Association of neonatal blood levels of brain-derived neurotrophic factor with development of autism spectrum disorder:a systematic review and meta-analysis

Ya-Kun Liu ·Hua Gao ·Shao-Bin Jin ·Wen-Jun Tu ·Ya-Jun Chen

Abstract Background Our goal was to evaluate the association between neonatal blood brain-derived neurotrophic factor (BDNF)level and autism spectrum disorder (ASD) diagnosis later in life.Methods MEDLINE and Web of Science databases were searched from inception until September 16,2020.Reference lists of all relevant articles also were reviewed.Mean blood BDNF concentrations,standard deviations,sample sizes,and other data needed for calculation of effect sizes were extracted by two independent investigators.The quality of the included studies was appraised using the Newcastle-Ottawa Scale for case-control studies.Data were pooled using the random-effects model.Results Five case-control studies involving 1341 cases and 3395 controls were included in the meta-analysis.The metaanalysis of all included studies showed no significant difference in blood BDNF levels between neonates diagnosed with ASD later in life and healthy controls [standardized mean difference (SMD)=0.261;95% confidence interval (CI)-0.052 to 0.573; P =0.102],with high level of heterogeneity (Q =64.346; I 2 =93.784; P ＜ 0.001).A subgroup analysis by assay type showed decreased blood BDNF levels in ASDs compared to controls (SMD=-0.070;95% CI-0.114 to-0.026;P =0.002),with high level of homogeneity (Q =0.894; I 2 =0.000; P =0.827).No evidence of publication bias was observed.Conclusions Neonates diagnosed with ASD later in life have decreased blood levels of BDNF measured by double-antibody immunoassay.More studies are warranted to facilitate a more robust conclusion.

Keywords Autism spectrum disorder·Autistic disorder·Brain-derived neurotrophic factor·Neonates·Neurotrophins

Introduction

Autism spectrum disorder (ASD) refers to a group of complex neuro-developmental disorders characterized by impairments in socialization,communication,and repetitive behaviors [1].Epidemiological data show an upward trend of ASD prevalence in the past two decades,estimated at 0.62-0.70%worldwide currently [2].ASD affects males about 4-5 times more than females [2].The impairments can range from very mild to extraordinarily severe;however,most patients cannot live independently and require some degree of the support during their entire life [1].One study shows that patients with ASD have a mortality risk that is 2.8 times higher than that of unaffected patients [3].

Evidence suggests that genetic factors play an essential role in the etiology of ASD [4].However,the hypothesis that ASD results from the constant interplay between genetic and environmental factors is commonly accepted [5,6].The heterogeneity of the broad spectrum of disorders renders their pathogenesis difficult to be elucidated.Early diagnosis allows early treatment and better outcomes [1].However,the treatment may still last a long period and can place a heavy economic burden on the family as well as the health care system [7].The need for advancements in early diagnosis facilitated the exploration of possible biomarkers,such as neurotrophic factors in the placenta and in infants’ blood samples,which also help to infer the pathogenesis of ASD[8-12].

Brain-derived neurotrophic factor (BDNF) is a prominent member of the neurotrophin family and is widely expressed throughout the central nervous system and peripheral blood.BDNF plays an essential role in neurocyte differentiation,neurons’ survival,synapse formation,neurotransmitter release,and cognitive functions [13-15].Several studies have indicated that children with ASD had higher blood BDNF levels [16-19],but other studies have showed that children with ASD had lower or normal blood BDNF levels [9,11].Meta-analyses of these studies indicated that children with ASD had increased peripheral blood levels of BDNF [16,20-22],while neonates diagnosed with ASD later in life had not [16].However,all of these studies contained with significant statistical heterogeneity,which yield less robust conclusions.The latest study in 2019 included a large sample size (801 ASD cases and 2421 matched controls) and drew a robust conclusion that decreased neonatal blood levels of BDNF are correlated with the diagnosis of ASD later in life [11].Given the inconsistent conclusions from these studies and their relatively limited sample sizes,a meta-analysis on the association between neonatal blood BDNF level and diagnosis of ASD later in life is warranted.To our knowledge,this is the first meta-analysis exclusively analyzing the relationship between neonatal blood BDNF level and the development of ASD.

