Hello!
This Is The Blog For The Resilience Research Community

On March 20 ~ 21^st, Prof. Eric-Jan Wagenmakers from the Department of Psychological Methods, Amsterdam University, gave a workshop of Bayesian Statistics and JASP for CRC PhD students and Post-Docs in Mainz.

The two-day workshop started with a gentle introduction to the theoretical background. Prof. Wagenmakers’s friendly and patient teaching style helped us a lot to wrap our heads around the concepts of Bayes factor and prior and posterior distributions. Subsequently, we immediately started to apply the new concepts hands on to Bayesian binomial tests, correlation analyses and t-tests in JASP. One of the most important lessons of the first day: always look at your raw data, otherwise, you may get statistically “strong evidence” for nonsense data. The first day ended with a discussion about how to control for multi-comparisons in Bayesian statistics.

In the morning of the second day, Prof. Wagenmakers started by explaining more about one-side tests and model-averaging. We were challenged to apply all new knowledge immediately to again to contingency tables, ANOVA, and linear regression analysis in JASP. The pros and cons of specified prior distributions were also introduced. Finally, we discussed how to use Bayesian sequential test to optimize experimental design and sample size estimations.

The workshop introduced all of us to a new, intuitive and reasonable way of testing research hypotheses and the powerful functions and the elegant GUIs of JASP. For many it dampened the anticipatory fear of the seemingly complicated Bayesian framework and motivated us to take up the challenge of quantifying the evidence for our hypotheses in the future. We are optimistic that this workshop helped in making our research more resilient in the midst of replication crisis.

Chuan-Peng & Anna

We have all learned that social support is an important resilience factor. We have also all become aware of the distinction between perceived and objective social support, perceived social support (how you feel supported) being said to actually matter for mental health. Finally, we have learned to distinguish between positive social support (where we are encouraged by others, or feel cared about) and negative social support (your friends or family members making demands, criticizing you, or arguing with you). So, what else could we wish than a) perceived b) positive social support (PPSS)? Sounds like a magic bullet.

Unfortunately, as pointed out by Nickerson et al. in their recent very elegant paper in Psychological Medicine, studies investigating the relationship between PPSS and PTSD symptoms have yielded rather mixed results. If anything, it looks like PTSD symptoms may lead to reductions in PPSS, while there is less evidence that PPSS might protect against the development of PTSD symptoms (i.e., that PPSS is a resilience factor).

To shed light on the issue, Nickerson and colleagues collected a unique prospective-longitudinal data set from injury survivors (N=1150 at study inclusion), who reported on their PTSD symptoms (CAPS) and their PPSS and PNSS (perceived negative social support) immediately as well as 3, 12, 24, and 72 months after the trauma.

These data allowed the authors to ask, among others: does a score on one of the three variables at time point t statistically affect the change in the scores on the other two variables from t to t+1. Meaning, for instance: if I enjoy high PPSS now, is this making it likely my PTSD symptoms will subsequently decrease (a cross-lagged bivariate change)?

The result? Subjects with high PTSD symptoms at t were likelier to perceive their NSS to subsequently increase (more PNSS at t+1 than at t - bad news 1). Such subjects also tended to find their PSS to decrease (less PPSS at t+1 than at t - bad news 2). And: neither PNSS nor PPSS predicted PTSD symptom changes. While it is perhaps relieving to see that negative social support perceptions (PNSS) do not enhance subsequent PTSD symptoms, it is clearly a disappointment for a resilience researcher that perceived positive support (PPSS at t) is not related to PTSD symptom reduction (from t to t+1 - very bad news 3). In other words: positive support perceptions do not protect against post-traumatic symptoms. This questions whether PPSS can be considered a resilience factor.

The authors are careful enough to discuss the limitations of their study, including a sample of subjects having experienced only a specific type of trauma and not living in apparently instable social environments, where social support might play a more important role; the focus on PTSD symptoms as single outcome measure; their lack of information on other potentially important factors, such as offered vs. experienced support, source of support, emotional vs. instrumental support, or moderating personality traits. Hence, the question clearly warrants further research.

Nevertheless, the study by Nickerson et al. reminds us that we should be cautious in the interpretation of cross-sectional associations between assumed resilience factors and mental health outcomes after adversity. After all, the cross-sectional correlation analyses also reported in the paper showed significant negative associations between PPSS and PTSD at all measurement time points (as well as positive associations between PNSS and PTSD). Hence, at first sight, a good reason to conclude that PPSS is a resilience factor. Unfortunately, it is still mostly these types of cross-sectional analyses that most of our information on resilience factors is derived from. The study therefore is a very nice example of the importance to conduct prospective-longitudinal studies in resilience research, as recently highlighted by a large international consortium of authors (Kalisch et al.). The study also highlights the value of advanced statistical methods for the analysis time-variant multivariate data, such as latent difference score (LDS) structural equation modelling used here.

