In this study, we first demonstrated that ESUS patients in the retrospective HYSR were younger than patients with well-defined stroke etiology and had less cardiovascular risk factors than the other subgroups. Based on these findings, we planned and launched the prospective SECRETO study to explore more systematically these differences and other characteristics of early-onset cryptogenic ischemic stroke. By using data gathered in SECRETO, we showed a strong association between any type of migraine and particularly MA and early onset cryptogenic stroke. Furthermore, we demonstrated an association between endothelial dysfunction and cryptogenic stroke in men and slightly older patients.
Study I: ESUS in young patients
We found that in HYSR, 21% of all young ischemic stroke patients could be retrospectively classified as ESUS. These patients were more frequently women, significantly younger, and had fewer cardiovascular risk factors and comorbidities as well as milder strokes (lower NIHSS score on admission) than patients with well-defined stroke etiology. However, the prevalence of migraine and PFO was higher in ESUS patients, and they were also using COCs more frequently. Regarding long-term outcome, with a median 10.1-year follow-up, young ESUS patients had better cardiovascular outcome and lower all-cause mortality than other TOAST subgroups. However, although ESUS and CE patients had a few differences in the baseline risk factor profile, they possessed similar adjusted risk of recurrent stroke. In 2018, our study was the largest one reporting the characteristics of young ESUS patients and potentially inspired others to explore their registries and early-onset cryptogenic strokes.
Other studies also support our findings. In the Global ESUS Registry (n=2144), ESUS patients were also significantly younger than patients in other subgroups and a similar proportion (~43%) were female.47 Prior to our study performed in 2018, only one report had focused on young ESUS patients alone, including 100 patients aged between 26 and 55 years.45 That study reported, similarly to ours, that 4/5 of early-onset cryptogenic ischemic stroke patients could further be classified as ESUS. That study also showed that 43% of the patients were female, as
68 in our study. In the last two years, a few newer studies have reported results from young ESUS registries. Perera et al. compared younger ESUS patients with older ESUS patients and showed that, as in our study, the median age of younger ESUS was 40 years, but only 33% of them were women.257 Compared with older ESUS patients, younger ESUS patients had expectedly a lower burden of cardiovascular risk factors, but the severity of the index stroke was similar in older and younger ESUS patients (NIHSS 4 vs. 3, P=0.06). In contrast, a study of all-aged ESUS patients from Greece reported that ESUS patients had a frequency of risk factors similar to other etiologic subgroups.46 Differences in these studies are most probably due to the limitations of retrospective identification of ESUS patients from pre-existing stroke registries, differing population characteristics, long inclusion periods, and improved diagnostics and control of risk factors over time. Moreover, the prevalence of PFO in ESUS patients in our study is supported by other case-control studies.79
The cumulative 15-year risk of recurrent stroke or TIA (secondary outcome in our study, shown in the supplement) was 15.4%. This is in accordance with a later study on young ESUS patients with a 15-year cumulative incidence rate of 15.0%. That study also showed that the 15-year cumulative rate of new-onset AF was 5.5%.258 Our registry suggests that ESUS patients might have an overall better long-term outcome than LAA, CE, and SVD patients after adjustment for relevant confounders. However, the cumulative risk for recurrent strokes was similar for both ESUS and CE. This might be due to more optimal and targeted secondary prevention in CE than in ESUS, and vascular risk profile may have changed over time independently from the etiology of the first-ever stroke.
