Early elevated neutrophil‐to‐lymphocyte ratio associated with remote diffusion‐weighted imaging lesions in acute intracerebral hemorrhage

Abstract Aims To explore the relationship between the circulating neutrophil‐to‐lymphocyte ratio (NLR) and the remote diffusion‐weighted imaging lesions (R‐DWILs) after spontaneous intracerebral hemorrhage (ICH). Methods Consecutive patients with spontaneous ICH were prospectively collected from November 2016 to May 2018 and retrospectively analyzed. We included subjects who presented within 24 hours after symptom onset and were free of detectable infections on admission or in hospital. Blood samples were obtained at 24‐48 hours after ICH ictus, while all complete MRI scans were performed at 5‐8 days. R‐DWILs were defined as focal hyperintensities remote from the site of the ICH or the peri‐hematoma regions. NLR was calculated by dividing the absolute neutrophil counts by the absolute lymphocyte counts. Multivariate binary logistic regression models were generated to evaluate the relationship between NLR and R‐DWILs. Results One hundred sixty‐three subjects met eligibility criteria (age 62.3 ± 13.6 years, 60.7% males), of whom 31(19.0%) experienced R‐DWILs. Higher circulating NLR was documented in patients with R‐DWILs. With the best cutoff value of 6.01, elevated NLR was independently associated with the presence of R‐DWILs (OR = 3.170, 95% CI 1.306‐7.697, P = .011) in the bivariate logistic regression analysis with adjustment for age, sex, atrial fibrillation, previous ischemic stroke/TIA, SBP on admission, hematoma volume, and IVH. Conclusions This study provides significant evidence of the association between circulating NLR and R‐DWILs in spontaneous ICH patients. Patients with NLR > 6.01 at 24‐48 hours after ICH ictus should be paid more attention to when evaluating R‐DWILs.


| INTRODUC TI ON
Intracerebral hemorrhage (ICH) accounts for 6.5%-19.6% of cases of stroke. Despite advances in neurocritical care, it remains a devastating disease for high morbidity, mortality, and recurrence rate among survivors. 1 The predictive factors for the detrimental outcome are increasing age, clinical severity neurological grade at initial presentation, hematoma volume, infratentorial location of ICH, and intraventricular extension. [2][3][4] Moreover, the presence of intracerebral remote diffusion-weighted imaging lesions (R-DWILs) has been known to worsen the stroke outcome. 5 R-DWILs are defined as focal hyperintensities remote from the site of the ICH or the peri-hematoma region on diffusion-weighted images, which are detected in about 11%-41% of ICH patients. [6][7][8][9] It is an essential complication that requires attention owing to its significant effect on the increased risk of recurrent ICH, cognitive impairment, and mortality.
The exact pathophysiology of R-DWILs in ICH patients remains uncertain. Apart from the progression of microangiopathy, inflammatory response following ICH has also been implicated to exert influence on the development of R-DWILs. 5,10 However, prior studies examining the effect of inflammation, in the form of leukocytes, had failed to find a correlation with R-DWILs. 8,11 It was possible that they analyzed leukocytes as an entirety instead of a subset and neglected the respective implications of neutrophils or lymphocytes following ICH. Recently, the neutrophil-to-lymphocyte ratio (NLR) emerges as an indicator of inflammatory status in patients with ST-segment elevation myocardial infarction, cerebral venous sinus thrombosis, and ischemic stroke. [12][13][14] It integrates neutrophils and lymphocytes and reflects the shift between them. In terms of ICH, NLR has been shown to be better than leukocytes alone for predicting peri-hematoma edema growth 15 and adverse clinical outcomes. 16,17 The primary aim of this study was to analyze the relationship between the circulating NLR and the R-DWILs in patients of ICH. Our findings showed that patients with an elevated NLR in acute stage should be intensely noted for the increased risk of R-DWILs.

| Participants
We prospectively collected a consecutive series of patients with ICH admitted to the Department of Neurology of Second Affiliated Hospital of Zhejiang University between November 2016 and May 2018. Criteria for inclusion were as follows: age ≥ 18 years; admitted within 24 hours of initial symptom onset (onset was defined at the time a patient was last known to be without stroke symptoms);

