Letters to the Editor
120
generate such large literatures of inconclusive results. We wondered whether one source of the difficulties in the interpretation of genetic association studies might lie with the journal that published the initial finding. Studies published in journals with a high impact factor typically attract considerable attention. However, it is not clear that these studies are necessarily more robust than those published in journals with lower impact factors. We used data from three meta-analytic reviews of gene–disease associations in the psychiatric genetics literature, resulting in a total of k = 81 studies published between 1990 and 2008. We divided the individual study odds ratio (OR) by the pooled OR, to arrive at an estimate of the degree to which each individual study over- or underestimated the true effect size, as estimated in the corresponding metaanalysis. We have recently used this method to identify a biasing effect of research location and resources.6 Additional data on the impact factor of the journal in which each individual study was published were collected, using 2006 data. These data were unavailable in the case of two studies, resulting in a final sample of k = 79 studies. Data were analysed using meta-regression of individual study bias score against journal impact factor. This indicated a significant correlation between impact factor and bias score (R2 = þ 0.13, z = 4.27, P = 0.00002). Our results are presented graphically in Figure 1. We also note that journals with high impact factors tend to publish studies with high bias scores and small sample sizes (as indicated by the smaller circles in the figure). Genetic association studies offer the advantages of being numerous and highly comparable, allowing
2.500
LnOR (bias)
2.000 1.500
analyses of the kind described here. Our results indicate that genetic association studies published in journals with a high impact factor are more likely to provide an overestimate of the true effect size. This is likely to be in part due to the small sample sizes used and the correspondingly low statistical power that characterizes these studies. Initial reports of genetic association published in journals with a high impact factor should therefore be treated with particular caution. However, although we cannot necessarily generalize our findings to other research domains, there are no particular reasons to expect that genetic association studies are unique in this respect. MR Munafo`1, G Stothart1 and J Flint2 Department of Experimental Psychology, University of Bristol, Bristol, UK and 2Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, UK E-mail:
[email protected] 1
References 1 Chanock SJ, Manolio T, Boehnke M, Boerwinkle E, Hunter DJ, Thomas G et al. Nature 2007; 447: 655–660. 2 Blum K, Noble EP, Sheridan PJ, Montgomery A, Ritchie T, Jagadeeswaran P et al. JAMA 1990; 263: 2055–2060. 3 Lesch KP, Bengel D, Heils A, Sabol SZ, Greenberg BD, Petri S et al. Science 1996; 274: 1527–1531. 4 Munafo MR, Freimer NB, Ng W, Ophoff R, Veijola J, Miettunen J et al. Am J Med Genet B Neuropsychiatric Genet 2008; e-pub ahead of print. 5 Munafo MR, Matheson IJ, Flint J. Mol Psychiatry 2007; 12: 454–461. 6 Munafo MR, Attwood AS, Flint J. Psychol Med 2008; 38: 1213–1214.
The Met allele of the BDNF Val66Met polymorphism is associated with increased BDNF serum concentrations
1.000 0.500
Molecular Psychiatry (2009) 14, 120–122; doi:10.1038/mp.2008.80
0.000 -0.500 -1.000 0.00
5.00
10.00 15.00 Impact Factor
20.00
25.00
Figure 1 Meta-regression of individual study bias score and journal impact factor. Bias score is plotted against the 2006 impact factor of the journal in which the study was published. Meta-regression indicates a positive correlation between journal impact factor and bias score (R2 = þ 0.13, P = 0.00002), suggesting that genetic association studies published in journals with a high impact factor are more likely to provide an overestimate of the true effect. Circles, representing individual studies, are proportional to the sample size (that is, accuracy) of the study. Source: Thomson Scientific. Molecular Psychiatry
The brain-derived neurotrophic factor (BDNF) hypothesis of depression postulates that a loss of BDNF function is directly involved in the pathophysiology of depression and its restoration may underlie the therapeutic efficacy of antidepressant treatments (for recent review see Groves1). The hypothesis is in line with observations that BDNF concentrations in humans are decreased in major depressed patients and healthy humans with depression-related personality traits and are increased after antidepressant treatment.2,3 A common single nucleotide polymorphism in the human BDNF gene (c.196G > A, dbSNP: rs6265) has been identified causing an amino-acid substitution
Letters to the Editor
121 11
BDNF serum concentration (ng ml –1) log transformed
