Low Circulating Acute Brain-Derived Neurotrophic Factor Levels Are Associated With Poor Long-Term Functional Outcome After Ischemic Stroke
Background and Purpose—Brain-derived neurotrophic factor (BDNF) plays important roles in brain plasticity and repair, and it influences stroke outcomes in animal models. Circulating BDNF concentrations are lowered in patients with traumatic brain injury, and low BDNF predicts poor recovery after this injury. We sought to investigate whether circulating concentrations of BDNF are altered in the acute phase of ischemic stroke and whether they are associated with short- or long-term functional outcome.
Methods—Serum concentrations of BDNF were measured in the Sahlgrenska Academy Study on Ischemic Stroke. The main outcomes were modified Rankin Scale (mRS) good (mRS score of 0–2) versus poor (mRS score of 3–6) at 3 months and 2 years after stroke, and good (mRS score of 0–2) versus poor (mRS score of 3–5) at 7 years after stroke.
Results—Acute concentrations of BDNF were significantly lower in ischemic stroke cases (n=491) compared with controls (n=513). BDNF concentrations were not significantly associated with 3-month outcome. However, patients with BDNF in the lowest tertile had an increased risk of experiencing a poor outcome both at 2-year and 7-year follow-up, and these associations were independent of vascular risk factors and stroke severity (odds ratio, 2.6; confidence intervals, 1.4–4.9; P=0.002 and odds ratio, 2.1; confidence intervals, 1.1–3.9; P=0.028, respectively).
Conclusions—Circulating concentrations of BDNF protein are lowered in the acute phase of ischemic stroke, and low levels are associated with poor long-term functional outcome. Further studies are necessary to confirm these associations and to determine the predictive value of BDNF in stroke outcomes.
Brain-derived neurotrophic factor (BDNF) has a role in neurogenesis and influences functional motor recovery after an ischemic brain lesion in animal models.1 Recently, the prognostic value of circulating BDNF levels has received attention in some brain disorders, including traumatic brain injury and poststroke depression. Acute serum concentrations of BDNF predicted severity and outcome of a traumatic brain injury, and patients with the lowest BDNF concentrations had the highest odds of incomplete recovery.2 Similarly, patients with stroke who developed poststroke depression had low admission levels of serum BDNF.3 No study has yet looked at acute BDNF protein levels in relation to functional outcome after ischemic stroke (IS).
Levels of circulating BDNF correlate with several vascular risk factors,4,5 and low BDNF concentrations have recently been found to associate with an increased risk of incident stroke/transient ischemic attack.6 We hypothesized that circulating BDNF concentrations are lowered in the acute phase of IS and that low BDNF concentrations associate with poor short- (3 months) or long-term (2 and 7 years) functional outcome after stroke.
Additional details can be found in the online-only Data Supplement.
Study Population, BDNF, and Outcome Measurements
Participants were from the Sahlgrenska Academy Study on Ischemic Stroke (SAHLSIS), which includes patients and controls aged 18 to 69 years, the design of which has been reported.7 Serum samples were collected from 514 patients in the acute phase and from 514 matched controls. BDNF levels were measured using the BDNF Emax ImmunoAssay System (Promega, Madison, WI). Functional outcome was assessed by the modified Rankin Scale (mRS). During the first month after IS, death is often due to stroke-related complications and can be hypothesized to associate with recovery. Therefore, outcome was defined as good (mRS score of 0–2) versus poor (mRS 3–6) at the 3-month time point and to take a conservative approach also at the 2-year time point. With regard to recovery in the long-term perspective, there are many confounding causes of death, such as malignancies, diabetes mellitus and cardiac causes. Thus, for the 7-year follow-up, good versus poor outcome was defined as mRS score of 0 to 2 versus 3 to 5, respectively.
Associations between BDNF and IS were investigated using unconditional logistic regression analysis. Model A was adjusted for age and sex. Model B was additionally adjusted for hypertension, hyperlipidemia, diabetes mellitus, smoking status, and atrial fibrillation. Associations between BDNF and functional outcome were evaluated by binary logistic regression. Receiver operating characteristic analysis was used to determine the optimal BDNF cutoff that maximized the sum of sensitivity and specificity. Model A was adjusted for age and sex. Model C included the same risk factors as in model B and initial stroke severity. Comparisons of areas under the receiver operating characteristic curve (c-statistic) from models including clinical variables (model C) with or without BDNF were performed. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indices were also used to assess the added value of BDNF over clinical parameters for poor functional outcome. Statistical analyses were performed using SPSS for Windows version 20 (IBM Corporation, Armonk, NY) and R version 3.2.2. The statistical significance level was 0.05 and P values were 2 tailed.
Baseline characteristics for SAHLSIS have been described previously.7 Acute serum concentrations of BDNF were significantly lower in patients with IS compared with controls (geometric mean 18.3 [95% confidence intervals, 17.0–19.7] ng/mL versus 23.9 [95% confidence intervals, 23.1–24.8] ng/mL; P<0.001). In multivariable analyses, lower BDNF was independently associated with IS and each of the 4 main pathogenic subtypes (Figure 1).