Study II. SECRETO protocol
Globally, every year at least half a million early-onset ischemic strokes remain cryptogenic despite extensive and timely diagnostic work-up. These patients must live in uncertainty since data on optimal secondary prevention are scarce. Thus, early-onset cryptogenic stroke should be considered as one of the most important focuses in stroke medicine. The ongoing SECRETO study will cover several key domains of early-onset cryptogenic ischemic stroke, including chemical, physical, and psychological triggers, disturbances in thrombosis and hemostasis pathways, oral health, and long-term outcome. It could be hypothesized that well-documented risk factors are highly prevalent in cryptogenic ischemic stroke, but it seems justified to assume that the mechanisms and strength of associations of these risk factors differ from those of
older-69 onset stroke.52 However, these patients may also have many less well-documented risk factors associated with the stroke such as migraine, PFO, and high alcohol consumption. Furthermore, SECRETO will probably generate novel data on pathways of thrombosis and hemostasis, and these biomarkers are collected at two time points, during the sub-acute phase and at 3 months.
We also aim to apply exome- and whole-genome sequencing techniques and add them to a traditional genetic family study setting, thus expecting to efficiently reveal novel variants contributing to the higher risk for early-onset cryptogenic ischemic stroke. Substudies performed in Finland alone, such as SECRETO Extended Cardiovascular Study, including measurement of endothelial function, and SECRETO Oral, may provide us with interesting results regarding associations between cryptogenic stroke and endothelial dysfunction, arterial stiffness, increased carotid intima-media thickness, and periodontitis. Finally, each patient will be followed for ten years. In general, SECRETO will thus provide us with further knowledge of several key risk factors and triggers in early-onset cryptogenic ischemic stroke. It will also improve our understanding of long-term outcome in this subgroup of patients, including recurrent stroke and transient ischemic attack and other cardiovascular endpoints, as well neuropsychological, functional, and social outcomes. This information is crucial in order to improve the treatment of these patients and further to identify patients with unfavorable prognosis and ultimately to prevent incident and recurring strokes in the young.
Study III. Migraine and early-onset cryptogenic ischemic stroke
This study demonstrated an independent association between any migraine and MA and early-onset cryptogenic ischemic stroke when adjusted for several baseline characteristics and vascular comorbidities. The results remained unchanged when stratified by sex. Furthermore, MA was associated with cryptogenic stroke irrespective of the presence of PFO. We found no association between MO and stroke, most probably due to inadequate sample size and power.
Our migraine screener performed well against an experienced senior headache-neurologist, showing relatively high sensitivity and specificity especially in patients with any migraine or MA. Sensitivity and specificity were lower for MO patients and controls. One previously reported screener validated for stroke, the 5-question Migraine Screener for Stroke (MISS),259 performed poorly in identifying MA, and the overall sensitivity was 47% and specificity 97%.
These differences compared with our screener might be explained by screener properties, different populations (ischemic and hemorrhagic stroke in all-aged patients vs. early-onset
70 cryptogenic ischemic stroke patients), and differing degree of neurological deficits. However, our results should be validated in other patient subgroups as well, including subgroups where the prevalence of MO is higher.
Previous meta-analyses have demonstrated an association between MA and any type of ischemic stroke in all-aged individuals with a pooled adjusted OR of 2.3 (95% CI 1.5-3.3).
