Long‐term risk of recurrent vascular events and mortality in young stroke patients: Insights from a multicenter study

Although the incidence of stroke in the young is rising, data on long‐term outcomes in these patients are scarce. We thus aimed to investigate the long‐term risk of recurrent vascular events and mortality in a multicenter study.


INTRODUC TI ON
While the worldwide age-standardized incidence of stroke is decreasing, stroke incidence and prevalence are increasing among younger individuals [1]. The impact of stroke in younger patients may be even more devastating than in older patients, since aside from immediate lasting consequences including disability and death, long-term sequelae may include social, psychological, and vocational issues [2].
Prediction and prevention of recurrent vascular events are essential in young patients with ischemic stroke (IS) because of their long remaining life expectancy, although secondary stroke prevention may be challenging due to different etiologies of stroke and lacking evidence from high-quality randomized controlled trials compared to older patients [3]. Furthermore, recurrent strokes and cancer are associated with long-term mortality in IS and transient ischemic attack (TIA) patients [4][5][6][7].
Few previous studies have investigated the long-term outcome of young IS patients. However, those that did were either singlecenter studies [8][9][10][11][12][13] or had short follow-up periods [14,15]. The Stroke in Young Fabry Patients (SIFAP) study is, to our knowledge, the largest multicenter, prospective study of young stroke patients with 5023 patients from 47 centers [16]. Due to the need for additional information on long-term outcomes including specific risk factors for recurrent vascular events in young stroke patients, we performed a 10-year follow-up study among patients originally included in the SIFAP study in three centers. In this multicenter cross-sectional study, we aimed at investigating the rate of recurrent strokes, other vascular events, cancer, and death as well as factors associated with recurrent vascular events at 10-year follow-up among young adults with IS or TIA as the index event.

Study population
The SIFAP study prospectively recruited 5023 patients aged 18 to 55 years with an acute cerebrovascular event in 15 countries and 47 centers in Europe between 2007 and 2010 and has been described in detail elsewhere [16,17]. Detailed clinical data were assessed, and all included patients underwent brain magnetic resonance imaging. Most of the included patients (n = 4467) had an ischemic event (either IS or TIA), others had intracerebral hemorrhage (ICH), subarachnoid hemorrhage (SAH), or cerebral venous thrombosis (CVT).
In the present study, three of the original participating centers where performing a long-term follow-up study was feasible (Graz/ Austria, Greifswald/Germany, and Helsinki/Finland), decided to re-invite original participants of the SIFAP study to an in-person follow-up in 2018-2020. Exclusion criteria were: (i) non-ischemic cerebrovascular event as the index event, (ii) a false primary diagnosis of IS or TIA in retrospective review, (iii) living outside of the hospital catchment area, and (iv) loss to follow-up. All included patients were invited to an in-person follow-up investigation with a written letter and/or telephone call, depending on the local center.

Data assessment
At the in-person follow-up, comprehensive data were collected according to a prespecified case report form, which included demographic and clinical data, as well as data on current medication, comorbidities, operations done, and stroke risk factors such as smoking and alcohol abuse (Appendix S1). Information regarding recurrent vascular events and cancer were assessed using a combination of patient history and review of patient records. In all patients attending the in-person follow-up, the modified Rankin Scale (mRS), National Institutes of Health Stroke Scale (NIHSS), and clinical status (weight, height, blood pressure, and pulse rate) were also assessed.
Outcome events that were explored included any cerebrovascular event, including fatal/non-fatal recurrent IS or ICH, TIA, CVT, non-traumatic SAH, and other cardiac (acute coronary syndrome), arterial (systemic embolism or peripheral arterial thrombosis), or venous events (deep venous thrombosis, pulmonary embolism; Appendix S1). Recurrent IS was classified according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification and ICH according to the SMASH-U classification [18,19].
Furthermore, we assessed diagnoses of malignant cancer and recorded mortality as well as reasons of death. We defined recurrent stroke as rapid onset of a new persistent focal neurological deficit attributed to an obstruction in cerebral blood flow and/or ICH with no apparent non-vascular cause (e.g., trauma, tumor, or infection) using neuroimaging studies to support the clinical impression and to determine if there is a demonstrable lesion compatible with an acute stroke [16,20]. Recurrent strokes due to both arterial and venous origin were assessed. TIA was defined as a transient episode of neurological dysfunction caused by focal brain or retinal Conclusions: This multicenter study shows a considerable risk of recurrent vascular events in young IS and TIA patients. Further studies should investigate whether detailed individual risk assessment, modern secondary preventive strategies, and better patient adherence may reduce recurrence risk.

