Skip to main content
NCBI home page
Translational Psychiatry logoLink to Translational Psychiatry
. 2026 Jul 23;16:573. doi: 10.1038/s41398-026-04289-8

Dynamic gene expression signatures predict psychotherapy response in PTSD

Changxin Xu 1, Heather N Bader 1,2, Janine D Flory 1,2, Linda M Bierer 1, Mitali Chattopadhyay 1,2, Frank Desarnaud 1,2, Ruoting Yang 3, Amy Lehrner 1,2, Iouri Makotkine 1,2, Rachel Yehuda 1,2,✉
PMCID: PMC13680502  PMID: 42897160

Abstract

Static blood-based biomarkers provide limited insight into psychiatric illness and recovery, largely because they capture only a single, unperturbed snapshot of inherently dynamic biological systems. In this study, we applied a dynamic, longitudinal dual-state approach to characterize peripheral blood mononuclear cell (PBMC) transcriptional trajectories associated with psychotherapy outcomes in post-traumatic stress disorder (PTSD). PBMC transcriptional profiles were generated for veterans with PTSD (n = 46, 21, and 17, respectively) and controls (n = 26, 26, and 23) at pre-treatment, post-treatment, and follow-up. The PTSD cohort with profiles across all three timepoints was divided into responders (n = 6) and non-responders (n = 11) based on their Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) trajectories. All subjects were profiled in both unstimulated and dexamethasone-stimulated states. At pre-treatment, unstimulated responders and non-responders exhibited pronounced transcriptional differences exceeding those observed between PTSD and controls. Longitudinally, unstimulated responders showed a marked shift toward control-like profiles, whereas non-responders retained persistent divergence. Pathway analyses identified a prominent temporal signal within Class C/3 G-protein-coupled receptor pathways driven largely by bitter taste receptors (TAS2Rs), along with subgroup-specific regulation of GABAB, opioid, and NMDA receptor signaling. Dexamethasone-induced responses further differentiated subgroups and showed opposing associations with recovery trajectories, implicating glucocorticoid sensitivity as a potential predictor of treatment outcomes. These findings indicate that biological responsiveness itself constitutes a measurable phenotype of recovery, and that dynamic functional profiling of living immune cells can reveal clinically relevant trajectories and candidate mechanisms not accessible through static biomarkers.

Subject terms: Predictive markers, Prognostic markers, Human behaviour

Introduction

Diagnosing the presence of posttraumatic stress disorder (PTSD) and assessing its severity, is generally done on the basis of clinical interviews—an approach that depends on the insight and transparency of the person being evaluated and the skill of the evaluator. The impetus for developing biomarkers for PTSD arose not only from the need to find objective ways to assess the disorder and track recovery, but also from the recognition that such information might uncover the biological mechanisms that give rise to, or sustain, symptoms.

While compelling in concept, it has proven difficult to develop reliable biomarkers for PTSD. Although neuroimaging has yielded important insights into the neural circuits affected in PTSD, these findings are not specific to PTSD and remain limited to guide individualized treatment [1]. Peripheral blood, by contrast, offers molecular and cellular information that complements brain imaging, providing access to genome-wide gene expression patterns, upstream regulators, downstream inflammatory and metabolic products. Yet despite its accessibility, blood-based findings have been noisy, inconsistent, and difficult to interpret [2–4]. One reason for this may be that blood biomarkers, in addition to being peripheral, are typically derived from static samples that capture only a single point in time.

This limitation motivated us to develop a functional biomarker assay that probes the transcriptomic landscape of peripheral cells both longitudinally and across two distinct conditions: the unstimulated and the stimulated states. By removing the confounding “noise” of the complex in vivo milieu, these profiles from in vitro cultured peripheral blood mononuclear cells (PBMCs) become more comparable across subjects and timepoints. Complementing this, the stimulation of PBMCs with dexamethasone (DEX), a synthetic glucocorticoid, allows us to activate dynamic biological pathways involved in stress response. This longitudinal dual-state framework represents an evolution of a prior approach we developed using live lymphocytes to assess lysozyme activity following DEX exposure [5–7]. We enhanced this strategy by applying transcriptome-wide profiling, and have recently validated the method by demonstrating concordant (and discordant) results in DEX-stimulated PBMCs and induced neurons derived from the same individuals [8, 9].

In the current study, we apply this novel biomarker method to address a different question: What biological changes accompany recovery from PTSD? To do this, we examined a longitudinal cohort of individuals undergoing cognitive processing therapy (CPT)—an evidence-based psychotherapy that does not involve pharmacological intervention [10, 11]. This design allowed us to isolate biological changes associated with clinical recovery, free from the confounding effects of medication. To address the broader question of transcriptomic stability over time, we also included a trauma-exposed control group of combat veterans without PTSD—an often-overlooked comparison group in biomarker research.

We hypothesized that:

  1. There would be baseline transcriptional differences between PTSD patients and control subjects;

  2. There might be baseline differences between treatment responders and non-responders that could predict clinical outcome;

  3. Transcriptomic measures would preferentially change over time in the responder group compared to non-responders and controls.

This study was designed as a focused pilot to evaluate the potential utility of a dynamic, stimulation-based biomarker in identifying molecular pathways associated with treatment response and recovery from PTSD. By targeting patients undergoing a non-biological therapy and adding a longitudinal, trauma-exposed control group, we aimed to clarify not only which pathways change during recovery, but also whether these changes reflect treatment-specific effects or broader biological variability.

Materials and methods

Participants

A total of 92 combat-exposed veterans (55 with current PTSD and 37 trauma-exposed controls) were enrolled in the study following screening. All participants were 23–62 years of age (36.3 ± 8.38), had a history of deployment-related trauma, and were free of current substance use disorders, moderate to severe traumatic brain injury (TBI), or other psychiatric conditions requiring immediate clinical intervention. Exclusion criteria included a lifetime history of psychotic disorder, bipolar disorder, or obsessive-compulsive disorder, assessed by a clinical psychologist using the Structured Clinical Interview for DSM-5 (SCID-5) [12].

42 participants (23 controls and 19 PTSD) completed all three assessments, while 8 participants (3 controls and 5 PTSD) missed the last assessment. The remaining 42 participants (31 PTSD and 11 controls) discontinued participation during treatment, primarily due to disruptions related to the COVID-19 pandemic.