Like every year, speakers at our 3^rdInternational Symposium on Resilience Research (taking place end of September 2017 in Mainz, Germany) showed lots of unpublished data, some of which have meanwhile made it into high-profile journals. One of those data sets, presented by Scott Russo from Mount Sinai, has recently come out in Nature Neuroscience (Menard et al.).

Like many studies on mouse resilience, the study used the Chronic Social Defeat Stress (CSDS) paradigm, in which a mouse is exposed once a day over ten days to the attacks of a much heavier, aggressive “resident” mouse, whose cage it is forced to enter. CSDS induces long-lasting social aversion (reduced social interaction with other, non-aggressive mice) and anhedonia (e.g., reduced sucrose preference) in the intruder mouse, behavioral dysfunctions reminiscent of some of the symptoms of human depression. Hence, CSDS is considered by many an interesting depression model. However, CSDS does induce these dysfunctions in all intruder mice. Already in 2007, Krishan et al. could show in Cell that a substantial fraction of the intruders (the “resilient” ones) show social and reward-related behavior comparable to non-stressed control mice. This finding was probably a key moment for resilience research in rodent models and has since led to many important insights into the neurobiological underpinnings of resilience.

Many of those studies focussed on the brain, but there have also been intriguing findings about a role for the peripheral immune system in resilience (e.g., Hodes et al., in PNAS, 2014). Now, Menard et al. find that CSDS induces loss of expression of the protein claudin 5 (Cldn5) specifically in intruder mice susceptible to the detrimental behavioral long-term effects of CSDS – but not in the resilient mice. Cldn5 lives in tight junctions, the cellular structures that knit together the endothelial cells that line the blood vessels in the brain. Tight junctions are crucial components of the so-called Blood-Brain Barrier (BBB) that keeps molecules from the blood from freely entering the brain. Cldn5 reduction in the present study was mainly observed in blood vessels in the nucleus accumbens (a key region of the reward system, dysfunctions of which have been linked to depression) and was also accompanied by ultrastructural vessel abnormalities and permeability of the BBB in this brain area even for large blood proteins, in susceptible mice. Interestingly, post-mortem patient data indicated a similar loss of Cldn5 in depression.

When the authors knocked down Cldn5 in the accumbens, this exaggerated the effects of a milder variant of social stress into full-blown social aversion and anhedonia, and when Cldn5 expression was allowed to recover in these animals, this also rescued the behavioral deficits. Hence, Cldn5 expression might have a pro-resilience, protective function.

Enters the immune system. The authors (see Hodes et al.) had previously observed a drastic increase in blood levels of the pro-inflammatory cytokine IL-6 after CSDS in susceptible mice. They now extend this finding to IL-6 levels in the accumbens, and they provide evidence that stress-induced accumbal IL-6 is most likely derived from the blood, not produced locally. Downregulation of Cldn5 expression is apparently crucial for allowing the passage of IL-6 into the accumbens. Finally, they show that infusion of IL-6 into the accumbens before a mild variant of social stress is sufficient to induce marked social aversion.

Hence, inflammatory processes play a key pathogenic role in a mouse model of depression, in line with an increasing number of findings of inflammatory processes in depressed patients. For resilience research, the data indicate an important role for an intact Blood-Brain Barrier in protecting against detrimental influences of peripheral inflammatory processes. It will be interesting to see whether future research can identify methods to preserve or repair BBB integrity, which should have preventative effects in chronically stressed individuals.

From a broader perspective, the study is important because it finally provides a bridge between the immune system, whose role in stress-related pathology has aroused much interest in recent years (not least by the work of the Russo lab), and the brain in explaining the effects of chronic stress. This bridge seems to be the vasculature of the accumbens. What this also means is: if it turns out impossible to find ways to dampen peripheral inflammation of entry of cytokines into the accumbens, other ways of protecting or improving accumbal function might as well also help. In any way, the brain is back on center stage.

Hear the latest news, get alerted about interesting papers, conferences, calls, grants or job openings, read comments and opinions, take part in discussions. 

This new blog is a service for the international resilience community by researchers of the DFG-funded Collaborative Research Center CRC 1193 “Neurobiology of Resilience“,  based in Mainz and Frankfurt; the International Resilience Alliance (www.intresa.org); the EU project DynaMORE; and the Deutsches Resilienz Zentrum in Mainz (www.drz.uni-mainz.de). We hope to develop this place into a true virtual home for all those interested in promoting this fascinating  and flourishing field of research.

New blog entries are announced via Twitter (ResilienceResearch@ResilienceRes, #resilience). Follow us and share!

And please comment and discuss with us. Or provide us with news and  information you feel might be interesting for everyone. Suggestions for blog entries  can be sent to rkalisch@uni-mainz.de and will be assessed by the editorial team.

Greetings from the team. And hope to see you again here.

Raffael Kalisch, Beat Lutz, Michèle Wessa, Oliver Tüscher, Klaus Lieb, Marianne B Müller