2025 Volume 266 Issue 2 Pages 153-160
Histon deacetylase 4 (HDAC4) modulates memory and cognitive impairment, but its association with post-stroke cognitive impairment (PSCI) is unclear. This study aimed to investigate the potential of HDAC4 for predicting PSCI risk. Sixty-nine PSCI patients and 70 control post-stroke (CPS) patients were enrolled in this case-control study. In all stroke patients, HDAC4 in peripheral blood mononuclear cells was detected by quantitative polymerase chain reaction; T-helper 17 (Th17) cells were detected by flow cytometry, and interleukin-17A was detected by enzyme-linked immunosorbent assay. HDAC4 was reduced in PSCI patients compared with CPS patients (P = 0.001). In total stroke patients, HDAC4 showed negative linkages with age (P = 0.003), history of diabetes (P = 0.012), stroke recurrence (P = 0.001), Th17 cells (P = 0.027), and interleukin-17A (P = 0.002). Additionally, multivariate logistic regression analysis revealed that HDAC4 (per unit) [odds ratio (OR) = 0.438, P = 0.024] was independently associated with a lower PSCI risk, but age (per unit) (OR = 1.061, P = 0.016) and multifocal disease (yes vs. no) (OR = 2.490, P = 0.014) were independently associated with a higher PSCI risk. By receiver operator characteristic curves, HDAC4 had an acceptable value for predicting PSCI risk [area under the curve (AUC) = 0.656, 95% confidence interval = 0.566-0.746]. The combination of HDAC4, age, and multifocal disease showed a good value for predicting PSCI risk (AUC = 0.728, 95% confidence interval = 0.644-0.811). HDAC4 may serve as a potential biomarker for predicting PSCI risk, which could facilitate the early screening and prevention of PSCI, thus promoting the management of stroke.
Stroke is a major cause of disability and mortality in human beings. This disease attacks over 12 million cases and induces over 6.5 million deaths annually worldwide (GBD 2019 Stroke Collaborators 2021). The treatments for stroke mainly include thrombolysis or thrombectomy within a certain period of time window for ischemic stroke, as well as blood pressure and blood glucose management plus surgery (if necessary) for hemorrhage stroke (Mendelson and Prabhakaran 2021; Magid-Bernstein et al. 2022). However, patients with stroke may still develop cognitive impairment or even dementia under appropriate treatment (Rost et al. 2022). The occurrence of post-stroke cognitive impairment (PSCI) is associated with neuron cell death, change of synaptic plasticity, inflammatory response, oxidative stress, and dysregulated immune response including high level and infiltration of T-helper 17 (Th17) cells (Mijajlovic et al. 2017; Gottfried et al. 2021; Chi et al. 2023; Goncharov et al. 2024). PSCI may induce sleep disorder, depression, and even physical disability, which also poses a huge burden for caregivers (Lanctot et al. 2020; Stolwyk et al. 2024). Therefore, early prediction of the risk of PSCI is vital for its prevention and intervention.
Histone deacetylase 4 (HDAC4) modulates the transcription of genes through regulating histone acetylation. It participates in the pathogenesis of various diseases, including cancers, diabetes mellitus, and cardiovascular diseases (Gong et al. 2019; Jiang et al. 2020; Cuttini et al. 2023). Moreover, it is also reported that HDAC4 could modulate memory and cognitive impairment (Fitzsimons et al. 2013). For instance, a previous study reported that either overexpression or knockdown of HDAC4 in the mushroom body, a vital structure for memory formation, impaired long-term memory in drosophila. This study suggested that a normal level of HDAC4 was required for the maintenance of long-term memory (Fitzsimons et al. 2013). Another study showed that a high level of HDAC4 in the hippocampus induced neuronal apoptosis, thus facilitating cognitive impairment (Xu et al. 2023). Apart from that, HDAC4 could also regulate cognitive impairment through potentially inhibiting Th17 differentiation. Several studies have revealed that HDAC4 was negatively correlated with Th17 proportion and response (Lu et al. 2015; Pei et al. 2018; Dou et al. 2022). Meanwhile, Th17 cells are responsible for neuroinflammation and infiltration of neutrophiles in the brain to cause cognitive impairment (Cipollini et al. 2019). Based on the above evidence, we supposed that HDAC4 might serve as a biomarker for PSCI.