PTSD diagnoses were confirmed using the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) [13], and a battery of clinical and biological assessments. All procedures were approved by the Institutional Review Board (IRB) of the JJP VAMC, and written informed consent was obtained from all participants.

Study design and procedures

This is a longitudinal observational study consisting of three timepoints: pre-treatment (T1), post-treatment (T2), and follow-up (12 weeks post-treatment; T3). The PTSD group received 12 sessions of Cognitive Processing Therapy (CPT), delivered by trained clinicians following standardized protocols [14]. Clinical assessments and blood draws were conducted at all three timepoints. Control participants underwent the same clinical and biological assessments at matched time intervals.

The longitudinal design allowed us to assess group differences, identify potential predictors of treatment outcomes, and examine whether stimulated transcriptional responses changed preferentially in treatment responders compared to non-responders and controls.

Clustering of PTSD recovery trajectories

Exploratory data-driven clustering was performed using Latent Growth Mixture Modeling (LGMM) to identify PTSD subgroups based on trajectory changes of CAPS-5 scores to provide a descriptive characterization of response heterogeneity. The lcmm R package (v2.2.1) was used for this analysis [15]. The optimal number of latent classes was determined using a combination of fit indices, including the Bayesian Information Criterion (BIC) for model parsimony, entropy for classification precision, and the bootstrapped likelihood ratio test (BLRT) to compare k versus k − 1 class solutions [16–18]. Only PTSD participants who completed all assessments were included (n = 19).

PBMC isolation, cell culture and DEX treatment

PBMCs were isolated and stimulated in vitro with DEX to probe stress-sensitive transcriptional responses. As previously demonstrated, 2.5 and 5 nM DEX were utilized as suitable concentrations for PBMC stimulation without inducing undue cytotoxicity [8, 9]. For each DEX concentration, 18 wells were prepared to ensure sufficient RNA yield. After 72 hours of culture, cells were harvested in 100 μL TRIzol reagent and stored at -80°C.

PBMC samples available for downstream processing included 46 PTSD participants (46 at T1, 21 at T2, and 17 at T3) and 26 controls (26, 26, and 23, respectively). The reduction from the total clinical enrollment was due to technical factors, including blood draw failures and cell culture complications.

RNAseq data analysis

Raw count data were imported and analyzed with the Bioconductor limma [19] package (v3.64.0) using limma-voom framework for data normalization and transformation. Immune cell type frequencies were estimated using the CIBERSORTx cell type deconvolution tool [20]. The cibersort result was grouped into lymphocyte, monocyte, macrophage, and others. The variancePartition [21] R package (v1.38.0) was used to assess the proportion of variance attributable to biological and technical factors. Major variables, including lymphocyte, age, and gender, were used as confounding variables. As a repeated measures design, the individual variable was used as a blocking variable. Gene set enrichment analysis (GSEA) was performed using the fgsea [22] R package (v1.34.0), with Reactome [23] pathway gene sets. -log10(Pvalue)*sign(logFC) was used as the ranking metric. Signed co-expression networks were constructed using the WGCNA [24] R package (v1.73) to help interpret pathway enrichment results. Module eigengenes were then correlated with time, PTSD diagnosis, and responder classification. For consistency, Reactome pathway gene sets were also used for module functional annotation, using the WebGestaltR [25] R package (v0.4.6).

Statistical analysis

Data analyses were performed using R version 4.5.0. Descriptive statistics were calculated for study variables. Continuous variables were summarized with means ± standard deviations (SD) and t-test. Categorical variables were summarized with frequencies and χ2 tests. The Wilcoxon signed-rank test was used to compare the distributions of |logFCs| values.

Results

Clinical and demographic characteristics

Ninety-two combat-exposed veterans were enrolled in the study: 55 participants met diagnostic criteria for PTSD (PTSD + ), and 37 trauma-exposed veterans without PTSD served as controls. Clinical and demographic characteristics are summarized in Table S1.

There were no significant differences between PTSD+ and control groups in gender distribution (χ² = 0.48) or racial composition (χ² = 0.122). Body mass index (BMI) was slightly higher in the PTSD group (30.79 ± 5.61) compared to controls (28.79 ± 4.79), though the difference did not reach statistical significance (p = 0.08).

As expected, PTSD+ participants showed significantly higher scores on the PTSD Checklist (PCL; 54.5 ± 11.7) and CAPS-5 (46.3 ± 5.6), compared with the control group (PCL = 11.3 ± 11.5; CAPS-5 = 4.73 ± 6.6), with both comparisons reaching high statistical significance (p < 2.2e-16).

Clinical outcomes of cognitive therapy

In the PTSD group, CAPS-5 scores significantly declined over the course of CPT, indicating a reduction in PTSD symptom severity:

T1, 47.13 ± 9.79 (24 PTSD + );

T2, 28.54 ± 12.56 (12 PTSD + , 12 PTSD − );

T3, 30.16 ± 13.92 (11 PTSD + , 8 PTSD − ).

This reduction in CAPS-5 scores from T1 to T3 was statistically significant (p = 8.799e-05). In contrast, the control group didn’t show significant change in CAPS-5 scores across timepoints:

T1, 4.73 ± 6.60 (26 PTSD − )

T2, 3.96 ± 4.46 (26 PTSD − )

T3, 3.65 ± 3.77 (23 PTSD − )

To ensure longitudinal comparability of the data, only participants who completed the first two assessments were included.

PTSD participants clustering

Latent growth mixture modeling of longitudinal CAPS-5 trajectories from PTSD participants who completed therapy (n = 19) identified two distinct clinical trajectories. A 2-class solution provided the optimal fit, supported by a significant bootstrapped likelihood ratio test (BLRT; p = 0.024), the lowest BIC, and high classification precision (Entropy = 0.889; Table S2). Responders (n = 8) showed a significantly greater reduction in CAPS-5 scores compared with the non-responder group (n = 11) (−26.7 vs. −12.1 from T1 to T3; Table S3). The longitudinal growth plot (Fig. S1) showed a notable separation between the two classes, which effectively reflected distinct responses to psychotherapy. This plot also revealed high concordance with the T3 diagnosis-based classification. Only two subjects (indicated by dot-dash lines) showed discrepancies, where LGMM more effectively captured the clinical trajectories than a single endpoint diagnosis.