of valine to methionine at amino-acid residue 66 (Val66Met). It alters intracellular trafficking and packaging of pro-BDNF and consequently, the regulated secretion of the mature peptide.4 We tested the hypothesis if BDNF serum concentrations are associated with BDNF Val66Met genotype in 114 healthy subjects (age: 42.3±13.2 years, 53 women). The study was approved by the ethics committee of the Charite´ University Medicine Berlin. Medical, neurological and psychiatric disorders and a family history (first degree) of axis I disorder lead to exclusion from the study. For further methodological details see Lang et al.5 Allelic discrimination of BDNF polymorphism and BDNF serum levels were determined as described previously.3,5 Mean BDNF serum concentrations from 114 subjects (16.7±7.5 ng ml1) did not differ between male (16.8±7.5 ng ml1) and female (16.6±7.5 ng ml1) volunteers. BDNF serum concentrations were not normally distributed. After logarithmic transformation a normal distribution was achieved (9.6±0.5 ng ml1) and further analyses were carried out with transformed concentrations. The genotype frequency Val/Val (n = 79, mean age 43.4±13 years, 50.6% women), Val/Met (n = 35, mean age 39.8±12 years, 37.1% women) and Met/Met (n = 0) was 69.3, 30.7 and 0%. The allele frequency was 0.85 (Val) and 0.15 (Met). An analysis of covariance showed that the dependent variable BDNF concentration was significantly affected by the factor ‘genotype’ (Val/Val versus Val/ Met: F = 5.391, d.f. = 1, P = 0.022), the covariate ‘age’ (F = 9.457, d.f. = 1, P = 0.003) but not by the covariate ‘gender’ (F = 0.063, d.f. = 1, P = 0.803; Figure 1). In conclusion, we found decreased BDNF concentrations in healthy Val/Val subjects when compared to Val/Met individuals. This is the first study showing a significant association between an important genetic variation of the BDNF gene and commonly investigated serum BDNF concentration in humans. Our data are in line with several studies reporting an over transmission of the common Val allele in bipolar and depressive human samples (for example, Sklar et al.6) and studies reporting decreased BDNF concentrations in depressive patients, which increase after antidepressant treatment.2 The data also fit in studies, showing BDNF Met allele moderates the effect of early adversity on depressive symptoms7 and Val/Val genotypes increase levels of trait anxiety and neuroticism scores.5 However, Met allele of BDNF genotype has been associated with abnormal intracellular secretion of BDNF and abnormal hippocampal structure and function.8 In line with this report, in a variant BDNF mouse (BDNF(Met/Met)), the secretion of BDNF from neurons was defective and BDNF(Met/Met) mice exhibited increased anxiety-related behaviours that were not normalized by the antidepressant, fluoxetine.9 Moreover, the Met variant has been hypothesized to decrease the activity dependent but not the
10
9
8
Val/Val
Val/Met
Figure 1 Brain-derived neurotrophic factor (BDNF) serum concentrations (log transformed) for different BDNF genotypes. The analysis of variance (ANOVA) showed a significant group difference.
constitutive BDNF secretion.10 Therefore, a constitutive upregulation of peripheral BDNF concentrations in carriers of the Met allele might compensate a defective intracellular protein signalling. As depressive disorders are typically state dependent and depressive symptoms are completely remitting in most cases, patients with low BDNF serum concentrations and carrying the Val/Val genotype might be more vulnerable to develop depressive symptoms but might respond well to antidepressant treatment by BDNF upregulation. In conclusion, we detected for the first time an association between serum BDNF protein concentration and Val66Met genotype. UE Lang1, R Hellweg1, T Sander2 and J Gallinat1 1 Department of Psychiatry and Psychotherapy, Charite´—University Medicine Berlin, Campus Mitte, Germany and 2Max Delbrueck Centrum, Berlin, Germany E-mail:
[email protected]
References 1 Groves JO. Mol Psychiatry 2007; 12: 1079–1088. 2 Shimizu E, Hashimoto K, Okamura N, Koike K, Komatsu N, Kumakiri C et al. Biol Psychiatry 2003; 54: 70–75. Molecular Psychiatry
Letters to the Editor
122
3 Lang UE, Hellweg R, Gallinat J. Neuropsychopharmacology 2004; 29: 795–798. 4 Egan MF, Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A et al. Cell 2003; 112: 257–269. 5 Lang UE, Hellweg R, Kalus P, Bajbouj M, Lenzen KP, Sander T et al. Psychopharmacology (Berl) 2005; 180: 95–99. 6 Sklar P, Gabriel SB, McInnis MG, Bennett P, Lim YM, Tsan G et al. Mol Psychiatry 2002; 7: 579–593.
Molecular Psychiatry
7 Wichers M, Kenis G, Jacobs N, Mengelers R, Derom C, Vlietinck R et al. Am J Med Genet B Neuropsychiatr Genet 2008; 147: 120–123. 8 Hariri AR, Goldberg TE, Mattay VS, Kolachana BS, Callicott JH, Egan MF et al. J Neurosci 2003; 23: 6690–6694. 9 Chen ZY, Jing D, Bath KG, Ieraci A, Khan T, Siao CJ et al. Science 2006; 314: 140–143. 10 Tramontina J, Frey BN, Andreazza AC, Zandona M, Santin A, Kapczinski F. Mol Psychiatry 2007; 12: 230–231.
