Low HALP (Hemoglobin, Albumin, Lymphocyte, and Platelet) Score Increases the Risk of Post-Stroke Cognitive Impairment: A Multicenter Cohort Study

Introduction

Stroke is one of the most common causes of deaths and disability, and it is highly associated with an increased risk of cognitive impairment.¹ Non-disabling cerebrovascular events including mild ischemic stroke (median NIHSS=2, median mRS=2) and transient ischemic attack (TIA) generally result in either short-lasting or mild neurological symptoms,² but these patients are at an increased risk of a recurrent cerebrovascular event and often suffered from cognitive impairment. Inflammation and malnutrition may be involved in post-stroke cognitive impairment (PSCI).^3,4 Some studies have demonstrated that lymphocytes could repair the injury via inflammation⁵ and were important immunomodulators for protection after acute ischemic stroke.⁶ Decreased lymphocytes level might indicate a poor outcome in systemic inflammatory disease. Recently, low lymphocyte to monocyte ratio was reported to be novel predictors of morbidity and mortality after ischemic stroke.^7,8 One study found that the cardiovascular surgery process could increase monocyte and decrease lymphocyte quantities, and the decreased Lymphocyte-to-monocyte ratio was associated with cognitive dysfunction after cardiovascular surgery.⁹ Ischemic stroke may result in impaired platelet function and excessive activation and accumulation of platelets could hamper stroke recovery,¹⁰ while a high platelet-to-lymphocyte ratio has been observed in patients with symptomatic internal carotid artery stenosis,¹¹ poor functional stroke outcomes,¹² and cognitive impairment.¹³ One study found a link between increased platelet prothrombotic potential at the time of stroke and subsequent development of vascular cognitive impairment.¹⁴

Malnutrition is associated with PSCI. Stroke patients with low hemoglobin levels have been found to harbor increased risks of PSCI.¹⁵ Studies reported that decreased hemoglobin levels might lead to brain hypoxia, mitochondrial dysfunction, and neuronal damage. Patients who have ischemia stroke with hypoxic conditions are much more susceptible to Alzheimer’s disease.¹⁶ Moreover, a population-based study demonstrated that anemia was associated with the incidence of dementia and cognitive decline, and severe anemia may lead to an increased risk of dementia.¹⁷ Hemoglobin level is a significant independent predictor of PSCI.¹⁵ Serum albumin is widely considered a surrogate of nutrition where deficiencies in serum albumin have been associated with poorer stroke outcomes,¹² explained in part by an impaired antioxidant capacity to protect neurons from ischemic injury.¹⁸ The geriatric nutritional risk index based on serum albumin has been similarly implicated in PSCI.¹⁹ Ischemic stroke patients with low albumin levels could not maintain normal neuronal metabolism and experience unfavorable outcomes. Moreover, albumin has a strong antioxidative activity that might protect the neurons from ischemic injury.¹⁸

In view of the above, we aggregated these inflammatory and nutritional indicators to explore the association between inflammation, nutrition, and PSCI in this study. In patients with multiple malignancies, the HALP (hemoglobin, albumin, lymphocyte, and platelet) score is a readily accessible measure of systemic inflammation and nutritional status linked with prognosis.^20,21 Recent studies have shown that the HALP score could predict stroke recurrence and poor outcome.^22,23 There is, however, only one study reporting the association between the HALP score and short-term PSCI, ie, within 2 weeks at subacute stroke phase.²¹

Malnutrition and inflammatory states can develop frailty. Frailty is a multidimensional geriatric syndrome manifested with accumulated deficits (energy, physical ability, cognition, and health) that give rise to vulnerability. A higher frailty index (FI) indicates more severe frailty.²⁴ Recent reports suggest that frailty index based on common laboratory tests (FI-Lab) may serve as a diagnostic tool for frailty and is associated with mortality.²⁵ Acute cerebral infarction subjects patients to adverse events, such as pneumonia²⁶ and cognitive disorders.²⁷ Pre-stroke frailty is often measured by the modified Rankin scale (mRS).

Hence, we aim to explore the relationships between the HALP score and PSCI both at subacute stroke phase and 1 year late.

