Recent alcohol intake impacts microbiota in adult burn patients

Abstract

Alcohol use is associated with an increased incidence of negative health outcomes in burn patients due to biological mechanisms that include a dysregulated inflammatory response and increased intestinal permeability. This study used phosphatidylethanol (PEth) in blood, a direct biomarker of recent alcohol use, to investigate associations between a recent history of alcohol use and the fecal microbiota, short chain fatty acids, and inflammatory markers in the first week after a burn injury for nineteen participants. Burn patients were grouped according to PEth levels of low or high and differences in the overall fecal microbial community were observed between these cohorts. Two genera that contributed to the differences and had higher relative abundance in the low PEth burn patient group were Akkermansia, a mucin degrading bacteria that improves intestinal barrier function, and Bacteroides, a potentially anti-inflammatory bacteria. There was no statistically significant difference between levels of short chain fatty acids or intestinal permeability across the two groups. To our knowledge, this study represents the first report to evaluate the effects of burn injury and recent alcohol use on early post burn microbiota dysbiosis, inflammatory response, and levels of short chain fatty acids. Future studies in this field are warranted to better understand the factors associated with negative health outcomes and develop interventional trials.

Introduction

Burn injuries are sudden, traumatic events that trigger systemic and localized responses in an effected individual. In the United States (US), nearly a half a million burn injuries occur each year, with 50,000 injured individuals requiring emergency care. Immediately after burn injury, an inflammatory response ensues that is associated to the severity of the burn (Davis, Janus, et al., 2013; Xanthe & Allison, 2017). Severe burns result in a dysregulated inflammatory response within hours of injury followed by a hyper-metabolic state (Bergquist et al., 2019; Jeschke et al., 2020; Mulder et al., 2022). The cascading systemic inflammation permeates throughout the body and organs (Evers, Bhavsar, & Mailänder, 2010), including the gastrointestinal tract that hosts trillions of bacterial cells (Sender, Fuchs, & Milo, 2016).

An added complexity in burn patients outcomes is seen in up to 50% of burn injury patients that are either intoxicated at the time of injury or have a history of alcohol use disorder (Davis, Esposito, et al., 2013). Indeed, the connection between alcohol and burn injury is well documented (Chen, O’Halloran, Ippolito, Choudhry, & Kovacs, 2015; Jones, Barber, Engrav, & Heimbach, 1991; Silver et al., 2008). The presence of alcohol at time of a burn injury increases intubation time (Silver et al., 2008), delays wound healing (Curtis, Hlavin, Brubaker, Kovacs, & Radek, 2014; Guo & Dipietro, 2010), increases hospital length of stay (Silver et al., 2008), increases rates of infection, and results in higher mortality rates compared to patients that abstain from alcohol (Thombs, Singh, Halonen, Diallo, & Milner, 2007). A direct biological marker of alcohol consumption is phosphatidylethanol (PEth), a group of phospholipids present in cell membranes. PEth is produced after alcohol intake (Isaksson, Walther, Hansson, Andersson, & Alling, 2011), is correlated to recent history of alcohol ingestion (Aradottir, Asanovska, Gjerss, Hansson, & Alling, 2006; Schröck, Wurst, Thon, & Weinmann, 2017), and remains detectable for several weeks (Hahn et al., 2012; Helander, Böttcher, Dahmen, & Beck, 2019). Even in the absence of a burn injury, alcohol use and alcohol related conditions are associated with chronic inflammation (Bishehsari et al., 2017; H. J. Wang, Zakhari, & Jung, 2010) and increased intestinal permeability (Parlesak, Schäfer, Schütz, Bode, & Bode, 2000) that can have bidirectional impacts on intestinal microbial communities. Moreover, the combination of alcohol and burn injury (i.e. “two hit” model) can further damage intestinal barrier function (Hammer, Morris, Earley, & Choudhry, 2015) as shown in animal models (Earley et al., 2015; Luck, Li, Herrnreiter, Cannon, & Choudhry, 2022; Zahs et al., 2012).

