Posttraumatic stress disorder is associated with altered gut microbiota that modulates cognitive performance in veterans with cirrhosis

Abstract

Posttraumatic stress disorder (PTSD) is associated with cirrhosis in veterans, and therapeutic results are suboptimal. An altered gut-liver-brain axis exists in cirrhosis due to hepatic encephalopathy (HE), but the added impact of PTSD is unclear. The aim of this study was to define linkages between gut microbiota and cognition in cirrhosis with/without PTSD. Cirrhotic veterans (with/without prior HE) underwent cognitive testing [PHES, inhibitory control test (ICT), and block design test (BDT)], serum lipopolysaccharide-binding protein (LBP) and stool collection for 16S rRNA microbiota composition, and predicted function analysis (PiCRUST). PTSD was diagnosed using DSM-V criteria. Correlation networks between microbiota and cognition were created. Patients with/without PTSD and with/without HE were compared. Ninety-three combat-exposed male veterans [ (58 yr, MELD 11, 34% HE, 31% combat-PTSD (42 no-HE/PTSD, 19 PTSD-only, 22 HE-only, 10 PTSD+HE)] were included. PTSD patients had similar demographics, alcohol history, MELD, but worse ICT/BDT, and higher antidepressant use and LBP levels. Microbial diversity was lower in PTSD (2.1 ± 0.5 vs. 2.5 ± 0.5, P = 0.03) but unaffected by alcohol/antidepressant use. PTSD (P = 0.02) and MELD (P < 0.001) predicted diversity on regression. PTSD patients showed higher pathobionts (Enterococcus and Escherichia/Shigella) and lower autochthonous genera belonging to Lachnospiraceaeae and Ruminococcaceae regardless of HE. Enterococcus was correlated with poor cognition, while the opposite was true for autochthonous taxa regardless of PTSD/HE. Escherichia/Shigella was only linked with poor cognition in PTSD patients. Gut-brain axis-associated microbiota functionality was altered in PTSD. In male cirrhotic veterans, combat-related PTSD is associated with cognitive impairment, lower microbial diversity, higher pathobionts, and lower autochthonous taxa composition and altered gut-brain axis functionality compared with non-PTSD combat-exposed patients. Cognition was differentially linked to gut microbiota, which could represent a new therapeutic target.

NEW & NOTEWORTHY Posttraumatic stress disorder (PTSD) in veterans with cirrhosis was associated with poor cognitive performance. This was associated with lower gut microbial diversity in PTSD with higher pathobionts belonging to Enterococcus and Escherichia/Shigella and lower beneficial taxa belonging to Lachnospiraceaeae and Ruminococcaceae, with functional alterations despite accounting for prior hepatic encephalopathy, psychoactive drug use, or model for end-stage liver disease score. Given the suboptimal response to current therapies for PTSD, targeting the gut microbiota could benefit the altered gut-brain axis in these patients.

INTRODUCTION

Chronic liver disease and cirrhosis are epidemic in the veteran population (13). Patients with cirrhosis have an altered gut-liver-brain axis that manifests itself as hepatic encephalopathy (HE) (1, 8). HE consists of a spectrum ranging from subclinical covert HE (CHE) and the clinically apparent overt HE (OHE) (52). HE is associated with altered gut microbial composition and function, impaired intestinal barrier, and systemic inflammation, which results in the neuroinflammation underlying its clinical symptoms (50). These negatively affect daily function and prognosis and can result in persistent impairment (11, 45). Chronic liver diseases and cirrhosis in veterans are most often associated with lifestyle and addiction-related diseases such as substance and alcohol use disorders and obesity, which can be exacerbated by posttraumatic stress disorder (PTSD) (13, 27, 28, 51). PTSD is epidemic in the combat veteran population and is often associated with chronic liver disease and greater mortality from chronic liver disease (15, 16, 20).

PTSD is associated with varying degrees of autonomic hyperreactivity, depression, and anxiety, which can result in cognitive impairment even in the absence of chronic liver disease and cirrhosis (25). There is evidence of an altered gut-brain axis in animal and human studies with PTSD without concomitant cirrhosis (26, 33). A prior study shows that veterans with PTSD and cirrhosis have poorer cognitive function compared with model for end-stage liver disease (MELD)-balanced veterans with cirrhosis without PTSD (14). However, the impact of gut microbial change related to PTSD in the setting of cirrhosis is unclear. This may be relevant as a new target for therapy since current therapy success rates in PTSD are suboptimal.

