Relationship between estimated glomerular filtration rate and outcome of ischemic stroke patients after mechanical thrombectomy

Abstract Aims We aimed to determine whether preprocedural renal function affects the outcome of acute ischemic stroke (AIS) patients with mechanical thrombectomy (MT) and whether this effect is modified by the onset‐to‐reperfusion time (OTR) and mediated by collateral status. Methods Eligible patients with anterior circulation large vessel occlusion (LVO) who underwent MT between August 2018 and August 2020 were reviewed. The main outcome was good functional outcome [defined as modified Rankin Scale (mRS) of 0–2] at 3 months. Multivariable logistic regression analyses were conducted to explore the relationship between renal function and good functional outcome. A moderation analysis and the Johnson‐Neyman technique were performed to assess the interaction between renal function and OTR to predict the outcome of AIS patients with MT. Results Among 100 enrolled patients, 36 (36%) exhibited good functional outcome. A decreased preprocedural estimated glomerular filtration rate (pre‐eGFR) was significantly associated with worse functional outcome [adjusted OR, 1.059 (1.012–1.108); p, 0.014], and this effect was partly mediated by collateral circulation. An interaction between OTR and pre‐eGFR on functional outcome was observed (P for interaction, 0.22), and pre‐eGFR only had a significant effect on functional outcome when OTR exceeded 455.8 min. Moreover, the adverse effect of OTR on functional outcome became no longer significant when the pre‐eGFR was higher than 89.0 mL/min/1.73 mL/min/1.73 m². Conclusions Renal function was related to functional outcome at 3 months, and this relationship could be modified by OTR. The results suggested that reducing OTR and improving collateral circulation may mitigate the adverse effect of reduced kidney function on functional outcome.


| INTRODUC TI ON
Mechanical thrombectomy (MT) is a mainstay in acute stroke treatment and benefits patients with acute ischemic stroke with large vessel occlusion (LVO). 1 However, not all patients have a good outcome following MT. It is necessary to have a deeper understanding of the factors associated with the outcome after MT to more accurately predict the benefits to individual patients and help clinics in decision-making regarding whether to perform MT. Renal impairment, characterized by a reduced estimated glomerular filtration rate (eGFR), is prevalent in patients with acute ischemic stroke (AIS). Recently, two studies demonstrated that renal dysfunction was an independent predictor of the functional outcome for patients treated with MT. 2,3 However, Laible et al. found that reduced eGFR was not associated with poor functional outcome of AIS patients receiving MT. 4 The relationship between renal function and functional outcome may be affected by other factors.
The onset-to-reperfusion time (OTR) is generally considered a critical important determinant of functional outcome in patients receiving MT, and every 30-minute delay may yield a 12-21% decrease in the likelihood of regaining functional independence. 5 Given the growing evidence on the impact of renal function on cerebral vascular structure 6,7 and hemodynamics, 8,9 patients with worse renal function would be expected to be more vulnerable to ischemic status, which might be related to the rapid evolution of ischemia to infarction. From this perspective, patients with worse renal function would need to acquire successful recanalization within a shorter time than patients with normal renal function to salvage the function of the ischemic penumbra to achieve similar functional outcomes. The impact of OTR on functional outcome might also differ according to renal function.
It is well known that collateral circulation is related to the clinical and radiological outcomes of AIS patients. 10,11 Collateral circulation is an established key factor determining the pace of evolution of ischemic penumbra into the ischemic core. 12,13 Similar anatomical and functional characteristics of vessels have been shown in the kidney and brain, and some shared vascular risk factors have been found in renal dysfunction and AIS. Whether cerebral collateral circulation affects the relationship between renal function and MT outcome needs to be studied.
In the present study, AIS patients receiving MT between August 2018 and August 2020 at our stroke center were included. We investigated the relationship between renal function and functional outcome at 3 months and the potential effects of collateral circulation and OTR on the relationship. We hypothesized that renal function was independently associated with functional outcome following MT, and this link was partially mediated by collateral flow and could be modified by OTR.