Methods

Search strategy and study selection

We attempted to follow the proposed Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA)guidelines to report this meta-analysis.Two independent investigators searched MEDLINE,Web of Science for peer reviewed articles published in English through September 16,2020.The search terms were“brain-derived neurotrophic factor”plus free text synonyms;“autism spectrum disorder”or“autistic disorder”plus free text synonyms;and“infant,newborn”or“infant”plus free text synonyms.No restrictions were imposed.We also checked the reference lists of all potentially relevant studies.Articles reported only as abstracts were included only if sufficient information was available.The detailed search strategy is appended in Supplementary Table 1.

Studies were included if they met the following criteria:(1) subjects with a diagnosis of ASD;(2) provided sufficient data:number of residents in cases and controls,mean and standard deviation (SD) of blood BDNF levels,or data for calculations of means or standardized mean differences(SMD);(3) blood drawn age ＜ 31 days.

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional ethics committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.We reported the data without providing any individual details.

Data extraction

One investigator extracted data from all included studies.Mean blood BDNF concentrations,SDs,Pvalues,sample sizes,and other data types needed for calculation of effect sizes were extracted.Sample source,analytical technology,mean patient age at blood drawn,and sex distribution also was extracted.A second investigator checked the data extraction.A third investigator was consulted if disagreements could not be resolved by discussion.

Quality assessment

The quality of the included studies was appraised using the Newcastle-Ottawa Scale (NOS) for case-control studies.NOS has been widely used,and its content validity and interrater reliability have already been examined.The assessment for case-control studies consists of three main perspectives:(1) the selection of the study and control group;(2) the comparability of the study and control group;(3) the ascertainment of exposure.A score of 7-9 stars represents a high-quality study,while scores of 4-6 and 0-3 represent moderate and low-quality studies,respectively.

Statistical analysis

Statistical analyses were conducted using the Comprehensive Meta-Analysis software,version 3.3.070 (Biostat Inc).The eligible data were converted into SMD and 95% confi-dence intervals (CI).If the data were presented with median and interquartile range (IQR),the formula SD=IQR/1.35 was used to calculate SD.The formula SMD=1.732LnOR/πwas used to convert odds ratio (OR) to SMD when the data were presented with OR and 95% CI.Effect sizes of ASD and control groups were calculated as SMDs.The pooled effect sizes were computed using the random-effects model in the presence of heterogeneity.Heterogeneity across studies was tested byQstatistic (significance level,P＜ 0.10),andI2statistic.An Egger linear regression test (significance level,P＜ 0.05) and a Begg rank correlation test (signifi-cance level,P＜ 0.05) were used to assess potential publication bias.

Results

Literature search

We initially retrieved 39 articles from MEDLINE,Web of Science,and reference review.Of these,17 items were excluded for duplication.After title and abstract review,13 studies were excluded because they were irrelevant studies,reviews,or meta-analysis.After full-text review,a total of five articles,including 1341 cases and 3395 controls,fulfilled all the inclusion criteria and were included in this meta-analysis.The detailed flowchart is presented in Fig.1.

Study characteristics

In all of the included studies,blood samples were dried blood spot drawn in the first month of life and then stored frozen.All studies reported sample sizes,blood sample source,and assay type.Most studies reported gender distribution and blood drawn time after birth.Three studies[9,12,23]reported mean BDNF concentrations,SDs,and the other two reported medians,IQRs [24],OR and 95%CIs [11].The characteristics of the included studies are presented in Table 1.

Fig.1 Flowchart of study selection to select studies eligible for metaanalysis

All the cases were from either USA or Denmark.Three studies were exempted from the possibility of samples run in replicate based on patients’ birth year [9,12,23].Of the other two studies [11,24],blood samples were acquired from the Danish Neonatal Screening Biobank.The authors of the later study [11]were not sure about whether there was some overlap of the cases between the two studies.Given that the Danish Neonatal Screening Biobank contains more than 10 million biosamples and that the patients in these two studies were chosen from different programs(one from Historic Birth Cohort utilizing the Danish Psychiatric Central Register;the other from iPSYCH Danish case-cohort),the possibility of samples run in replicate is considered very low.

Quality assessment

All of the included studies were of high quality (NOS scoring rank 7-9).The detailed assessment of the included studies is appended in Supplementary Table 2.

