Pathoclinical associations between panic disorders and the brain-derived neurotrophic factor Val66Met polymorphism: an updated meta-analysis

Abstract

Prior studies have indicated the pathological role of brain-derived neurotrophic factor (BDNF) gene polymorphism in panic disorders (PD). A functionally less active BDNF Val66Met mutant was previously detected in PD patients with different ethnic backgrounds. However, the results remain inconclusive or inconsistent. A meta-analysis was used to explore the consistency of the BDNF Val66Met mutant’s association with PD irrespective of the subject’s ethnicity. Relevant case-controlled full-length clinical and preclinical reports were retrieved by database searching, and 11 articles involving 2203 cases and 2554 controls were systematically selected per the standard inclusion criteria. Eleven articles were finally included that explored the relationship between the Val66Met polymorphism and PD risk susceptibility. Statistical analysis revealed a significant genetic association of the mutation, allele frequencies, and genotype distributions of BDNF with PD onset. Our findings demonstrated that the BDNF Val66Met is a susceptibility factor of PD.

Introduction

Panic disorder (PD) is characterized as one of the common types of anxiety disorders that affect at least 5% of the global population at any point of their lifetime (^{Roy-Byrne et al., 2006}). Besides, PD has known pathological connections with increased predispositions to cardiovascular diseases and associated deaths (^{Albert et al., 2005; Smoller et al., 2007}), which can extremize the suffering and economic burden of patients and their caregivers (^{Greenberg et al., 1999}). Therefore, revealing the etiopathological factors and strategizing targeted treatment options are urgently required for this disorder.

In preclinical studies, conditional deletions of mutant brain-derived neurotrophic factor (BDNF) variants in response to stress-inducing environmental or physiological stimuli, such as anxiety, have been linked to hyperactivity in the experimental animal model (^{Rios et al., 2001}). In the clinical examination, serum concentrations of BDNF are significantly higher in PD patients compared with age-matched healthy control subjects and are found to be associated with high levels of anxiety susceptibility and sensitivity, symptomatic severity, and poor clinical outcomes, as well (^{Karege et al., 2002; Shimizu et al., 2005; Ströhle et al., 2010; Kurita et al., 2012}), suggesting that the BDNF gene products may have etiological linkage to PD crisis. The reduced BDNF activity in PD has been correlated with a single-nucleotide polymorphism (SNP) at Val66Met of exon XIII, resulting in a G (Val) to A (Met) substitution in the coding sequence of the exon XIII, which may impact the structure-function relationship of the BDNF protein, contributing to the susceptibility of several neuropsychiatric disorders, including PD (^{Egan et al., 2003}a; ^{Chen et al., 2004; Zhao et al., 2015; Chen et al., 2017}).

Although the correlation of BDNF Val66Met with PD onset and aggressiveness has been explored in subjects from a wide range of ethnic backgrounds; however, these findings are inconsistent or inconclusive due to multiple confounding factors such as different ethnic backgrounds, small sample sizes, and inconsistent sampling strategies. To overcome these shortcomings, we conducted a meta-analysis testifying to the correlative strength of BDNF SNPs with PD susceptibility.

Methods

Search strategy

we used full-text searching of the terms “Panic Disorder(PD)” OR “Panic Attacks” AND “brain-derived neurotrophic factor (BDNF)” OR “Brain-derived neurotrophic factor” OR “BDNF” and “Val66Met” OR “G196A” OR “rs6265.” Databases/online libraries mainly used for searching articles were PubMed, Medline, Web of Science, APA, Cochrane, and Embase. The search deadline was 9 November 2022, and the review-type articles were further searched and included in the analysis.

Inclusion criteria

Study reports were included if they had (a) mentioned diagnostic criteria (e.g. International Classification of Diseases-10, Diagnostic and Statistical Manual of Mental Disorders-4th edition, Diagnostic and Statistical Manual of Mental Disorders-5th edition), (b) detailed genotyping and gene frequency data, (c) reported genotypes met the Hardy-Weinberg equilibrium principle, and (d) proper case-controls.

Data extraction

Basic demographic information including age, sex, and race; sample sizes; study results; study-specific inclusion and exclusion criteria, diagnostic criteria, and severity assessment criteria; numbers of genotyped samples, and allelic distributions in both case and control groups were extracted from the included reports (Tables 1,2 and 3).

