The current study focused on the brain mechanism of remitted PTSD children and adolescents after earthquake compared with current PTSD and a trauma-exposed control using trauma-unrelated and trauma-related stimuli. As hypothesized, in the comparison between the PTSD group and the remitted group, under the trauma-unrelated condition, PTSD exhibited lower activity in the left hippocampus. However, no significant difference was found under the trauma-related condition. In the comparison between the PTSD group and the control group, PTSD exhibited lower activity in the left hippocampus under the trauma-unrelated condition and higher activity in the right middle occipital gyrus under the trauma-related condition. There was a similar pattern in the comparison between the remitted group and the control group, but the activity of the left hippocampus of the remitted group was in the middle between the other groups under the trauma-unrelated condition. In trauma-related vs. trauma-unrelated contrast, contrary to our hypothesis, there was no significant neural activity difference in the PTSD group, while in the remitted group, different brain activity was found in the right parahippocampal gyrus and right lingual gyrus. We also explored a group analysis of trauma-related vs. trauma-unrelated contrast and found that the activity difference in the left hippocampus was PTSD = removed > control. No difference in mPFC/ACC/amygdala was found in any comparison.
The pathological mechanism of PTSD is closely related to the hippocampus. The largest PTSD brain image study included 1868 participants from 16 cohorts confirmed the relationship between PTSD and a smaller hippocampus volume.(Logue et al., 2018) Klaming et al. also found the correlation between right hippocampus morphology and symptom severity in 70 trauma-exposed veterans.(Klaming, Spadoni, Veltman, & Simmons, 2019) Furthermore, in 2017, Malejko et al. reviewed the research on PTSD remission and found that the remission of PTSD was often accompanied by a change in hippocampus activity.(Malejko et al., 2017) In the current study, using whole-brain analysis, the only significant difference between the PTSD group and the remitted group was in the left hippocampus under the trauma-unrelated condition. To be specific, the remitted group exhibited higher activity than the PTSD group, and lower than the control group. This finding supported the significance of the hippocampus, as well as trauma-unrelated stimuli, in the remittance mechanism of PTSD. In the following exploratory analysis, we also found that this cluster was the only one with inter-group differences under trauma-related vs trauma-unrelated contrast.
A number of studies have found that the volume of the occipital lobe in the PTSD population is smaller than that of the control group,(Cwik et al., 2019; Sussman, Pang, Jetly, Dunkley, & Taylor, 2016) In adult(Gudrun et al., 2013; Ke et al., 2015) and children(P. Yang, Wu, Hsu, & Ker, 2004) PTSD studies, this region also showed abnormal high activity under trauma-related conditions. This may be due to the involvement of the occipital lobe in the transformation of traumatic memory in visual form into narrative trauma-related memories.(Lanius, Bluhm, Lanius, & Pain, 2006) In the current study, the hyperactivity in right middle occipital gyrus in the PTSD and remitted group may reflect the re-experiencing under trauma-related stimuli condition.
Contrary to our intuition, the remitted individuals still had PTSD-like brain responses to trauma stimuli. This state-independent feature may be interpreted as the endophenotype of PTSD.(McAuley, Crosbie, Charach, & Schachar, 2014) However, there is still a question from the other side of this phenomenon: how can we define an individual who still exhibits 'abnormal' brain activity under trauma stimuli as remitted? This dichotomy may occur because our definition of the state of psychiatric disorder highly depends on the influence or potential influence of symptoms on an individual's social function, and the 'abnormal' brain activity that existed only during trauma-specific stimulation had relatively little influence on social function. The PTSD diagnostic criterion G in DSM-5 is 'The disturbance causes clinically significant distress or impairment in social, occupational, or other important areas of functioning', which directly points to the evaluation of social functions.(Bovin et al., 2015) During the K-SADS interview, psychiatric clinicians do not simply judge the presence or absence of a symptom but spend a large proportion of time verifying the possible impact of suspicious symptoms on patients' functions. If an individual only have 'abnormal response' to stimuli that are highly specific to the original trauma (trauma-related stimulus), then the chance of exposure in daily life to cause functional damage will be relatively low, and the patient will consequently be defined as 'remitted', even if the possibility of an abnormal response under very special circumstances exists. Schnurr et al. studied the relationship between symptom improvement benchmarks and the quality of life of PTSD patients and found that remission (defined as loss of diagnosis plus a severity score < 20 in the Clinician-Administered PTSD Scale) did not yield more benefit than the loss of diagnosis, although remission is considered the most desirable outcome for relieving PTSD symptom burden.(Schnurr & Lunney, 2016) This result is not direct evidence, but it is also supported by results that the loss of diagnosis in PTSD patients does not always mean complete 'normalization' at the symptom level, but more meaningfully at the level of social function. Similar to the current finding, participants who were defined as remitted did not achieve complete 'normalization'.