Methods

Study Population

Data of this study are drawn from the Impairment of Cognition and Sleep (ICONS) study. ICONS is one of the research subgroups of China National Stroke Registry-III (CNSR-III) database. The CNSR-III is a nationwide prospective hospital-based registry for patients with acute ischemic stroke (AIS) or TIA within 7 days performed in August 2015 to March 2018 in China.²⁸ Patients’ stroke/TIA were diagnosed by neurologists and confirmed with brain computed tomography or magnetic resonance imaging. The acute ischemic stroke was diagnosed according to World Health Organization criteria.²⁹ TIA was defined by the American Stroke Association.³⁰ All the participating hospitals approved the protocol. All participants or their legal proxies signed written informed consent. A total of 40 study sites participated in ICONS subgroup. Finally, 1022 patients received the Montreal Cognitive Assessment (MoCA)-Beijing and HALP (hemoglobin, albumin, lymphocyte, and platelet) scores at baseline (within two weeks of index event) and at 1 year.

The study population included patients (1) aged > 18 years older; (2) with a diagnosis of AIS or TIA within 7 days;

The exclusion criteria included: (1) silent cerebral infarction diagnosed by MRI or CT without symptoms or signs; (2) patients who were illiteracy; (3) history of cognitive impairment, psychosis, or schizophrenia disease (documented in medical records); (4) physical disability affecting cognitive tests, eg, visual or hearing disorders, severe unilateral neglect, severe aphasia [defined as National Institutes of Health Stroke Scale (NIHSS) item 9>2].

Data collection for ICONS was approved by the ethics committee of all participating hospitals.²⁸ All patients or participants provided their written informed consent for this study.

Baseline Data Collection

Baseline data, including age, gender, body mass index (BMI calculated as weight in kilograms divided by the square of height in meters, km/m²), smoking, drinking, body mass index, educational level, and medical histories (hypertension, diabetes, hyperlipidemia, coronary heart disease, atrial fibrillation, heart failure, stroke, epilepsy, smoking, and drinking) were collected.

A detailed physical examination was completed, and several parameters, including the mRS, Trial of ORG 10172 in Acute Stroke Treatment (TOAST) type and NIHSS score. Participants’ anxiety and depression status were assessed by the 7-item Generalized Anxiety Disorder Scale (GAD) and Patient Health Questionnaire-9 (PHQ-9). The blood samples were gathered within 24h of admission and transported through the cold chain to the central laboratory. All serum samples were kept frozen at −80◦C until ready for assay.

Sample Collection and Measurement

Fasting whole blood samples were routinely processed within 24 hours of admission and were tested at study sites or the center laboratory in Beijing Tiantan Hospital. Platelets, lymphocyte counts, and hemoglobin concentration were measured by automated hematology analyzers at each participating hospital. Serum albumin was also tested in participating hospital within 24 hours of admission. Bromocresol purple assay or bromocresol green assay were used to analyze the serum albumin levels according to the test reagent of participating hospital. The following formula was used to calculate HALP scores: hemoglobin (g/L) × albumin (g/L) × lymphocytes (/L) / platelets (/L).³¹ The patients were categorized according to quartiles of the HALP score.

Cognitive Assessment

The original MoCA is the most commonly used cognitive-screening scale with 30 points, including Visuospatial/executive abnormal (5 points), Naming (3 points), Attention (2 points), Language (2 points), Delayed recall (5 points) Orientation (6 points). There were several official language versions of the MoCA, and the cutoffs are different. The MoCA-Beijing was developed with some modifications as previous study presented.³² According to our previous work, a MoCA-Beijing cutoff score of 22/23 is optimally sensitive and specific for detecting cognitive impairment after mild stroke and TIA in the acute stroke phase.³³ One point was added to the total score for those with education <12 years. All the participants were evaluated for cognitive status at 2-weeks, 3-months, and 1-year by trained neurologists at each center.

