Background
Multiple randomized controlled trials have demonstrated the efficacy of intravenous recombinant tissue plasminogen activator (IV rtPA) administered up to 4.5 h after the onset of symptoms of ischemic stroke [
1]. However, the risk of hemorrhagic transformation (HT) is increased by as much as tenfold after IV rtPA, largely due to reperfusion injury and the toxic effects of rtPA [
2]. In addition to the already known indicators for HT [
2], the detection of new paradigms is still worthwhile. In addition, an improved understanding of the prevention or early risk assessment of rtPA-related HT may also be applicable to other reperfusion strategies such as endovascular therapy.
In animal studies, neutrophils have been shown to contribute to intracerebral hemorrhaging after treatment with rtPA following cerebral ischemia, while depletion of neutrophils reduces blood–brain barrier (BBB) disruption and the rate of HT [
3,
4]. In humans, infiltration of matrix metalloproteinase-9 (MMP-9)-positive neutrophils is associated with BBB breakdown, basal lamina type IV collagen degradation, and HT [
5]. Recent studies suggested that the initial neutrophil to lymphocyte ratio (NLR) is associated with mortality and infarct size in ischemic stroke [
6,
7] and can predict the 90-day outcome after endovascular therapy [
8]. However, all of these studies mainly focused on static NLR values at baseline, which may not reflect the comprehensive dynamic changes of patients’ conditions. Furthermore, there is also a lack of information on the clinical value of the NLR in acute ischemic stroke patients treated with IV rtPA, especially its relationship with the most serious and common complication of IV rtPA treatment, HT.
Thus, we aimed to explore the temporal variation of the NLR in patients and its relationship with the most serious subtypes of HT, namely symptomatic intracranial hemorrhage (sICH) and parenchymal hematoma (PH), in patients with ischemic stroke treated with IV rtPA [
9].
Discussion
This study first shows that the NLR is a dynamic variable, and its variation is associated with HT after treatment with IV rtPA in patients with acute stroke. In addition, the best discriminating value of the NLR for PH and sICH was 10.59 or more at 12–18-h post-treatment, which was associated with an 8.50-fold increased risk for PH and a 7.93-fold increased risk for sICH.
In previous studies, a high NLR was found to be independently associated with an increased risk of stroke in atrial fibrillation [
13]. Moreover, the initial NLR is associated with infarct size and mortality rate in ischemic stroke [
6,
7], and it also has predictive value for 90-day outcome after endovascular therapy [
8]. However, the previous studies did not explore the clinical value of the NLR in acute ischemic stroke patients treated with IV rtPA.
Recently, Maestrini et al. reported that higher neutrophil counts and NLR before thrombolysis for cerebral ischemia are independently associated with sICH and worse outcome at 3 months [
14]. However, they did not exclude patients with previous infections, which might contribute to the difference in the predictive values of the baseline NLR for sICH between their study and ours. That is because infections can lead to poor outcome after stroke via many different mechanisms, which may include (1) increased BBB disruption and tissue damage by neutrophil-derived various proteases, reactive oxygen species (ROS), as well as numerous inflammatory mediators; (2) impaired tissue reperfusion through endothelia-dependent mechanisms; (3) increased platelet activation and microvascular coagulation; and (4) CRP-induced ischemic tissue injury via a complement-dependent mechanism [
15,
16]. Therefore, this shows that neutrophils are just one of many different mechanisms, and the association between the baseline NLR and endpoints may disappear after adjustment for infections. The reasons for such a difference may also relate to what the NLR represents. Changes observed in the baseline NLR could reflect the disease itself or external environment factors such as infection or cancer. Our results and their conclusions do not contradict per se because their study population is different to ours, with the NLR at baseline reflecting both the disease itself and the external environment factors in their study, but only the disease itself in our study.
The study by Maestrini et al. has a higher statistical power than our study based on the larger sample size [
14], and we also believe that their results are fully credible and reliable. Higher numbers of baseline neutrophils at baseline have greater potential to induce tissue damage via the release of various proteolytic enzymes, ROS, and numerous inflammatory mediators [
2]. Thus, in theory, patients with infection or other conditions that can potentially change baseline the NLR have a higher risk for the occurrence of HT after treatment with IV rtPA. Similarly, when we reintegrate the patients with infection or other conditions with potential to change the NLR into our analysis, the baseline NLR is higher in patients with HT (Additional file
1: Figure S3).
Moreover, the neutrophil and lymphocyte counts after ischemic stroke exhibit significant temporal variation [
17], which is also indirectly confirmed by Maestrini et al. in their evaluation of the influence of the onset-to-sample time on the neutrophil count, leukocyte count, and NLR [
14]. This suggests that the neutrophil count, lymphocyte count, and NLR are “dynamic” variables. However, Maestrini et al. mainly focused on static neutrophil count and NLR values at baseline [
14], which may not have dynamically and comprehensively reflected the patients’ conditions. The clinical application value of dynamically testing the neutrophil count, lymphocyte count, and NLR in sICH and worse outcomes may be meaningful. Holding strict exclusion criteria in our study, we found that the NLR changed dynamically and a high NLR at 12–18 h after treatment with IV rtPA was independently associated with HT after IV rtPA. In addition, although there were no differences between the NLR at 12–18 h and the NLR at 36–48 h for auxiliary diagnosis of PH and sICH, we hold that the NLR at 12–18 h may be more valuable than that at 36–48 h. First, the NLR at 12–18 h could be obtained earlier than the NLR at 36–48 h, which could allow for better monitoring and could better reflect the severity and progression of disease, helping clinicians to adjust medication regimens and apply related auxiliary examination in time. Moreover, the NLR at 12–18 h was not inferior to the NLR at 36–48 h, and it exhibited a tendency to rise superior to the NLR at 36–48 h for diagnosing sICH. Therefore, we believe that the NLR at 12–18 h is the appropriate selection based on the main concerns in the present study.