Our aim was to determine changes in gut microbiota and cognitive ability in veterans with cirrhosis and PTSD compared with those without PTSD.

MATERIALS AND METHODS

We prospectively enrolled outpatient veterans with cirrhosis diagnosed using biopsy, radiological evidence, or endoscopic evidence of varices in chronic liver disease. All subjects provided informed consent. We excluded patients with active illicit drug or alcohol abuse within the last 3 mo defined by current urine drug screens and interviews per standard of care, those unable to provide consent or samples, with a minimental status exam <25, with bipolar disorder, seizures, schizophrenia, or on antipsychotic and antiseizure medications. We also excluded patients who were currently on or had been on absorbable antibiotics over the last 3 mo, which included patients on spontaneous bacterial peritonitis prophylaxis. Patients on chronic antidepressants and opioids were included provided their regimen was stable over 3 mo before enrollment. The protocol was approved by the Institutional Review Board at the McGuire VA Medical Center, and all patients provided written informed consent.

Demographics, cirrhosis (etiology, MELD score), and HE (prior OHE) were recorded. PTSD was diagnosed using validated Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-V) criteria (51a), and the current treatments for this condition were recorded. We recorded a detailed 3-day dietary recall for all subjects. Serum was also collected for lipopolysaccharide binding protein (LBP) levels (ng/ml) that were analyzed using published techniques at Assaygate (Ijamsville, MD) (10). All subjects underwent three cognitive testing strategies that have been used for covert HE testing: the inhibitory control test (ICT), psychometric hepatic encephalopathy score (PHES), and block design test (BDT) (52). ICT is a test of psychomotor speed, response inhibition, and working memory with the output being lures and weighted lures (9, 22). A low lure and weighted lure number indicate good performance. PHES is a composite of five tests, and the total SDs beyond healthy controls are calculated (53). A high score indicates good performance. BDT is a test of nonverbal problem solving and visual spatial ability (35), and a low score indicates poor performance. Covert HE based on ICT and PHES was defined using published US-based criteria (2).

Microbiota.

Stool was collected on the same date as the cognitive tests using published techniques on the same day (5). 16S rRNA microbiota analysis was performed using Multitag sequencing on an Ion Torrent PGM as previously published (24). Microbiota diversity was calculated using the Shannon Diversity Index, with a higher score indicating greater diversity, and individual taxa were compared using linear discriminant effect size (LEfSe) (49). Predicted function of microbiota using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PiCRUST) was performed for all samples and compared between PTSD vs. no-PTSD and subgroups with and without prior OHE using LEfSe (32). Correlation networks were created separately for the four subgroups based on the presence or absence of PTSD and prior OHE between microbiota at the genus level and cognitive tests (41). Only those interactions that were P < 0.05 and r > 0.6 or <−0.6 were included in the final analysis.

Statistical analysis.

We compared demographics, cirrhosis details, cognitive performance, and microbiota outputs 1) between patients with or without PTSD; 2) in those without prior OHE, again PTSD vs. no-PTSD; and 3) in those with prior OHE, PTSD vs. no-PTSD.

RESULTS

We enrolled 93 male veterans with cirrhosis, of whom 29 were diagnosed with PTSD. Of the total patient number, 32 had prior OHE, which was controlled on lactulose and rifaximin.

Clinical comparisons.

As shown in Table 1, cirrhotic subjects with PTSD had similar demographics compared with those without PTSD. Cirrhosis severity, prior OHE, and etiologies of cirrhosis were also similar between groups. PTSD patients had greater use of psychoactive drugs. Cognitive testing on PHES was similar between groups, but PTSD patients had significantly worse performance on block design and ICT. All veterans were from the Vietnam era, and all had experienced active combat. All patients followed a nonvegetarian diet and were born and brought up on East Coast of the United States. There were 42 patients without OHE or PTSD, 19 with PTSD without OHE, 22 with OHE without PTSD, and 10 with both (Table 2). Patients with either PTSD or OHE or both were older, but their educational attainment was similar compared with those without PTSD or OHE. As expected, cirrhosis severity was greater in those with prior OHE, as was cognitive impairment on PHES and its subtests. On the other hand, block design and weighted lures were worse in patients with PTSD with/without OHE compared with those who did not have PTSD or OHE. As reported before, covert HE prevalence was higher in those with prior OHE compared with those without on ICT (65 vs. 25%, P = 0.001) and PHES (45 vs. 29%, P = 0.001), but did not differ between those with and without PTSD on ICT (58 vs. 54%, P = 0.76) or PHES (39 vs. 35%, P = 0.71) (14). Psychoactive medication prevalence was higher in the PTSD subgroups regardless of OHE. No dietary differences between groups were seen (Tables 1 and 2).