K E Y W O R D S
long-term follow-up, mortality, prognosis, recurrent vascular events, young stroke

| 2677
RECURRENT VASCULAR EVENTS IN YOUNG STROKE ischemia lasting less than 24 h, and without acute infarction on brain imaging. We defined cancer as a history of malignant cancer according to the patient or legal representative and/or such data obtained from hospital records.
For patients that did not participate in the in-person follow-up, outcome events were assessed using electronic records for the respective hospitals/regions. In Helsinki, recurrent vascular events as well as dates and causes of deaths were identified from patient records and death certificates obtained from Statistics Finland [21]. In Graz, recurrent vascular events and mortality were assessed from the electronic health records encompassing all acute-care hospitals in the state of Styria. In Greifswald, the outcome events were assessed from the electronic health records of the University Medicine Greifswald.

Statistical analyses
The baseline characteristics of patients are presented with descriptive statistics. We reported numbers of observations and percentages for categorical variables, and medians and interquartile ranges for continuous variables. Characteristics at baseline were compared between patients who did and did not attend in-person follow-up Univariable and multivariable Cox regression models were fitted on the study population to study baseline factors and the hazard of any recurrent vascular event during follow-up. We identified factors considered as potential confounders in previous literature [8,22,23] and those variables from a univariable Cox regression model with p-values<0.10 to be further studied in the multivariable analysis. We considered age and sex as essential to control for in the multivari-

RE SULTS
After applying exclusion criteria ( Figure 1), a total of 396 patients (62.4% male, median age 47 years) were included in this study. In those who attended the in-person follow-up-visit, alcohol use and smoking, as well as NIHSS and mRS scores, had decreased compared to baseline. However, prevalence of arterial hypertension, hyperlipidemia, diabetes mellitus, coronary heart disease, congestive heart failure, atrial fibrillation, body mass index (BMI) and waist circumference, as well as treatment with antihypertensives, statins, oral antidiabetics, and insulin, had increased at follow-up (Table 1). Those who did not come to an in-person follow-up were more often male, smoked cigarettes more often, had higher rates of hypertension and diabetes mellitus, and more frequently used antihypertensives and insulin at baseline. These patients also more often had an IS and a higher NIHSS score at the index event.  Table 2.
Cumulative risks and numbers of patients at risk for endpoint events and mortality are shown in Figure 2.
In univariable Cox regression analysis, age, arterial hypertension, hyperlipidemia, and SVD as the etiology of the index event were associated with recurrent vascular events (Table 3). In the age-and sex-adjusted model, only age was associated with recurrent events. The final multivariable model included age, sex, hyperlipidemia, atrial fibrillation, and TOAST. In this model, atrial fibrillation at baseline was found to be independently associated with recurrent vascular events in the follow-up (Table 3). Previous studies demonstrated that young IS patients remain at significant risk of recurrent vascular events in the long term, with a large Dutch study reporting a vascular recurrence rate of 20% during a mean follow-up period of 9 years [13], while a Finnish study reported a cumulative 10-year risk of recurrent vascular events of 36%, and for recurrent stroke 19% [23]. However, both these studies were performed at single centers over very long time periods, ranging back to 1980 and 1994, respectively, which may influence the results due to extensive changes in stroke diagnosis, treatment, and secondary prevention since those times. In a more recent large nationwide registry study, the 10-year overall risk of stroke recurrence following IS was 11% when restricting the study population to patients under 50 years of age [24]. In comparison, we found somewhat higher recurrence rates (10-year risk of 14% for a recurrent cerebrovascular event and 19% for any recurrent vascular event), possibly also due to a higher sensitivity to confirm recurrent events in our study setting.
Index event characteristics at baseline  Previous studies have also shown that the risk of death in young stroke survivors remains higher over the long-term compared with the general population, reporting 10-year mortality rates of 12%-16% in young IS patients [4,25]. In contrast, only 7% of the young IS or TIA patients died in our study during the follow-up period of 10 years. Possible explanations for the difference might include the registry-based study design in previous studies, or that acute stroke treatment and secondary prevention have improved over the years, leading to lower mortality rates. Furthermore, our study focused on IS and more benign TIA patients and unlike previous studies [13,25] excluded patients with ICH as index event, which has the highest rates of dependency and mortality of all stroke subtypes. In our study, atrial fibrillation at baseline was the only variable independently associated with recurrent vascular events. However, the total number of patients with atrial fibrillation at baseline was quite low (n = 15, 4%), so the results should be interpreted with caution. Similarly, in a Finnish study, young stroke patients with highrisk sources of CE as IS etiology had the highest hazard for recurrent cardiovascular events compared with other TOAST categories [23].
The association between cancer and stroke has been increasingly studied recently [27]. A large case-control study detected a cancer diagnosis after IS in 17% of young patients and found that stroke patients had twice the risk for developing cancer within 10 years after the index IS compared to the non-stroke control group [5]. In our multicenter study, 6% of young IS or TIA survivors had cancer, from which the vast majority of cases were diagnosed after the index event after a median time of 7 years. One registrybased study reported cancer in nearly 8% of young IS patients, but only approximately half were detected after IS, with a median time of 7 years from the index IS, similar to our findings [6]. A direct causal relation is unlikely in most of the patients in our study due to the large time gap between stroke and cancer diagnosis in most of the cases. However, common risk factors may be one of the reasons for the apparently increased risk of cancer among young stroke patients.
TA B L E 2 Number of outcome events at follow-up, person-years, cumulative 10-year risk, and cumulative incidence rates per 1000 person-years with 95% confidence intervals.