| Study population
We performed a retrospective analysis of a prospective registry of consecutive AIS patients who underwent MT for anterior circulation LVOs at our stroke center between August 2018 and August 2020. The patients were deemed candidates for MT according to national guideline recommendation and neuro-interventionist's evaluation. They also received standard treatment for AIS. During this period, patients were considered eligible for this study if they met the following criteria: modified Rankin Scale (mRS) of 0 or 1 before stroke; age ≥ 18 years; treatment initiated (ie, onset-topuncture) within 24 h of symptom onset. Study criteria further required availability of serum creatinine values on admission.
Participants were excluded if they lack of computed tomographic angiography (CTA) image or the quality of CTA image was poor due to motion artifact. This research was approved by the Ethics Committee of Nanjing Drum Tower Hospital, and the individual informed consent was waived. Program Early CT Score (ASPECTS)], was collected from our electronic database (see Table 1 for details). OTR refers to the time to recanalization during the MT, or the end of the procedure for those without reperfusion. 14 Preprocedural estimated glomerular filtration rate (pre-eGFR) was calculated by using creatinine-based Chronic Kidney Disease Epidemiology (CKD-EPI) equation 15 ; successful recanalization was defined as TICI scores 2b and 3.

| Imaging analysis
All participants had a multimodal stroke imaging, including noncontrast head CT and CT angiography (CTA) performed in the emergency room, and all CT sequences were acquired on a 64-slice CT device (Discovery CT750 HD, GE Healthcare, Milwaukee, WI, USA).
3D orthogonal maximum intensity projection (MIP) images were created in three planes. Collaterals were evaluated on MIP images of CTA by two experienced neuroradiologists using the Tan score: 0 = no collaterals visible; 1 = diminished collaterals in >50% of the occluded arterial territory; 2 = absent collaterals in <50% of the occluded territory; and 3 = collateral equal to the contralateral normal hemisphere. 16 For analysis purpose, the scores were dichotomized into either good collaterals (grades 2 and 3) or poor collaterals (grades 0 and 1).

| Outcome measures
Our main outcome was functional status, assessed at 3 months by trained staff via phone interview using a modified Rankin Scale (mRS) questionnaire. Functional independence (ie, good functional outcome) was defined as mRS 0-2, while poor functional outcome was defined as an mRS score 3 or higher.

| Statistical analyses
Patients were dichotomized based on their clinical outcome using mRS. Data were expressed as mean (SD), median (interquartile range), or number (percentage), as appropriate. Univariable comparison of good and poor outcome patients in baseline demographic, clinical, and procedural variables was performed using Student' s t test or Mann-Whitney U test for metric variables and chi-square or Fisher's exact test for qualitative variables, as appropriate.
Kolmogorov-Smirnov tests were applied to test the normality of variables.
There are three main statistical steps. First, to test the association between pre-eGFR and functional outcome, the multivariable regression model was established based on p < 0.1 in the univariate analysis plus the variables that were generally considered to be associated with functional outcome in prior literature: age, diabetes, hypertension, collateral status, NIHSS, gender, TICI grade, ASPECT, and OTR. Covariates were assessed for interaction effects.
Multicollinearity was checked by variance inflation factor (VIF) tests.
All VIF values were less than 1.746, which implies absence of severe multicollinearity. 17 When pre-eGFR was a significant predictor, the optimal sensitivity-specificity cutoff point of pre-eGFR associated with good functional outcome was calculated using receiver operating characteristic curve (ROC) analysis.
Subsequently, we explored the mediating role of collateral status on the relationship between pre-eGFR and good functional outcome at 3 months. Given that both the dependent variable and mediator were categorical, we adopted the statistical strategy for assessing mediating effects of categorical variables via Sobel test proposed by Iacobucci, 18 which comprises three separated logistic regression equations. In the first regression equation, the mediator (ie, collateral status) was regressed on the independent variable (ie, pre-eGFR) by logistic regression analysis. In the next regression equation, the dependent variable (ie, good functional outcome) was regressed on the independent variable (ie, pre-eGFR) by logistic regression analysis. In the last regression equation, the dependent variable (ie, good functional outcome) was regressed on both the independent variable (ie, pre-eGFR) and the mediator (ie, collateral status). All the three equations were adjusted for the following potential confounders: age, baseline NIHSS score, ASPECT, diabetes, hypertension, gender, OTR, and TICI grade. We also include OTR×pre-eGFR interaction as a covariate term. The test is significant at α = 0.05, if TA B L E 1 Comparisons of baseline characteristics in study population with poor and good outcome

| Demographic and clinical characteristics
Of

| Prediction of good outcome at 3 months
The NIHSS, diabetes, collateral status, and pre-eGFR were significantly related to 90-day outcome and remained significant in multivariable logistic regression (

| Collateral circulation on the association between pre-eGFR and functional outcome
Mediation analyses were performed to estimate the direct and indirect association between pre-eGFR and good outcome at 3-month follow-up, with the collateral status as a proposed mediator. The collateral status significantly mediated the association of pre-eGFR with good outcome, and the percentage of mediated effect was 30.92% ( Figure 1; Table 3).

