The Prevalence and Comorbidity of PTSD and Depression

Post-Traumatic Stress Disorder (PTSD) is a condition in which exposure to traumatic experience(s) causes prolonged psychological distress, which manifests in four clusters of symptoms, including recurrent intrusive memories of the trauma experienced, avoidance of stimuli and memories related to the trauma, changes in mood and cognition, and an increase in emotional reactivity (American Psychiatric Association, 2013). As many as half (50%) of individuals living with PTSD have a concurrent diagnosis of MDD (Price et al., 2019), characterized by depressed mood, loss of interest or pleasure, tiredness and low energy, and, similarly to PTSD, changes in cognition (American Psychiatric Association, 2013). Despite these troubling findings, minimal research on this comorbidity has been conducted. A central goal of this paper is to elucidate how varying levels of depression symptoms relate to memory abilities in individuals with PTSD.

Cognition and Memory Deficits in PTSD

Deficits in executive functioning refer to any impairment in one’s ability to carry out adaptive behavior, including the ability to plan, update information mentally in working memory, tune out irrelevant stimuli, and self-correct (Rabinovici et al., 2015). PTSD has been linked to significant executive functioning deficits, like attention regulation, working memory, and inhibition of irrelevant stimuli (Aupperle et al., 2012). Although not explicitly listed in the diagnostic criteria, people with PTSD often report memory issues with emotionally neutral material not related to their trauma (Samuelson, 2011). Many studies have corroborated these self-reported memory issues, including a timed test measuring information processing speed and working memory that showed significant impairments in participants with PTSD (Cohen et al., 2013). Furthermore, it has been found that memory deficits in PTSD include both deficits in encoding (Vasterling et al., 1998; Yasik et al., 2007; Pederson et al., 2004) and deficits in delayed recall (Carrión et al., 2010; Cisler et al., 2015; Geuze et al., 2008).

Cognition and Memory Deficits in Depression

Chronic stress promotes the development of depression symptoms and increases the chance of depressive episodes (Monroe & Harkness, 2005), which is a possible explanation for why many of the cognitive deficits described above in PTSD also exist in individuals with depression. Overall, people with depression exhibit impaired memory (Ramponi et al., 2004), but specifically, they have better memory for negative material (Burt et al., 1995) and poorer memory for positive material (Matt et al., 1992). Further, it is known that depressed individuals pay more attention to negative stimuli compared with positive or neutral stimuli and exhibit better memory for negatively-valenced autobiographical events (Mennen et al., 2019).

Memory Deficits in Comorbid PTSD and Depression

PTSD and depression have been linked to significant deficits in learning and memory processes when they occur comorbidly. Comorbid depression symptoms helped to account for the relationship between PTSD and deficits in the ability to remember information in the form of a story and the ability to remember paired words in a sequence (Burriss et al., 2008). Even more, PTSD and depression have been linked to more severe memory impairment when they occur together compared to when they occur alone (Sachinvala et al., 2000; Johnsen et al., 2008; Nijdam et al., 2013). While there is a growing number of studies examining several types of memory impairments in comorbid PTSD and depression, no studies to our knowledge have investigated the relationship between pattern separation abilities and comorbid PTSD and depression. The current study seeks to address this gap in the existing literature.

Hippocampal Involvement in Pattern Separation and Completion

The hippocampus is essential for pattern separation and pattern completion. The cortical projections to the dentate gyrus (DG) and CA3 via the entorhinal cortex (EC) underlie pattern separation, in which distinct representations of things are formed. In contrast, pathways in the CA1 region underlie pattern completion, in which relationships between similar things are recognized and consolidated into an updated memory of that pattern (Norman & O’Reilly, 2003; Becker et al., 2009). Many studies have examined pattern separation performance and corresponding hippocampal activation in normal aging (Reagh et al., 2018; Yassa et al., 2010), mild cognitive impairment (MCI) (Yassa et al., 2010; Sinha et al., 2018; Tran et al., 2017), and Alzheimer’s Disease (AD) (Roher et al., 2017). Each of these conditions has been linked with impairments in lure discrimination ability (pattern separation), and these impairments have further been linked to increased hippocampal activation compared to healthy controls. It has been proposed that this finding could be due to increased mental effort employed by these individuals while completing these tasks (Dillon et al., 2017). Although the above studies are not directly related to hippocampal function in PTSD and depression, they are important because they establish that impairments specifically in pattern separation have been linked with differences in hippocampal function in other conditions.