There could be a potential variable related to social function of earthquake induced PTSD - the “Commonness” of the trauma, referencing the probability of exposure to related cues in daily life. In the current study, earthquake-related cues should be relatively rare in daily life, and correspondingly, the chance to affect individuals was low. Frequent stimulation does cause pain to individuals, but also provides more opportunities for extinguish learning. The 'Commonness' of cues in PTSD should be a factor between the dynamic balance of persist–remit and a potential difference between various traumas. Therefore, 'commonness' could be one direction for future research.
In the remitted group, the right parahippocampal gyrus and the right lingual gyrus exhibited higher brain activity under the trauma-related condition than under the trauma-unrelated condition. These clusters were highly consistent with the findings from another earthquake-induced PTSD study on youths, in which PTSD youths exhibited higher activity than controls in the same brain regions only.(P. Yang et al., 2004) The parahippocampal gyrus is involved in the storage and retrieval of emotional memory.(P. Yang et al., 2004) And lingual gyrus is related to narrative memory.(Lanius et al., 2006) In previous study, reduction of the lingual gyrus is related to the severity of PTSD symptoms.(Wrocklage et al., 2017) The current findings in the remitted group may also reflect the re-experiencing under trauma condition, and there is no such response under the trauma-unrelated condition.
In contrast, there is no significant brain activity difference in the PTSD group between the two conditions. This result, however, needs to be discussed very carefully. We previously considered whether this statistically negative result could be a supportive evidence for the claim in previous studies that abnormalities of PTSD patients are "generalized" to trauma unrelated condition.(van Rooij et al., 2015; Zinchenko et al., 2017) However, inspection of Fig. 5 shows that this trauma related vs. trauma unrelated difference between the groups is rather small, and was not statistically significant. Exploratory correlation analysis also failed to find a significant correlation between brain activity and generalization effect. In the future, bigger sample with better statistical power will be necessary to be applied to verify these findings. In addition, if researchers want to study the generalization effect, a more specific material should be considered, such as hieratically larger circles, to get a more convincing conclusion.
The current findings support that trauma-unrelated negative stimuli, rather than trauma stimuli, play an important role in PTSD remission. However, most effective therapies for PTSD are trauma-focused.(Helpman et al., 2016; Hinton et al., 2009; Maguen et al., 2019; Mavranezouli et al., 2020) Therefore, negative emotional stimuli (with high commonness) must be taken into account in future therapies. Kaczkurkin et al. also argued that because of the important role of maladaptive generalization in the pathological mechanism of PTSD, future psychotherapy may also involve trauma-like stimulation (generalized stimulation).(Kaczkurkin et al., 2017) Therefore, nontrauma-focused therapy may a future direction of PTSD treatment.
