Predicting prognosis in patients with stroke treated with intravenous alteplase through the 24‐h trajectory of blood pressure changes

Abstract Blood pressure (BP) monitored within 24 h from the beginning of intravenous thrombolysis (IVT) with alteplase, is one of the important factors affecting the prognosis of patients with acute ischemic stroke (AIS). This study aimed to explore longitudinal BP trajectory patterns and determine their association with stroke prognosis after thrombolysis. From November 2018 to September 2019, a total of 391 patients were enrolled consecutively during the study period, and 353 patients were ultimately analyzed. Five systolic (SBP) and four diastolic blood pressure (DBP) trajectory subgroups were identified. The regression analysis showed that when compared with the rapidly moderate stable group, the continuous fluctuation‐very high level SBP group (odds ratio [OR]: 2.743, 95% confidence interval [CI]: 1.008–7.467) was associated with early neurological deterioration (END). Both the rapid drop‐high level SBP (OR: 0.448, 95% CI: 0.219–0.919) and DBP groups (OR: 0.399, 95% CI: 0.219–0.727) were associated with early neurological improvement (ENI). Moreover, there was a U‐shaped correlation between the OR value of SBP trajectory group and favorable outcome (the modified Rankin Scale [mRS] score 0–2) at 3 months: the slow drop‐low level SBP group represent a well‐established unfavorable outcome risk factor (OR:5.239, 95% CI: 1.271–21.595), and extremely high SBP—the continuous fluctuation‐very high level SBP group, are equally associated with elevated unfavorable outcome risk (OR:3.797, 95% CI: 1.486–9.697). The continuous fluctuation‐very high level DBP group was statistically significant in mRS (OR: 3.387, CI: 1.185–9.683). The BP trajectory groups show varying clinical features and risk of neurological dysfunction. The findings may help identify potential candidates for clinical BP monitoring, control, and specialized care.


INTRODUCTION
The clinical significance of monitoring and managing blood pressure (BP) during intravenous thrombolytic (IVT) therapy for the prevention of adverse prognosis in patients with acute ischemic stroke (AIS) remains an area of active investigation. Several observational studies have found that approximately 75%-80% of patients have elevated BP in response to stroke episodes, 1,2 and a high BP during the thrombolysis period is a risk factor for poor clinical outcomes, such as poor rate of recanalization, intracerebral hemorrhage, and neurological dysfunction. [3][4][5] Interventional studies that used antihypertensive drugs to maintain BP within 141-150 mmHg or 130-140 mmHg have not reached a consensus on whether this leads to an improved prognosis. 6,7 In order to determine the ideal therapy, it is still necessary to accurately describe BP variations and their relationship with prognosis in patients undergoing venous thrombolysis.
In comparison, much less is known about the importance of changes in BP over time (ie, BP trajectories) in patients with AIS. A study on this topic showed that different BP trends were associated with the occurrence of cerebrovascular events. 8,9 Unsupervised functional principal components analysis was used to characterize SBP trajectories in patients with spontaneous intracerebral hemorrhage over first 24 h and their relationship to the unfavorable shift on modified Rankin scale (mRS). 10 The method used to analysis BP trajectory has gradually received researchers' attention and has a broad research prospect.
For BP monitoring during IVT, current guidelines only indicate that BP should be maintained at < 180/105 mmHg within 24 h of treatment onset, mainly to prevent serious complications, such as symptomatic intracranial hemorrhage. 11 Even if the ischemic penumbra is not affected by changes in infarction or massive hemorrhage, a varying BP trajectory causes unstable cerebral blood flow and cerebral perfusion pressure fluctuations, which will also affect the long-term prognosis.
Moreover, it is unreasonable to use a single BP value, the average BP, or BP variation-related indicators at a single time point to represent the BP status within that time period. Studies describing the relationship between BP trajectories and early stroke outcomes in patients treated with IVT are lacking.
Hypertension is a heterogeneous condition in patients with AIS, and available data support the importance of both systolic (SBP) and diastolic (DBP) BP monitoring. We designed our analysis using an unsupervised cluster approach, group-based trajectory modeling (GBTM) approach that may provide an alternative method for summarizing long-term BP values accounting for the dynamic nature of BP over time, to group similar longitudinal BP response patterns. We then evaluated the associations of these clusters, or most commonly BP parameters in similar articles, with neurological function changes and status using a standard multivariate regression approach. Improved knowledge of BP trajectories is critical in understanding the role of BP as a risk factor for adverse outcomes.

WHAT IS ALREADY KNOWN ABOUT THE TOPIC?
• Stroke guidelines indicate that monitoring BP within 24 h from the beginning of IVT in patients with AIS is essential due to the high incidence of complications. • Several studies have shown that high SBP or DBP levels are associated with the prognosis of patients with AIS, such as changes in neurological function, hemorrhagingrelated complications, and mRS scores.
• In most similar studies, the BP values at a single time point, its mean value, the variation coefficient, and other indicators at multiple time points were used to describe BP and to explore the correlation between BP and stroke prognosis.