Table 1 - Genotype and allele frequency in the included studies

Authors	Year	Race	Diagnostic criteria	Results	Case				Control				Case		Control
					Total	Val/Val	Val/Met	Met/Met	Total	Val/Val	Val/Met	Met/Met	Val	Met	Val	Met
Lam et al.	2004	China	DSM-IV	No association	103	30	53	20	180	34	107	39	113	93	175	185
Shimizu et al.	2005	Japan	DSM-IV	No association	109	33	56	20	178	59	91	28	122	96	209	147
Lim et al.	2007	Korean	DSM-IV	No association	106	28	46	32	160	47	82	31	102	110	176	144
Ishii et al.	2009	Japan	DSM-IV	Association	138	42	56	40	242	55	145	42	140	136	255	229
Otowa et al.	2009	Japan	DSM-IV	No association	638	223	294	121	589	200	283	106	740	536	683	495
Konishi et al.	2014	Japan	DSM-IV	No association	470	158	216	96	458	159	221	78	532	408	539	377
Han et al.	2015	Korean	DSM-IV	No association	136	42	63	31	263	81	135	47	147	125	297	229
Zou et al.	2019	China	DSM-IV	No association	223	47	118	58	218	60	104	54	212	234	224	212
Wang et al.	2020	China	DSM-V	No association	85	34	33	18	91	32	36	23	101	69	100	82
Yang et al.	2021	China	DSM-V	No association	79	24	41	14	76	25	36	15	89	69	86	66
Chu et al.	2022	China	DSM-V	No association	116	42	48	26	99	35	45	19	132	100	115	83

Table 2 - Basic information characteristics of the included studies

Characteristics	Lam et al. (2004)	Shimizu et al. (2005)	Lim et al. (2007)	Ishii et al. (2009)	Otowa et al. (2009)
Sample size (case/control)	283 (103/180)	287 (109/178)	266 (106/160)	380 (138/242)	1227 (638/589)
Sex (male/female)	45/58	39/70	63/43	Not mentioned	216/470
	68/112	75/103	52/108		277/312
Age (mean ± SD) (case/control)	38.5 ± 10.3 years, 37.4 ± 7.7 years	37.4 ± 13.3 years, 28.7 ± 10.2 years	41.87 ± 10.23 year, 31.61 ± 9.04 year	Not mentioned	38.7 ± 10.5 years, 35.6 ± 11.6 years
Design	Case–control study	Case–control study	Case–control study	Case–control study	Case–control study
Exclusion criteria or Inclusion criteria	Inclusion criteria: (a) patients met the DSM-IV criteria for panic disorder and (b) the patients who had hyperthyroidism were excluded.	Inclusion criteria: (a) patients met the DSM-IV criteria for panic disorder and (b) controls who did not have any medical or psychiatric diagnosis after our clinical interview was selected.	Exclusion criteria: (a) patients with comorbid current major depression, bipolar disorder, post-traumatic stress disorder, or any kind of psychotic disorder, including schizophrenia, by administrating the Mini-International Neuropsychiatric Interview.	Inclusion criteria: (a) patients met the DSM-IV criteria for panic disorder.	Inclusion criteria: (a) patients met the DSM-IV criteria for panic disorder; and (b) the controls received a short interview to exclude the history of major psychiatric illness.
Diagnosis	DSM-IV criteria	DSM-IV criteria	DSM-IV criteria	DSM-IV criteria	DSM-IV criteria
Measure to assess symptom severity	Not mentioned	Not mentioned	HAMA/BDI	Not mentioned	Not mentioned

DSM-IV, Diagnostic and Statistical Manual of Mental Disorders-4th edition.

Table 3 - Basic information characteristics of the included studies

Characteristics	Konishi et al. (2014)	Han et al., (2015)	Zou et al. (2019)	Wang et al. (2020)	Yang et al. (2021)	Chu et al. (2022)
Sample size (case/control)	928 (470/458)	399 (136/263)	287 (223/218)	380 (138/242)	155 (80/75)	215 (116/99)
Sex(male/female)	178/292	79/57	89/134	Not mentioned	33/47	49/67
	195/263	138/125	98/120		33/45	42/57
Age (mean ± SD) (case/control)	18–75	40.6 ± 9.6 years, 31.2 ± 8.4 years	Not mentioned	16–60 years	46.54 ± 10.61 years, 49.36 ± 10.66 years	47.62 ± 10.56 years, 49.95 ± 10.43 years
Design	Case–control study	Case–control study	Case–control study	Case–control study	Case–control study	Case–control study
Exclusion criteria or Inclusion criteria	Inclusion criteria: (a) drug-free, no previous diagnosis of a psychiatric disorder, and no family history of psychiatric disorder; and (b) the healthy control subjects were screened for the presence or absence of DSM-IV axis I disorder	Inclusion criteria: (a) patients met the DSM-IV criteria for panic disorder.	Inclusion criteria: (a) patients met the DSM-IV criteria for panic disorder; and (b) patients with neurological diseases, or past or current episodes of generalized anxiety disorder or other psychiatric disorders were excluded.	Inclusion criteria: (a) had a diagnosis of PD according to the DSM-V criteria; and (b) had the score of the Panic Disorder Severity Scale–Chinese Version (PDSS-CV) at least 10 and the HAMA-14 at least 14.	Inclusion criteria: (a) aged from 18 to 65 years; (b) meeting the diagnosis of PD defined by the DSM-V independently by two experienced psychiatrists; (c) were free of any antidepressants for at least 4 weeks before enrollment; and (d) a negative pregnancy test for women.	Inclusion criteria: (a) aged from 18 to 60 years; (b) diagnosis of PD was conducted according to the DSM-V (SCID) criteria through a psychiatric interview; (c) total score of Hamilton Anxiety Scale (HAMA-14) ≥14; and (d) total score of PDSS-CV ≥10.
Diagnosis	DSM-IV criteria	DSM-IV criteria	DSM-IV criteria	DSM-V criteria	DSM-V criteria	DSM-V criteria
Measure to assess symptom severity	Not mentioned	Not mentioned	PDSS score	PDSS-CV, HAMA-14	PDSS-CV, HAMA-14	PDSS-CV, HAMA-14

