Sex effects for the interaction of dopamine related genetic variants for COMT and BDNF on declarative memory performance

Genetic factors are assumed to contribute to memory performance, especially genes affecting the dopaminergic neurotransmission. We aimed to evaluate leading functional genetic variants of the dopamine system, Catechol‐O‐methyltransferase (COMT) SNP rs4680 and Brain‐derived neurotropic factor (BDNF) SNP rs6265, previously found to be associated with memory performance. In two independent general population cohorts (total N = 5937) we investigated direct and interaction effects between COMT and BDNF SNPs on declarative memory performance. We found significant two‐way interactions for COMT and BDNF in both cohorts but no direct genetic effects. Sensitivity analyses revealed that an interaction between COMT and BDNF was mainly carried by females. While direct associations of COMT and BDNF on memory have been reported previously, we could demonstrate that the interaction of COMT and BDNF is sex‐dependent and more complex and needs further investigation. Our results could be demonstrated in two independent cohorts of valuable size.

population cohorts (total N = 5937) we investigated direct and interaction effects between COMT and BDNF SNPs on declarative memory performance. We found significant two-way interactions for COMT and BDNF in both cohorts but no direct genetic effects. Sensitivity analyses revealed that an interaction between COMT and BDNF was mainly carried by females. While direct associations of COMT and BDNF on memory have been reported previously, we could demonstrate that the interaction of COMT and BDNF is sex-dependent and more complex and needs further investigation. Our results could be demonstrated in two independent cohorts of valuable size. Neurocognitive performance comprises a wide range of cognitive tasks including reading and mathematics performance, visual and verbal memory, problem solving, or learning. 1 This neurocognitive performance generally declines with age 2 and is also associated to a wide range of psychiatric and neurological disorders like schizophrenia, depression, 3,4 Alzheimer's, or Parkinson's disease. 5 The individual cognitive decline is likely attributed to a wide range of causes including environmental and genetic influences. On the genetic side evidence suggests that the neurotransmitter dopamine plays an important but complex role in several cognitive processes. 6 A major candidate gene effect allele cannot provide a full picture of the BDNF*COMT interaction on cognition. Moreover, sample size varied between N = 32 and N = 928 which is still far too small for genetic interaction analyses, especially in three-way analyses. Another influencing factor is the putative sex-dependency of the dopaminergic system, especially in association with substance use disorders. 19,20 We aim to explore the interaction between COMT rs4680 and BDNF rs6265 in two independent samples with overall sample size N = 5937 to analyze the still inconclusive findings. We hypothesize that the COMT and BDNF genetic polymorphisms as well as their combination have an impact on memory function in the general population. We further hypothesize that effects are sex-dependent.

| Study sample and phenotypes
We analyzed data from the Study of Health in Pomerania (SHIP) 21 comprising adult German residents in northeastern Germany. A two-stage stratified cluster sample of adults aged 20-79 years was randomly drawn from local registries. 4308 Caucasian subjects participated at baseline SHIP-0 (1997SHIP-0 ( -2001. As part of the SHIP baseline study 2400 subjects were included in the "Life-Events and Gene-Environment Interaction in Depression (LEGEND)" sample, a sub-sample with extended examination of various psychosocial, clinical and metabolic factors. 22 Among others, subjects were administered the auditory Verbal Learning and Memory Test (VLMT), a German adaption of the widely used Rey Auditory Verbal Learning Test. 23 The VLMT was used to assess short-term learning as well as delayed retrieval. Subjects were asked to remember neutral semantically unrelated nouns from an orally presented list in three consecutive encoding runs. Participants have 120 s after each encoding run for immediate retrieval. A sum score of the correctly remembered terms of the three encoding runs was formed to measure short-term learning.
Late retrieval was tested 20 min after the first encoding trial. The number of correctly remembered words reflected long-term retrieval.
In 2008 a new independent sample called SHIP-TREND (N = 4420) from the same area was drawn, encompassing similar examinations like SHIP-LEGEND. 21 However, instead of the VLMT, the word list of the Nuremberg Age Inventory (NAI) was used as a measure for immediateand delayed memory performance. The NAI is a German test developed to measure the cognitive abilities during brain aging. 24 It consists among others subtests of a list of eight neutral words. Eight words are read to the participant, who is asked to recall as many words as possible immediately. After 20 min the participant is asked to retrieve the eight words previously learned from a list containing eight additional distractor words in a passive recall. The number of correctly identified words is summarized to a sum score minus the number of identified distractor words.
The investigations in both studies were carried out in accordance with the Declaration of Helsinki, including written informed consent of all participants. The survey and study methods were approved by the institutional review boards of the University of Greifswald.