| Interactions between pre-eGFR and OTR in predicating functional outcomes
After adjusting for potential confounders, there was a significant interaction between pre-eGFR and OTR (p interaction , 0.022; Table 4).
Since there was evidence of OTR to pre-eGFR interaction, we further quantified the interaction effect using the Johnson-Neyman analysis. Figure  The moderating role of pre-eGFR in the association between OTR and functional outcome is shown in Figure 2B. The range of pre-eGFR is 24.0-128.8 mL/min/1.73 m² in the current sample. According to the range of simple slopes obtained from the Johnson-Neyman analysis, the relationship between OTR and functional outcome was insignificant for pre-eGFR 88.9888 mL/min/1.73 m² or higher. However, once the pre-eGFR was lower than this threshold, shorter OTR was significantly associated with higher possibility of good outcome, and the association became stronger as the pre-eGFR decreased.

| DISCUSS ION
Concordant with our hypothesis, our findings confirmed that pre-  NIHSS indicates National Institutes of Health Stroke Scale; pre-eGFR, preoperative estimated glomerular filtration rate.

TA B L E 2
Prediction of good functional outcome (mRS 0-2) F I G U R E 1 Mediation analysis. Mediation model displaying standardized path coefficients between pre-eGFR, collateral status, and good outcome at 3 months, controlling for potential covariates (namely, age, baseline NIHSS score, diabetes, hypertension, gender, TICI grade, ASPECT, OTR, and OTR×pre-eGFR). Note: The path from pre-eGFR to good outcome at 3 months includes the coefficient with adjustment for collateral status (c′) and without adjustment for collateral status (c). All individual pathways were satisfied leptomeningeal collaterals in cases of ischemia. Patients with worse renal function are also found to have reduced production of nitric oxide 30 and a decreased number and function of endothelial progenitor cells, 31 which contribute to impaired collateral angiogenesis.
We also found that the effect of pre-eGFR on functional outcome weakened with shorter OTR and disappeared in patients with OTR less than 455.8 min. In line with our study, Laible et al. conducted a cohort study (median onset-to-puncture time, 227 min) and reported that eGFR was not related to functional outcome. 4 This time-dependent phenomenon may be partly explained by the hypothesis that collateral flow requires a relatively longer time to develop and thus mediates the association between eGFR and functional outcome. 32 Hence, it is conceivable to hypothesize that renal function may exert effects on outcome in a time-dependent manner. Further studies specifically designed to address this issue are needed.
An additional novel finding of the current study is that the like- Our findings also provide a rationale for selecting eligible patients for MT in an extended time window. Our results suggested

| CON CLUS IONS
To conclude, our study indicated that decreased pre-eGFR is related to poor 90-day functional outcomes in patients with anterior circulation LVO AIS with MT and can serve as a prognostic marker for selecting eligible patients for MT. Although all patients need to achieve recanalization as soon as possible, the time window may close earlier for patients with reduced kidney function than for patients with good renal function. These findings furthered the understanding of the relationship between renal function and the functional outcome of ischemic stroke patients receiving MT and may contribute to planning treatments in patients with decreased renal function.

ACK N OWLED G EM ENTS
We thank the participants of this study for their participation. This research was supported by the National Natural Science Foundation and Jiangsu Province Key Medical Discipline (ZDXKA2016020).

CO N FLI C T S O F I NTE R E S T
The authors declare that they have no conflicts of interest.

AUTH O R S' CO NTR I B UTI O N S
YX contributed to conceptualization; ZY and HX collected the data; ZY and YC made statistical analysis; ZY wrote original draft; YX reviewed and edited the manuscript; YX acquired funding; and YX supervised the study.

DATA AVA I L A B I L I T Y S TAT E M E N T
Anonymous data that support the findings of this study are available on reasonable request from the corresponding author.