Study inclusion
The flowchart of literatures searched and included at each stage of analysis were shown in Figure 1. After the initial keywords search, 126 articles were included, of which 39 repetitive literatures were removed, and 50 irrelevant literatures were excluded by reading titles and abstracts. Read the full text of the remaining 37 articles, excluding 5 abstracts, 3 reviews and 17 irrelevant results. Finally, 12 literatures were included in this meta-analysis[11, 12, 19-28].
Study characteristics
A total of 12 studies (n=3641 patients) published between 2015 to 2020 were included in this meta-analysis. All 12 studies used standardized intravenous thrombolytic therapy for AIS patients: rt-PA (0.9 mg/kg body weight, maximum 90mg) was injected intravenously within 4.5 h after the onset of ischemic stroke, with 10% of the total dose as a bolus and the rest by a 60-minute infusion. Blood samples were collected on admission (n=5)[19, 21, 23, 27, 28] and within 24 hours after IVT (n=3)[12, 24, 26], 4 studies were collected at both two times[11, 20, 22, 25]. HT (including symptomatic intracerebral hemorrhage and parenchymal hemorrhage), mRS at 3 months, mortality were reported in 6, 10, and 4 articles. The best cut-off values of NLR ranging from 2.2 to 10.59. Since all the 12 studies were cohort studies, we used NOS for quality assessment, with a score range 6 to 8 points. For the basic characteristics and quality assessment of the 12 included studies were demonstrated in Table1.
Meta-analysis
Hemorrhagic transformation
Six studies reported the relationship between NLR and HT after IVT[11, 12, 19, 22, 25, 27]. Higher NLR levels were associated with an increased risk of HT (OR=1.33,95%CI=1.14-1.56, P<0.001). Significant heterogeneity between studies were observed (I²=71.8%, P<0.001) (Figure2).
Functional outcome
Ten studies reported the association between NLR and 3-month poor functional outcome (mRS≥3) after IVT[11, 19-26, 28]. Higher NLR levels were associated with higher risk of 3-month poor functional outcome (OR=1.64,95%CI=1.38-1.94, P<0.001), significant heterogeneity between studies were found (I²=86.3%, P<0.001) (Figure3).
Mortality
Four studies showed the relationship between NLR and 3-month mortality[19, 21, 22, 26]. There was no significant association between higher NLR levels and higher risk of 3-month mortality (OR=1.14, 95%CI=0.97-1.35, P=0.120). Moreover, significant heterogeneity between studies were observed (I²=81.0%, P<0.001) (Figure4).
Subgroup analysis
In subgroup analysis, studies were divided according to time of blood sample collection, age, country, presence or absence of infection, stroke severity and onset-to-IVT time. Subgroup analysis of HT suggested that admission NLR levels rather than post-IVT NLR levels was associated with increased risk of HT (OR=1.33,95%CI=1.01-1.75, P=0.039). Higher risk of HT was observed in elderly group (OR=1.32,95%CI=1.11-1.57, P=0.002). Higher NLR levels were associated with higher risk of HT in the studies excluded infection (OR=1.64,95%CI=1.14-2.35, P=0.008). Interestingly, the onset-to-IVT time less than 3 hours was associated with higher risk of HT (OR=1.32,95%CI=1.11-1.57, P<0.001). Regardless of the country and the stroke severity, the relationship between NLR and HT remains significant (Table2).
In addition, subgroup analysis of 3-month poor functional outcome suggested that blood sample collection time, age, country, onset-to-IVT time and presence or absence of infection had no significant influence on the overall results. Higher NLR levels were associated with higher risk of poor functional outcome in the studies with moderate stroke severity (NIHSS≥8) (OR=2.15,95%CI=1.50-3.09, P<0.001) (Table2).
Publication bias and sensitivity analysis
The evidence of publication bias in the studies of those provide HT and functional outcomes were detected by Egger test (P=0.019 and P=0.001, respectively). After the trim-and-fill test, the pooled OR were 1.31(0.99-1.72) and 1.42(0.86-2.36), respectively (Table S1).
Sensitivity analysis showed that no studies have affected the effects of the pooled OR, indicating that the results of this meta-analysis were stable (Figure S1, S2).