Since the implementation of the new approach to stroke as a neurological emergency, which has led to the progressive creation of Stroke Units and the development of new reperfusion therapies, short-term outcome has improved in developed countries (1–4) both in terms of mortality and functional outcome. The new guidelines, however, have focused mainly on patient care in the acute phase, but we have not seen many developments in post-hospital care, or secondary prevention, and some data suggest an increase in late disability in stroke patients (2).

A large part of early, medium and late morbimortality is associated with stroke recurrence, which affects 40% at 5 years and 50% at 10 years after the first cerebrovascular episode, both ischemic and hemorrhagic (5–9). The control of vascular risk factors, antiplatelet agents and statins has not significantly modified therapeutic strategy, although direct oral anticoagulant drugs have fundamentally demonstrated fewer hemorrhagic complications (10, 11).

The influence of reperfusion therapies in acute phase on stroke recurrence has not been well-established. It seems, however, that in patients undergoing mechanical thrombectomy early recurrence is lower, but in patients that receive intravenous thrombolysis medium and long-term recurrence is similar. In both cases, reperfused patients seem to have a better long-term progress as compared to non-reperfused patients (1, 12–14).

On the other hand, there is clinical evidence that moderate to severe leukoaraiosis ((LA) or white matter lesions) presence may be related with endothelial dysfunction and blood brain barrier (BBB) damage (15–18). LA presence is known to contribute to long-term functional decline, morbidity, and death in independent outpatients and in stroke patients (19). We have recently identified an endothelial dysfunction marker, the soluble tumor necrosis factor-like inducer of apoptosis (sTWEAK), as a possible biomarker independently associated with hemorrhagic transformation and poor functional outcome in patients with IS undergoing reperfusion therapies through the presence of LA (20). sTWEAK is constitutively expressed by monocytes, tumor cell lines, and endothelial cells. Via binding to fibroblast growth factor-inducible 14 [Fn14]), sTWEAK can function as an inflammatory cytokine. In this line, previous studies have shown that patients with IS had high sTWEAK levels. However, no correlation was found between sTWEAK and an ischemic area volume during acute stroke (21, 22).

At present, the primary goal of secondary prevention strategies after IS is to reduce the risk of recurrent stroke, and information on stroke recurrence and survival is useful to assess the effect of secondary prevention and risk factors for recurrence and death. In this scenario, it would be useful to identify biomarkers that could become therapeutic targets for developing future treatments or diagnostic indicators for stroke recurrence prevention; which would allow more accurate post-hospital follow-up/care, as this would lead to lower disability and mortality in medium and long-term outcome.

We hypothesized that elevated serum levels of sTWEAK might be involved in a higher frequency of stroke recurrence through the presence of LA. In the present study, we intend to investigate the possible relationship among sTWEAK—LA—stroke recurrence in reperfused IS patients; compare results with other inflammation biomarkers and evaluate the functional outcome at 3 months.

The first patient was enrolled in January 2008 and until the end of the enrollment period (December 2017) 986 reperfused IS patients were registered. Figure 1 lists flowchart of patient groups. We excluded 27 patients who died during the first 24 h and 84 patients for whom no follow-up through either personal interview or IANUS was available. Of the 875 patients eligible for this study (53.9% males; mean age 69.6 ± 11.8 years vs. 46.1% females; mean age 74.9 ± 12.6 years), 710 patients underwent intravenous thrombolysis, 87 endovascular therapy (intraarterial thrombolysis or mechanical thrombectomy) and 78 underwent both intravenous and endovascular therapy. According to the TOAST classification, 206 patients were classified as atherothrombotic (23.5%), 381 as cardioembolic (43.5%), 11 as lacunar (1.3%) and 277 as undetermined (31.7%). Symptomatic hemorrhagic transformation (HT) was noted in 280 (32%) patients during the first admission; of which, 127 suffered stroke recurrence.