The physiological alterations associated with both burn injury and alcohol (e.g. dysregulated inflammatory response, increased intestinal permeability) are known to influence the microbial communities in the gut (Al Bander, Nitert, Mousa, & Naderpoor, 2020; Braniste et al., 2014; Honda & Littman, 2012; Inczefi, Bacsur, Resál, Keresztes, & Molnár, 2022). Furthermore, a dysregulation of gut microbiota may alter immune responses (Kamada, Seo, Chen, & Núñez, 2013), and contribute to systemic physiologic dysfunction, including elevated inflammation and oxidative stress (Zheng, Liwinski, & Elinav, 2020). Individually, the human gut microbiota has been noted to shift following burn injury (Dyamenahalli et al., 2022; Shimizu et al., 2015; X. Wang et al., 2017) and chronic alcohol use (Bode & Bode, 2003; Engen, Green, Voigt, Forsyth, & Keshavarzian, 2015; Qamar et al., 2019). There is a gap in knowledge regarding whether the gut microbiome is more dramatically altered in burn patients who consume alcohol in excess relative to burn patients who drink more moderately or not at all. Importantly, the combined insult of burn injury and alcohol use may alter biological responses or clinical outcomes and may warrant a new approach to management.

To address the combined influence of burn injury and recent alcohol exposure history in patients, we conducted a study of 19 individuals admitted for emergency care following burn trauma. Stool and blood samples were obtained in the first week post-injury to examine the immediate response of the gut microbiota, fecal short chain fatty acids, and cytokines in burn patients with or without evidence of a history of alcohol misuse. Participants were grouped according to low or high PEth concentrations for the analysis. This study, to our knowledge, is the first to evaluate the effects of burn injury and alcohol on the fecal microbiota shifts, inflammatory response, intestinal permeability, and presence of short chain fatty acids in burn patients.

Material and Methods

Sample Description

The current study includes a cohort of burn patients enrolled at an American Burn Association verified burn center from July 2018 to February 2020 (Table 1). The patients were consented within the study guidelines and enrolled with Colorado Multiple Institutional Review Board (COMIRB) approval within the biorepository protocol established through Colorado Pulmonary-Alcohol Research Collaborative, protocol number 11–1525. This study was conducted according to the guidelines of the Declaration of Helsinki. Written informed consent was obtained before participation in any study procedures. Participants were included if they were over 18 years old, sustained a burn injury, admitted to the hospital as inpatient within 24 hours of initial burn injury (total burn surface area [TBSA] > 0%), and were expected to require inpatient care for greater than 48 hours. Exclusion criteria included: treatment with immunosuppressive medications; a history of known autoimmune or chronic inflammatory diseases; preexisting clinical or historical evidence of gastrointestinal pathologies or gastrointestinal infections; presence of perforated viscera or peritonitis; any form of ostomy, pica or similar psychological disorder; presence of a foreign body within the gastrointestinal tract, concurrent blunt/penetrating trauma, patients with comorbid malignancy; and pregnant or lactating women. The total body surface area (TBSA) was calculated from the standard Lund and Browder assessment (Murari & Singh, 2019) by trained clinical staff. Demographic information on participants were collected upon hospital admission (Table 1). Antibiotics were considered if they were administered by the hospital staff any time prior to the fecal sample being collected. Alcohol use disorder estimated based on AUDIT-C Questionnaire results.

Table 1.

Summary of Demographic and Collected Data (mean (± standard deviation))

Category	Low PEth (n = 12)	High PEth (n = 7)	Difference P value (test)
Age	49 (±21)	45 (±77)	0.799 (Wilcoxon)
Caucasian	92%	86%	0.314 (Chi-squared)
Male	83%	100%	0.714 (Chi-squared)
Days Sampled Past Injury	3.1 (±1.4)	4 (±1.6)	0.318 (Wilcoxon)
% TBSA	24 (±18.1)	38 (±24.1)	0.218 (Wilcoxon)
PEth concentration (mg/dL)	0.8 (±2.6)	134(±163)	>0.001(Wilcoxon)
BAC on Admission	0 (±0)	81(±131)	0.0304 (Wilcoxon)
Alcohol Use Disorder (+)	0%	100%	>0.001(Chi-squared)
Cigarette Use (+)	42%	43%	0.861 (Chi-squared)
Cannabis Use (+)	58%	40%	0.861 (Chi-squared)
Substance Use (+)	17%	43%	0.477 (Chi-squared)
Antibiotic Use (+)	25%	29%	1.0 (Chi-squared)
Total Hospital Days	31 (±30.6)	50 (±5.8)	0.237 (Wilcoxon)
Total ICU Days	18 (±32.1)	15 (±17.3)	0.832 (Wilcoxon)

Abbreviations: PEth = phosphatidylethanol; % TBSA = Percent total burn surface area, BAC = Blood alcohol content