The Relationship Between Hippocampal Function and PTSD and Depression

One study found that individuals with higher depression scores performed significantly worse on a verbal declarative memory task, and they exhibited reductions in hippocampal activation compared to healthy controls (Bremner et al., 2004). Another study observed that participants with PTSD exhibited impaired performance on the same task, but showed no significant differences in hippocampal activation during the encoding phase compared to controls. However, they did show decreased hippocampal activation during the retrieval phase compared to healthy controls (Carrión et al., 2010). In addition, a different study found that a PTSD diagnosis, while not associated with deficits in associative learning abilities, was associated with increased hippocampal activation during the encoding phase of an associative learning task (Werner et al., 2009). Interestingly, another study looked at the ability to remember unrelated (unassociated) word pairs and found that a PTSD diagnosis was associated with deficits in performance on the task but not with any difference in hippocampal activation during encoding. However, they did exhibit decreased hippocampal activation during the retrieval phase compared to healthy controls (Geuze et al., 2008). These findings suggest that hippocampal activation may not directly correspond with memory performance in the same way for all types of tasks. It is possible that in some cases, greater hippocampal activation during encoding is necessary for individuals with PTSD to perform as well as their healthy counterparts (Joshi et al., 2020). Furthermore, the observed decrease in hippocampal activation during retrieval, coupled with impaired performance, may point to retrieval-specific deficits in PTSD, so when increased activation during the encoding phase is observed, it could mean that the hippocampus is overcompensating for this deficit (Joshi et al., 2020). In another associative learning task, participants with depression exhibited impaired memory for associated word pairs and exhibited a decrease in hippocampal activation during the task compared to healthy controls. (Milne et al., 2012). The findings from this study suggest that higher levels of depression are related to more dysfunction in the hippocampus, which manifests as a decrease in hippocampal activity.

The Present Study

The primary purpose of this study was to examine how depression symptom severity is related to pattern separation performance and hippocampal activation in individuals diagnosed with PTSD. We also examined the relationship between hippocampal activation in PTSD and performance on the MST. Our first research question examined the relationship between pattern separation performance and hippocampal activation in people with PTSD. We hypothesized that we would observe a positive correlation between pattern separation performance and hippocampal activation during pattern separation in PTSD, meaning that poorer pattern separation performance would be related to lower levels of hippocampal activation. This prediction is based on evidence that impaired memory retrieval performance is related to decreased hippocampal activation in PTSD (Carrión et al., 2010; Geuze et al., 2008). Our second research question examined the relationship between pattern separation performance and depression symptom severity in people with PTSD. We hypothesized that we would observe a negative correlation between depression symptom severity and pattern separation performance in PTSD, meaning that higher depression scores would be related to poorer performance. This prediction is based on evidence that suggests that pattern separation is impaired in individuals with high levels of self-reported depression symptoms compared to individuals with lower levels of self-reported depression symptoms (Dohm-Hansen & Johansson, 2020; Shelton & Kirwan, 2013). Our third research question examined the relationship between depression symptom severity and hippocampal activation in people with PTSD. We hypothesized that we would observe a negative correlation between depression symptom severity and hippocampal activation during pattern separation, meaning that higher depression scores would be related to lower levels of hippocampal activation. This prediction is based on evidence that suggests that higher levels of depression were related to lower levels of hippocampal activation in a declarative memory task (Bremner et al., 2004) and an associative memory task (Milne et al., 2012).

Participants

This project involved secondary analysis of data collected as part of a larger study examining hippocampal-dependent learning and memory in people with PTSD and chronic pain. Right-handed adults aged 18 to 45 were recruited through print and online ads throughout the University of Michigan and the surrounding community. All participants included in the current study had PTSD. Eligibility was determined based on a participant’s ability to give informed consent, a current or prior diagnosis of PTSD, and normal or corrected-to-normal vision. Exclusion criteria included the presence of a significant neurological condition, learning disability, active suicidal ideation, drug abuse or dependence in the last year, current use of psychotropic medications known to interfere with our measures of interest, pregnancy, left-handedness or ambidextrousness, history of significant head trauma, or contraindication for fMRI (e.g., metal in the body). After two participants were excluded due to incomplete or poor-quality data, our final sample included 18 participants: 16 (88.9%) women and two (11.1%) men. 15 (83.33%) identified as white, one (5.5%) as black, and two (11.1%) as “Other.” The cohort had an age range of 19–44 (M = 25.7). Depression symptom severity ranged from minimal to severe. This study was approved by the University of Michigan’s Institutional Review Board (HUM00121812; PI: Duval), and all participants provided written informed consent before participation.