In the analysis of functional outcome, patients in the lowest tertile of BDNF (T1) were compared with patients in the 2 upper tertiles (T2+T3). The analysis of cutoffs for BDNF and the distribution of clinical characteristics based on BDNF tertiles are presented in the online-only Data Supplement (Results section and Table I in the online-only Data Supplement). Participants in T1 were significantly younger, had higher triglyceride levels, higher prevalence of atrial fibrillation, lower low-density lipoprotein cholesterol, lower blood pressure, and lower prevalence of hypertension.
The majority of patients who experienced poor outcome was in T1 (Figure 2A). In regression analyses, BDNF concentrations were not associated to functional outcome in the short-term (Figure 2B). However, low BDNF concentrations were associated with poor functional outcome both at 2- and 7-year follow-up, and these associations were independent of stroke severity and traditional risk factors (Figure 2B). When patients who died were included in the 7-year analysis the association was attenuated (Model A: odds ratio, 1.51 [1.00–2.32]; P=0.051 and Model C: odds ratio, 1.58 [0.96–2.60]; P=0.069). The additional predictive value of BDNF was next measured over clinical parameters by c-statistics, IDI, and NRI. Addition of baseline BDNF yielded marginal improvement in the clinical risk prediction models in all 3 measures of discrimination at both 2- and 7-year follow-up (2-year: c-statistic, 0.838 [0.795 to 0.883] to 0.848 [0.806 to 0.892]; P=0.311, Figure SII in the online-only Data Supplement; IDI, 1.3% [0 to 2.5]; P=0.035 and NRI, 18.9% [−3.0 to 40.8]; P=0.091 and 7-year: c-statistic, 0.811 [0.754 to 0.867] to 0.820 [0.765 to 0.875]; P=0.347, Figure SII in the online-only Data Supplement; IDI, 1.4% [−0.2 to 3.0]; P=0.099 and NRI, 21.3% [−5.0 to 47.5]; P=0.112).
Here, we report that circulating concentrations of BDNF are lower in the acute phase of IS compared with healthy controls. Low BDNF concentrations have been observed in patients with metabolic syndrome,8 atrial fibrillation,6 and acute coronary syndromes.4 Furthermore, low concentrations of BDNF were recently demonstrated to be associated with an increased risk of incident stroke/transient ischemic attack when adjusting for age, sex, and traditional risk factors.6 Although these studies support our findings, the mechanism of decreased serum BDNF in IS requires further study.
We also report that low acute concentrations of BDNF are associated with an increased risk of suffering poor functional outcome 2 and 7 years after IS. These findings are in line with a study on traumatic brain injury that assessed recovery after 6 months.2 However, in this study there was no significant association between BDNF and 3-month outcome. One potential explanation may be that for several impairments, such as aphasia, recovery continues much beyond 3 months.9 In the 7-year analysis, when patients who died were included, the association was attenuated and no longer significant. However, death in the long-term prospective is because of many different causes, and it is reasonable to assume that many of these do not associate with recovery.
Although the logistic regression models showed that BDNF was an independent predictor of poor functional outcome 2 and 7 years post stroke, the additional predictive value of BNDF was modest over clinical data in terms of c-statistic, IDI, and NRI. This suggests that BDNF on its own may have limited clinical use as a prognostic biomarker. Provided replication in other studies, it would be of interest in the future to evaluate BDNF together with other biomarkers related to recovery in a multimarker panel.
In conclusion, in this study low acute concentrations of BDNF are associated with poor long-term functional outcome after IS, and further studies are necessary to seek replication.
We thank Ingrid Eriksson for excellent work and assistance with study patients.
Sources of Funding
This study was supported by the Swedish Medical Society (Svenska Lakaresallskapet), grants from the Swedish Government (ALFGBG-11206, ALFGBG-147771, ALFGBG-148861, and ALFGBG-144341), the Swedish Research Council (2013–3595), the Swedish Heart Lung Foundation (20130315), the Swedish Stroke Association, the Goteborg Foundation for Neurological Research, and the Goteborg Medical Society (036/14).
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.012383/-/DC1.
- Received December 15, 2015.
- Revision received April 29, 2016.
- Accepted May 3, 2016.
- © 2016 American Heart Association, Inc.
- Schäbitz WR,
- Steigleder T,
- Cooper-Kuhn CM,
- Schwab S,
- Sommer C,
- Schneider A,
- et al
- Golden E,
- Emiliano A,
- Maudsley S,
- Windham BG,
- Carlson OD,
- Egan JM,
- et al
- Pikula A,
- Beiser AS,
- Chen TC,
- Preis SR,
- Vorgias D,
- DeCarli C,
- et al
- Jood K,
- Ladenvall C,
- Rosengren A,
- Blomstrand C,
- Jern C
- Chaldakov GN,
- Fiore M,
- Stankulov IS,
- Manni L,
- Hristova MG,
- Antonelli A,
- et al