The peripheral blood mononuclear cells (PBMCs), containing lymphocytes and monocytes, have been considered as a valuable and easy-acquired specimen of biomarkers (Alexovic et al. 2024). Various studies have used PMBCs as specimen to evaluate biomarkers in different diseases such as diabetes, stroke, and cancers (Mosallaei et al. 2022). Considering that HDAC4 might regulate cognitive function through Th17, we assume that HDAC4 in PMBCs might act as a potential biomarker for PSCI.
The current study detected HDAC4 in PSCI and control post-stroke (CPS) patients, aiming to investigate the potential of HDAC4 for predicting the risk of PSCI.
In this case-control study, a total of 69 PSCI patients were enrolled as the case group, which were identified according to a Chinese expert consensus (Chinese Society of Vascular Cognitive Impairment 2021). The inclusion criteria were: 1) diagnosed with stroke through clinical symptoms and imaging examinations; 2) identified as PSCI; 3) aged more than 18 years old; 4) cooperated with blood sample collection. A total of 70 CPS patients were enrolled as the control group. The inclusion criteria of the control group were the same as for the cases except the presence of PSCI. The exclusion criteria for both case and control groups were similar, which included: 1) patients with cognitive impairment before stroke; 2) patients with cancers or hematological malignancies; 3) female patients who were pregnant or lactational. Patients in both groups were enrolled from April 2022 to May 2023 with a ratio set as 1:1. Age and sex were not matched by intervention between PSCI and CPS groups. This study obtained Ethics Committee approval and written informed consent from patients.
Data and sample collectionClinical characteristics containing demographics and factors that commonly affected cognitive impairment in stroke patients (both groups) were collected. These characteristics included age, sex, body mass index (BMI), nationality, education degree, history of smoke, history of hypertension, history of hyperlipidemia, history of diabetes, multifocal, stroke recurrence, and stroke type. Besides, peripheral blood samples were collected after enrollment, and each sample was divided into two parts for further detection. One part of the blood sample was used to separate peripheral blood mononuclear cells (PBMCs) for detecting HDAC4 expression by quantitative polymerase chain reaction assay and Th17 cells by flow cytometry assay; while the other part was used to separate serum for detecting interleukin-17A (IL-17A) by enzyme-linked immunosorbent assay.
HDAC4 expression detectionThe HDAC4 expression in PBMCs was detected using the following kits: Trizol (No. Cat: R0016, Beyotime, China), cDNA Synthesis Kit with gDNA EZeraser (No. Cat: D7180S, Beyotime, China), and Talent qPCR PreMix (SYBR Green) (No. Cat: FP209, TianGen Biotech, China). GAPDH was set as a reference for normalization, and the PBMC HDAC4 expression was figured out via the 2-∆∆Ct method (the median value of CPS was set as control ‘1’). The list of primers used can be found in Supplementary Table S1.
Statistical analysisData analyses were completed using SPSS 26.0 (IBM, USA) software. Data were described as mean ± standard deviation (SD), median with interquartile range (IQR), or number with percentage. The comparison between groups was analyzed by the student t-test, Chi-square test, or Mann-Whitney U test. The correlation analyses were conducted using Spearman’s test. All the clinical characteristics of stroke patients were included in the univariate logistic regression analyses. The independent factors of PSCI risk were explored by the forward-stepwise multivariate logistic regression analysis. The predictive ability of HDAC4 expression or the combined factors (independent risk factors in the multivariate analysis) was shown by receiver operator characteristic (ROC) curves. A P value less than 0.05 was considered statistical significance.