The average follow-up time of the study was 25 ± 13 months, during which 127 patients (14.5%) showed stroke recurrence. In the second stroke, 24 patients were classified as atherothrombotic (19.3%), 74 as cardioembolic (59.0%), 25 as undetermined (19.5%) and 3 intracerebral hemorrhages (2.2%). Recurrence was lower in patients with effective reperfusion (4.5 vs. 10.0%, p < 0.0001), and in patients undergoing endovascular treatment (6.9%), than in those who received intravenous thrombolysis (14.9%) or those who received both reperfusion therapies (19.2%, p < 0.0001). Functional outcome after the second stroke was worse than after the first stroke (mRS at 3 months after the first stroke 1[0, 3] vs. 4 [3, 6] after the second, p < 0.0001). Consequently, rates of poor functional outcome (26.1 vs. 85.8%, p < 0.0001) and mortality (7.1 vs. 28.3%, p < 0.0001) at 3 months were higher after the second stroke.

Stroke recurrence is the first cause of increased mortality and non-motor sequelae, and this complication persists in patients undergoing reperfusion treatments (1, 5–9, 12–14, 28). In our series of patients with acute IS, who received the best possible treatment according to management guidelines, recurrence was 14.5% for an average follow-up time of 2 years. The outcome of patients with recurrence was poor in 86% of cases, with a mortality of 28%. Recurrence in our study is similar to that obtained by several authors (29), but higher than that referred in other studies. This may be explained by the fact that our follow-up time was longer, and the age of our patients was higher. Mortality, however, was similar in all the studies reviewed (6, 12, 30, 31).

The type of stroke did not influence the frequency of recurrence, although the second episode led to the reclassification of almost 50% of the undetermined strokes into cardioembolic, and three patients with cardioembolic strokes recurred as intracerebral hemorrhage. Recurrence has been significantly lower in patients undergoing thrombectomy than in the case of systemic thrombolysis, and much lower than when the procedure was combined. Previously published data are uneven (1, 12–14, 32, 33). In our cases, these results were not influenced by the time between the onset of symptoms and treatment (p = 0.108), or follow-up time (p = 0.424, data not shown). However, patients in whom an effective reperfusion was achieved presented with lower recurrence rates. In a previous research work, we found that the treatment with tPA without reperfusion is associated with a worse patient progression, possibly due to the toxic effect of the drug in these cases (34).

In our study, oral anticoagulation and white blood cell count in the first stroke were independent factors associated with stroke recurrence. Although in the first episode the frequency of cardioembolic subtype was similar in both groups, in the second episode half of undetermined strokes were reclassified into cardioembolic, which implies an undervaluation of the initial diagnosis of cardioembolic. The platelet count was similar in both groups and functional situation before stroke was worse in the patients who recurred, although this data did not reach independence in the multivariate model (13).

It is interesting that LA has been the strongest factor associated with stroke recurrence, and this association is directly related to the severity and extent of the white matter lesion. Despite the differences in the neuroimaging study and the method to quantify LA, this association is widely reported in the literature (35–41). There are, however, some differential data: (1) the association with lacunar infarctions (35) (in our series, only in reperfused patients, and none in the 11 lacunar infarctions recurred), and (2) the relationship with cardioembolic infarctions, which is not found in any study (38, 39). In our case, the association between LA and recurrence was similar in atherothrombotic, cardioembolic and undetermined strokes (p = 0.383). A possible explanation for this discrepancy might be that in our series the patients with cardioembolic strokes had a more advanced age (atherothrombotic 69.6 ± 12.6 years, cardioembolic 73.8 ± 11.8 years, lacunar 67.3 ± 11.8 years and undetermined 71.8 ± 12.9 years). Aside from theses discrepancies, LA is currently an important factor of poor outcome after a stroke.