DSM-V, Diagnostic and Statistical Manual of Mental Disorders-5th edition.

Data analysis

Statistical software STATA version 15.0 (Stata Corp. 2015. Stata Statistical Software, Release 15, College Station, Texas, USA) was used for data analysis, following the procedure published elsewhere (^{Zhao et al., 2013}).

Results

Relating BDNF Val66Met and PD association, a total of 95 articles were initially retrieved from the selected databases. Among these, 39 articles were further screened to eliminate duplicated searches. Finally, there were only 11 articles that met the inclusion criteria with available data (^{Lam et al., 2004; Shimizu et al., 2005; Lim et al., 2007; Ishii et al., 2009; Otowa et al., 2009; Konishi et al., 2014; Han et al., 2015; Zou et al., 2019; Wang et al., 2020; Yang et al., 2021; Chu et al., 2022}), involving 2203 cases and 2554 controls. The flow chart of literature selection is shown in Fig. 1.

Fig. 1: Flow chart of literature selection process.

Concerning the 11 included articles, detailed clinical characteristics, patients’ demographics, and other relevant information can be identified from Tables 1 and 2. We assumed that the Val66Met allele could be the primary risk factor for PD onset, and then six genetic models of allelic distributions were explored to reveal the relationship between this particular missense mutation and PD susceptibility, as follows: Val versus Met, Val/Met versus Met/Met, Val/Val + Val/Met versus Met/Met, Val/Met versus Val/Val + Met/Met, Val/Val versus Val/Met, and Val/Val versus Met/Met. According to the results of heterogeneity, first, the fixed-effects model was adopted in this meta-analysis to measure the combined odds ratio (COR), and corresponding 95% confidence intervals (CIs) evaluating the genetic association in PD, otherwise, the random-effects model was tested. The study heterogeneity analysis revealed that the effect size in each study was NS (I² = 0.00%; P = 0.71). The COR derived from these studies was 0.94 (95% CI, 0.87−1.02; z = 1.45; P = 0.15) (Fig. 2), suggesting marginal significant association between the Val66Met allele and PD pathogenesis. Likewise, other models also yield significant genetic correlations with PD. In the Val/Met versus Met/Met model, OR was 0.81 (95% CI, 0.70–0.95; z = 2.67; P = 0.01; Fig. 3). For the Val/Val + Val/Met versus Met/Met model, OR was 0.84 (95% CI, 0.73–0.97; z = 2.42; P = 0.02; Fig. 4); and for the Val/Met versus Val/Val + Met/Met model, OR was 0.88 (95% CI, 0.78–0.98; z = 2.23; P = 0.03; Fig. 5). In contrast, other models yield no significant genetic correlations with PD. In the Val/Val versus Val/Met model, OR was 1.08 (95% CI, 0.95–1.24; z = 1.16; P = 0.25; Fig. 6); for the Val/Val versus Me/Met model, OR was 0.88 (95% CI, 0.75–1.04; z = 1.47; P = 0.14; Fig. 7). Results from the six genetic models demonstrated that the Met/Met genotype might be a susceptibility factor for PD onset.

Fig. 2: Results of the fixed analysis for the BDNF Val66Met allele (Val vs. Met) in the PD and control groups. BDNF, brain-derived neurotrophic factor; PD, panic disorders.

Fig. 3: Results of the fixed analysis for the BDNF Val66Met genotype (Val/Met vs. Met/Met) in PD and control groups. BDNF, brain-derived neurotrophic factor; PD, panic disorders.

Fig. 4: Results of the fixed analysis for the BDNF Val66Met genotype (ValVal + Val/Met vs. Met/Met) in PD and control groups. BDNF, brain-derived neurotrophic factor; PD, panic disorders.

Fig. 5: Results of the fixed analysis for the BDNF Val66Met genotype (Val/Met vs. ValVal + Met/Met) in PD and control groups. BDNF, brain-derived neurotrophic factor; PD, panic disorders.

