The Multicenter Clinical Trial of Revascularization Treatment for Acute Ischemic Stroke (TRAIS) study retrospectively collected data from all consecutive patients with acute stroke admitted to 14 Chinese stroke centers, including the Wuhan Union hospital, the Wuhan Union Hospital West Campus, the Wuhan Union Hospital Jinyinhu Campus, the People’s Hospital of Dongxihu District, the Hubei Provincial Hospital of Integrated Chinese and Western Medicine, the Central People’s Hospital of Yichang, the First People’s Hospital of Yichang, the Second People’s Hospital of Yichang, the Central Hospital of Hefeng County, the People’s Hospital of Honghu, the People’s Hospital of Jingshan, the First People’s Hospital of Jiangxia District, Wuhan Red Cross Hospital, Puren Hospital of Wuhan.

Patients were included in the study if they met the following criteria: (1) retrospective clinically and radiologic confirmed AIS; (2) received rt-PA injections according to the indications for thrombolytic therapy; and (3) were aged 18 years or older. Exclusion criteria included: (1) mental disorders or severe cognitive dysfunction; and (2) incomplete clinical data. All centers follow the same patient screening process.

The analysis included patients admitted between January 1, 2018 and June 1, 2021. All patients were informed of their participation in the study.

We collected the following data: (1) clinical assessments, including the National Institute of Health Stroke Scale (NIHSS) score evaluated on admission and 24 h after thrombolysis, baseline blood pressure, and time from onset to treatment (OTT); (2) demographic information, such as gender and age; (3) vascular risk factors, including alcohol consumption, smoking history, hypertension, diabetes mellitus, coronary heart disease, and previous stroke; (4) laboratory tests, including admission blood glucose level, C-reaction protein (CRP), white blood cell count (WBC), red blood cell count (RBC), neutrophil count, lymphocyte count, platelet count (PLT), D-dimer, international normalized ratio (INR), prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen (FIB), total bilirubin (TBil), direct bilirubin (DBil), glutamic pyruvic transaminase (ALT), glutamic oxiracetam transaminase (AST), blood urea nitrogen (BUN), creatinine (Cr), uric acid (UA), lactic dehydrogenase (LDH), creatine kinase (CK), creatine kinase myocardial band (CK-MB), cardiac troponin I (cTNI).

According to the European Cooperative Acute Stroke Study, END was defined as an increase in NIHSS score of 4 or more points or death within 24 h (20).

Categorical variables were reported as counts and percentages, while continuous variables were presented as means and standard deviations. We used ordinal logistic regression to calculate the odds ratio (OR) and 95% confidence interval (CI) for patients with and without END in the derivation cohort. Briefly, to assess the univariate relationship between baseline variables and END, we used t-tests or Mann–Whitney U tests for continuous variables, and χ² or Fisher exact test for categorical variables, as appropriate. With the dependent variable END, we conducted a stepwise multivariable binary logistic regression analysis. We used the C statistic to assess the discriminative ability of the score in predicting END. We used SAS version 9.4 (SAS Institute) and SPSS version 16.0 (SPSS Inc.) for statistical analysis. We considered statistics to be significant at a two-tailed p-value of less than 0.05.

Between January 2018 and June 2022, 1,425 patients with AIS were treated with IVT in the TRAIS study. Of these, 212 patients were excluded (126 received endovascular therapy and 86 patients lacked clinical data), and the remaining 1,213 patients were included in the final analysis (Figure 1). We set the patients in Wuhan Union Hospital as the training cohort, and the patients in other 13 medical centers as the validation cohort. The derivation cohort consisted of 751 patients, while the validation cohort consisted of 462 patients. In the derivation cohort, 487 (63.4%) patients were male, the mean age (±SD) was 64.6 (±12.5) years, and the median NIHSS (interquartile range [IQR]) score was 3 (1–7). In the validation cohort, 315 (67.9%) patients were male, the mean age (±SD) was 66.9 (±11.8) years, and the median NIHSS (IQR) score was 5 (3–10). Patients in the derivation cohort had a higher prevalence of cigarette consumption, lower frequency of previous stroke, ischemic heart disease, hyperlipidemia, and lower admission NIHSS score. Table 1 provides detailed baseline characteristics of patients in the derivation and validation cohort.

Table 2 displays the results of the univariate analysis of variables related to END in the derivation cohort. Out of the derivation cohort, 49 patients (7.3%) experienced END. Among the patients who experienced END, 29 (51.8%) were male. The median NIHSS (IQR) score for these patients was 13 (6–21). To facilitate clinical application, we classified continuous and sequential variables and selected cut-off values based on common clinical values and actual results. In the univariate analysis, cigarette smoking, higher admission systolic blood pressure, baseline NIHSS score, white blood cell count, neutrophil ratio, D-dimer, and low lymphocyte ratio were found to be significantly correlated with END. Diabetes, lower, and higher AST were also associated weak to moderately associated with END.