Due to sample attrition during PBMC profiling, we had 17 PTSD participants with profiles across all three timepoints, including 11 non-responders and 6 responders, for subsequent analyses.

Transcriptomic profiles in unstimulated PBMCs at T1

To identify diagnostic and prognostic signals prior to treatment, we first analyzed gene expression differences across the following groups: PTSD (n = 46), controls (n = 26), responders (n = 6), and non-responders (n = 11).

PTSD vs. controls

24 differentially expressed genes (DEGs) were identified between the PTSD and control groups (adj.P.Val < 0.05; Fig. 1A–B), although none exhibited a |logFC| greater than 1 (Table S4). While these DEGs may reflect diagnostic markers or molecular changes associated with PTSD, hierarchical clustering can’t clearly separate PTSD and control samples. GSEA identified a range of significantly elevated immune pathways in the PTSD group (Table S5). In addition, several neurotransmitter pathways showed enrichment approaching statistical significance, including GABAB receptor activation (padj = 0.06), opioid signaling (padj = 0.17), and negatively enriched Class C/3 (Metabotropic glutamate/pheromone receptors) (padj = 0.06) (Fig. 1C).

Fig. 1. Group comparisons of unstimulated PBMCs at pre-treatment stage.

Fig. 1

Open in a new tab

A–B The number of DEGs of group comparisons. PTSD: PTSD vs. controls; rVSnr: responders vs. non-responders; nrVSctl: non-responders vs. controls; rVSctl: responders vs. controls. C Heatmap of enrichment of representative neural pathways. *: 0.05 < = padj < 0.2; **: padj < 0.05. D Hierarchical clustering of the PTSD group by top DEGs between responders (r) and non-responders (nr). E LogFC concordance analysis between the responders vs. controls and non-responders vs. controls contrasts. F Enrichment analysis of Class C/3 (Metabotropic glutamate/pheromone receptor) pathway. G TAS2Rs expression in groups at T1 without stimulation.

Responders vs. non-responders

Next, we compared baseline gene expression between treatment responders and non-responders, and identified 248 DEGs (padj < 0.05). A heatmap constructed using 43 DEGs with |logFC | > 1.5 showed good separation except for one subject (Fig. 1D). GSEA revealed many enriched immune pathways in non-responders, including hemostasis, neutrophil degranulation, and Toll-like receptor cascades. Additionally, several neurotransmitter pathways were differentially enriched, including enriched opioid signaling, GABAB receptor activation (padj = 0.08), dopaminergic neurotransmission, and NMDA receptor activation (padj = 0.13) in non-responders. By contrast, responders showed potential enrichment in Class C/3 (Metabotropic glutamate/pheromone receptors) (padj = 0.08) (Fig. 1C).

Responders/non-responders vs. controls

Finally, responders and non-responders were each compared with the control group (Fig. 1A–B). DEGs from the non-responder vs. control comparison showed strong overlap with both the PTSD vs. control and responder vs. non-responder comparisons, whereas the responder vs. control comparison had the fewest DEGs. Neurotransmitter pathway analysis revealed opposite enrichment patterns between the responder vs. control and non-responder vs. control comparisons (Fig. 1C), especially on opioid signaling, GABAB receptor activation, and Class C/3 (Metabotropic glutamate/pheromone receptors). LogFC concordance analysis between responders vs. control and non-responder vs. control comparisons confirmed the low concordance (Fig. 1E).

Class C/3 (Metabotropic glutamate/pheromone receptors) is a class of G-protein coupled receptors that includes the metabotropic glutamate receptors and several other receptors. It’s crucial to determine which specific Class C/3 receptors were enriched in our analysis. In addition to the significant negative enrichment in the non-responders vs control comparison, the responders vs non-responder comparison also showed enrichment close to significance (padj < 0.2) (Fig. 1F). The leading edge genes from these two comparisons are mostly overlapping, and all belong to the taste 2 receptors (TAS2Rs), a family of bitter taste receptors, including TAS2R13, TAS2R31, TAS2R20 and others. Expression analysis confirmed that non-responders had the lowest expression of these TAS2Rs compared with both the control and responder groups (Fig. 1G).

CPT treatment reversed PTSD at the unstimulated transcriptional level

We then examined whether transcriptional profiles changed over time in a way that reflected recovery from PTSD. To do this, we performed a logFC concordance analysis between the T1 vs. T2 (and T3) contrasts and the T1 PTSD vs. control contrast at the unstimulated level.

A negative correlation in logFC between the PTSD group T2 vs. T1 contrast and the T1 PTSD vs controls contrast (Fig. 2A) suggested that CPT shifted expression in a direction opposite to the disease profile—reversing the molecular signature of PTSD. This trend was even more pronounced at T3, 12 weeks after therapy ended, reinforcing the idea that these changes persisted and were consistent with sustained recovery.

Fig. 2. CPT treatment reverses PTSD on transcriptional level.

Fig. 2

Open in a new tab

A LogFC concordance analysis between PTSD vs. controls and PTSD temporal change. T21: T2 vs. T1; T31: T3 vs. T1. B LogFC concordance analysis between PTSD subgroups vs. controls and subgroups temporal change.

In Fig. 2B, we further stratified the PTSD group into responders and non-responders, and assessed how each subgroup’s longitudinal transcriptional changes related to their differences from controls at T1. Both subgroups showed significant negative correlations between the T3 vs. T1 contrast and the T1 PTSD subgroup-control differences (responders: r = −0.59; non-responders: r = −0.61).

Longitudinal transcriptional dynamics in unstimulated PBMCs

We next examined how transcriptional profiles changed longitudinally within each group.

PTSD and control groups

Distinct patterns were observed between the PTSD (n = 46 at T1, 21 at T2, and 17 at T3) and control groups (n = 26, 26, and 23, respectively).

From T1 to T2, the control group showed 160 DEGs (adj.p.Val < 0.05), while the PTSD group showed a much larger transcriptional response, with 5476 DEGs (Fig. 3A). At T3, the number of DEGs increased in the control group to 1004, whereas no significant DEGs were detected in the PTSD group. The sharp dynamic changes revealed the strong impact of CPT on the PTSD group, and also suggested that the acute transcriptional response to CPT may not be sustained. In contrast, the control group showed a gradual increase in responsive genes across timepoints, suggesting the trauma-exposed control group may experience normal fluctuations or effects of study engagement.