Measures

To determine eligibility and quantify symptoms, participants completed the Clinician Administered PTSD Scale for DSM-5 (CAPS-5; Blake & Charney, 1995) to verify PTSD diagnosis, and the Beck Depression Inventory-II (BDI-II; Beck et al., 1961) to quantify the severity of depressive symptoms. Higher scores indicate greater symptom severity (Beck et al., 1961). Of the 18 participants included in the final analysis, 5 (27.7%) reported minimal, 3 (16.6%) reported mild, 5 (27.7%) reported moderate, and 5 (27.7%) reported severe depression symptoms. MRI scans were conducted using a 3.0 Tesla GE Discovery MR750 System (Waukesha, WI) using a 32-channel radiofrequency coil and updated software (Discovery 20.0, Neuro-optimized gradients). MRI scanning included the collection of T1-weighted anatomic images with a 3D MPRAGE sequence (FOV = 256 × 256 mm, slice thickness = 1 mm, 0 mm gap). Functional scans included gradient echo blood oxygen level-dependent (BOLD) scans with standard parameters: TR/TE = 2000/28 ms, flip angle = 90, FOV = 192 × 192 mm, slice thickness = 1.2 mm.

Task Procedures

Over the course of two consecutive days, participants completed a series of tasks to assess learning and memory. The MST was completed on the second day during MRI scanning. The MST is a two-phased task evaluating pattern separation and pattern completion (Stark et al., 2013, 2019). During the encoding phase, participants were shown 192 pictures of everyday objects and prompted to categorize them as indoor or outdoor objects (Figure 1). Immediately after the encoding phase, participants completed the test phase, where they were shown the same images as in the encoding phase (targets), an image that was similar but not identical to an image from the encoding phase (lure), or a new image (foil). Participants were instructed to indicate whether each image was “old” (target), “similar” (lure), or “new” (foil). There were 64 images each in the target, lure, and foil categories. Each image appeared on the screen for 2 seconds, and the interstimulus interval (ISI) was 0.5 seconds (Stark et al., 2013). There were four versions of this task with different images, and participants were randomized to receive one of the four versions. Pattern separation ability is quantified by calculating a lure discrimination index (LDI), which is the difference between the probability of correctly categorizing the similar images as “Similar,” minus the probability of incorrectly categorizing the old images as “Similar” to account for any existing bias a participant could have to use the “Similar” response. (Similar as Similar/Total Similar) - (Old as Similar/Total Old; Stark et al., 2013). This task takes approximately 20 minutes to complete in the MRI scanner.

Figure 1A.

Examples of images presented during the encoding (left) and test (right) phases; 1B: Examples of lure images shown during encoding and test phases. This figure is reproduced from Stark et al., 2013.

Data Processing and Statistical Analysis

We implemented a bivariate correlation using IBM SPSS Statistics (Version 28.0) to test the relationship between depression symptom severity and pattern separation performance. We used Statistical Parametric Mapping (SPM12) software for MATLAB (Version 9.10) to analyze the fMRI data. Functional images were preprocessed using standard methods, including slice-time correction, realignment, and coregistration to the structural images, normalization to the Montreal Neurological Institute (MNI) standard brain, and smoothing with a 2mm kernel. No participants were excluded for motion or sleeping.

To test the first hypothesis that poorer pattern separation performance would be related to lower levels of hippocampal activation during pattern separation in people with PTSD, we conducted a correlation analysis in which LDI score was the predictor variable and hippocampal activation was the outcome variable. To test the second hypothesis that higher depression scores would be related to poorer pattern separation performance, we conducted a bivariate correlation analysis in which BDI score was the predictor variable and LDI Score was the outcome variable. To test the third hypothesis that higher depression scores would be related to lower levels of hippocampal activation, we conducted a correlation analysis in which the BDI score was the predictor variable and hippocampal activation was the outcome variable. Due to our small sample size, the significance threshold for all fMRI analyses was set to a liberal cutoff of p < .01, uncorrected. Thus, all results reporting hippocampal activation should be treated as preliminary.