| Data collection
Demographic information and clinical history were retrieved through patients or family members' interviews. We collected age, sex, smoking and alcohol drinking history, and comorbid conditions such as the F I G U R E 1 Section flowchart of this cohort · Secondary cause of ICH: n = 16 · Isolated IVH or SAH: n = 9 · Surgical evacuation of hematoma: n = 8 · No blood cell counts at 24-48 h: n = 8 · No MRI sequences: n = 68 · Cancer, hematologic disease, or use of immunosuppresant: n = 3 · Severe hepatic or renal diseases: n = 7 · Active infections before or in hospital: n = 19 Final analysis (n = 163) · Presenting R-DWILs: n = 31 · Without R-DWILs: n = 132 history of hypertension, diabetes mellitus, atrial fibrillation, previous ICH, or previous ischemic stroke/TIA. Prestroke medications such as antiplatelet agents, antihypertensive drugs, and hypoglycemic treatments were also collected. Hypertension was defined by the known elevation of blood pressure on at least 2 separate occasions according to the medical history or the use of antihypertensive medications. Diabetes mellitus was described by the patients had been informed of this diagnosis by a physician before admission or was receiving hypoglycemic treatments (dietary, oral antidiabetic agents, or insulin) or patients who with serum hemoglobin A1c levels ≥ 6.5%.
Previous ICH ischemic stroke/TIA was defined as a history of related syndromes or documented information in the medical record.
Smoking and alcohol drinking status was ascertained according to medical history. Moreover, initial evaluation parameters (systolic

| Laboratory variables
Peripheral complete blood samples were obtained at 24-48 hours after ICH onset in all participants, and the blood samples were immediately centrifuged (2264 g for 10 minutes at 37°C) after being collected in a calcium EDTA tube. Total leukocytes, neutrophil, and lymphocyte counts were determined using an autoanalyzer (XN-9000, Sysmex). NLR was computed by dividing the absolute neutrophil counts by the absolute lymphocyte counts. Other hematologic and biochemical data-fasting blood glucose, erythrocyte sedimentation rate (ESR), and high sensitive C-reactive protein (hs-CRP)-were obtained together at 24-48 hours.

| Neuroradiology variables
All patients underwent a baseline CT brain to verify spontaneous ICH and the initial CT brain scans after admission were reviewed for analysis. The site of hematoma was categorized as lobar, deep, brainstem, or cerebellum, and the volume of hematoma was measured using the ABC/2 method. 19  hyperintense distinct from the focal hematoma (>20 mm) on DWI sequence (b = 0/1000 s/mm 2 ), measuring <20 mm in diameter. 20 The corresponding regions on the ADC map were also viewed to confirm that the diffusion coefficient was reduced relative to the adjacent nonlesional brain parenchyma ( Figure 2). All images were reviewed by an experienced clinical neurologist (Lu-sha Tong) and an experienced radiologist (Jin-song Cai) without knowledge of clinical information. The two researchers reached high interrater reliability (Kappa = 1.0).

| Ethical approval
Informed consent was obtained from all participants or legal representatives. And the study protocol was approved by the Ethics Review Board of Second Affiliated Hospital of Zhejiang University.

| Statistical analysis
Patients were divided into two groups based on the presence vs absence of R-DWILs. Kolmogorov-Smirnov test was used to figure out the distribution pattern. Continuous variables with normal distributions were shown as the mean ± SD while other F I G U R E 2 In a 43-y-old man with left frontal lobe hemorrhage (A), diffusion-weighted imaging (DWI) shows a small remote ischemic lesion on the right occipital lobe (B), with corresponding low signal intensity in apparent diffusion coefficient (ADC) map (C) variables were presented as the median with interquartile range (IQR). Categorical data were expressed as number (proportion). were undertaken using SPSS version 23.0 (IBM). Two-tailed significance values were applied, and statistical significance was defined as P < .05. 68.0 ± 11.9, P = .005), and were found to be prone to experience IVH (61.3% vs 28.8%, P = .001) and elevated body temperature (38.2°C vs 37.8°C, P = .006). Moreover, higher NLR (7.7 vs 5.0, P = .001) and

| Clinical characteristics of the patients
higher blood glucose (6.7 vs 5.8, P = .002) were documented in patients with R-DWILs.

| Receiver operating characteristic curve analysis
The results and AUCs were shown in Figure 3. Based on the ROC curve, the best optimal cutoff value of NLR was projected to be 6.01 at 24-48 hours after ICH onset, which yielded a sensitivity of 71% and a specificity of 64.4%, with the area under the curve at 0.700 (95% CI 0.592-0.789; P = .001).