Fig. 3. Unstimulated transcription profiles over time: Controls and PTSD.

Fig. 3

Open in a new tab

A The number of DEGs over time. T21: T2 vs. T1; T31: T3 vs. T1. B LogFC concordance analysis between PTSD and controls on temporal change. C Heatmap of longitudinal pathway enrichment of representative neural pathways in controls (Ctl) and PTSD. D TAS2Rs expression over time in the controls and PTSD subgroups.

Despite the differences in DEG counts, logFC concordance analysis showed positive correlations between the PTSD and control groups (r = 0.4 for T2 vs T1 and r = 0.42 for T3 vs T1; Fig. 3B), suggesting shared directional gene expression trends over time. The high positive correlation may explain why only 3 differentially responsive genes (DRGs) were identified for T2 vs. T1 and none for T3 vs. T1.

Pathway-level analyses supported this positive correlation. Both groups shared numerous significantly negatively enriched pathways, many related to cell cycle and immune pathways (Table S6). Among neurotransmitter signaling pathways (Fig. 3C), the control group showed continued enrichment of Class C/3 receptors and negative enrichment of opioid, NMDA, and AMPA signaling pathways over time, while the PTSD group exhibited the peak enrichment at T2. At T3, both groups showed a similar enrichment pattern compared with T1.

Expression analysis of the leading edge genes from Class C/3 (Metabotropic glutamate/pheromone receptors) pathway again showed that TAS2Rs were primary drivers of enrichment, and their expression levels increased over time in both groups (Fig. 3D).

Responders and non-responders

We next examined gene expression changes across time within the responder and non-responder subgroups of the PTSD cohort. Compared with T1, the non-responder group exhibited over 2,000 differentially expressed genes at T2, and dropped sharply to 49 DEGs at T3, a pattern that paralleled the broader PTSD group (Fig. 4A). The responder group, on the other hand, showed fewer but gradually increased DEGs from 1 to 33. LogFC concordance analysis between responders and non-responders revealed shifting correlations of 0.508 for T2 vs T1, and −0.23 for T3 vs. T1 (Fig. 4B). These values suggest divergent transcriptional responses between the two groups, particularly during follow-up.

Fig. 4. Unstimulated transcription profiles over time: responders and non-responders.

Fig. 4

Open in a new tab

A The number of DEGs over time in responders (r) and non-responders (nr). T21: T2 vs. T1; T31: T3 vs. T1. B LogFC concordance analysis between responders and non-responders on temporal change. C Potential prognostic marker, and differentially responsive genes (DRGs) between responders and non-responders on T3 vs. T1 (T31). D NES concordance analysis between responders and non-responders on temporal change. E Heatmap of longitudinal pathway enrichment of representative neural pathways in responders (Res) and non-responders (Nonres). F LogFC concordance analysis between the T2 vs. T1 (T21) and T3 vs. T2 (T32) comparisons within responders and non-responders. G DEGs of PTSD vs. controls (PTSDvsCtl) and responder vs. non-responders (resVSnon-res) at all timepoints. H Heatmap of pathway enrichment of representative neural pathways between groups at T3.

420 DRGs were identified for T3 vs T1 (none for T2 vs. T1), which further confirmed the divergence. Ninety of these overlapped with previously identified potential prognostic genes (Fig. 4C).

Pathway-level NES concordance analysis confirmed the difference between the responders and non-responder groups: T3 vs. T1 showed a low negative correlation (r = −0.107) (Fig. 4D). While Class C/3 (Metabotropic glutamate/pheromone receptors) signaling was enriched in both groups, GABAB receptor activation, opioid signaling, dopamine neurotransmitter release, and NMDA receptor signaling showed opposite enrichment directions (Fig. 4E), confirming the divergence of transcriptional trajectory.

Finally, logFC correlation analysis between T2 vs. T1 and T3 vs. T2 revealed negative correlations in responders and non-responders (Fig. 4F), indicating that the groups underwent distinct transcriptional changes during and after CPT treatment. These differences were more pronounced in non-responders.

Group comparison at T3

To evaluate the impact of CPT therapy on group-level differences, we compared transcriptional profiles between groups at each timepoint. The number of DEGs between the PTSD vs Control and responder vs non-responder comparisons markedly decreased by T3 (Fig. 4G), suggesting a reduction in group-level differences following treatment. Despite this general convergence, group differences persisted at the pathway level. These differences were consistent with residual symptom differences as reflected in CAPS-5 scores.

Importantly, neurotransmitter pathway enrichment analysis revealed ongoing divergence between responders and non-responders from T1 to T3 (Fig. 4H). At T3, the non-responder group continued to show enriched neurotransmitter pathways, including GABAB receptor activation, opioid signaling and NMDA receptor signaling pathways compared with controls, which is similar to observations at T1 (Fig. 1C), suggesting a weak response to CPT therapy. In contrast, the responder group exhibited substantial shifts. Relative to controls, many pathways showed opposite enrichment at T3, suggesting a potential normalization of molecular activity following successful treatment.

Weighted gene co-expression network analysis (WGCNA) on unstimulated PBMCs

A WGCNA co-expression network analysis was performed to identify coordinated patterns of gene expression associated with time, PTSD status, and treatment responsiveness in unstimulated samples and across all timepoints. A total of 18906 genes were clustered into 30 co-expression modules. Module eigengene (ME) values were used to find correlated modules to time, PTSD diagnosis, and treatment responsiveness within the PTSD subgroup (Table S7).

For temporal association, Module 1 showed the strongest association with time, and was significantly enriched for the Class C/3 (Metabotropic glutamate/pheromone receptors) pathway, and sensory perception processes. Thirteen TAS2Rs were identified in this module, which is consistent with our finding that the Class C/3 (Metabotropic glutamate/pheromone receptors) pathway enrichment was mainly driven by TAS2Rs receptors.

Among PTSD-associated modules, Module 26 and 16 both relate to immune regulation, while Module 15 was enriched for G-protein-coupled receptor signaling, including activation of GABAB receptors and opioid signaling, both of which showed enrichment in our group differential analyses.