PSCI patients had a mean age of 68.5 ± 7.5 years, consisted with 35 (50.7%) males and 34 (49.3%) females; there were 43 (62.3%) patients with multifocal disease and 21 (30.4%) patients with stroke recurrence. CPS patients had a mean age of 63.9 ± 8.4 years with 45 (64.3%) males and 25 (35.7%) females; 26 (37.1%) patients had multifocal disease and 11 (15.7%) patients had stroke recurrence.
PSCI patients had a higher mean age (P = 0.001), less advanced education degree (P = 0.034), and higher proportions of history of diabetes (P = 0.026), multifocal disease (P = 0.003), and stroke recurrence (P = 0.039) compared with CPS patients. Moreover, the proportion of Th17 cells [median (IQR): 3.0 (2.4-4.2) % vs. 2.4 (2.0-3.6) %, P = 0.015] and IL-17A level [median (IQR): 71.7 (56.5-94.2) pg/mL vs. 54.9 (42.5-95.1) pg/mL, P = 0.005] were both higher in PSCI patients compared with CPS patients. Other characteristics, including sex, BMI, nationality, history of smoke, history of hypertension, history of hyperlipidemia, and stroke type, were not different between PCSI patients and CPS patients. More detailed information is presented in Table 1.
Clinical characteristics of stroke patients.
§1. To test the normality of continuous variables, the Kolmogorov-Smirnov test was employed. Age and BMI were described as mean ± standard deviation, which followed a normal distribution. Th17 cells and IL-17A were described as median (interquartile range), following nonnormal distribution. Other dichotomous variables left in the table were described as numbers (percentages).
§2. The superscript ‘a’ represented that the student t-test determined the P value between PSCI patients and CPS patients, and the corresponding effect size was ‘t’. The superscript ‘b’ represented that the Chi-square test determined the P value between PSCI patients and CPS patients, and the corresponding effect size was ‘Chi-square’. The superscript ‘c’ represented that the P value was determined by the Mann-Whitney U test between PSCI patients and CPS patients, and the corresponding effect size was ‘Z’.
§3. The data for Th17 cells were partially missing. In PSCI patients, data for 6 patients were missing, whereas in CPS patients, data for 12 patients were missing.
PSCI, post-stroke cognitive impairment; CPS, control post-stroke; BMI, body mass index; Th17 cells, T helper 17 cells; IL-17A, interleukin-17A.
The median (IQR) HDAC4 expression in stroke patients was 0.90 (0.68-1.36). Comparison analysis showed that HDAC4 expression was reduced in PSCI patients compared with CPS patients [median (IQR): 0.80 (0.55-1.08) vs. 1.00 (0.76-1.66), P = 0.001] (Fig. 1).
HDAC4 expression in total stroke patients, PSCI patients, and CPS patients.
Mann-Whitney U test was used to analyze the difference in HDAC4 expression between CPS and PSCI patients.
HDAC4 was negatively correlated with age in stroke patients (r = 0.247, P = 0.003). Meanwhile, lower HDAC4 expression was associated with history of diabetes (P = 0.012) and stroke recurrence (P = 0.001). However, no relationship was found in HDAC4 expression with BMI, sex, nationality, education degree, history of smoke, history of hypertension, history of hyperlipidemia, multifocal disease, or stroke type (all P > 0.05) (Table 2).
Moreover, HDAC4 expression was negatively correlated with Th17 cells (P = 0.027) and IL-17A (P = 0.002). Th17 cells and IL-17A were positively correlated with each other (P < 0.001) (Table 3).
Correlation of HDAC4 expression with characteristics of stroke patients.
§1. To test the normality of HDAC4 expression, the Kolmogorov-Smirnov test was employed. HDAC4 expression was described as median (interquartile range), which followed a nonnormal distribution.
§2. The superscript ‘a’ represented that the Spearman test determined the P value between HDAC4 expression and continuous variables, and the corresponding effect size was ‘r’. The superscript ‘b’ represented that the P value was determined by the Mann-Whitney U test between HDAC4 expression and categorical variables, and the corresponding effect size was ‘Z’.