Of the determined inflammatory markers (white blood cells, fibrinogen, C-reactive protein, sedimentation rate, IL-6 and TNFα), only white blood cells and TNFα maintained statistical significance in the first regression models but they lost it when including all clinical factors and of neuroimaging. However, sTWEAK (levels ≥ 7,000 pg/mL) was associated independently with an increased risk of stroke recurrence. The strong relationship between the sTWEAK levels in the first stroke and the severity of LA suggests that sTWEAK is a surrogate marker for LA, and thus, when we included the severity of LA in the regression model, sTWEAK disappeared as an independent recurrence factor (Table 4, Model B).

sTWEAK is a type II transmembrane glycoprotein of the TNF (tumor necrosis factor) superfamily that acts by binding to Fn14 which is a small transmembrane type I protein. TWEAK-Fn14 is expressed in all cells that act in the Neurovascular Unit and overexpresses within a few hours of establishing a cerebral ischemia (42–45). TWEAK-Fn14 overexpression induces an inflammatory profile in brain endothelial cells with increased secretion of proinflammatory cytokines, production and activation of matrix metalloproteinases that will participate in the disruption of the blood-brain barrier and expression of intercellular adhesion molecules involved in the union of white blood cells to the endothelium (46, 47). This maintained expression could condition the development and progression of LA and could be the molecular marker associated with white matter disease associated with chronic cerebral ischemia. This hypothesis, however, remains to be demonstrated.

From a clinical point of view, the importance of sTWEAK as predictor of LA progression associated with the increase of stroke recurrence does not seem preferred, since neuroimaging is more sensitive and specific, at least with the method used (we have exclusively determined sTWEAK, and no sTWEAK-Fn14). However, the possibility of blocking the activation of the sTWEAK-Fn14 system (anti-sTWEAK or anti-Fn14 monoclonal antibodies, or through sTWEAK-Fn14 fusion blockade) makes this marker a hopeful therapeutic target that could decrease the progression of LA and stroke recurrence (48, 49).

This study has some limitations. First, our study presents the weaknesses of any retrospective study, even if its origin is prospective. Bias in the enrolment of patients was reduced as we enrolled all those registered in our hospital and followed-up in any hospital of the public system (in Galicia, the network of private hospitals is small). Second, sTWEAK measurements were not simultaneous and were made by different researchers, although measurements were always blind to the clinical and neuroradiological data and supervised by the same senior researchers, and the same is true of clinical and neuroradiological data. Three, it is important to note that LA is a gradual disease affected by different risk factors, and not associated with a unique pathological process (16, 50). There is the possibility that LA may be associated with factors in the study population other than stroke. It is known that in regions corresponding to LA on neuroimaging, the wall of penetrating arteries is thickened and hyalinized, and there is often narrowing, elongation, and tortuosity of small vessels, potentially leading to reduced cerebral blood flow, and permanent BBB damage. Furthermore, after the first stroke, Wallerian degeneration (WD) could develop and cause new white matter hyperintensities related with LA progression (51). Four, serum levels of sTWEAK do not represent a specific marker of a particular process; patients with multiple sclerosis, heart failure, or atherosclerosis show also variations in the sTWEAK levels (52). However, we investigate the possible relationship among sTWEAK—LA—stroke recurrence in reperfused IS patients. The strong points of this work are the unbiased screening of individuals, the high number of enrolled patients, and the large number of biomarkers assessed.

Stroke recurrence is associated with increased mortality, non-motor sequelae. Currently, preventive efficacy is limited. The presence of an advanced degree of LA, as well as its progression, is the main neuroimaging factor associated with stroke recurrence. sTWEAK (≥ 7,000 pg/mL) is a biomarker correlated with the progression of LA and stroke recurrence. sTWEAK could become a diagnostic boimarker and a potential therapeutic target in reducing stroke recurrence but further studies will be necessary.

The original contributions generated for this study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

The studies involving human participants were reviewed and approved by This research was carried out in accordance with the Declaration of Helsinki of the World Medical Association (2008) and approved by the Ethics Committee of Santiago de Compostela (2019/616). The patients/participants provided their written informed consent to participate in this study.

PH, RI-R, and JC: conception and design of the study. FC, AdS-C, TS, and MP-M: data acquisition and analysis. JC, MR-Y, JP, and IL-D: clinical data acquisition and analysis. RI-R, TS, JC, PH, and AdS-C: handled funding and supervision. PH and JC: statistical analysis. RI-R, PH, and JC: manuscript drafting. MP-M, TS, JC, and FC: critical revision for important intellectual content. JC, MR-Y, JP, IL-D: supervision. All authors reviewed and approved the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.