Table 1.

Comparison between PTSD and no-PTSD groups as a whole

	No-PTSD (n = 64)	PTSD (n = 29)
Age, yr	58.8 ± 8.2	59.9 ± 8.4
Education, yr	13.4 ± 2.4	13.1 ± 2.7
Total calories/day	2,513 ± 396	2,489 ± 451
Percent calories from protein/day	16 ± 8	15 ± 9
Lipopolysaccharide-binding protein, ng/ml	9.4 ± 2.9	12.7 ± 4.3*
MELD score	12.0 ± 7.5	10.5 ± 7.9
Etiology (HCV, alcoholic, Alc+HCV, NASH, others)	28/12/7/10/7	10/10/4/3/2
Prior overt HE	22 (34%)	10 (34%)
Past alcohol use disorder	40 (63%)	19 (66%)
PTSD Details
Current psychotherapy	0 (0%)	3 (10%)
Current psychoactive medications	12 (19%)	15 (51%)*
Cognitive Testing
Number connection-A	46.2 ± 22.1	49.5 ± 27.3
Number connection-B	130.0 ± 75.6	153.0 ± 106.0
Digit symbol	46.1 ± 16.2	45.1 ± 14.9
Serial dotting	80.4 ± 36.1	75.7 ± 33.1
Line tracing time	116.5 ± 67.9	114.2 ± 42.5
Line tracing errors	36.9 ± 33.8	36.8 ± 33.6
PHES total	−3.7 ± 3.4	−4.2 ± 4.1
Block design test	27.0 ± 13.0	22.0 ± 11.9*
Weighted lures	15.0 ± 11.0	22.1 ± 13.9*
Shannon Diversity Index	2.5 ± 0.5	2.1 ± 0.5*

Values are means ± SE; n, number of patients. PTSD, posttraumatic stress syndrome; PHES, psychometric hepatic encephalopathy score; HE, hepatic encephalopathy; MELD, model for end-stage liver disease; HCV, hepatitis C virus; NASH, nonalcoholic steatohepatitis.

^*P < 0.05 on unpaired t test or Man-Whitney as appropriate between groups; high score on digit symbol and block design tests indicates good performance, while low score on the rest indicates good performance. Higher Shannon diversity score indicates greater microbial diversity.

Table 2.

Comparison between subgroups divided according to PTSD and prior OHE

	No-PTSD or OHE (n = 42)	PTSD Without OHE (n = 19)	OHE Without PTSD (n = 22)	PTSD and OHE (n = 10)	P Value All Subgroups*
Age, yr	55.8 ± 8.3	60.6 ± 6.2	60.6 ± 7.3	60.4 ± 5.7	0.03
Education, yr	13.7 ± 1.9	12.4 ± 2.9	14.3 ± 2.6	13.2 ± 2.1	0.10
Total calories/day	2,589 ± 401	2,480 ± 405	2,390 ± 327	2,510 ± 297	0.28
Percent calories from protein/day	17 ± 6	16 ± 9	16 ± 7	15 ± 5	0.57
Lipopolysaccharide-binding protein, ng/ml†‡	8.1 ± 2.3	9.9 ± 3.1	11.7 ± 3.1	14.3 ± 5.1	0.005
MELD score†	10.3 ± 4.5	8.7 ± 2.1	18.6 ± 8.7	15.4 ± 5.4	<0.0001
Etiology (HCV, alcoholic, Alc+HCV, NASH, others)	21/5/3/7/6	7/5/4/2/1	7/7/4/3/1	3/5/0/1/1	0.09
Past alcohol-use disorder	23 (54%)	13 (68%)	17 (77%)	6 (60%)	0.27
PTSD Details
Current psychotherapy		2 (11%)		1 (10%)	1.0
On psychoactive medications‡	9 (14%)	10 (51%)	3 (14%)	5 (50%)	0.006
Cognitive Testing
Number connection-A†	37.6 ± 12.7	45.4 ± 28.4	60.9 ± 25.2	61.8 ± 29.3	<0.0001
Number connection-B†	106.9 ± 47.2	146.7 ± 114.3	171.2 ± 98.5	173.0 ± 93.1	0.01
Digit symbol†	52.8 ± 14.5	47.8 ± 12.8	33.8 ± 11.0	38.8 ± 17.6	<0.0001
Serial dotting†	70.1 ± 27.9	64.0 ± 22.6	102.1 ± 41.3	96.5 ± 41.1	<0.0001
Line tracing time†	98.8 ± 41.1	99.5 ± 38.4	152.3 ± 91.8	142.0 ± 91.8	0.002
Line tracing errors	31.1 ± 31.2	40.8 ± 39.9	47.6 ± 35.6	28.3 ± 13.7	0.24
PHES total†	−3.1 ± 3.3	−3.8 ± 4.1	−8.6 ± 4.3	−7.4 ± 5.4	<0.0001
Block design test	30.0 ± 13.7	22.6 ± 12.7	20.0 ± 8.9	22.6 ± 10.2	0.01
Weighted lures‡	15.2 ± 11.0	22.3 ± 13.6	22.4 ± 19.2	22.3 ± 14.3	0.05
Shannon Diversity Index†	2.5 ± 0.5	2.3 ± 0.6	2.0 ± 0.8	1.7 ± 0.5	0.001