For treatment responsiveness, Module 25, 21, and 15 showed the strongest association. Module 25 and 21 were enriched for immune-related signaling. Module 15 overlapped with the PTSD-associated network.

DEX stimulation on PBMCs

Similar analyses were performed on DEX-stimulated samples.

At T1: we observed more DEGs (n = 248) between PTSD and controls at 5 nM DEX than in the unstimulated state. However, a high concordance between the 5 nM DEX and unstimulated states indicated persistent differences across both conditions (Fig. 5A). High concordance was also observed between responders and non-responders (Fig. 5B). Comparing stimulated (Fig. 5C) and unstimulated profiles (Fig. 1C), neurotransmitter signaling analysis showed overall stronger enrichment in the stimulated condition within PTSD subgroups—especially responders—relative to controls, suggesting varied glucocorticoid responsiveness among groups.

Fig. 5. Dex stimulation on PBMCs at T1.

Fig. 5

Open in a new tab

A FC concordance analysis between 0 nM DEX (unstimulated) and 5 nM DEX stimulated-PBMCs on PTSD vs. Ctl. B FC concordance analysis between 0 nM DEX and 5 nM DEX stimulated-PBMCs on responders vs. non-responders (ResVSnonRes). C Heatmap of group comparisons in 5 nM DEX-stimulated PBMCs at T1. D Heatmap of enrichment of representative neural pathways on DEX stimulation in the controls and PTSD groups. D41: DEX 2.5 nM vs. unstimulated; D51: DEX 5 nM vs. unstimulated. E Heatmap of enrichment of representative neural pathways on DEX stimulation in responders (res) and non-responders (nonres). F–G The expression level of FKBP5 and NR3C1 at T1 in PTSD and control; responders and non-responders. D0nM: unstimulated; D2.5 nM: DEX 2.5 nM; D5nM: DEX 5 nM. H–J FC concordance analysis between DEX effect and PTSD recovery within PTSD, non-responders (non-res), and responders (Res). D2.5nM: DEX 2.5 nM vs. unstimulated; T21: T2 vs. T1; T31: T3 vs. T1.

To confirm this, we compared DEX effects on neurotransmitter pathways between groups. While the PTSD and control groups (Fig. 5D) showed no significant difference on DEX stimulation, responders showed stronger enrichment than non-responders (Fig. 5E). This included significantly enriched GABAB, opioid, and NMDA receptor signaling, confirming stronger glucocorticoid responsiveness in responders. We also examined the expression levels of the glucocorticoid responsive gene FKBP5 and the glucocorticoid receptor (GR) gene NR3C1, genes highly related to GR signaling [26]. As expected, FKBP5 upregulation and NR3C1 autoregulation occurred in a dose-dependent manner (Fig. 5F–G). While FKBP5 and NR3C1 levels were similar between PTSD and controls, NR3C1 expression was slightly lower in non-responders at all DEX levels compared with responders although not significant.

Importantly, we assessed whether DEX-response profiles at T1 correlated with PTSD characteristics and longitudinal transcriptional changes. In the PTSD group, logFC concordance analysis showed a positive correlation of the DEX effect with the PTSD vs. control comparison at T1, suggesting a possible negative effect on PTSD recovery. Consistently, DEX effects were negatively correlated with the T2 vs. T1 and T3 vs. T1 comparisons (Fig. 5H). In responders, DEX effects were positively associated with recovery trajectories (Fig. 5J), while in non-responders, they were negatively associated (Fig. 5I), suggesting that baseline glucocorticoid responsiveness may be associated with therapeutic response.

Longitudinally: the stimulated PTSD group showed a strong response to CPT that then subsided during follow-up (Fig. 6A), a pattern similar to that observed in the unstimulated state. High concordance in temporal change was observed between the unstimulated and stimulated conditions, especially at the T2 vs. T1 comparison in the CPT-treated PTSD group [Fig. 6B], indicating strong and persistent CPT responses across conditions. Neurotransmitter pathway analysis again revealed a distinct enrichment pattern in responders at T3 vs. T1 compared with non-responders and controls (Fig. 6C), with a similar pattern observed at the unstimulated level [Fig. 4E]. However, non-responders showed less negative enrichment, while responders showed attenuated positive enrichment under the stimulated state. To understand the basis of these opposite directional changes, glucocorticoid responsiveness was compared longitudinally to evaluate whether specific responsiveness trajectories differentiate responders from non-responders.

Fig. 6. Glucocorticoid responsiveness over time.

Fig. 6

Open in a new tab

A The number of DEGs over time at 5 nM DEX within controls (Ctl) and PTSD. T21: T2 vs. T1; T31: T3 vs. T1. B FC concordance analysis between the unstimulated and the 5 nM DEX-stimulated PBMCs on longitudinal changes within the PTSD and control groups. C Heatmap of longitudinal pathway enrichment in 2.5 nM DEX-stimulated PBMCs. ctl: controls; nonres: non-responders; res: responders. D The expression of FKBP5 and NR3C1 over time in unstimulated PTSD and control groups. E The expression of FKBP5 and NR3C1 over time in unstimulated responder and non-responder groups. F DEX responsiveness in the non-reponders (NR) and responders (R) groups across timepoints. P values were evaluated by Wilcox.test on their |logFC| values.

The expressions of FKBP5 and NR3C1 were tracked longitudinally at the unstimulated level (Fig. 6D–E). NR3C1 showed a consistent downregulation from T1 to T2 in all groups, suggestive of enhanced glucocorticoid receptor autoregulation in treatment-responsive individuals. Group differences were revealed from T2 to T3, with continued downward regulation in responders, while upward in non-responders. FKBP5 expression levels also showed group differences over time. From T1 to T2, FKBP5 expression changed in different directions between the control and PTSD groups.

The number of DEGs and the logFC distribution were compared to evaluate their DEX responsiveness across timepoints (Fig. 6F) within PTSD subgroups. Both responders and non-responders showed significantly lower responsiveness at T2, which is consistent with decreased NR3C1 expression. At T3, responders showed a continued downward trend in responsiveness, while the DEX responsiveness rose in non-responders. This distinct pattern suggests that long-term suppression of glucocorticoid responsiveness may be one of the characteristics of responders, and raises the possibility of using glucocorticoid responsiveness to track a patient’s recovery trajectory.