These meta-analyses included both case-control and control studies.100,260 A recent Danish population-based cohort study found a similar magnitude of heightened risk for both MA and MO (adjusted HRs 2.5, 2.2-2.9 and 1.8, 1.6-2.1, respectively).101 Compared with these studies, the stronger association between any migraine and MA and early-onset cryptogenic ischemic stroke shown in our study indicates that migraine is an even more profound risk factor in this highly-selected population. Only one of the previous three case-control studies shown in Table 3 showed a weak association (OR 1.6, 1.1-2.3) between probable MA and stroke of undetermined cause in women aged <50 years. However, this association diminished after adjustment for age, ethnic background, geographic region, diabetes, and hypertension.104 Another study found an association between any migraine in older patients and cryptogenic ischemic stroke or transient ischemic attack but not in younger patients. However, that study included only 89 younger patients.102 Several mechanisms might explain the association between MA and ischemic stroke, including increased platelet aggregation, cortical spreading depression leading to cerebral microcirculatory vasoconstriction, increased concentrations of procoagulant factors, and endothelial dysfunction.87,261 Furthermore, estrogen use and habitual risk factors, such as smoking, might explain the stronger association in women, as shown also in our study. In contrast to a previous study,86 we did not, however, find a significant interaction between MA and female sex or MA, estrogen use, and smoking in women. This might have been due to the low frequency of female smokers in our study or perhaps lower estrogen doses were used than a few decades ago. Finally, previous studies have demonstrated an association between prevalence of PFO and MA.91,92 However, in our study especially the magnitude of PFO was associated with MA in cryptogenic ischemic stroke. Potential mechanisms explaining this association are, among others, paradoxical embolism and serotonin-induced platelet activation.87
To the best of my knowledge, this study is the first to explicitly demonstrate the strong association between MA and early-onset cryptogenic stroke, regardless of sex and independent
71 of vascular risk factors. This association remained robust independently of PFO, although larger PFO was associated with a higher prevalence of MA. Future studies should further explore the significance of activity, recent onset, and lifetime burden of migraine (especially MA) in early-onset cryptogenic ischemic stroke. Moreover, studies on shared genetics and coagulation abnormalities in MA and stroke are warranted.
Study IV. Endothelial dysfunction and early-onset cryptogenic ischemic stroke
In this study, an association between endothelial dysfunction, measured with EndoPAT, was found in men and in a subgroup of patients aged ≥41 years. This association remained significant after adjusting for demographic factors, vascular comorbidities, and laboratory measurements, including admission Total-C/HDL-C ratio. However, in the overall population of 136 case-control pairs, we did not find a significant association when LnRHI was explored as a continuous variable or in tertiles. There are only a few previous studies reporting an association between endothelial dysfunction and stroke in younger patients, and these studies registered endothelial function with FMD. One study showed that young patients with either spontaneous carotid artery dissection (mean age 41.8 years) or with vertebral artery dissection (mean age 45.6 years) had a significantly lower FMD than young patients with cryptogenic ischemic stroke (mean age 43.6 years).179 Another study with 50 middle-aged patients with both ischemic and hemorrhagic strokes found that stroke patients had a lower FMD index than healthy controls. In that study, FMD index was even lower in ischemic stroke patients than in patients with hemorrhagic stroke. In ischemic stroke patients, no differences were present between etiologic subtypes. However, the number of patients (n=50) was rather low.180 Endothelial dysfunction has been shown to be associated with male sex, increasing age, smoking, diabetes, higher body mass, binge drinking, and Total-C/HDL-C ratio.262,263 This was also shown in our results where patients in the lower lnRHI tertile were more often men, obese, slightly older, and had a higher Total-C/HDL-C ratio than in other tertiles. Furthermore, adjusting for study visit lipid status (after initiation of statins in 4/5 patients), the association between LnRHI and stroke in men was no longer significant. There was a significant change in lipid status between admission and study visit, and this might serve as a surrogate for statin initiation. It is well-known that such drugs as statins, angiotensin receptor blockers, and angiotensin-converting enzyme inhibitors can improve endothelial function,178 which perhaps was shown in our study, as even a short-term exposure to especially statins often improved
72 endothelial function. Furthermore, glycated albumin enhances platelet activation and aggregation and has been shown to be associated with ischemic stroke in diabetic patients.
264-266 In prediabetic patients with acute ischemic stroke, glycated albumin is also associated with early neurological deterioration.267 Also in our study, the independent association of LnRHI was dampened by replacing history of diabetes with measurement of glycated albumin.
Again, this study was the first to explore the association between endothelial dysfunction and early-onset cryptogenic ischemic stroke. In addition, we explored more specifically several demographics and vascular risk factors associated with endothelial dysfunction in this patient group. These results with other future studies might eventually guide our treatment decisions, e.g. whether to prescribe statins for young patients with first-ever cryptogenic stroke. However, we still need direct evidence on whether improving endothelial function also reduces the risk of recurrent stroke and other cardiovascular events.