WHAT THIS PAPER ADDS
• According to BP data obtained at multiple time points in patients with AIS treated using thrombolysis, the groupbased trajectory model can be used to assess patterns of BP fluctuations. • SBP and DBP showed different patterns as the time to thrombolytic therapy increased.
• There were differences in clinical characteristics among patients with different patterns of BP changes as well as in the degree of correlation with stroke prognosis, which was one of the independent influencing factors.
• Compared with these parameters, the BP values at a single time point, its mean value, the variation coefficient, and other indicators at multiple time points, which were used to describe BP in similar articles, BP trajectories are equally important values for predicting stroke prognosis.

Outcome measures
The primary functional outcome was the mRS score at 90 days after stroke onset. Three months after the occurrence of ischemic stroke, the neurological impairment caused by stroke can be basically terminated. Walking function is the necessary function basis for patients to carry out daily living activities, and the improvement of walking ability in stroke patients is positively and linearly correlated with the future quality of life. Since the mRS scale takes the walking ability as a clear scoring standard, which was used the mRS score > 2 as the demarcation value of disability, it is widely used in stroke-related studies and is one of the gold standards for evaluating the independent living ability of stroke patients. A favorable outcome was defined as an mRS score of 0-2 points, whereas an unfavorable outcome was defined as an mRS score of 3-6 points. 12 The secondary outcome events included early neurological deterioration (END) and improvement (ENI). END was defined as an increase in the National Institutes of Health stroke scale (NIHSS) score ≥4 points or as an increase ≥2 points in one sub-item that occurred at 24 h following alteplase infusion; these criteria were previously used to define significant deterioration. 13 ENI was defined as an improvement in the NIHSS score ≥8 points or as a 0 or 1 score at 24 h following alteplase infusion. 14

Covariates
Medical records were retrospectively reviewed by a nurse who was blinded to patients' outcomes and the following information was

Ethics statement
The study protocol was approved by the Ethical Committee of Xuanwu Hospital and conformed to the principles outlined in the Declaration of Helsinki. Written informed consent was obtained from all patients.

Group-based trajectory modeling
During the 24 h after thrombolysis, we adopted a group-based trajectory modeling approach using the traj procedure in SAS (SAS Institute, Cary, NC) to identify superior BP trajectories. 17

Statistical analysis
Continuous data were reported as means ± standard deviation and were analyzed using one-way ANOVA or Kruskal-Wallis tests as appropriate. Categorical data were presented as frequency and percentages and were analyzed using the chi-square test. Logistic regression analysis and area under the curve (AUC) were used to determine the association between neurological damage and different BP trajectory groups, previous BP parameters. The strengths of the associations were determined by estimating the odds ratios (OR) and their 95% confidence interval (CI). To detect changes in associations between outcome and main exposures, the following multivariate logistic models were con-

RESULTS
During the 11-month study period, 353 out of 391 patients met the study criteria and were enrolled in the study. Among the 12 patients with missing data, 1 patient was discharged from the hospital with the main diagnosis other than ischemic stroke, and 25 patients were lost to follow-up. Among the included patients, 257 (72.8%) were male, 96 (27.2%) were female, and the mean age was 62.49±11.79 years.

Trajectory groups based on systolic blood pressure
Five groups were identified based on the SBP trajectory during the first 24 h after the beginning of IVT ( Figure 1A). The groups describe the BP level and trend: group 1 (slow drop-low SBP group, 102-114 mmHg, n = 22, 6.2%); group 2 (rapid drop-low SBP group, 120-128 mmHg, n = 76, 21.8%); group 3 (rapid drop-medium SBP group, 134-143 mmHg, n = 124, 34.8%); group 4 (rapid drop-high SBP group, 150-157 mmHg, n = 84, 24.0%); and group 5 (continuous fluctuationvery high SBP group, 162-173 mmHg, n = 47, 13.2%). In group 1, BP was relatively low with the highest point being below normal, and it fluctuated widely during the first 3 h. Groups 2, 3, and 4 showed stable BP after a rapid and steady decline within 1.5-2 h, whereas groups 2 and 3 had normal to high BP; the BP in group 4 remained stable at about 150 mmHg. In group 5, BP declined rapidly and steadily for the first 1.5 to 2 h, followed by wavy fluctuations.
Patients in these five SBP trajectory groups had distinct clinical profiles and laboratory results (Tables 1-2). Patients in group 1 were significantly younger by about 10 years on average and had a lower prevalence of hypertension and diabetes by 23.4%-57.1% and 11.6%-30.7%, lower than the other groups, respectively. Patients with high BP were more likely to have chronic diseases; those in groups 4 and 5 had high BP and a high prevalence of both intravenous antihypertensive treatment and large artery atherosclerosis TOAST subtype. Blood glucose, glycosylated hemoglobin, and erythrocyte sedimentation ratio showed an increasing trend from group 1 to 5, whereas triglycerides and total cholesterol levels showed a "W" and "V" pattern, respectively. Both the international normalized ratio and prothrombin time decreased from group 1 to 5. The maximum or minimum values generally appeared in group 5.