Relationship Between Hippocampal Activation and Pattern Separation Performance

Figure 2.

Activation only in right hippocampus is positively correlated with LDI (A). Negative correlation between LDI and hippocampal activity (B). Activation only in right hippocampus is negatively correlated with LDI (C).

We hypothesized that there would be a positive correlation between hippocampal activation during the presentation of similar images and pattern separation (LDI). We found that less right hippocampus activation (coordinates [x, y, z] = 29, -22, -10; T = 3.74, p = 0.001 uncorrected; Figure 4A) related to poorer pattern separation performance. We also found in a similar area of the right hippocampus that more hippocampal activation was related to poorer pattern separation performance (coordinates [x, y, z] = 28, -24, -13; T = 4.14, p = 0.000 uncorrected; Figures 4B and 4C). When we extracted beta weights from the activation map around these coordinates and plotted them against pattern separation scores (Figure 4C), we observed a negative relationship, suggesting that more voxels in this area exhibit activation that is negatively correlated with pattern separation performance.

Relationship Between Depression Symptoms and Pattern Separation Performance

Contrary to our hypothesis, there was no significant relationship between depression symptoms and pattern separation (p = 0.357).

Relationship Between Hippocampal Activation and Depression Symptoms

Figure 3.

Activation in left hippocampus is negatively correlated with BDI score (A). Negative correlation between BDI score and bilateral hippocampal activity (B). Activation in right hippocampus is negatively correlated with BDI score (C).

We hypothesized that there would be a negative correlation between hippocampal activation during the presentation of similar images and depression symptoms. We found that less activation in the left hippocampus (coordinates [x, y, z] = -31, -20, -13; T = 5.06, p = 0.000 uncorrected; Figure 5A and 5B); and right hippocampus ([x, y, z] = 29, -25, -14; T = 3.12, p = 0.003 uncorrected; Figures 5B and 5C) was associated with more severe depression. These findings should be treated as preliminary, due to our liberal cutoff of p < .01, uncorrected.

Relationship between Hippocampal Activation and Pattern Separation Performance

Our first hypothesis was that we would find that poorer pattern separation performance would be related to lower levels of hippocampal activation in people with PTSD. This hypothesis was partially supported with a notable caveat. There was evidence that poorer pattern separation performance was related to both lower and higher levels of activation in the same region of the hippocampus, with the evidence primarily suggesting that poorer pattern separation performance was associated with greater hippocampal activation. Given our small sample size, liberal threshold for determining significance, and inconsistent findings, this result should be interpreted with caution. In addition, prior literature is also inconclusive on the relationship between memory performance and hippocampal activation in PTSD. There is one study that found that a PTSD diagnosis was associated with increased hippocampal activation during an associative learning task, but this increased hippocampal activation was not found to be related to impaired performance on the task (Werner et al., 2009). Other research suggests impaired pattern separation performance is related to decreased hippocampal activation in PTSD. For example, multiple studies have found that impaired memory retrieval performance is related to decreased hippocampal activation in PTSD (Carrión et al., 2010; Geuze et al., 2008). There is also evidence that there is no relationship between hippocampal activation and memory performance in PTSD, which has been shown by the finding that there is no difference in hippocampal activation during memory encoding when there are deficits in performance during retrieval (Carrión et al., 2010).

There has been much more research conducted on pattern separation and memory performance in other conditions in which the hippocampus is altered, including normal aging (Reagh et al., 2018; Yassa et al., 2010), MCI (Yassa et al., 2010; Sinha et al., 2018; Tran et al., 2017), and AD (Roher et al., 2017). In each of these conditions, pattern separation impairment has been linked with increased hippocampal activation. While the reason for this is currently unknown, some have postulated that the increased hippocampal activation is due to increased mental effort required to complete pattern separation (Dillon et al., 2017). There is also some theoretical support for the idea that increased hippocampal activation coupled with poorer pattern separation could be explained by a decreased ability of the vmPFC to exert top-down inhibition on the hippocampus (Lanius et al., 2010). Interestingly, findings of reductions in activation in the vmPFC (Quirk & Beer, 2006; Etkin & Wager, 2007; Koenigs & Grafman, 2009), the size of the vmPFC (Rauch et al., 2003; Milad et al., 2005), and gray matter in the vmPFC (Richert et al., 2006; Kasai et al., 2008) in individuals with PTSD are well-replicated, lending support to the idea that the vmPFC could be implicated in pattern separation in PTSD.