Discussion

This study introduces a novel, longitudinal dual-state approach to characterize biological changes in PTSD through dynamic transcriptional profiling of PBMCs. This method revealed significant time-dependent transcriptional changes following cognitive therapy. A key strength of this longitudinal design is the inclusion of a trauma-exposed control group assessed at matched intervals. Across analyses, responders showed a marked shift in transcriptional profile that moved closer to the profile of controls, suggesting that recovery from PTSD is accompanied by measurable normalization at the transcriptional level. The differences between responders and non-responders in transcriptional trajectories, especially at the follow-up stage, highlight the heterogeneity within PTSD and underscore the potential of this approach to distinguish individuals on different recovery trajectories.

One of the most striking findings was the divergence between responders and non-responders, even at baseline. Differences in gene expression and pathway enrichment prior to treatment suggest that these two groups may represent biologically distinct subtypes of PTSD. This was also apparent in neurotransmitter pathways—particularly GABAB receptor activation, opioid signaling, and Class C/3 (Metabotropic glutamate/pheromone receptors) activity. These findings reinforce the growing recognition that PTSD is not a single disorder but a heterogeneous syndrome, and raise the possibility that transcriptional profiling could help predict who is likely to benefit from therapy.

The Class C/3 (metabotropic glutamate/pheromone) pathway was consistently enriched across groups and timepoints. Our results show that TAS2Rs are the primary drivers of this enrichment. These receptors are expressed broadly outside the gustatory system, including in lymphocytes, monocytes, and macrophages, where they act as sensors for endogenous or environmental stressors, capable of dampening inflammatory responses—including the release of TNF-ɑ and other pro-inflammatory mediators—through GPCR-mediated signaling [27–33].

Their upregulation over time in our study, especially in responders, may reflect an adaptive mechanism for constraining peripheral inflammation during recovery. Conversely, the relative downregulation of TAS2Rs in non-responders could contribute to a sustained pro-inflammatory state. These patterns suggest that TAS2Rs are not merely correlates of PTSD symptom change, but may actively participate in biological processes that enable or impede recovery.

TAS2R signaling is a less-explored but potentially important node in the intersection of immune and neural regulation. Our findings raise the possibility that this family of receptors may be viable therapeutic targets for enhancing treatment response in PTSD. While this is the first report linking TAS2Rs to PTSD recovery trajectories, their role in immune regulation, their differential expression across treatment responses, and their placement within recovery-linked gene networks make them intriguing candidates for further investigation. Future work should focus on validating their expression, testing their functional impact on immune tone, and assessing whether known TAS2R agonists modulate biological or clinical markers of recovery.

In addition to TAS2Rs, GABAB receptor activation and opioid signaling also emerged as prominent neurotransmitter pathways showing subgroup-specific enrichment patterns over time. Both pathways involve G-protein-coupled receptors (GPCRs) that mediate inhibitory signaling [34, 35], and are broadly implicated in regulating stress reactivity, affective tone, and synaptic plasticity—all relevant to PTSD [36].

At baseline, both pathways were enriched in non-responders, suggesting these inhibitory systems may be functionally enhanced in persons less likely to benefit from cognitive behavioral psychotherapy. Importantly, their enrichment patterns shifted differentially at follow-up: these pathways were enriched in responders, while declined in non-responders.

Accordingly, NMDA signaling, an excitatory form of neural signaling [37], showed a similarly divergent enrichment pattern. From T1 to T3, responders showed enrichment, while declined in non-responders, echoing the changes in inhibitory pathways.

This divergence of enrichment patterns between the two subgroups further confirms the heterogeneity of the PTSD group and a group-dependent response to CPT therapy, suggesting the possibility of more informed treatment matching for patients with PTSD.

The similar enrichment patterns of inhibitory and excitatory pathways over time suggest a subtle tuning of their balance. The imbalance between inhibitory and excitatory signaling is one of the features of persistent PTSD pathology [38, 39]. While inhibitory GPCR-mediated systems, like GABAB and opioid pathways, may serve as a brake on arousal and reactivity, NMDA overactivation could perpetuate maladaptive circuits associated with fear, stress sensitization, and intrusive memory processing.

A key challenge in PTSD treatment is identifying biological mechanisms that not only mark the presence of illness, but also differentiate the trajectories of recovery. In this context, the GABAB, opioid, and NMDA receptor pathways may represent adaptive systems that are either re-engaged or remain impaired following therapy depending on treatment outcome.

Finally, DEX stimulation of live PBMCs provided a dynamic window into GR signaling, and its potential role in recovery trajectories. Divergent DEX effects were shown on the PTSD subgroups, with a positive correlation to recovery trajectory in responders while negative in non-responders, suggesting different GR responses between the PTSD subgroups. Although NR3C1 and FKBP5 expression did not show large group differences at baseline, we demonstrated more pronounced DEX responding pathways in responders. Moreover, only responders showed a persistent decline of DEX responsiveness at T2 and T3. These results align with our prior findings in a randomized trial of hydrocortisone augmentation of psychotherapy, where elevated GR sensitivity prior to treatment predicted clinical response, and a decline in GR responsiveness over time was associated with symptom improvement [40]. These converging findings suggest that GR sensitivity may serve not only as a state marker of recovery, but may also reflect baseline differences. These data support the potential of GR-targeted interventions for enhancing outcomes in persons less responsive to psychotherapy alone.

In summary, our results demonstrate that transcriptional profiling of in vitro PBMCs is a robust platform for longitudinal study on PTSD therapy. The enrichment of TAS2R signaling appears to be one of the main forces for PTSD recovery, while GABAB, opioid, and NMDA receptor pathways contribute in a subgroup-specific manner. DEX treatment results implicate GR involvement in PTSD recovery, and reveal subgroup-specific effects.

While our findings offer novel insights into the molecular dynamics of PTSD recovery, several limitations warrant consideration. The sample size—particularly for subgroup analyses—was modest, and attrition limited our ability to evaluate all participants longitudinally. Additionally, while in vitro PBMC culture and DEX stimulation provide controlled models for probing biological processes, they may not fully capture the complexity of in vivo neural-immune interactions. Furthermore, per SAGER guidelines [41], treating sex and age as additive covariates adjusts for baseline differences but not moderation, assuming uniform transcriptional responses and potentially obscuring differences across life stages or sexes. Despite these constraints, the convergence of transcriptional, pathway, and glucocorticoid signaling findings supports the value of this approach for identifying biomarkers and therapeutic targets. Future studies with larger, more diverse cohorts and integrated peripheral and central measures will be essential to validate and extend these findings.