Assessment of Pre-Stroke Frailty Status

In stroke, the mRS is often used for measuring pre-stroke dependency. Pre-stroke mRS > 2 was also considered as pre-stroke frailty.^34,35

Outcome Evaluation

The MoCA-Beijing was used to assess global cognition and detect PSCI. The MoCA-Beijing was developed with the following modifications as shown by a previous study.³² The primary outcome was PSCI at 2 weeks and at 1 year after AIS or TIA. Patients who have MoCA-Beijing ≤22 were regarded as having cognitive impairment, determined by formal neuropsychological tests.³³ The optimal cutoff point of MoCA-Beijing ≤22 has excellent discriminant ability (sensitivity: 85% and specificity: 88%).³³

Statistical Analysis

SAS 9.4 (SAS Institute Inc, Cary, NC) was used for data analysis. Measurement data were tested for normality using the Kolmogorov–Smirnov test. Continuous variables with normal distribution were expressed as mean± standard deviation (mean±SD), and statistical differences were calculated by analysis of variance. Skewed data were described as median (interquartile range) and analyzed using nonparametric tests. Categorical variables were analyzed by the Chi square test. Patients in this study were categorized into 4 groups by HALP quartiles. Multivariable logistic regression analysis was used to evaluate the associations among HALP and PSCI. All independent variables included in the multivariate analysis were independent from each other, and there was no multicollinearity (tolerance of all independent variables was greater than 0.1). The Box–Tidwell method calculated that there was a linear relationship between the independent variables included in the multivariate model and the logit transformation value of the dependent variable Two analysis models were performed with adjustment for demographics only and together with clinical factors. P < 0.05 was considered statistically significant.

Results

Baseline Characteristics

There were 2625 patients enrolled in the ICONS study. We excluded 1603 patients missing HALP on admission and MoCA results during the 1-year follow-up. Finally, 1022 patients were included in the current analysis. Most patients were at their 60s (61.59±11.00 years old), and 746 (72.99%) participants were male. The HALP score ranges based on quartiles (Q1–Q4) were defined as follows: Q1<36.56, Q2 36.56–48.83, Q3 48.83–65.08, and Q4>65.08. The differences in demographics and clinical characteristics of patients among different HALP scores is shown in Table 1. Patients in Q1 quartile are significantly older, have a more prevalent smoking history, have lower BMI, and lower scores in scores at 2 weeks and 1 year (Table 1). They also had lower scores on pre-stroke mRS, 2 week-mRS and 1-year mRS (Table 1). In cognitive sub-test domains, patients with HALP scores in Q1 layer had significantly lower scores than Q4 layer in executive function (3.16±1.65 vs 3.60±1.48, p=0.003), naming (2.66±0.73 vs 2.82±0.56, p=0.017) and attention (5.18±1.28 vs 5.42±1.03, p=0.044) (Table 1).

Table 1 Demographic, Clinical Characteristics and Outcomes of Patients with Different HALP Score Ranges

Logistic Regression Analysis

A total of 503 participants (49.21%) suffered from PSCI at 2 weeks and 331 participants (32.39%) suffered from PSCI at 1-year follow-up (Table 2). The 2-week incidence of PSCI with different HALP scores (from low to high, ie, Q1 to Q4) were 54.86%, 46.90%, 49.80%, and 45.31%, respectively. After adjustment, the HALP score was not significantly associated with PSCI at 2-weeks follow-up (Table 2). The 1-year incidence of PSCI with different HALP scores (from low to high, ie, Q1 to Q4) were 40.84%, 31.78%, 30.28%, and 26.56%, respectively (Table 3). After adjustment for conventional covariables (Model 1) and further adjustment for smoking, history of stroke, hypertension, diabetes, coronary heart disease, atrial fibrillation, TOAST type, NIHSS score, mRS, and MoCA scores at baseline in model 2, patients in the lowest quartile of HALP score (Q1) had worse cognitive function (MoCA≤22) compared with the highest quartile [quartile (Q) 1 vs Q4: adjusted odds ratio (aOR) 1.13, 95% confidence interval: 0.74–1.75, p=0.022; Q2 vs Q4: aOR 1.16, 95% confidence interval: 0.75–1.77, p = 0.32; Q3 vs Q4: aOR 1.57, 95% confidence interval: 1.02–2.41, p=0.38]. (Figure 1).

Table 2 Logistic Regression Analysis of Cognitive Outcomes at 2 Weeks in AIS/TIA Participants

Table 3 Logistic Regression Analysis of Cognitive Outcomes at 12 Months in AIS/TIA Participants

Figure 1 Percentage of patients with PSCI at 1 year.

Subgroup Analysis

HALP score for 1-year incidence of cognitive impairment and ORs for cognitive impairment at 1-year by age, sex, and etiology of stroke are shown in Table 4. After adjusting for all potential confounding variables, the hazards or odds of PSCI were not modified by age, sex, and etiology of stroke (all P values for interaction >0.05). Significant associations were found between lower HALP scores and 1-year PSCI. Such associations are more pronounced in those who are older (≥60 years), male, and with the large artery atherosclerosis subtype (Figure 2).