Values are means ± SE; n, number of patients. PTSD, posttraumatic stress syndrome; PHES, psychometric hepatic encephalopathy score; OHE, prior overt hepatic encephalopathy; MELD, model for end-stage liver disease; HCV, hepatitis C virus; NASH, nonalcoholic steatohepatitis.

^*P < 0.05 ANOVA or Kruskal-Wallis tests.

^†P < 0.05 between OHE vs. no-OHE groups.

^‡P < 0.05 between PTSD vs. no-PTSD groups; high score on digit symbol and block design tests indicates good performance, while low score on the rest indicates good performance. Higher Shannon diversity score indicates greater microbial diversity.

PTSD details.

PTSD was diagnosed using DSM-V criteria, and in all cases was combat related. PTSD in affected patients was treated using psychotherapy in the minority and psychotropic medication in the rest. Of the 15 patients on psychoactive medications, the majority were on antidepressants (trazodone n = 6, bupriopion n = 4, sertraline n = 3, citalopram n = 2, fluoxetine n = 2, and mirtazapine n = 1) followed by benzodiazepines (n = 2). Five patients were on more than one medication. The remaining patients who were not actively treated for PTSD specifically were not on disability at the time of testing and were able to complete the testing and analysis.

In the 12 patients without PTSD that were on psychoactive medications, the major ones were anti-depressants (buproprion n = 3, trazodone n = 2, duloxetine n = 1, fluoxetine n = 1, venlafaxine n = 1, sertraline n = 1, citalopram n = 1) followed by benzodiazepines (n = 2).

Microbiota changes.

For Shannon diversity scores, in the entire group, advancing disease was associated with lower diversity. The MELD score was negatively correlated (r = −0.54, P < 0.0001), and there was a lower Shannon in those with prior OHE vs. the rest (1.94 ± 0.70 vs. 2.45 ± 0.55, P = 0.001). Shannon was also correlated with PHES score (r = 0.33, P = 0.002), Weighted lures (r = −0.3, P = 0.04), and block design (r = 0.25, P = 0.05).

PTSD was associated with a lower Shannon (Table 1), which was lowest in the group with prior OHE and PTSD (Table 2). Diversity scores were statistically similar between patients with an alcoholic etiology/not (2.4 ± 0.6 vs. 2.4 ± 0.4, P = 0.58), hepatitis C virus (HCV) etiology/not (2.2 ± 0.7 vs. 2.4 ± 0.6. P = 0.52), or those on psychoactive medications or not (2.4 ± 0.6 vs. 2.3 ± 0.7, P = 0.17).

We performed a linear stepwise regression with Shannon diversity as the dependent variable using age, education, MELD score, prior OHE, PTSD, HCV etiology, and prior alcohol use as the potential predictors. The MELD score (coefficient −0.054, P < 0.001) and PTSD (coefficient −0.28, P = 0.02) were independently associated with Shannon diversity scores.

For individual microbiota differences, our primary comparisons were between PTSD vs. no-PTSD in those with prior OHE and those without prior OHE on LEfSe, which are shown in Table 3. Regardless of prior OHE, PTSD patients had lower relative abundance of potentially beneficial taxa belonging to families Ruminococcaceae and Lachnospiraceae. In addition, PTSD patients without prior OHE had higher relative abundance of potentially pathogenic taxa belonging to Enterococcaceae and Streptococcaceae.

Table 3.