Total (n = 396) (n (%)) Person-years Cumulative 10-year risk (%)
Cumulative incidence rates per 1000 person-years (95% CI)  their medical treatment increased, which is likely at least in part due to increasing age. Furthermore, the treatment guidelines for dyslipidemia have become more stringent over the years [28].

years 10 years
However, the fact that those individuals that did not come for a follow-up visit had more risk factors for stroke and more severe strokes, makes the interpretation of these results difficult -and necessitates new strategies to reach patients who appear to be less motivated to keep contact with health care despite harboring significant vascular risk factors.
We found that one-seventh of patients in our cohort did not use antihypertensives. The use of these medications in our study appeared to be more optimal than what was found in earlier registrybased studies on young IS survivors, where fewer than half of the patients used statins [22] and use of antihypertensives was suboptimal in one-third of patients for whom these medications were prescribed [23].
Our study has several strengths. It is a fairly large multicenter study with young IS and TIA patients and with detailed information on stroke characteristics and other clinically important factors at baseline as well as at follow-up and with a relatively long follow-up period. Furthermore, as a study providing also in-person follow-up data, the risk of information bias is lower compared to purely registry-based studies.
This study also has some limitations. First, we lacked partial clinical information in those study participants who did not attend the in-person follow-up visit, and those that did attend had less severe index strokes than those who did not attend. However, we collected data on outcome events recorded also for these patients from electronic health records and registries. Second, methods to assess recurrent events and mortality in patients who did not attend the follow-up visit varied somewhat between the centers.
Even though electronic health record assessment was thorough,  conceptualization, methodology, writing -review and editing. J.P.: supervision, conceptualization, data curation, methodology, writing -review and editing. K.A.: supervision, conceptualization, data collection and curation, methodology, writing -review and editing.

ACK N OWLED G M ENTS
We thank all the patients, their families, and caregivers for their contribution to the study and the study nurses who supported this study. Hospital District (Y12493001, Y211200004, Y2112NEMIL).

CO N FLI C T O F I NTER E S T S TATEM ENT
All the authors have stated explicitly that there are no conflicts of interest in connection with this article.

DATA AVA I L A B I L I T Y S TAT E M E N T
Due to country-specific data protection legislation, the individual patient data from this study cannot be made available.

TA B L E 3
Baseline factors and their association with recurrent vascular events at follow-up. Results of multivariable model are from the last step of backwards stepwise elimination. Note: Data missing in: a Total population is 379 due to missing data in 17 (4.3%) patients, b 5 (1.3%), c 3 (0.8%), d 9 (2.3%), e 6 (1.5%), and f 1 (0.3%) patients.
Abbreviations: aHR, adjusted hazard ratio; CI, confidence interval; HR, hazard ratio; NIHSS, National Institutes of Health Stroke Scale; Ref., reference; TOAST, Trial of Org 10172 in Acute Stroke Treatment.