HDAC4, histone deacetylase 4; BMI, body mass index.
Correlation of HDAC4 expression with Th17 cells and IL-17A of stroke patients.
§. The Spearman test determined the P value, and the corresponding effect size was ‘r’. HDAC4, histone deacetylase 4; Th17 cells, T helper 17 cells; IL-17A, interleukin-17A.
HDAC4 expression (per unit), which means every increase in HDAC4 expression by 1, was associated with a lower PSCI risk [odds ratio (OR) = 0.355, P = 0.003]. Age (per unit) (OR = 1.076, P = 0.001), education degree (middle school and below vs. high school and above) (OR = 2.074, P = 0.035), history of diabetes (yes vs. no) (OR = 2.260, P = 0.028), multifocal disease (yes vs. no) (OR = 2.799, P = 0.003), stroke recurrence (yes vs. no) (OR = 2.347, P = 0.042), and IL-17A (per unit) (OR = 1.013, P = 0.022) were associated with a higher PSCI risk (Table 4). By multivariate logistic regression analysis, HDAC4 expression (per unit) (OR = 0.438, P = 0.024) was independently associated with a lower PSCI risk; while age (per unit) (OR = 1.061, P = 0.016) and multifocal disease (yes vs. no) (OR = 2.490, P = 0.014) were independently associated with a higher PSCI risk (Table 5).
Univariate logistic regression analysis of PSCI risk.
PSCI, post-stroke cognitive impairment; OR, odds ratio; CI, confidence interval; HDAC4, histone deacetylase 4; BMI, body mass index; Th17 cells, T helper 17 cells; IL-17A, interleukin-17A.
Multivariate logistic regression analysis of PSCI risk.
§. HDAC4 expression, sex, body mass index (BMI), nationality, education degree, history of smoke, history of hypertension, history of hyperlipidemia, history of diabetes, multifocal, stroke recurrence, stroke type, and IL-17A were included in the analysis.
PSCI, post-stroke cognitive impairment; OR, odds ratio; CI, confidence interval; HDAC4, histone deacetylase 4.
ROC analysis showed that the area under the curve (AUC) [95% confidence interval (CI)] of HDAC4 expression for predicting PSCI risk was of 0.656 (0.566-0.746), indicating an acceptable predictive value (Fig. 2A). Nevertheless, after combining HDAC4 expression with age and multifocal disease, the AUC (95% CI) for predicting PSCI risk increased to 0.728 (0.644-0.811), indicating a good predictive value (Fig. 2B).
The ability of HDAC4 to predict PSCI risk.
HDAC4 to predict PSCI risk (A). The combination of HDAC4, age, and multifocal disease to predict PSCI risk (B). ROC analysis with AUCs were used to analyze the predictive value of variables for PSCI risk.
Previous studies have identified that HDAC4 was dysregulated in patients with diseases associated with cognitive impairment (Shen et al. 2016; Wang et al. 2023a). For instance, it was discovered that HDAC4 was markedly elevated in the brain of patients with Alzheimer’s disease (Shen et al. 2016). Another study observed that HDAC4 was increased in the locus coeruleus and substantia nigra of patients with Braak stage 4 and 5 Parkinson’s disease (Wang et al. 2023a). The current study revealed that HDAC4 was reduced in PSCI patients compared with CPS patients, which was different from the findings of previous studies (Shen et al. 2016; Wang et al. 2023a). The possible explanation for this discrepancy could be that: previous studies detected HDAC4 in the brain tissues of patients with Alzheimer’s disease or Parkinson’s disease (Shen et al. 2016; Wang et al. 2023a), but the current study compared HDAC4 in the PBMCs of PSCI and CPS patients. The difference in tissue/cell where HDAC4 was hosted could result in the diverse effect of HDAC4 on cognitive impairment. In detail, HDAC4 overexpression in the brain tissues promoted apoptosis of hippocampal neurons and impairs synaptic plasticity to induce cognitive impairment (Mielcarek et al. 2013; Xu et al. 2023); while HDAC4 insufficiency in the cluster of differentiation (CD)4+ T cells might facilitate the differentiation of Th17 cells to induce cognitive impairment (Dou et al. 2022; Wang et al. 2023b).