LEfSe changes in stool of patients with cirrhosis based on PTSD and prior overt hepatic encephalopathy

Family_ Genus	Higher in PTSD	Higher in No-PTSD
Without prior OHE	Enterococcaceae_Pillibacter Streptococcaceae_Streptococcus Acidaminococcaceae_Acidaminococcus Gracilibacteriaceae_Lutispora Clostridiaceae_Proteiniclasticum Thermodesulfobiaceae_Thermodesulfobium	Lachnospiraceae_Ruminococcus Lachnospiraceae_Roseburia Lachnospiraceae_Anaerostipes Lachnospiraceae_ClostridiumXIVa Lachnospiraceae_Eisenbergiella Lachnospiraceae_Lachnospira Ruminococcaceae_Pseudoflavonibacter
With prior OHE		Ruminococcaceae_Subdoligranulum

Higher in one group means lower in the other and vice versa based on linear discriminant analysis score effect size (LEfSe). OHE, overt hepatic encephalopathy; PTSD, posttraumatic stress syndrome.

PiCRUST analysis.

Tables 4, 5, and 6 show comparisons between groups with/without PTSD as a whole, without prior OHE, and with prior OHE, respectively.

Table 4.

PiCRUST values in all patients

	Higher in Group	LDA
Phenyl-propanoid biosynthesis	No-PTSD	2.35
Cyano amino acid metabolism	No-PTSD	2.31
Terpenoid biosynthesis	PTSD	2.21
Alanine metabolism	PTSD	2.07
Genetic Information Processing: nucleotide excision repair	PTSD	2.03
Genetic Information Processing: homologous recombination	PTSD	2.18
Genetic Information Processing: base excision repair	PTSD	2.04
Genetic Information Processing: ribosome biogenesis	PTSD	2.39
Genetic Information Processing: DNA repair and recombination proteins	PTSD	2.57

PiCRUST, Phylogenetic Investigation of Communities by Reconstruction of Unobserved States. Higher linear discriminant analysis (LDA) score indicates greater difference between the groups.

Table 5.

PiCRUST values in patients without prior overt HE

	Higher in Group	LDA
Glyoxylate and dicarboxylate metabolism	No-PTSD	2.11
Inorganic ion transport and metabolism	No-PTSD	2.04
Phenyl-propanoid biosynthesis	No-PTSD	2.21
Arginine and proline metabolism	No-PTSD	2.49
Pentose and glucuronate interconversions	No-PTSD	2.55
Starch and sucrose metabolism	No-PTSD	2.50
Sphingolipid metabolism	No-PTSD	2.49
Oxidative phosphorylation	No-PTSD	2.34
Cyano amino acid metabolism	No-PTSD	2.26
Tyrosine metabolism	PTSD	2.06
Terpenoid biosynthesis	PTSD	2.14
Peptidoglycan biosynthesis	PTSD	2.24
Alzheimer’s disease	PTSD	2.01
Genetic Information Processing: homologous recombination	PTSD	2.15
Genetic Information Processing: DNA repair and recombination proteins	PTSD	2.65
Genetic Information Processing: ribosome biogenesis	PTSD	2.47
Membrane transport secretion system	PTSD	2.65
Alanine metabolism	PTSD	2.03

PiCRUST, Phylogenetic Investigation of Communities by Reconstruction of Unobserved States. Higher linear discriminant analysis (LDA) score indicates greater difference between the groups.

Table 6.

PiCRUST values in patients with prior overt HE

	Higher in Group	LDA Score
Flagellar assembly	No-PTSD	2.94
Bacterial chemotaxis	No-PTSD	2.85
Bacterial motility proteins	No-PTSD	3.20
Linoleic acid metabolism	PTSD	2.44
Glycan degradation	PTSD	2.98
Cellular antigen processing	PTSD	2.76
Alanine, aspartate, and glutamate metabolism	PTSD	2.89
Sphingolipid metabolism	PTSD	2.80
Glutamatergic synapse	PTSD	2.50
Alzheimer’s disease	PTSD	2.84

PiCRUST, Phylogenetic Investigation of Communities by Reconstruction of Unobserved States. Higher linear discriminant analysis (LDA) score indicates greater difference between the groups.

Correlation network analysis.