We did not find differences between groups in PFC/ACC or amygdala, which were reported in previous PTSD-remittance studies.(Cisler et al., 2015; Negreira & Abdallah, 2019) This discrepancy may be caused by differences in age, gender, trauma experience, use of ROI and the paradigms applied.(Malejko et al., 2017; Negreira & Abdallah, 2019) A previous PTSD study using trauma-related stimulus found different brain activities in the amygdala and ACC in different genders.(Shin et al., 2004) In the current study, the mixed genders of subjects may erase the effect of some brain regions, especially the amygdala and mPFC. Furthermore, the classic pathological model of PTSD is mostly based on adult studies. However, there should be differences between youths and adults.(Herringa, 2017) Thus, the findings in PTSD research in youths could be different from the classical theory. In addition, considering the special original trauma (earthquake) in the current study, and thus differences in paradigm/stimulus, the results will also be affected. Interestingly, the brain regions revealed by the current study are similar to the only PTSD study on mixed gender youths after an earthquake, which included occipital lobe, hippocampus/parahippocampus, but no prefrontal lobe/ACC (earthquake).(P. Yang et al., 2004)
Our study had several limitations. First, a small sample size was used due to the common difficulties of task-state fMRI studies on natural disaster-induced PTSD. Among all 10 previous related studies, the sample sizes of the PTSD groups ranged from 5–16 subjects, with an average of 10.9 subjects.(Du et al., 2014; Piccardi et al., 2016; P. Yang et al., 2004) Small groups provide relatively low statistical power and make it difficult to strictly control the comorbidity and treatment of subjects; therefore, expanded sample size and controlled comorbidity and treatment are needed in the future to validate the findings of the current study. Second, the statistical thresholding was liberal, which increased the risk for false-positive results. However, it also makes our negative findings more robust, such as the similarity of brain activities between the remitted and PTSD groups under trauma-related conditions. In addition, the P value of GRF correction in this study is less than 0.01, which is enough to make the false-positive rate lower than 0.05, even if the spatial distribution does not satisfy the Gaussian distribution.(Eklund, Nichols, & Knutsson, 2016) Third, the generalizability of the current findings may be limited because of the particularity of earthquake-induced PTSD. Previous studies showed that the type of original trauma affected the heterogeneity of PTSD research.(Boccia et al., 2016; Negreira & Abdallah, 2019) The low "Commonness" of earthquake-specific stimuli may also have a potential impact on the remittance processing of PTSD patients. Future PTSD experiments should measure the "Commonness" of stimuli and study their influence on PTSD. Fourth, this study did not include the questionnaires commonly used in PTSD research, such as the childhood trauma questionnaire (CTQ).(Jiang et al., 2018) Future study should consider multiple questionnaires for a comprehensive evaluation. Last, the earthquake pictures used in this study are from Wenchuan. Although there are no landmark buildings and the ruins are difficult to identify (Fig. 1), there may still be a component of personal relevant experiences. In contrast, the negative pictures from IAPs do not have this component, which could be an issue. However, in some previous studies, similarly, the picture stimulus from the particular original trauma was adopted for symptom provocation, such as the Vietnam war rather than other wars,(Wolf, Miller, & McKinney, 2009) and the mine accidents that subjects experienced instead of other mine accidents.(Hou et al., 2007) Elements related to personally relevant experiences in such pictures may also be types of trauma cues. For example, a trauma cue could be "war-related" or "Vietnam-related" or the combination of both. For the current study, in the between-group comparison, since all the subjects came from Wenchuan, this factor may have relatively little influence. However, it may still complicate comparisons between trauma-related and trauma-unrelated stimuli, especially for the hippocampus, which is responsible for episodic memory. Therefore, in future research, trauma pictures from other regions should be adopted to exclude the potential impact of personally relevant experiences. Moreover, instead of scrambled pictures, future studies could adopt pictures depicting people and/or buildings that are as neutral as possible as a control condition to distinguish between general object recognition processes and more specific affective processes. Furthermore, presentation of stimulus in a fixed order may lead to a risk of systematic spill-over effects. Thus, (quasi-)randomization of blocks or trials could be considered in future studies.