Meta-analysis of blood BDNF concentrations in ASD patients compared to controls

A random-effects model was used for the meta-analysis.Among the included studies,one study showed a signifi-cantly decreased blood BDNF concentration in children diagnosed with ASD later in life.Another study suggested increased BDNF levels in these children.Overall,no significantly difference was observed in this meta-analysis(SMD=0.261;95% CI-0.052 to 0.573;P=0.102).High level of heterogeneity was observed (Q=64.346;I2=93.784;P＜ 0.001).Figure 2 presents the results from the random-effects model and the forest plot.

Subgroup analysis

We conducted a subgroup analysis to investigate the potential sources contributing to the high level of heterogeneity.We noticed that in four studies [9,11,12,24],the BDNF concentrations were analyzed by the immunoassay entailing two kinds of antibodies (capture antibody plus detection antibody),while the other one study [23]used a single antibody system (recycling immunoaffinity chromatography,RIC).We conducted the subgroup analysis based on different assay types,double antibody assay,and single antibody assay.The result is presented in Fig.3.When analyzed by double-antibody immunoassay,the ASD group had a significantly decreased blood BDNF concentrations compared with control group (SMD=-0.070;95% CI-0.114 to-0.026;P=0.002),with high level of homogeneity (Q=0.894;I2=0.000;P=0.827).

Table 1 Characteristics of studies included in this meta-analysis

Fig.2 Forest plot for meta-analysis of blood BDNF concentrations,ASD patients compared to controls.BDNF brain-derived neurotrophic factor,ASD autism spectrum disorder

Publication bias

Both Begg’s test (with continuity correction,P=0.086)and Egger’s test (P=0.254) did not show significant publication bias.

Discussion

Fig.3 Forest plot for subgroup analysis.DAA double antibody assay,SAA single antibody assay

Our study’s main finding is that newborns diagnosed with ASD later in life have decreased blood BDNF concentrations when analyzed by double-antibody immunoassay(U-PLEX or Luminex).However,the pooled analysis of all the included studies showed no significant difference in BDNF concentrations between neonates diagnosed with ASD later in life and healthy controls,with an extremely high level of heterogeneity (I2=93.784).We noticed that the immunoassay type might contribute to the significant heterogeneity.The subgroup analysis based on assay type showed very good homogeneity in both subgroups,suggesting that the studies’ effect sizes in subgroups could be synthesized for meta-analysis.All of the included studies were ranked high quality and had a low risk of bias,which further increased our conclusion’s robustness.

Immunoassay type for analysis of neurotrophins may contribute to the bias influencing research results [12,23].Nelson et al.used a single-antibody immunoassay (RIC)to analyze newborn blood specimens and found increased BDNF level in newborns with a later diagnosis of autism spectrum disorder.Subsequently,they failed to achieve a consistent conclusion using Luminex and enzyme-linked immunosorbent assays (ELISA) methods.Double-antibody immunoassays,such as Luminex and ELISA,are proved sensitive,effective,and reliable for analyzing cytokines and neurotrophins [25-27].RIC is rarely seen in literature in the past decade.Therefore,subgroup analysis by assay type is reasonable from this perspective.However,the robustness of our conclusion is weak.Samples from the research by Skogstrand et al.comprised 59.8% of the ASD population and 71.3% of the controls.This study may play a decisive role in the overall result.Therefore,the random-effects model was used to reduce the impact of sample sizes on the pooled results.The relative weights were 24.4% and 84.68% in the overall meta-analysis and the subgroup analysis,respectively.As a result,our conclusions were not sufficiently robust.

BDNF plays a critical role in neurocyte differentiation,neurons’ survival,synaptic plasticity and repair,and neurotransmitter release [13,14,28].It is also essential for cognitive functions,learning,and memory [29]and may be responsible for several neuropsychiatric disorders,such as schizophrenia and Alzheimer’s disease [30-32].In patients with ASD,most research on blood BDNF levels includes cross-sectional studies.Therefore,the role of BDNF in the development of ASD is unclear.

Studies suggest that blood BDNF concentrations decline across the lifespan.The blood BDNF concentration is high in infants and then falls gradually in adolescence,puberty,early adulthood,and further [33,34].This pattern may be different in patients with ASD.Former studies indicated higher blood BDNF level in ASD children (non-neonates)than that in healthy controls [16,22].However,the heterogeneity was statistically significant in all of these studies.Our conclusion is that the blood BDNF levels in neonates diagnosed with ASD later in life is lower than matched neonates measured by U-PLEX or Luminex immunoassay.This may suggest that insufficiency of blood BDNF in early life lead to structural or functional deficiencies,which eventually facilitate the development of ASD.This needs to be clarified by further studies about the effects of BDNF on the development of ASD.However,another possibility should not be neglected:low blood BDNF concentration may be an element of ASD’s pathophysiology.This means ASD leads to low blood BDNF level.This possibility calls for further research as well.