We created subnetworks based on potentially beneficial taxa related to Ruminococcaceae Fecalibacterium, and other potential negative taxa, such as Enterococcaceae and Enterobacteriaceae constituents with other microbiota and cognitive tests in patients with PTSD with/without OHE, and similarly for those without PTSD (Fig. 1, A–C). Regardless of subgroup, Fecalibacterium was positively correlated with other beneficial taxa and good performance on cognitive tests, especially in groups with prior OHE (Fig. 2, A–D). In patients without OHE or PTSD, there were no significant correlations between Shigella/Escherichia and Enterococcus and other bacteria and cognitive tests. Regardless of whether patients had OHE or not, PTSD patients demonstrated a significant positive correlation between Shigella/Escherichia and Enterococcus and a negative correlation between them and taxa belonging to Lachnospiraceae and Ruminococcaceae (Fig. 1, A and C). When patients without PTSD with OHE were studied, Enterobacteriaceae constituents were not prominent; rather, Enterococcus was associated with poor cognition and negatively with taxa belonging to Lachnospiraceae and Ruminococcaceae (Fig. 1B).

Fig. 1.

Correlation network between microbiota and cognitive tests were performed using R, and significant interactions with P < 0.05 and r > 0.6 or <−0.6 are shown. Subnetworks centered around Enterococcus and Escherichia/Shigella are shown in A–C. Microbiota are presented as Family I Genus and are shown as red nodes, while cognitive tests are shown as blue nodes. Red lines indicate negative, while blue lines indicate positive correlations. OHE, prior overt hepatic encephalopathy currently on lactulose and rifaximin; PTSD, posttraumatic stress disorder. A: no-OHE PTSD group showing negative correlations between Enterococcus and Escherichia/Shigella and several members of the Lachnospiraceae and Ruminococcaceae families. B: OHE no-PTSD group showing Enterococcus associated negatively with several members of the Lachnospiraceae and Ruminococcaceae families. Escherichia/Shigella were not meaningfully linked. C: OHE PTSD group showing negative correlations between Enterococcus and Escherichia/Shigella and several members of the Lachnospiraceae and Ruminococcaceae families. Cognitive performance on lures (higher = worse) was linked positively with Enterococcus and negatively with Lachnospiraceae and Ruminococcaceae genera. The block design test (BDT; high = better) was positively linked with Lachnospiraceae and Ruminococcaceae genera and negatively with Escherichia/Shigella. The linkage was similar with the number connection test-A (NCT-A; high = worse) with positive correlation with Escherichia/Shigella and negatively with Lachnospiraceae and Ruminococcaceae genera.

Fig. 2.

Correlation network between microbiota and cognitive tests were performed using R, and significant interactions with P < 0.05 and r > 0.6 or <−0.6 are shown. Subnetworks centered round Fecalibacterium are shown in A–D. Microbiota are presented as Family I Genus and are shown in red nodes, while cognitive tests are shown in blue nodes. Red lines indicate negative, while blue lines indicate positive correlations. OHE, prior overt hepatic encephalopathy currently on lactulose and rifaximin; PTSD, posttraumatic stress disorder. A: no-OHE no-PTSD group showed positive correlations between Fecalibacterium and other Lachnospiraceae and Ruminococcaceae genera. B: no-OHE PTSD group again showed positive correlations between Fecalibacterium and other Lachnospiraceae and Ruminococcaceae genera. Line tracing time (high = worse) was negatively correlated with Lachnospiraceae and Ruminococcaceae genera and positively with Fusobacterium. C: OHE no-PTSD demonstrated negative correlations with Fecalibacterium and Escherichia/Shigella, Enterococcus, Prevotella, and positive with other Lachnospiraceae and Ruminococcaceae genera. D: OHE PTSD group correlation network showed similar negative correlations between Fecalibacterium and Escherichia/Shigella, Enterococcus, Streptococcus, and Veillonella. Block design test (BDT high = better) was positively linked with Fecalibacterium. Positive linkages were demonstrated with other Lachnospiraceae and Ruminococcaceae genera.

DISCUSSION

The current study demonstrates more severe cognitive impairment in cirrhotic veterans with PTSD than those without PTSD. This is accompanied by a reduced microbial diversity, increased relative abundance of pathobionts belonging to Enterococcus and Proteobacteria, and a reduction in taxa belonging to Ruminococcaceae and Lachnospiraceae in PTSD patients. Changes in the microbiota are linked with cognitive dysfunction, especially in those with prior OHE.

PTSD is a major contributor toward morbidity and mortality in veterans, especially related to chronic liver disease (20). Therefore, the current study results demonstrating greater dysbiosis and cognitive impairment in patients with PTSD regardless of prior OHE, can help us understand gut microbial modulation as a target in future studies. Replicating prior studies, there was additional cognitive impairment and psychoactive drug use in veterans with PTSD (14, 25). However, the groups were statistically similar with respect to cirrhosis severity, prior OHE, and demographic data. Given the major impact of prior OHE and its therapy on the gut-liver-brain axis, we split the comparison by OHE and PTSD and despite that found that greater impairments in the PTSD groups compared with their non-PTSD counterparts persisted.