Table 4 Subgroup Analysis of Association Between HALP Score and Cognitive Function

Figure 2 Subgroup analysis of association between HALP score quartiles and 1-year post-stroke cognitive impairment (PSCI) in patients with acute ischemic stroke/transient ischemic attack recurrence.

Discussion

In this prospective study, we explored the relationships between HALP score and short-and long-term cognitive impairment after AIS/TIA. Patients with a lower HALP score at admission have more risk of frailty and were more likely to suffer from short-and long-term PSCI, suggesting that low HALP score might be a potential risk factor for cognitive impairment in the short and long term following stroke. And there was no heterogeneity in the effects of HALP score levels on the PSCI between subgroups classified by age, gender, TOAST subtypes, and educational level.

Inflammation plays a critical role in the pathophysiology of stroke and dementia. Systemic inflammation may increase morbidity and mortality and cause cognitive decline in patients with stroke.³⁶ Systemic inflammation processes are thought to be closely associated with endothelial dysfunction, damaged blood–brain barrier and reduced cerebral blood flow, thus lead to cognitive impairment.³⁷ Leucocytes are important components of systemic inflammation and play a key role in immune regulation. Studies have shown that neuroprotective subtypes of lymphocytes decrease in response to stress-induced corticosteroids and lead to poor prognosis and neurological impairment after ischemic stroke.⁶ Previous studies have shown that immunosuppression and decreased levels of lymphocytes after a stroke lead to a poorer prognosis.³⁸

Platelets have a primary role in the thrombo-inflammation of stroke^10,12 and predict poor functional outcomes. When a stroke occurs, massive production of platelets accumulated to the injured endothelial regions. The inflammatory substances secreted by them can further recruit more inflammatory cells such as leukocytes to the site of injury, amplify the inflammatory response, and cause injury to both inside and outside blood vessels, as well as to neurons, eventually damaging brain tissue.³⁹ In contrast, lymphocytes are known to control inflammatory response by modulating and repairing inflammation during cerebral ischemia. The platelet-to-lymphocyte ratio has recently been reported as a potential novel biomarker in acute ischemic stroke intravenous thrombolysis treatment.^40,41 Apart from this, the platelet-to-lymphocyte ratio was found to be associated with lower cognitive performance in patients with breast cancer survivors and type 2 diabetes.⁴² This study has also shown the role of inflammation in long-term cognitive problems.

The prevalence of malnutrition risk in AIS patients ranged from 15.99% to 57.86%, which has been considered to adversely affect the prognoses of stroke and PSCI.^43,44 A total of 34 dietary factors were associated with stroke susceptibility.⁴⁵ Our study showed that the nutritional conditions evaluated at acute stroke stage might affect the cognitive performance 1 year after stroke. The findings are consistent with previous studies. They revealed that malnutrition was associated with poor cognitive improvement after an ischemic stroke.³ Furthermore, another study revealed that the calorie-protein supplements could improve global cognition after a stroke.⁴⁶ There are several potential mechanisms underlying the association between malnutrition and PSCI, including its influence on the stroke severity, and the premorbid vulnerability of specific brain structures. Malnutrition adversely affects neuronal plasticity and protein synthesis. It also correlates with increased risk for white matter hyperintensities, microbleeds, and mesial temporal lobe atrophy in the general population, and in patients with mild cognitive impairment or dementia.⁴⁷ Further, malnutrition often results in an immunosuppressive state, increasing the chance of infection during the acute stroke phase.