Relationship between Depression Severity and Pattern Separation Performance

Our second hypothesis was that higher depression scores would be associated with poorer pattern separation performance, and this hypothesis was not supported by our results. This finding was unexpected and in contrast with previous findings that pattern separation performance on the MST is worse in individuals with high levels of self-reported depression symptoms compared to individuals with lower levels of self-reported depression symptoms (Dohm-Hansen & Johansson, 2020; Shelton & Kirwan, 2013). Our findings are also inconsistent with previous research that comorbid depression symptoms helped to account for impairments in declarative memory in individuals with PTSD, and that general memory impairments in PTSD were significantly worse when PTSD and depression were experienced together versus PTSD alone (Sachinvala et al., 2000; Johnsen et al., 2008; Nijdam et al., 2013). However, a recent study reported that individuals with low depression symptoms showed deficits specifically in pattern separation, while individuals with high depression symptoms did not exhibit these deficits in pattern separation (Grupe et al., 2022). In fact, when controlling for perceived stress, they found that participants with higher levels of depression actually had enhanced pattern separation performance based on lure discrimination ability. The proposed explanation for this finding by Grupe et al. (2022) is that enhanced LDI could be evidence of perceptual biases. For example, individuals with high levels of depression could perceive neutral MST imagery as negative more often with increasing depression symptoms. Since it is known that people with high levels of depression have enhanced memory for negative imagery (Mennen et al., 2019), this could make sense.

Relationship between Depression Severity and Hippocampal Activation during Pattern Separation

Our third hypothesis was that higher depression scores would be related to lower levels of hippocampal activation, and this was supported by our results in both the left and right hippocampus. This finding is consistent with evidence that higher levels of depression were related to lower levels of hippocampal activation during a declarative memory task (Bremner et al., 2004) and an associative memory task (Milne et al., 2012). There is no evidence to our knowledge that depression has ever been linked with hyperactivation in the hippocampus during any memory task. This suggests that higher levels of depression are related to more dysfunction in the hippocampus, which manifests as a decrease in activity. Similarly to what has been found in PTSD, dysfunction in the prefrontal cortex could underlie the decreased activation observed with increasing depression symptoms in our sample. There have been findings of a significant increase in ventrolateral prefrontal cortex (vlPFC) activity (Etkin et al., 2015) and increased functional connectivity between the vlPFC and the hippocampus in depression (Hao et al., 2020). It is possible that in depression, the dysfunctional prefrontal cortex is impaired in its ability to exercise control over the hippocampus.

Limitations

Our small sample size of 18 participants and the lack of racial and gender diversity in our sample limit the applicability of our study. A larger sample of people with comorbid PTSD and depression is needed to determine if these findings could be considered generalizable to the larger population of people with these comorbid conditions. Furthermore, having a small sample size was compounded by the fact that 88.9% of our participants identified as women and 83.3% identified as white. While it is true that women are more than twice as likely as men to have PTSD in the United States (Olff, 2017), our results may not generalize to the entire US population with PTSD. Additionally, it has been shown that Black individuals in the US are significantly overrepresented in national prevalence rates for PTSD (Roberts et al., 2011). Thus, our findings from a predominantly White sample are likely not representative of Americans with PTSD.

Conclusions

We found that impaired pattern separation performance was related to higher levels of hippocampal activation, but this finding was inconsistent. We also found no evidence that higher depression scores were associated with poorer pattern separation performance. Lastly, we found that higher depression scores were associated with lower levels of hippocampal activation. Overall, these results suggest hippocampal dysfunction in people with PTSD and comorbid depression symptoms. These results are the first to our knowledge that have examined pattern separation performance and hippocampal activity in people with PTSD and depression. Therefore, they highlight the importance of studying comorbid PTSD and depression and how the combination of these two conditions could relate to memory impairments and brain activity differently than each one alone.