Multivariate logistic regression analysis was performed to identify the independent predictors of END after IVT. Variables that were found to be significant in the univariate analyses were entered simultaneously into the multivariate model. After adjusting for confounding factors, CK-MB ≥ 5 ng/mL, NIHSS score, SBP ≥ 160 mmHg, LDH ≥ 220 U/L, ALT≥40 U/L, Neutrophil ratio ≥ 80%, D-dimer≥0.5 mg/L were all independent predictors of END (as shown in Table 3).

The CNS-LAND scores (CK-MB, NIHSS score, Systolic blood pressure, LDH, ALT, Neutrophil ratio, D-dimer) were developed based on the multivariate logistic analysis in the derivation cohort, incorporating seven independent predictors of END after IVT: CK-MB, NIHSS score, systolic blood pressure, LDH, ALT, neutrophil ratio, and D-dimer. The weight of each predictor was determined by the strength of its association with the beta coefficient, and points were assigned accordingly. The total score ranged from 0 to 9, with higher scores indicating a greater risk of END (refer to Table 4 for details).

In both the derivation and validation cohorts, there was a significant increase in the proportion of patients with END as the CNS-LAND score increased, as shown in Table 5. To facilitate clinical application, we classified patients with low risk of functional dependence as having a score of 0–3 and those with high risk as having a score of 4–9. The cut-off value was determined from the receiver operating characteristic (ROC) curve analysis to achieve optimal performance. In the derivation cohort, 122 patients (16.2%) and in the validation cohort, 112 patients (81.2%) had a score ≥ 4, with a predicted sensitivity of 73.5 and 72.7%, respectively, and a predicted specificity of 87.7 and 80.8%, respectively. Further details of the classification test are provided in Table 5.

The CNS-LAND score demonstrated good discrimination and calibration in both the derivation cohort (c statistic, 0.862; p value of Hosmer-Lemeshow test, 0.123) and the validation cohort (c statistic, 0.851; p value of Hosmer-Lemeshow test, 0.685).

We compared the discriminative performance of the CND-LAND with three existing AIS prognostic scores including DRAGON, ISCORE, and THRIVE, in the validation cohort (n = 456). Six patients were excluded from the analysis due to missing information on elements of other scores. For END, the area under the curves (AUCs) ranged from 0.700 to 0.851 (DRAGON score, 0.700; THRIVE, 0.710; ISCORE score, 0.739; CNS-LAND score, 0.851). The CNS-LAND score had the highest AUC. The pairwise difference in AUCs between the CNS-LAND and other scores showed statistical significance (all p < 0.001) (see Supplementary Table S1). Figure 2 displays the ROC curves of the aforementioned scores for END.

Research on peripheral biomarkers for END risks remains relatively limited, although AIS is widely recognized as a systemic disease that initiates excitotoxicity, neuroinflammation, and oxidative stress in the brain parenchyma, leading to peripheral immune disorders, autonomic nervous disorders, neuroendocrine overstimulation, and even damage to distant organs (11–14, 16). In this study, we assessed and validated the association between biomarkers reflecting excessive neuroinflammation, coagulation abnormalities, hemorheological disorders, and remote organ injury, and END risks.

Excessive neuroinflammation has been identified as a significant cause of stroke progression after IVT. Following AIS, neuronal death and microglial activation release large amounts of DAMPs and chemokines, which activate and recruit peripheral immune cells. Neutrophils, with their strong pro-inflammatory phenotype, are the first immune cells to infiltrate the brain (21, 22). They release not only pro-inflammatory cytokines and reactive oxygen species that cause secondary neuronal damage, but also matrixmetalloproteinase 9, which damages the blood–brain barrier, leading to acute vasogenic cerebral edema (23, 24). Moreover, neutrophil adhering to the microvascular walls can block the flow of red blood cells, resulting in the no-reflow phenomenon in the brain (25). Additionally, the excessive activation of neutrophils causes immunodepletion, making the host susceptible to secondary infections (26). These factors collectively contribute to a sharp increase in END risk.

D-dimer is general activated in response to thrombotic events, including AIS, acute myocardial infarction (AMI), pulmonary embolism, and venous thromboembolism (27). Recent studies have confirmed its association with stroke progression and early mortality (28). In studies on post-stroke complications during hospitalization, D-dimer levels are used for the early prediction of stroke recurrence, acute decompensated heart failure and respiratory tract infections (29–31). It is reasonable to speculate that elevated D-dimers not only contributed to cerebral thrombus expansion, but also increase the risk of thrombosis in the heart, liver, and even throughout the body, resulting in END.