Fig. 6: Results of the fixed analysis for the BDNF Val66Met genotype (Val/Val vs. Val/Met) in PD and control groups. BDNF, brain-derived neurotrophic factor; PD, panic disorders.

Fig. 7: Results of the fixed analysis for the BDNF Val66Met genotype (Val/Val vs. Met/Met) in PD and control groups. BDNF, brain-derived neurotrophic factor; PD, panic disorders.

Sensitivity analysis

Sensitivity analysis showed that the estimated value was 0.81, and the 95% CI was in the range of 0.70–0.95, which did not affect the analysis results.

Publication bias testing

The Begg-Mazumdar rank correlation test was used to verify the publication bias, which showed z = 1.09 and P = 0.28, indicating no incidence of publication bias in this analysis.

Discussion

This updated meta-analysis suggests that the BDNF allele frequencies and genotype distributions are significantly different between healthy controls and PD patients and that the Val66Met mutation in BDNF is a risk factor for PD, which are consistent with the previous meta-analysis (^{Chen et al., 2017}). Moreover, our study, including 11 full-length reports involving 2203 cases and 2554 controls, presented a relatively larger sample size than that of previous investigations(involving 1230 cases and 1612 controls) and, thus, might indicate precise and validated results.

Previous studies have demonstrated that the BDNF Met/Met carriers exhibit an increased not only the sensitivity to anxiety but also the severity of clinical features of panic attacks compared with those of Val/Val or Val/Met carriers (^{Monteleone et al., 2006; Elzinga et al., 2011}). Similarly, the Met allele is believed to play a role in anxiety-associated personality trait disorders, such as higher levels of trait anxiety as well as higher mean neuroticism scores (^{Sen et al., 2003; Lang et al., 2005}). The ^{Egan et al.’s (2003}b) report shows that, compared with the BDNF Val allele, the Met allele has a significant negative impact on the activity-dependent secretion of BDNF. Furthermore, a meta-analysis has revealed that the Met allele carriers might have diminished hippocampal volumes compared with that in Val/Val homozygotes (^{Molendijk et al., 2012}). Therefore, the Met/Met homozygous genotype can modulate the activity of BDNF, resulting in a decreased number of hippocampal neuroblasts, as observed in PD patients.

This finding has been further verified in animal studies, consistently showing that the disruption of BDNF’s brain-specific activities could be involved in altered hippocampus structure and associated anxiety disorder phenotypes (^{Dias et al., 2014}).

Based on these results, we may speculate that the reduced activity of BDNF Val66Met may induce hippocampal volume changes in patients with PD, as well, thus, involved in the cause of PD.

As per previous findings, the discrepancy in the BDNF mutation and PD onset could be explained as follows. First, the linkage disequilibrium between the Val66Met and another SNP within the BDNF gene could be causally related to PD pathogenesis, but this condition may not be similarly applicable across the diverse ethnic populations, which may partly explain the reason for inconsistent findings in various studies. Second, the Met allele distribution frequency largely depends on the ethnic background of an individual and most likely varies across diverse ethnic populations influencing the statistical power of the study. For example, Ishiguro et al. (^{Ishiguro et al., 2011}) have reported that the genotypic distribution of rs6295G/G is 3.1%, whereas that in the ^{Yevtushenko et al.’s (2010}) study is 19.6%. ^{Shimizu et al. (2004}) have proposed that the ethnic difference-based distribution variation of the BDNF G196A (Val66Met) allele may help explain the reason for differential prevalence rates of illness in different ethnic populations. Third, BDNF Met/Met carriers are reportedly more sensitive to stress. The early life stress exposure in this population may predict elevated anxiety levels and its association with panic attacks (^{Chen et al., 2006; Elzinga et al., 2011}). Thus, future studies should consider genetic factors, recent stress exposures, childhood abuse, and age at onset to accurately analyze their roles in the cause of PD.

One major drawback of this study was the small number of included articles and their limited sample sizes. Hence, we could not assess multiple subclinical factors as well as sex differences to elucidate their roles in the etiopathology of PD, warranting further studies to comprehensively analyze their effects. Additionally, we were not able to include the combined effects of BDNF G196A polymorphism versus environment factors’ interactions in PD pathology. It is shown that the gene-environment interactions are critical players in the onset of complex psychiatric disorders, such as PD.

In summary, this updated meta-analysis could identify pathological linkage between the BDNF Val66Met mutation and the cause of PD patients. However, further studies with statistically larger sample sizes and involving individuals from different ethnic backgrounds with additional clinical factor assessments are needed to precisely elucidate the nature of this etiopathology.

Acknowledgements

The authors would like to thank all the volunteers for taking part in this study. This study was supported by the Young Teacher Foundation of CMU (XZR20160010) for Dr. Yinglin Huang.

Conflicts of interest

There are no conflicts of interest.