Predictive value of glycoprotein DKK3 for early neurological deterioration after ischemic stroke

Abstract

Objective: To explore the value of serum Dickkopf-3 (sDKK3) in predicting Early Neurological Deterioration (END) and in-hospital adverse outcomes in acute ischemic stroke (AIS) patients.

Methods: AIS patients (n = 200) were included and assessed by the National Institutes of Health Stroke Rating Scale. Serum Dkk3 levels were assessed by ELISA. END was defined as an increase of ≥ 4 points in NIHSS score within 72h. The biological threshold of sDKK3 level and END occurrence were predicted based on X-tile software. Primary outcomes were END and all-cause death, and the secondary outcome was ICU admission during hospitalization. The logistic regression model and Cox risk regression model were applied to evaluate the relationship between DKK3 level and END incidence, all-cause in-hospital mortality, and in-hospital adverse outcomes (ICU admission).

Results: During hospitalization, the incidence of END in patients with AIS was 13.0 %, and the mortality rate within 7 days after END was 11.54 % (3/26). In patients below the serum DKK3 cutoff (93.0 pg/mL), the incidence of END was 43.5 % (20/48). Patients with lower sDKK3 levels were associated with a 1.188-fold increased risk of developing END (OR = 1.188, 95 % CI 1.055‒1.369, p < 0.0001). However, there was no significant association with admission to the ICU. sDKK3 below the threshold (93.0 pg/mL) was a risk factor for death.

Conclusion: Predictive threshold levels of serum DKK3 based on X-tile software may be a potential predictive biomarker of in-hospital END in patients with AIS, and low levels of DKK3 are independently associated with increased in-hospital mortality.

Keywords: Acute ischemic stroke; Dickkopf-3; Early neurological deterioration; Predictive value; Risk of death.

Fig. 1

Serum DKK3 levels in (A) healthy people and AIS patients; (B) patients with END and patients without END; (C) Patients who died during their stay (n = 19) and those who did not (n = 181); (D) patients admitted to the ICU during their stay and those not admitted to the ICU.

Fig. 2

Unsupervised principal component analysis (A) and load analysis (B) of serum DKK3 levels and clinical features.

Fig. 3

Factors associated with all-cause death during hospitalization in univariate (A) and multivariate (B) Cox risk regression analyses.

Fig. 4

The construction of Receptor Operating Characteristic (ROC) curves was employed to assess the predictive ability of serum DKKs in identifying END events in patients with AIS).