Elevated systolic blood pressure may indicate a disorder of vascular autoregulation (3, 32–34). High systolic blood pressure can cause vasodilation of the microvessels and hyperperfusion in the infarct area, leading to cerebral edema and blood–brain barrier damage (35, 36). Currently, the management of blood pressure after acute ischemic attack is controversial. We found that admission systolic blood pressure ≥ 160 mmHg was significantly associated with END after IVT, supporting the notion that active intervention of blood pressure at admission improves patient outcomes.

Moreover, we identified novel END independent predictors: CK-MB, LDH, and ALT, which represent specific damage to distant organs after stroke. Previous studies have shown that AIS often results in cardiac abnormalities such as arrhythmias, autonomic nervous disorders, AMI, and paroxysmal arterial hypertension, which cause secondary damage to the brain. Heart failure or arrhythmias such as atrial fibrillation and flutter may lead to blood stasis in the left atrium, resulting in cardiogenic stroke (13, 14). It has been found that 4.9% of AIS patients experienced AMI, with worse outcomes than patients without AMI, and elevated cardiac parameters were associated with functional recovery at 3 months and 1 year in patients with AIS (37). Here, CK-MB and LDH, as traditional cardiac markers, were found to be associated with END risk. In addition, LDH may originate from ischemic cerebral hemispheres in addition to the myocardium. Intracellular LDH is upregulated for energy utilization and adaption to the ischemia in all injured brain cell, which was released into the extracellular space and then into the peripheral circulation through the damaged BBB. Our previous research has validated that serum LDH levels are associated with cerebral infarct size and cerebral edema and predicted neurological changes and 90-day outcomes (38).

Following stroke, chemokines CXCL1and CCL2 levels are elevated in liver, which lead to an increase in neutrophils and monocytes and an excessive inflammatory response, resulting in the release of ALT from necrotic hepatocytes. At the same time, brain-derived glutamate toxicity also promotes the synthesis of ALT, which metabolizes blood glutamate to provide protection. But more importantly, ALT has been found to be associated with symptomatic intracranial hemorrhage and mortality after AIS in recent studies (39, 40). Liver dysfunction (including fibrosis and cirrhosis), significantly impaired platelet aggregation, and diminished antifibrinolytic activity are hypothesized to cause the syndrome. In other words, ALT elevation reflects reduced liver function and damaged coagulation system, resulting in hemorrhagic transformation and END after IVT.

In conclusion, our study has identified several novel biomarkers associated with END risks, including CK-MB, LDH, ALT, D-dimer, neutrophil ratio, NIHSS score, and SBP. These biomarkers reflect different pathological processes, including cardiac damage, remote organ injury, excessive neuroinflammation, coagulation abnormalities, and hemorheological disorders, which could contribute to the development of END. Further studies are needed to validate these biomarkers’ clinical significance in predicting END risks and guide personalized treatment strategies.

CNS-LAND score is the first scale for predicting the END after IVT based on systemic injury and responses. With increasing CND-LAND scores, the risk at the END was gradually increased in the derivation and validation cohort. The score is generally divided into two categories: low-risk patients (0–3 points) who can benefit from reduced medical costs, and high-risk patients (4–9 points) who require multiple interventions based on close follow-up monitoring.

Currently, there are several scales based on large sample, multi-center cohorts for evaluating the hemorrhagic transformation risks and 3-month prognosis of patients with AIS, such as the DRAGON score, ISCORE score, and THRIVE score. However, these scales have little external verification of their discriminant capacities for END risk. In comparison, CNS-LAND is superior to these scales in predicting END. First, CNS-LAND includes END-specific hematologic indicators and clinical features. Second, it is based on a large multicenter TRAIS study sample, with patients continuously recruited from large medical centers. Additionally, CNS-LAND is based on hematological examination and clinical characteristics that are widely used in routine practice, has a short turnaround time, and does not require special laboratory equipment, making it convenient for primary hospitals. Lastly, we verified the differential power and diagnostic accuracy of the score in an independent cohort of 426 patients recruited from 13 centers in Hubei Province, China. Therefore, CNS-LAND scoring is currently the most feasible and applicable in clinical settings in China.

The study has several limitations: (1) the inclusion of patients from only large hospitals in China introduces inevitable selection bias, as patients from primary care institutions were not included; (2) imaging information, such as the thrombus location, core infarction and ischemic penumbra was not analyzed, because these data were not routinely collected for AIS patients; (3) as an observational and retrospective study, confounding factors could not be completely eliminated, although baseline data were refined to the extent possible.