| D ISCUSS I ON
In this study, we evaluated the circulating NLR in a cohort of 163 patients with spontaneous ICH who presented to our department of neurology. Here, we showed that there was an increased NLR level in patients with R-DWILs at 24-48 hours after ICH onset. The best discriminating value of NLR for R-DWILs was 6.01, which was associated with a 3.17-fold risk for R-DWILs. Since the time interval was critical for the occurrence of ischemic lesions associated with ICH, 11 we had the strength that all patients underwent MRI imaging at 5-8 days after ictus. R-DWILs were detected among 19% of patients with spontaneous ICH in our cohort. Our results were corroborating with the prevalence reported in previous studies, ranging from 11.1%-41%. [6][7][8][9] Here, we provide further evidence that R-DWILs occur frequently, which therefore require our further attentions. Past studies linked R-DWILs to aggressive early blood pressure lowering, 8,9 remote extension of hematoma, 20 or the progression on cerebral microangiopathy. 7,11,21,22 However, the underlying mechanisms of R-DWILs remained uncertain.
The inflammatory reactions following acute hemorrhagic stroke occur rapidly and typify a highly intricate interaction between the resident cells in the brain and those in the peripheral immune systems. 23,24 Previous experimental and clinical evidence supported that neuroinflammation played an essential role in the secondary brain injury after ICH and may lead to deleterious outcome. 25 Circulating leukocytes were elevated immediately after ictus, which reflects the inflammation activation at some extend. Therefore, several studies implied that leukocytes could be involved in the pathogenesis of R-DWILs after ICH. In a post hoc analysis of ERICH study, 8 Kidwell, et al found that patients who developed ischemic lesions showed a higher circulating leukocyte counts; and after controlling for confounding factors, there was a trend for an association between them (OR = 1.050; 95% CI 0.993-1.111; P = .084). Whereas in another prospective study enrolled 97 patients with acute hypertensive ICH, the circulating leukocytes on admission did not differ between patients with or without ischemic lesions. 11 This inconsistency between the findings from these studies and those in our patients might be explained by the lack of strict exclusion criteria to control confounding factors, like infections. The relatively small sample size in these studies may be another reason for concealing the association between ischemic lesions and circulating leukocytes.
Moreover, since they analyzed leukocytes as the target entity, the changes within different subtypes of leukocyte were neglected.
Thus in the present study, we rigorously evaluated the NLR levels in the patients with ICH and excluded the infections and other comorbidities, which might lead to bias.
Peripheral sterile inflammatory response has been described in the context of stroke-induced stress reaction, which usually is accompanied by neutrophils demarginated, activation, 26 90-day morbidity, 16,17 and it also associated with stroke severity and peri-hematoma edema growth. 15,32 The recent remarkable observations indicated that NLR was an independent predictor for delayed cerebral ischemia in patients with a SAH, 33 making it a prime candidate for assessing the inflammatory aspects of ICH.
Here, we found that NLR was independently associated with the occurrence of R-DWILs. The roles of neutrophils in promoting blood coagulation and thrombosis may enhance the occurrence of ischemic lesions. Neutrophils originated from the blood circulating could be found within the brain as early as 4 hours in mice ICH models. 34 And the results from the literature suggest that neutrophils could externalize decondensed nucleosomes and granule proteins to form neutrophil extracelluar traps (NETs) in some cases. 35 In this context, with the pre-existing cerebral microangiopathy in ICH, elevated NLR was prone to have NET increasing and thus led to microvascular thrombosis. Another potential mechanism could be the NET-induced activation of both the extrinsic and contact pathway of blood coagulation. 35,36 In our cohort, the higher NLR group showed a trend for higher fibrinogen accumulation in peripheral. A recent study found that patients with elevated neutrophil counts had a lower risk of hematoma expansion following acute ICH, 37 may also be explained by the increased coagulation by neutrophils. These findings suggested that higher NLR indicated a tendency of coagulation or forming of microembolism. This might explain the relationship between high NLR and R-DWILs.
Besides elevated NLR, the presence of IVH was the other correlated factor of R-DWILs in our cohort. Analyzing the CSF in patients with IVH showed an early and modest elevation in leukocytes. 38 IVH induced the blood-brain barrier disruption, the more rigorous immune alterations in CSF, and the extension of hemoglobin degradation product might contribute to the presence of the higher R-DWILs and NLR. Still, the mechanisms underlie in R-DWILs in IVH requires further study.
Some limitations of this study should be noted. First, the generalizability of our conclusions is limited by our single-center retrospective study design. Second, the low in-hospital mortality in relation to our ward-based approach and a low proportion of patients with ICH enrollment due to the strict inclusion and exclusion criteria might add selection bias into the analysis. Third, the clinically unstable cases cannot tolerate an MRI examinations are excluded, and therefore, there is likely to introduce bias toward patients with less severe ICH as well. Forth, we excluded patients with active infections prior to stroke or with early infection signs on admission. Inevitably, we cannot rule out those with only subclinical symptoms. Moreover, we had no detailed microbiological and viral data confirming potential infections occurring before. Fifth, there were no serial MRI scan data within our cohort, which makes it possible that R-DWILs presented before or concurrently with ICH onset and had reversed by the time of undergoing MRI; the possibility also existed that patients without R-DWILs in our study might present later. Finally, we neither confirmed these remote hyperintense lesions on DWI were truly cellular infarction nor investigated prothrombotic mediators, such as NETs, indeed elevated in those patients with R-DWILs. These will be the key points in our following work.
In conclusion, the current study showed that the elevated NLR