The levels of hemoglobin and albumin, which were markers of nutritional health, had also been found to be essential indicators for predicting the development of PSCI.¹⁹ Hemoglobin transported oxygen to the whole tissues, and the brain consumed nearly 20%. The decreased hemoglobin levels might lead to brain hypoxia, mitochondrial dysfunction, neuronal injury, oxidative stress, and inflammation. Both oxidative stress and inflammation were correlated with cognitive impairment in patients with ischemic stroke.⁴⁷ It is reported that anemia was associated with cognitive decline in general population, and severe anemia might

increase the risk of dementia.¹⁷ Another study showed that premorbid anemia was independently associated with an increased risk for PSCI after 1-month follow-up.⁴⁸

The HALP score, as an index, combines hemoglobin, albumin, lymphocyte, and platelet, which results in the unique advantage of linking key pathways in both inflammation and nutritional condition.²² The previous study proved that lower HALP score correlated with an increased risk of early-onset PSCI.⁴⁹ However, we did not find the relationship between HALP score and PSCI at 2 weeks in our study. Firstly, cognitive evaluation scales were different in the previous study and the present study [Mini-Mental State Examination (MMSE) vs MoCA], the MoCA has been reported to be sensitive to changes in acute cognitive impairment after mild stroke/TIA, whereas the MMSE is reportedly not.¹¹ Secondly, patients’ cognition was poorer in the previous study (average MMSE=18) than the present study (average MoCA=23), indicating HALP is related to poorer cognitive status at the subacute phase and the residual cognitive impairment over a longer term. Additionally, the cognitive status was not stable and could be affected by many factors such as delirium and mood at acute phase. Some patients might have deteriorated, while others improved or keeping stable, respectively. Our novel finding is that the HALP score could predict long-term PSCI at 1 year, which was relatively stable and residual cognitive impairment, prompting for early intervention and targeted management.

The hemoglobin, albumin, lymphocyte, and platelet are important components of laboratory tests for frailty index assessment.⁵⁰ Frailty represents multiple systems deficits such as comorbidity, weakness, malnutrition, and cognitive impairment. In stroke patients, the mRS>2 is often used for measuring stroke frailty.^34,35 In this study, the prevalence of pre-stroke frailty (defined as mRS>2) were high in Q1 group than that in other groups. We suggest that the HALP score could partly reflect frailty, as it is a stronger predictor of post-stroke cognitive impairment and also prognosis. In subgroup analysis, the relationship between HALP and PSCI was more significant in patients aged ≥60 years old. Individuals age biologically at different rates. Frailty increases multisystem decline and its vulnerability with age. In frailty, this decline is accelerated in older individuals, which is consistent with our study.⁵¹ The relationship between HALP and PSCI was more significant in males, suggesting possible sex bias with HALP score. Females usually live longer than men.⁵² In previous studies on older individuals, a greater risk for mortality in males than females was found.⁵³ Further study is required in this area.

Another novel finding of our study is to show differences in HALP scores under stroke etiology. In particular, among patients with large artery atherosclerosis, a lower HALP score was significantly associated with 1-year cognitive impairment. This might be due to different post-ischemic inflammatory responses and nutritional states in patients with different TOAST classifications. Atherosclerosis is due to chronic inflammation. In patients with large artery atherosclerosis, ischemic stroke might trigger a much stronger inflammatory response based on this chronic inflammatory process. Malnutrition may aggravate existing inflammation and accelerate atherosclerosis, worsening patients’ cognitive decline.⁵⁴

There are some limitations to our study. Firstly, the HALP score was measured according to the data derived on admission. However, the HALP score was likely to fluctuate, so we were unable to examine the associations of HALP dynamic changes with 1-year PSCI; Secondly, as we all known, the platelet and lymphocyte count were not stable at acute phase, they might be different and fluctuate during 7 days. And it might have an effect on the results. In future studies, we will refine the analysis at 1 and 7 days after stroke to get a better understanding of the association of peripheral inflammation with PSCI. Thirdly, all patients with AIS or TIA were enrolled consecutively, and a minority of patients with TIA were included in the study, leading to results less generalizable to TIA patients. Fourthly, patient with a history of severe cognitive impairment were excluded from this study. Moreover, we did not assess the cognitive function before the stroke. Cognitive assessment was mainly evaluated in mild stroke patients (median NIHSS=3, median mRS=1). Hence, patients with severe stroke or severe aphasia could not undertake cognitive evaluation, which may lead to the underestimate of cognitive function and impede results generalization.

Conclusions

To sum up, the risks from malnutrition and inflammation, represented as low HALP score at subacute stroke phase, were independently associated with PSCI at 12 months after the occurrence of ischemic stroke. The HALP score may help identify high-risk patients for PSCI, enhance early targeted stroke care (swallowing assessments, nutritional intervention) for these patients, and follow-up with further management, such as anti-inflammatory therapy and nutritional intervention.