The mechanisms underlying these observations are not well established, but they seem to be related to the roles of neutrophils and lymphocytes in ischemic insult and the disruption of the BBB. Circulating neutrophils are recruited to the site of cerebral injury shortly after ischemia occurs and then further contribute to BBB disruption and tissue damage via a variety of mechanisms [
18‐
20]. Neutrophils have been shown to be an important source of MMP-9, which may open the BBB inside the lumen of the blood vessel by acting directly on tight junction proteins or may be absorbed into endothelial cells and act on the basement membrane [
2]. In a rat cerebral ischemia model, treatments preventing neutrophil infiltration reduced MMP-9 released in the brain [
21]. Moreover, inhibition or depletion of neutrophils can reduce the BBB breakdown and the rate of HT in ischemic stroke [
3,
22]. In contrast, when neutrophils are increased via lipopolysaccharide or granulocyte colony-stimulating factor administration, there is an increase in BBB disruption in a mice model [
23] and an increase in MMP-9 and rtPA-related HT in a rat stroke model [
4]. In humans with ischemic stroke, early neutrophilia is associated with larger infarct volumes [
24], and MMP-9-positive neutrophil infiltration has also been associated with disruption of the BBB, basal lamina type IV collagen degradation, and HT [
5]. Thus, neutrophils may mediate HT through neutrophil-derived MMPs in ischemic stroke. In addition to neutrophil-derived MMPs, other factors released from neutrophils after stroke [
20], including ROS, myeloperoxidase, elastase, cathepsin G, proteinase 3, cytokines, and chemokines, can also disrupt the neurovascular unit and ultimately result in increased BBB permeability and HT [
17].
In addition to the mechanisms mediated by factors released from neutrophils, novel aspects of neutrophil biology may also contribute to ischemic brain injury. Recently, activated neutrophils have been described to form neutrophil extracellular traps (NETs), a web-like structure composed of DNA, histones, and specific granule proteins, such as neutrophil elastase and myeloperoxidase, in response to various stimuli [
25]. Recent evidence indicates that a lack of NETs during myocardial and liver ischemia/reperfusion (I/R) injury offers significant cardioprotective, hepatoprotective, and anti-inflammatory effects [
26,
27]. Furthermore, extracellular chromatin and histones exacerbate cerebral I/R injury in mice [
28]. These results suggest that NETs may play a role in BBB disruption and tissue damage. Further studies are needed to explore whether NETs have deleterious effects on HT.
The lymphocyte counts might serve as an index for general health, influenced by acute physiologic stress [
29]. Relative lymphopenia on the other hand reflects the cortisol-induced stress response and sympathetic tone [
30], which can increase the production of proinflammatory cytokines that aggravate ischemic injury [
31]. This means that low lymphocyte counts in patients with HT are not merely an initial response to severe stroke, but that lymphocytes may be actively involved in a protective mechanism in the ischemic brain. Experimental evidence suggests that specific subtypes of lymphocytes (namely, regulatory T cells) play key roles in abrogating the inflammatory response and are major cerebroprotective immunomodulators in acute stroke [
32]. Our findings suggest that lower lymphocyte counts in patients with HT (data not shown) might have been the result of fewer regulatory T cells being available to curtail the inflammatory response, thereby leading to greater tissue damage. However, other subtypes of lymphocytes (namely, proinflammatory lymphocytes) may have a deleterious effect on I/R injury [
33]. It is uncertain which subtype of lymphocytes has a dominant role in the pathophysiology of cerebral ischemia, and we demonstrated that a decrease in lymphocytes as a whole has a negative effect on HT. Further studies are also needed to elucidate the complex immunomodulatory interactions that occur after stroke.
The NLR reflects the balance between neutrophil and lymphocyte levels, which may be comprehensively represent the immunological conditions. In this sense, the NLR is superior to only the neutrophil count or lymphocyte count for distinguishing the occurrence of HT, and this may also explain why the AUC for the NLR appeared to be greater than those for neutrophil and lymphocyte counts at each corresponding time point (Additional file
1: Table S4). On the other hand, inflammatory cytokines released by neutrophils may trigger lymphocyte apoptosis [
34]. This suggests that the NLR may not simply reflect the neutrophil and lymphocyte counts but also overactivation of neutrophils, thus leading to a wider gap between the two leukocyte types, and this also supports the superiority of the NLR.
The main strength of our study is that the clinical information and blood samples taken at different time points from all patients were collected in a prospective fashion with strict exclusion criteria. Previous infections and early hospital infections in stroke are associated with an increase in leukocytes and poor outcomes [
14,
35]. We limited these potential confounders by ruling out patients with infection. In addition, our observations of elevated neutrophil counts in patients with HT (data not shown) may offer a partial explanation as to why stroke patients with infection may have poor outcomes. Nonetheless, our current findings also have some limitations. First, the small sample size weakens the statistical strength of our conclusions. Therefore, further studies with larger samples are needed. Second, the study population included patients receiving a bridging strategy of the use of IV rtPA followed by endovascular therapy, which may interfere with our results. Nevertheless, similar results were found for patients experiencing a bridging strategy even if the sample was small (data not shown). On the other hand, this shows that the NLR may also be applicable to distinguish the presence and absence of HT in other reperfusion strategies such as endovascular therapy. Regrettably, a control group that did not experience reperfusion therapy was not included in our study. Third, we neither explored the mechanisms by which neutrophils and lymphocytes affect the BBB breakdown and HT nor investigated what factors regulate the dynamic changes in neutrophil and lymphocyte counts after ischemic stroke in animal studies. These will be the focus of our next work.