;19(10):1093–101. [DOI] [PubMed] [Google Scholar]
- 36.Vemmos K, Ntaios G, Spengos K, Savvari P, Vemmou A, Pappa T, et al. Association between obesity and mortality after acute first-ever stroke: the obesity-stroke paradox. Stroke. 2011;42(1):30–6. [DOI] [PubMed] [Google Scholar]
- 37.Scherbakov N, Dirnagl U, Doehner W. Body weight after stroke: lessons from the obesity paradox. Stroke. 2011;42(12):3646–50. [DOI] [PubMed] [Google Scholar]
- 38.Rodriguez-Castro E, Rodriguez-Yanez M, Arias-Rivas S, Santamaria-Cadavid M, Lopez-Dequidt I, Hervella P, et al. Obesity paradox in ischemic stroke: clinical and molecular insights. Transl Stroke Res. 2019;10(6):639–49. [DOI] [PubMed] [Google Scholar]
- 39.Kimura Y, Yamada M, Kakehi T, Itagaki A, Tanaka N, Muroh Y. Combination of low body mass index and low serum albumin level leads to poor functional recovery in stroke patients. J Stroke Cerebrovasc. 2017;26(2):448–53. [DOI] [PubMed] [Google Scholar]
- 40.Khan RS, Bril F, Cusi K, Newsome PN. Modulation of insulin resistance in nonalcoholic fatty liver disease. Hepatology. 2019;70(2):711–24. [DOI] [PubMed] [Google Scholar]
- 41.Francisco V, Sanz MJ, Real JT, Marques P, Capuozzo M, Ait Eldjoudi D, et al. Adipokines in non-alcoholic fatty liver disease: Are we on the road toward new biomarkers and therapeutic targets? Biology. 2022. 10.3390/biology11081237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Nasiri-Ansari N, Androutsakos T, Flessa C, Kyrou I, Siasos G, Randeva HS, et al. Endothelial cell dysfunction and nonalcoholic fatty liver disease (NAFLD): a concise review. Cells. 2022. 10.3390/cells11162511. [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
Data can be accessed and downloaded from the “PLoS One” database (10.1371/journal.pone.0228738).