| Genetic data
The SHIP sample was genotyped using the Affymetrix Human SNP Array 6.0. The overall genotyping efficiency was 98.55%. Genotyping of a subset of the SHIP-TREND subjects (N = 986) was performed using the Illumina HumanOmni 2.5-Quad. The final sample call rate was 99.51%. The remaining SHIP-TREND sample (N = 3133) was genotyped using the Illumina GSA-24. Arrays with a genotyping call rate < 94% were removed. For further details see. 21 Imputation of genotypes was performed using the HRCv1.1 reference panel and the Eagle and minimac3 software implemented in the Michigan Imputation Server for pre-phasing and imputation, respectively. SNPs with a Hardy-Weinberg-Equilibrium p-value <0.0001, a call rate < 0.95, and a MAF <1% were removed before imputation.
After exclusion of subjects due to missing data, at least 2098 subjects from SHIP LEGEND and 3839 subjects from SHIP-TREND were included in our analyses.

| Statistical analyses
In both samples, linear regression analyses assuming an additive effects of the SNPs were applied in order to calculate predictive effects of the genetic variants of COMT and BDNF on early as well as delayed recall of neutral words. Additional gene-gene-interaction analyses were performed. To account for the nonnormal distribution of the outcome variables from VLMT and NAI, confidence intervals, and p-values were assessed through bootstrap with 1000 replicates.
To ease comparison of effects, the word recall scores have been ztransformed prior to analyses. All analyses were adjusted for age, sex, age*sex interaction, and education level. SHIP-TREND was additionally adjusted for genetic array. As two genetic variants were tested for their effects on verbal memory in two independent cohorts, significance p-value was set to p < 0.0125. Analyses were performed using STATA/MP software, version 13 (StataCorp LP, College Station, TX).

| RESULTS
Demographic results and allele frequencies for BDNF and COMT.
There were no significant group differences between SHIP-LEGEND and SHIP-TREND for COMT or BDNF frequencies (Chi 2 -test: p = 0.12 and p = 0.44). Both SNPs were in Hardy-Weinberg equilibrium in both samples (all p > 0.1). The minor allele frequencies were 14% and 19% for BDNF and 47% and 46% for COMT in SHIP-LEGEND and SHIP-TREND, respectively. A detailed demographic characteristic is given in Table 1. In general, females performed significantly better than males in all verbal memory tests, age was highly associated with lower verbal memory score (p < 0.001) as well as higher education level.

| Direct effects of COMT and BDNF
In SHIP-LEGEND there was a positive nominal significant association between the Val-allele of COMT and delayed recall of words (β = 0.06, p = 0.021) but missed significance for immediate recall (p = 0.054). A direct association could not be replicated in SHIP-TREND. BDNF exhibited no direct effect on recall of words neither in SHIP-LEGEND nor in SHIP-TREND (see Table 2).

| Sex-separated sensitivity analyses
As previous studies have found sex differences in dopamine neurobiology, 19 we additionally analyzed sex differences regarding the association with verbal memory. In sex-separated analyses we could see  Figure 1 with the Val-allele for COMT associated with higher verbal memory score.

| DISCUSSION
Our results showed that the interaction of COMT rs4680 and BDNF rs6265 genotype has an influence on memory performance in the general population especially with regard to sex. These results could be supported in two independent samples of valuable size (each sample N > 2000).
It is well known that genetic polymorphisms are associated with memory function as episodic memory is a heritable and polygenic trait. 26 homozygotes. This is somehow counterintuitive as most findings demonstrate that better cognitive performance was associated with the COMT Met-allele 28 which is associated with higher dopamine levels in cognition relevant brain areas. 28 For BDNF most of the literature suggests that the Met-allele was associated with less BDNF and dopamine availability and impaired cognitive performance. 10,29 In our gene-gene interactions we observed that in the female sample the effect of COMT Val/Met polymorphism on memory function changed depending on the BDNF allele. When carrying the Met/Met genotype for COMT, females showed the highest scores for delayed recall of words when also being Met-allele carriers for BDNF.
When being Val homozygous for COMT, being Val homozygous for BDNF was associated with higher scores (see Figure 1). In the combined sample of both sexes we observed a linear association between COMT polymorphism and delayed memory only in carriers of BDNF Val-allele homozygotes (see Figure 1) with higher verbal memory scores when carrying one or two Val-alleles of COMT.
Previous results showed that the COMT Met-allele and the BDNF Val-allele were associated with higher dopamine levels than their respective counterparts. 18   Finally, we can say that the interaction between COMT rs4680 and BDNF rs6265 has an impact on verbal memory performance. But this interaction is complex and also depending on sex differences in dopamine levels. This could be observed in two large independent samples (in total ≈ 6000 individuals) drawn from the same source population. Our findings of no direct, but interactive effects between these polymorphisms may contribute to explain the heterogeneous findings from previous studies.

CONFLICTS OF INTEREST
All authors declare that no competing interest exist. The interpretation of the data was not influenced by any funding.

DATA AVAILABILITY STATEMENT
The data used in this study are freely available after data application and a data transfer agreement on the study website (https://www. fvcm.med.uni-greifswald.de/dd_service/data_use_intro.php).