Trajectory groups based on diastolic blood pressure
Four groups were identified based on the DBP trajectory ( Figure 1B (Tables 1-2). Group 4 had a higher proportion of coronary heart disease, pneumonia, and catheter insertion; higher NIHSS score at admission; and a greater likelihood of disease changes. The prevalence of drinking and repetitive thrombolysis decreased linearly among the four groups. White blood cell count, blood glucose, and total cholesterol levels increased from group 1 to 4, whereas the platelet count showed an inverted "U" shape. Highsensitivity C-reactive protein levels increased linearly in groups 1 to 3, whereas the values in group 4 decreased slightly.
The trend of the graph in Figure 2 and 3, which are drawn by calculating the average 24-hour blood pressure of each track group, is similar to the result after applying GBT grouping.

Trajectory groups and stroke outcomes
After IVT, there were 67 (19.0%) cases of END, 131 (37.1%) cases of ENI, and 242 (68.6%) patients with an mRS score 0-2 at the 3-month follow-up. To examine the association between BP trajectory groups and outcomes, the groups were included as independent variables in a logistic regression model, and the moderate stable BP trajectory groups (SBP: group 3; DBP: group 4) were considered as the control groups (

DISCUSSION
Five SBP and four DBP trajectory subgroups were identified in the first 24 h after initiating IVT using the group-based trajectory model. In these groups, BP was classified according to its level as low, medium, and high and according to its changes as slow decline, rapid decline, and  .608  difference between the cerebrovascular and brain interstitium, and the cracks in the vessel wall increases risk of brain edema and bleeding. and steadily to about 160 mmHg later and to subsequently stabilize it.

TA B L E 2 Comparison of laboratory parameters after thrombolysis(n
Therefore, it is particularly important to monitor BP at multiple time points in the acute phase, and clinicians can more intuitively grasp the changes in patients' condition according to the trajectory map.
Our study both confirms and expands on the findings of previous BP studies. First, a high BP level was prevalent in patients with AIS. Harper and coworkers 1 found that 69.3%-82% of patients had BP > 140/90 mmHg and less than 5% of patients had a BP < 120 mmHg in the acute stage. In our study, 72% of patients had a higher-thannormal BP, and 6.2% of patients had a BP < 120 mmHg. Second, most patients' BP levels gradually declined and stabilized over time within the acute phase and were associated with neurological deficits. The decrease in BP is faster within the first 8 h and lasts up to 36 h. Gill and coworkers 22 found that an SBP decrease of 10 mmHg was related to a decrease of 0.51 points in the NIHSS scores. Zhang and coworkers 23 and Tsou and coworkers 24 et showed that SBP ≥160 mmHg or an increase of 15 mmHg could predict the risk of neurological deteriora-tion. Leonardi-Bee and coworkers 25 also found that the relationship between SBP and the 14-day mortality rate and 6-month mortality or disability rate was U-shaped rather than linear. The study showed that the risk of adverse prognosis increased by 5% for every 1-mmHg increase up to 90 mmHg. Third, current research on the association between BP and prognosis mainly focuses on data from single measurements, multiple measurements in a short time, or 24-h ambulatory blood pressure measurements. Current BP reporting methods may not adequately reflect individuals' accurate BP levels. Group-based trajectory modeling considers BP variations over time and the heterogeneity within multiple BP measurements, thus providing an effective approach to describe the relationship between BP changes and stroke outcomes. 26,27 Our study had several limitations that should be acknowledged.
First, the participants were recruited only from one hospital; our findings should be verified in other cohorts to determine generalizability to other ethnicities and populations with different backgrounds. Second, identifying stroke outcomes based on more objective indicators, such as imaging findings, is more accurate. Third, due to the observational study design, although BP management was carried out according to guidelines, how long and how much the individual patients' BP was controlled was left at the discretion of primary stroke physicians; future studies should aim to standardize this management.

CONCLUSION
Trajectory analysis models showed that the 24-h changes in BP in patients with AIS treated with alteplase can reflect the dynamic changes in BP over time. BP may effectively be grouped according to distinct trajectory patterns, which have differential clinical characteristics and risk of subsequent early neurological improvement or deterioration as well as different associations with mRS score at 3 months.
Being classified into the continuous low SBP (102-114 mmHg), fluctuating high SBP/DBP (162-173/92-101 mmHg), or rapidly high stable SBP/DBP (150-157/82-88 mmHg) groups was an independent predictor of adverse events. The clinical significance of this study is that our findings may help identify patients at a high risk of future vascular events and those requiring intervention.