Supplementary material for Lonsdorf et. al., A review on experimental and clinical genetic associations studies on aversive fear conditioning, extinction and cognitive behavioral treatment (CBT)

Materials and methods

As details have been published elsewhere,1,2 methods will only be described briefly here.

Procedure

Briefly, in the acquisition phase,participants saw 18 pseudo randomized 5-sec presentations each of two geometric symbols (acircle, a triangle) one of which (CS+) was paired in 80% of cases with a painful electricstimulus (UCS) applied to the back of the right hand. The other symbol served as a controlstimulus (CS-) for non-associative effects and was never paired with the UCS. Assignment ofsymbols as CS+ or CS- was counterbalanced across participants. In the extinction phaseboth stimuli were again presented 18 times each, but in the absence of the UCS. The subsequent reacquisition phase was identical to the acquisition phase. The inter-trialinterval (ITI) was jittered between 9 to 14 s, with an average of 11.5 s. We intermittentlyasked participants to give explicit ratingsof their CS-evoked stress/fear/tension (see below). Throughout the experiment, the participants had toperform a speeded decision task on the geometric symbols.UCS intensity was individually adjusted prior to the experiment to achieve maximumtolerable pain. For details see our prior publication 1.

SCR

SCRs were defined as the maximum skin conductancein a time window of 5 s after CS onset minus skinconductance at the time of CS onset. UCS-evoked SCRswere scored analogously over 10 s. SCR values were normalized to the first trial of a givenphase. For details on recording see our prior publication.1

Ratings

At the onset of the experimentand every 12 trials (6CS+ and 6 CS- trials) thereafter, participants rated howmuch stress/fear/tension the symbols had causedthem by moving a red dot to the corresponding position on a visual analog scale that ranged from 0(no stress at all) to 100 (very much stress), with thehelp of the keypad. Rating values were normalized to the lastrating of the preceding phase (that is, to the post-habituationrating for Acquisition, the last Acquisition rating for Extinction,and the last Extinction rating for Reacquisition). For further details see our prior publication 1.

Learning Rates (Rescorla Wagner [RW] Model)

In the RW model of associative learning, the prediction error δav is weightedby a constant α, the learning rate, that determines how much a deviation from prediction attrial t is taken into account when formulating the prediction for the next trial t+1. A highlearning rate signifies rapid prediction adjustment and thus quick learning. Detailed descriptions are described in our prior publication.2

Genotyping

Genotyping of the BDNFval66met (rs6265) was performed as described previously.1Due to the low frequency of the BDNFval66met met/met genotypecarriers of one or two 66met alleleswere combined to a met-carrier group. Genotype frequencies (46 val/val, 23 met-carriers whereof 2 met/met) did not deviate from Hardy-Weinberg equilibrium.

Statistics

As we previously reported significant associations of the NPSR1 and DAT1 genotypes within this dataset1, these genotypes were always included as covariates in all analyses.

Univariate Analyses with UCS SCRs and learning rate α were performed with BDNFgenotype as the fixed factor.

Separate repeated measurements ANOVAs per experimental phase with SCRs or fear ratings to the stimuli (2: CS+, CS-) as the within-subject factor and BDNF genotype (2: val/val vs. met-carrier) as the between-subject factor were performed. Additionally similar stimulus (2) * genotype (2) * time (3: early, middle, late per experimental phase) repeated measurements ANOVAs per experimental phase were performed to test for interactions with time.

Results

SCR

Of the 69 participants, 12 had to be excluded from SCR analyses (3 did not show any apparent SCR to the UCS and for 9 participants, data were not acquired or lost due to technical problems) leaving 57 participants for analyses.

A univariate ANOVA revealed no differences in UCS SCRs (raw data) between both BDNF genotype groups, F(1,46)<1.

Repeated measurements ANOVAs showed no main effect of genotype and no stimulus*genotype interaction on SCRs during acquisition, both F(1,52)<1, extinction, F(1,52)=2.89, p=0.095 and F(1,52)=2.83, p=0.098 respectively, or reacquisition, F(1,52)=2.54, p=0.12 and F(1,52)=2.08, p=0.15, respectively. The weak trend for a stimulus*genotype interaction during extinction was driven by numerically higher SCRs to the CS- as compared to the CS+ in val-homozygotes, while met-carriers showed numerically slightly higher SCRs to the CS+ as compared to the CS-.

Taken the factors time (3) and stimulus (2) into account, did not change results, except for a main effect of BDNF genotype that became significant during reacquisition, F(1,46)=5.33, p=0.025, η2=0.10. Met-carriers displayed significantly lower SCRs than homozygous val-carriers.

Ratings

Of the 69 participants, 4 had to be excluded from analyses of the ratings as data were not acquired or lost due to technical problems, leaving 65participants for analyses (43 val/val, 22 met-carrier, whereof 2 met/met).

No main effect of genotypeand nostimulus*genotype interaction on baseline ratings of CS+ and CS- were observed, both F(1,61)<1.3, both p>0.26.

During acquisition, a significant stimulus*genotype interactionindicated better discrimination in met-carriers than in val/val-homozygous, F(1,61)=5.18, p=0.026, η2=0.08. This interaction was due to significantly higher CS+ ratings in met-carriers than in val/val-homozygotes, F(1,61)=4.80, p=0.0232, η2=0.07, despite of comparable CS- ratings, F(1,61)<1 (see supplementary figure 1A). Furthermore, met-carriers tended to give higher fear ratings, as indicated by a trend wise main effect of genotype, F(1,61)=2.90, p=0.09, η2=0.05.

During extinction, no main effect of genotype or stimulus*genotype interaction was observed, both F(1,61)<2.4, both p>0.13.

During re-acquisition however, a significant stimulus*genotype interactionindicated again better discrimination in met-carriers than in val/val-homozygotes, F(1,61)=6.16, p=0.016, η2=0.09, in absence of a main effect of genotype, F(1,61)<2.1, p>0.15. Again, met-carriers tended to rate particularly the CS+ higher than non-carriers, F(1,61)=4.25, p=0.04, η2=0.07, despite of no differences in CS- ratings, F(1,61)<1.7 (see supplementary figure 1B).

Taking the factor time (3: early, middle, late experimental phase) into account did not change the results, despite of a significant time*genotype interaction during extinction, F(2,122)=4.94, p=0.013, η2=0.08, indicating a steeper decline in fear ratings during extinction in met-carriers as compared to non-carriers.

Learning rates

Univariate Analyses revealed no significant differences between the two genotype groups in learning rates during acquisition, extinction or reacquisition, all F(1,59)<1.7, all p>0.2.

REFERENCES

1.Raczka, K. et al. A neuropeptide S receptor variant associated with overinterpretation of fear reactions: a potential neurogenetic basis for catastrophizing. MOLECULAR PSYCHIATRY15, 1067-1074 (2010).

2.Raczka, Karolina Anna et al. Empirical support for an involvement of the meso-striataldopamine system in human fear extinction. Translational Psychiatry (in press).