The current study also revealed negative linkages of HDAC4 with Th17 cells, IL-17A, age, history of diabetes, and stroke recurrence in stroke patients. The possible explanations were as follows: (1) HDAC4 negatively regulated the differentiation of Th17 cells (Dou et al. 2022), which resulted in its negative correlation with Th17 cells and IL-17A in stroke patients. (2) HDAC4 was negatively correlated with Th17 cells. Meanwhile, Th17 cells were involved in age-associated senescence (Lee et al. 2011; Li et al. 2017). Therefore, our study found that HDAC4 was negatively correlated with age in stroke patients. (3) HDAC4 might inhibit the pathogenesis of diabetes. According to a study, HDAC4 knockdown mice had higher levels of circulating glucose, free fatty acid, and insulin compared with wild-type mice when treated with a high-fat diet. This study suggested that HDAC4 knockdown facilitated insulin resistance and contributed to the pathogenesis of diabetes (Luan et al. 2014). Therefore, our study revealed that HDAC4 was negatively associated with history of diabetes. (4) Th17 cells were associated with stroke recurrence according to a previous study (Yu et al. 2022). Therefore, the current study found that HDAC4 was negatively associated with stroke recurrence.
More importantly, our study also found that HDAC4 showed an acceptable ability to predict the risk of PCSI, and it was independently associated with a lower risk of PCSI. These findings could be explained by the ability of HDAC4 dysregulation to induce cognitive impairment as mentioned above (Mielcarek et al. 2013; Dou et al. 2022; Wang et al. 2023b; Xu et al. 2023). Apart from HDAC4, age and multifocal disease were independently and positively associated with PSCI risk in our study. These variables are well-recognized risk factors for PCSI (Mijajlovic et al. 2017). In order to improve the predictive performance of HDAC4 to the risk of PCSI, we further combined it with age and multifocal disease. The data revealed that performance for predicting PSCI risk was promoted after combining HDAC4 with age and multifocal disease. These findings suggested that the combination of HDAC4, age, and multifocal disease could serve as a predictive tool to identify PSCI risk. In clinical practice, the combination of HDAC4, age, and multifocal could be applied as a predictive model for PSCI risk. Clinicians could use this model to identify patients with a high risk of PSCI, thus implementing more radical treatments to these patients, thus improving their prognosis. Nevertheless, the AUC of HDAC4 combined with age and multifocal disease was still not great enough, which meant that predicting PSCI was not an easy issue. Therefore, further studies are required to identify biomarkers with higher performance to predict PSCI risk.
Our study revealed the potential of HDAC4 for predicting PSCI risk, which had not been reported before. Meanwhile, our study detected HDAC4 from PBMCs, which were easy to acquire. However, several limitations should be clarified. First, the sample size was not large enough, and the predictive value of HDAC4 for PSCI risk should be verified in a larger sample size. Second, the current study did not enroll a validation cohort to confirm the predictive value of HDAC4 for PSCI risk. Third, this study only enrolled stroke patients, and the predictive value of HDAC4 for cognitive impairment in other diseases, such as Alzheimer’s disease and Parkinson’s disease, should be further investigated. Fourth, the association of HDAC4 in other specimen with PSCI risk, such as cerebrospinal fluid, should be explored in further studies.
Collectively, HDAC4 is negatively linked with Th17 cells, age, history of diabetes, and stroke recurrence in stroke patients; meanwhile, its low expression reflects a high risk of PSCI. The findings of this study indicate that HDAC4 may be applied for the early screening and prevention of PSCI.
The authors declare no conflict of interest.