Ultimately, molecular tools like those used here—particularly approaches that incorporate dynamic stimulation paradigms—may prove far more informative than static blood-based assays. By revealing latent differences in transcriptional responsiveness, these methods could help identify individuals more likely to benefit from psychotherapy alone versus those who may require pharmacologic augmentation to achieve recovery.

Supplementary information

Supplemental material (6.8KB, docx)
Figure S1 (853.8KB, tif)
Supplemental Tables (49.4KB, xlsx)

Acknowledgements

The authors would like to acknowledge the contributions of all clinical research coordinators, evaluators, and study therapists.

Author contributions

RY, HNB, JDF, LMB, and AL conceived the study design and provided oversight for the clinical research study and sample collection. CX and RYa conducted the computational data analysis. IM, MC, and FD performed the laboratory experiments and along with RY, HNB, and RYa aided in interpretation of the data. CX drafted the manuscript, RY and HNB edited the manuscript and all authors reviewed and approved the final version.

Funding

This work was supported in part by a Merit Review Award (#5I01CX001219-02) from the United States (U.S.) Department of Veterans Affairs Clinical Sciences Research and Development Service, and by the U.S. Army Medical Research Material Command through the Department of Defense Broad Agency Announcement for Extramural Medical Research under Award No. W81XWH-16-1-0773.

Data availability

Aggregate gene expression data are available via an interactive visualization tool at https://lab-results.shinyapps.io/cpt_dex/. Raw transcriptomic data are restricted due to Department of Veterans Affairs (VA) data privacy and security policies. Access to restricted datasets may be requested from the corresponding author, pending institutional IRB approval and the execution of a formal Data Use Agreement (DUA).

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

We confirm that all methods were performed in accordance with the relevant guidelines and regulations. Approval for the conduct of this study has been obtained from the James J. Peters VAMC IRB and R&D under protocol number YEH-15-018 (IRB number 1574389) as well as the DoD OHRO committee. Informed consent was obtained from all participants.

Disclaimer

Material has been reviewed by the Walter Reed Army Institute of Research. There is no objection to its presentation and/or publication. Opinions, interpretations, conclusions, and recommendations are those of the authors and do not necessarily reflect the official position, policy, or views of the Department of Veterans Affairs, the U.S. Army, the Department of Defense, or the United States Government. The investigators have adhered to the policies for protection of human subjects as prescribed in AR 70–25.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41398-026-04289-8.