The gut-brain axis has been implicated in the pathogenesis and progression of several diseases including depression, schizophrenia, autism, and HE (17). In a small human study focusing on non-combat PTSD, there was a lower relative abundance of Actinobacteria, Lentisphaerae, and Verrucomicrobia phyla in PTSD compared with trauma-exposed controls. The current study extends these into combat-related PTSD in male veterans and found major differences at the genus level that was independent of the stage of liver disease and prior OHE. We found a significant reduction in the diversity in PTSD patients, which also persisted when the patients without OHE were analyzed separately. PTSD was an independent predictor of lower diversity despite controlling for cirrhosis severity, OHE, alcohol, and psychoactive drugs. The alterations in microbiota found in our study were lower relative abundance of autochthonous genera belonging to Lachnospiraceae and Ruminococcaceae and higher relative abundance of pathobionts such as Enterococcus and Escherichia/Shigella in patients with PTSD. The autochthonous taxa belong to Firmicutes, which in rodent studies were reduced after intruder exposure in a population of male mice (23). Lachnospiraceae and Ruminococcaceae contain genera that produce short-chain fatty acids, strengthen the intestinal barrier, and are associated with better outcomes in patients with cirrhosis and other gut-based diseases (3, 42). These findings underline the importance of microbiota in PTSD, although the effect of concomitant liver disease remains unclear.

There is currently an interest in the neural and molecular mechanisms responsible for extinction of learned fear, particularly since fear extinction is a translationally relevant animal model for the treatment of human disorders, such as PTSD (38, 44). Previous studies suggest that fear extinction does not erase the original fear memory, but rather promotes the formation of a new inhibitory memory that reduces fear to the conditioned stimulus. Research in rodent models and humans suggests that the main structures are involved in processes related to fear extinction, particularly prefrontal and hippocampal inputs to the amygdala (38, 44). In support of this theoretical model, functional imaging studies of PTSD patients demonstrate hypoactivity in the ventromedial prefrontal cortex, suggesting that, at least in part, frontal region impairment may account for symptoms of PTSD (30). Consistent with this hypothesis, in our study patients with PTSD performed poorly on block design (assessing nonverbal problem-solving) and the ICT (evaluating the integrity of the frontal lobe inhibitory circuits and working memory). Performance on both of these measures depends on frontal/parietal integrity (35). These data suggest that PTSD, and the associated microbial change, selectively impacts frontal/parietal region function.

Chronic liver disease is a major consequence of lifestyle factors such as substance use and obesity that leads to hepatitis C, nonalcoholic fatty liver disease, and alcoholic liver disease (48). These lifestyle factors are associated with PTSD and depression, which often coexist and worsen the prognosis of these patients (25). In most patients with cirrhosis, cognitive impairment and gut microbial alterations are assumed to be related to covert or overt HE (52). However, as our current and prior findings show, PTSD adds to the cognitive impairment and alteration in microbial composition and diversity even in the skewed background of cirrhosis and prior OHE (14). When the correlation networks were analyzed, pathobionts Enterococcus and Esherichia/Shigella were negatively associated with good cognition and Lachnospiraceae/Ruminococcaceae constituents in patients with PTSD regardless of prior OHE. Moreover, patients without PTSD or OHE did not have a rich correlation network between microbiota and cognitive impairment, again demonstrating the additive impact of PTSD on microbial composition and interactions. These observations were strengthened by the higher levels of LBP, a functional correlate of excess gram-negative taxa such as Esherichia/Shigella, which was higher in PTSD patients, especially those with OHE. In addition to the differential correlation patterns in patients with and without PTSD, predicted functionality using PiCRUST showed that patients with PTSD as a whole had greater microbial functionality related to Alzheimer’s disease, alanine, and terpenoids along with multiple levels of genetic information-processing functions. Terpenoids, which are usually secondary bile acids, whose excess presence is associated with intestinal barrier disruption, were higher in PTSD patients (19). On the other hand, indolic tryptophan derivatives, phenylpropanoids, which are associated with a functioning gut barrier and beneficial gut-brain effects, were higher in patients without PTSD (12, 46). A healthy gut barrier is also supported by arginine and proline metabolism, which were also higher in patients without PTSD (21). Mirroring the relatively higher Escherichia/Shigella and serum LBP, patients with PTSD also had higher predicted microbial functions related to glycan metabolism that is associated with endotoxin and gram-positive cell wall production. Additionally, there was also a higher excitatory amino acid glutaminergic synapse function association with PTSD-microbiota, which could relate to the anxiety present in these patients (29).