ASD constitutes a broad spectrum of neuro-develop-mental disorders.The genetic and environmental risk factors underlying the etiology of ASD and its phenotypes are heterogeneous [35].In addition,the pathophysiology of different phenotypes is diverse.BDNF levels probably differ among different phenotypes.The inconsistency of patients’blood BDNF levels among the include studies could result from different phenotype composite structure.Phenotype associated BDNF levels were not assessed in these studies,rendering subgroup analysis by phenotype impossible.Moreover,more than 70% of individuals with ASD have comorbidities [36,37].The high frequency of co-occurring conditions also may influence the measured BDNF levels in ASD patients.

ASD is a complex neurodevelopmental brain disorder.In most studies,the BDNF levels were measured from peripheral blood samples.Only on the condition that the BDNF level in the periphery can reflect that in the brain,the altered serum BDNF levels can be useful as a possible biologic marker for ASD.Animal studies have suggested that serum BDNF level correlates with that in central nervous system [38].Whereas,whether the altered BDNF levels in the periphery are able to reflect the BDNF levels in human central nervous system requires further investigation.

The primary strength of this study is a relatively large sample size of 1341 cases and 3395 controls.Also,a very high homogeneity level in the subgroups and the high quality of included studies further increased the strength.

This meta-analysis has several inherent limitations.First,the number of included articles is small.The subjects are neonates with a diagnosis of ASD later in life and were drawn blood samples in the first month after birth.Consequently,only five studies met the inclusion criteria.Furthermore,the main findings were drawn from a subgroup analysis.Second,in all included studies,blood samples were dried blood spots stored frozen for years.Skogstrand et al.demonstrated the degradation of BDNF over the years and decreased BDNF concentrations after a freeze-thaw cycle [11].As a result,the intrinsic features of the dried blood spot may contribute to the detection bias.Third,it is difficult to identify whether a causal relationship exists between decreased BDNF levels and the development of ASD through case-control studies.

In conclusion,our meta-analysis indicated decreased blood concentrations of BDNF measured by double-antibody immunoassay in neonates later diagnosed with ASD.More studies on the relationship of neonatal blood BDNF concentrations with the development of ASD are warranted to facilitate a more authentic conclusion and to evaluate the effect of BDNF in the development of ASD.

Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s1251 9-021-00415-2.

AcknowledgementsThis study was supported by grants from CAMS Innovation Fund for Medical Science,Natural Science Foundation of Tianjin,China Postdoctoral Science Foundation,Science Foundation for Post Doctorate Research of the Beijing,and Youth Backbone Project from Chinese Academy of Medical Science Institute of Radiation Medicine.We also thank Prof.Jian-Lei Cao from Zhongnan Hospital of Wuhan University for his help in technical support.

Author contributionsLYK and GH contributed equally to this work.LYK and JSB contributed to drafting of the manuscript,acquisition,analysis,or interpretation of data.GH contributed to statistical analysis,acquisition,analysis,or interpretation of data.TWJ contributed to study concept and design,critical revision of the manuscript,acquisition of funding,acquisition,analysis,or interpretation of data,and supervision.CYJ contributed to study concept and design,acquisition,analysis,or interpretation of data,and supervision.CYJ had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.All authors approved the final version of the manuscript.

FundingThis study was supported by grants from CAMS Innovation Fund for Medical Science (Nos.2017-I2M-1-016 and 2019-I2M-2-006),Natural Science Foundation of Tianjin (No.19JCYBJC26600),China Postdoctoral Science Foundation (Nos.2019M660921 and 2020T130436),Science Foundation for Post Doctorate Research of the Beijing (Nos.2017-ZZ-123 and 2020-ZZ-005)and Youth Backbone Project from Chinese Academy of Medical Science Institute of Radiation Medicine (No.2019043).

Compliance with ethical standards

Ethical approvalAll procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional ethics committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.We reported the data without providing any individual details.

Conflict of interestThe authors have no conflict of interest to declare.No financial or non-financial benefits have been received or will be received from any party related directly or indirectly to the subject of this article.