References

  • 1.Neria Y. Functional neuroimaging in PTSD: from discovery of underlying mechanisms to addressing diagnostic heterogeneity. Am J Psychiatry. 2021;178:128–35. [DOI] [PubMed] [Google Scholar]
  • 2.Whitney AR, Diehn M, Popper SJ, Alizadeh AA, Boldrick JC, Relman DA, et al. Individuality And Variation In Gene Expression Patterns In Human Blood. Proc Natl Acad Sci USA. 2003;100:1896–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Min JL, Barrett A, Watts T, Pettersson FH, Lockstone HE, Lindgren CM, et al. Variability of gene expression profiles in human blood and lymphoblastoid cell lines. BMC Genomics. 2010;11:96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Jaffe AE, Irizarry RA. Accounting for cellular heterogeneity is critical in epigenome-wide association studies. Genome Biol. 2014;15:R31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Yehuda R, Yang RK, Guo SL, Makotkine I, Singh B. Relationship between dexamethasone-inhibited lysozyme activity in peripheral mononuclear leukocytes and the cortisol and glucocorticoid receptor response to dexamethasone. J Psychiatr Res. 2003;37:471–7. [DOI] [PubMed] [Google Scholar]
  • 6.Yehuda R, Yang RK, Golier JA, Tischler L, Liong B, Decker K. Effect of topiramate on glucocorticoid receptor mediated action. Neuropsychopharmacology. 2004;29:433–9. [DOI] [PubMed] [Google Scholar]
  • 7.Yehuda R, Daskalakis NP, Desarnaud F, Makotkine I, Lehrner AL, Koch E, et al. Epigenetic biomarkers as predictors and correlates of symptom improvement following psychotherapy in combat veterans with PTSD. Front Psychiatry. 2013;4:118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Seah C, Breen MS, Rusielewicz T, Bader HN, Xu C, Hunter CJ, et al. Modeling gene × environment interactions in PTSD using human neurons reveals diagnosis-specific glucocorticoid-induced gene expression. Nat Neurosci. 2022;25:1434–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Breen MS, Bierer LM, Daskalakis NP, Bader HN, Makotkine I, Chattopadhyay M, et al. Differential transcriptional response following glucocorticoid activation in cultured blood immune cells: a novel approach to PTSD biomarker development. Transl Psychiatry. 2019;9:201. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Resick PA, Schnicke MK. Cognitive processing therapy for sexual assault victims. J Consult Clin Psychol. 1992;60:748–56. [DOI] [PubMed] [Google Scholar]
  • 11.Resick PA, LoSavio ST, Monson CM. State of the science of cognitive processing therapy. Behav Ther. 2024;55:1205–21. [DOI] [PubMed] [Google Scholar]
  • 12.First MB, Williams JBW, Karg RS, Spitzer RL Structured Clinical Interview for DSM-5—Research Version (SCID-5-RV). Arlington, VA: American Psychiatric Association; 2015.
  • 13.Weathers FW, Bovin MJ, Lee DJ, Sloan DM, Schnurr PP, Kaloupek DG, et al. The clinician-administered PTSD scale for DSM-5 (CAPS-5): development and initial psychometric evaluation in military veterans. Psychol Assess. 2018;30:383–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Resick PA, Monson CM, Chard KM Cognitive processing therapy for PTSD: A comprehensive manual. New York, NY: Guilford Press; 2017.
  • 15.Proust-Lima C, Philipps V, Liquet B. Estimation of extended mixed models using latent classes and latent processes: The R package lcmm. J Stat Softw. 2017;78:1–56. [Google Scholar]
  • 16.van de Schoot R, Sijbrandij M, Winter SD, Depaoli S, Vermunt JK. The GRoLTS-checklist: guidelines for reporting on latent trajectory studies. Struct Equ Modeling. 2017;24:451–67. [Google Scholar]
  • 17.Yang R, Kannan S, Gautam A, Powell TM, LeardMann CA, Hoke AV, et al. Long-Term miRNA changes predicting resiliency factors of post-traumatic stress disorder in a large military cohort—millennium cohort study. Int J Mol Sci. 2025;26:5195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Nylund KL, Asparouhov T, Muthén BO. Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct Equ Modeling. 2007;14:535–69. [Google Scholar]
  • 19.Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Newman AM, Steen CB, Liu CL, Gentles AJ, Chaudhuri AA, Scherer F, et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol. 2019;37:773–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Hoffman GE, Schadt EE. variancePartition: interpreting drivers of variation in complex gene expression studies. BMC Bioinforma. 2016;17:483. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Korotkevich G, Sukhov V, Sergushichev A. Fast gene set enrichment analysis. Preprint at https://www.biorxiv.org/content/10.1101/060012v2 (2019).
  • 23.Milacic M, Beavers D, Conley P, Gong C, Gillespie M, Griss J, et al. The reactome pathway knowledgebase 2024. Nucleic Acids Research. 2024;52:672–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Elizarraras JM, Liao Y, Shi Z, Zhu Q, Pico AR, Zhang B. WebGestalt 2024: faster gene set analysis and new support for metabolomics and multi-omics. Nucleic Acids Research. 2024;52:W415–W421. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Binder EB. The role of FKBP5, a co-chaperone of the glucocorticoid receptor in the pathogenesis and therapy of affective and anxiety disorders. Psychoneuroendocrinology. 2009;34:S186–S195. [DOI] [PubMed] [Google Scholar]
  • 27.Patel NN, Workman AD, Cohen NA. Role of taste receptors as sentinels of innate immunity in the upper airway. J Pathog. 2018;2018:9541987. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Kobayashi D, Watarai T, Ozawa M, Kanda Y, Saika F, Kiguchi N, et al. Tas2R signaling enhances mouse neutrophil migration via a ROCK-dependent pathway. Front Immunol. 2022;13:973880. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Medapati MR, Singh N, Bhagirath AY, Duan K, Triggs-Raine B, Batista EL Jr, et al. Bitter taste receptor T2R14 detects quorum sensing molecules from cariogenic streptococcus mutans and mediates innate immune responses in gingival epithelial cells. FASEB J. 2021;35:e21375. [DOI] [PubMed] [Google Scholar]
  • 30.Gopallawa I, Freund JR, Lee RJ. Bitter taste receptors stimulate phagocytosis in human macrophages through calcium, nitric oxide, and cyclic-gmp signaling. Cell Mol Life Sci. 2020;78:271–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Malki A, Fiedler J, Fricke K, Ballweg I, Pfaffl MW, Krautwurst D. Class i odorant receptors, tas1r and tas2r taste receptors, are markers for subpopulations of circulating leukocytes. J Leukoc Biol. 2015;97:533–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Dotson CD, Shaw HL, Mitchell BD, Munger SD, Steinle NI. Variation in the gene TAS2R38 is associated with the eating behavior disinhibition in old order Amish women. Appetite. 2010;54:93–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Moberg PJ, Li M, Kanes SJ, Gur RE, Kamath V, Turetsky BI. Association of schizophrenia with the phenylthiocarbamide taste receptor haplotype on chromosome 7q. Psychiatr Genet. 2012;22:286–9. [DOI] [PubMed] [Google Scholar]
  • 34.Terunuma M. Diversity of structure and function of GABAB receptors: a complexity of GABAB-mediated signaling. Proc Jpn Acad Ser B Phys Biol Sci. 2018;94:390–411. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Law PY, Wong YH, Loh HH. Molecular mechanisms and regulation of opioid receptor signaling. Annu Rev Pharmacol Toxicol. 2000;40:389–430. [DOI] [PubMed] [Google Scholar]
  • 36.Felice D, Cryan JF, O’Leary OF. GABAB receptors: anxiety and mood disorders. Curr Top Behav Neurosci. 2022;52:241–65. [DOI] [PubMed] [Google Scholar]
  • 37.Dupuis JP, Nicole O, Groc L. NMDA receptor functions in health and disease: old actor, new dimensions. Neuron. 2023;111:2275–460. [DOI] [PubMed] [Google Scholar]
  • 38.Michels L, Schulte-Vels T, Schick M, O’Gorman RL, Zeffiro T, Hasler G, et al. Prefrontal GABA and glutathione imbalance in posttraumatic stress disorder: preliminary findings. Psychiatry Res Neuroimaging. 2014;224:288–95. [DOI] [PubMed] [Google Scholar]
  • 39.Rasmusson AM, Pineles SL, Brown KD, Pinna G. A role for deficits in GABAergic neurosteroids and their metabolites with NMDA receptor antagonist activity in the pathophysiology of posttraumatic stress disorder. J Neuroendocrinol. 2022;34:e13062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Yehuda R, Bierer LM, Pratchett LC, Lehrner A, Koch EC, Van Manen JA, et al. Cortisol augmentation of a psychological treatment for warfighters with posttraumatic stress disorder: randomized trial showing improved treatment retention and outcome. Psychoneuroendocrinology. 2015;51:589–97. [DOI] [PubMed] [Google Scholar]
  • 41.Heidari S, Babor TF, De Castro P, Tort S, Curno M. Sex and gender equity in research: rationale for the SAGER guidelines and recommended use. Res Integr Peer Rev. 2016;1:2. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental material (6.8KB, docx)
Figure S1 (853.8KB, tif)
Supplemental Tables (49.4KB, xlsx)

Data Availability Statement

Aggregate gene expression data are available via an interactive visualization tool at https://lab-results.shinyapps.io/cpt_dex/. Raw transcriptomic data are restricted due to Department of Veterans Affairs (VA) data privacy and security policies. Access to restricted datasets may be requested from the corresponding author, pending institutional IRB approval and the execution of a formal Data Use Agreement (DUA).

Articles from Translational Psychiatry are provided here courtesy of Nature Publishing Group

RESOURCES