The gut-brain axis in cirrhosis has been modulated beneficially using probiotics, antibiotics such as rifaximin, laxatives such as lactulose, and a fecal microbiota transplant (FMT) (52). In those with prior OHE, however, despite lactulose and rifaximin being on board, our PTSD subjects still demonstrated increased Enterococcus and Proteobacteria and lower Ruminococcacaeae and Lachnospiracaeae compared with OHE without PTSD. This shows that perhaps in the OHE setting, further therapeutic options may be needed to alleviate PTSD. FMTs from mice which had a “defeat” PTSD phenotype were able to transmit inflammatory processes and their depression-like behavior to naive mice, indicating that the microbiota play a major role in behavior (43). Prior FMT studies in cirrhosis have demonstrated improvement in brain dysfunction in those with prior OHE (7, 10). In addition, microbiota from post-FMT humans with cirrhosis were able to quench neuroinflammation engendered by the microbiota before FMT (37). The therapy for PTSD currently consists of group or individual psychotherapy and medications to focus on specific symptoms, such as depression, anxiety, or flashbacks. However, despite these efforts, the disability and prevalence of noncombat PTSD continues to be a major burden, and many veterans with PTSD are not able to function to maintain competitive employment (31). Superimposed chronic liver disease makes these disabilities worse (47). Our findings related to altered microbial composition even in the earlier cirrhosis stages that separated PTSD from non-PTSD veterans could be another target for therapy (18). Specifically, we found that genera such as Fecalibacterium were positively related to good cognition, other autochthonous taxa, and negatively with pathobionts regardless of PTSD or prior OHE. Fecalibacterium and other members of Ruminococaceae have been associated with a strong intestinal barrier, are lower in those with depression, and could therefore be linked with better outcomes if studied as “psychobiotics” (18, 36, 39). However, the effect on PTSD needs to be studied.

Our study is limited by the relatively small sample size: male veterans and PTSD related to combat in those with cirrhosis. Therefore, the results may not be generalizable to women and noncombat PTSD. PTSD patients had a higher proportion on psychoactive drugs for depression, which was like the psychoactive medicine profile in those without PTSD, but we did not find any changes in microbial diversity between those on or not on these medications. Alcohol-use disorder was similar between groups, which could have been another modulator of the gut-brain axis (34, 40). Although we did not find a significant effect of alcoholic or HCV etiology on microbiota diversity, studies with sample sizes are needed. We only included OHE patients already on lactulose and rifaximin to avoid confounding, and studies of patients without rifaximin are needed (6). The cross-sectional nature makes it difficult to interpret whether the combat event that led to PTSD was associated with microbiota change first leading to PTSD or vice versa. Animal studies of FMT transmitting PTSD-like behavior into naive mice and the several decades between PTSD/combat events and development of cirrhosis make it likely that PTSD is a contributor to the microbial change that would worsen cirrhosis-related neurocognitive performance. However, longitudinal studies are needed (20, 37, 43).

We conclude that in male veterans with cirrhosis, combat-related PTSD is associated with cognitive impairment, lower microbial diversity, greater relative abundance of pathobionts such as Escherichia/Shigella, and lower relative abundance of beneficial gut microbial taxa such as Fecalibacterium. This was associated with functional changes in the gut microbiota with implications for the gut-brain axis. Microbial changes were independent of prior overt hepatic encephalopathy and could represent a new target to potentially benefit veterans with cirrhosis and PTSD.

GRANTS

This work was supported by Veterans Affairs Merit Review 2I0CX001076 and R21TR002024 to J. S. Bajaj.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.S.B. conceived and designed research; M.S., A.F., H.C.G., E.A.G., M.F., S.N., and P.M.G. performed experiments; J.S.B., M.S., A.F., S.N., P.M.G., and J.B.W. analyzed data; J.S.B., P.M.G., and J.B.W. interpreted results of experiments; J.S.B., S.N., and P.M.G. prepared figures; J.S.B., D.M.H., J.G.-M., P.M.G., and J.B.W. drafted manuscript; J.S.B., D.M.H., M.F., J.G.-M., S.N., P.M.G., and J.B.W. edited and revised manuscript; J.S.B., M.S., A.F., D.M.H., H.C.G., E.A.G., M.F., J.G.-M., S.N., P.M.G., and J.B.W. approved final version of manuscript.