Plasma Collection and Analysis

Blood collection tubes containing ethylenediaminetetraacetic acid (EDTA) were utilized to collect patient’s blood samples within 24 hours from hospital admission, as previously described (Davis, Janus, et al., 2013). Blood was immediately centrifuged to collect the plasma portion, aliquoted to ensure all samples only had one freeze-thaw cycle, and stored at −80°C in sterile cryovials. Cytokine levels were determined in the plasma samples using Bio-Plex Pro Human Cytokine Bio-48-Plex Screening Panel (Cat. No. 12007283, Bio-Rad, Hercules, California) in accordance with manufacturer guidelines at the Children’s Hospital Colorado Clinical Translational Research Centers Core Lab. Commercially available ELISAs for human intestinal fatty acid binding protein (iFABP) (Cat. No, DFBP20, R&D Systems, Minneapolis, Minnesota) and zonulin (Cat. No. MBS749365, MyBiosource, San Diego, California) were used to assess protein levels in the plasma. Samples were run in duplicate according to manufacturer’s instructions and absorbency at 450nm was measured using a Thermo Scientific MultiSkan GO plate reader. Phosphatidylethanol (PEth) concentrations were measured in whole blood samples spotted on cards and stored at room temperature prior to shipment to the United States Drug Testing Laboratories as described elsewhere (Afshar et al., 2017). Concentrations of PEth under 20 ng/mL were classified as “Low” and concentrations equal or greater than 20 ng/mL were grouped into a “High” category (Ulwelling & Smith, 2018). Statistical analysis of blood measures was conducted in R v. 3.5.1 (Team, 2021) with an additional statistical packages Hmisc (F, 2023) and visualized through ggplot2 (Wickham, 2016) and corrplot (Wei, 2021). Statistical comparisons between low and high categories of PEth and blood biomarkers were determined in ggpubr (Kassambara, 2023) with a generalized linear model (glm). Differentially abundant genera and blood measures was determined in a generalized linear and mixed model using MaAsLin2 (Mallick et al., 2021). Statistical testing was adjusted for age, gender, TBSA, antibiotic use, and substance use disorder (covariates). Multiple comparisons in p-values were considered for false discovery rate of multiple tests via the Benjamini-Hochberg (BH) method (Benjamini & Hochberg, 1995).

Microbiota Sample Collection, Processing, and Analysis

Fecal samples were collected from 19 patients at first available sampling point, between one to six days post-injury (mean = 3.5 days). Variation in bowel patterns due to injury precluded collection of samples at the same day post injury. Fecal samples were collected, transferred into sterile Eppendorf tubes, flash-frozen, and stored at −80°C until processed.

Sample DNA was extracted from microbiota samples using the QIAamp PowerFecal extraction kit (Cat. No. 51804, Qiagen, Valencia, California). Marker genes in isolated DNA were polymerase chain reaction (PCR)-amplified using GoTaq Master Mix (Cat.No. M5133, Promega, Madison, Wisconsin) with 338F F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806 R (5′- GGACTACHVGGGTWTCTAAT-3′) primer pair (Integrated DNA Technologies, Coralville, Iowa) targeting the V3/4 hypervariable region of the 16S rRNA gene, modified with a unique 12-base sequence identifier for each sample and the Illumina adapter. The thermal cycling program consisted of an initial step at 94 °C for 3 minutes followed by 35 cycles (94 °C for 45 seconds, 55 °C for 1 minute, and 72 °C for 1.5 minutes), and a final extension at 72 °C for 10 minutes. PCRs were run in duplicate and the products from the duplicate reactions were pooled and visualized on an agarose gel to ensure successful amplification. PCR products were cleaned, normalized, and paired-end sequenced an Illumina MiSeq with versions v2.4 of the MiSeq Control Software and MiSeq Reporter, using a 600-cycle version 3 reagent kit.

Sequences were processed and analyzed using the Quantitative Insights into Microbial Ecology program (QIIME2 v. 2022.11) (Bolyen et al., 2019). The DADA2 algorithm (Callahan et al., 2016) was used to denoise demultiplexed sequences with default parameters to process a mean of 105,068 sequences per sample (range 31,162 to 213,885). After quality control measures, a mean of 17,341 ASVs per sample (range 8,301 – 27,903) were aligned with mafft algorithm (Katoh & Standley, 2013) with a phylogenetic tree developed in fasttree (Price, Dehal, & Arkin, 2009). Taxonomic classification was conducted through a Native Bayes classifier, pre-trained on the 99% full 16S rRNA operational taxonomic units in the Silva database (v138) (Quast et al., 2012). Demultiplexed paired-end sequences were deposited in the NCBI Sequence Read Archive (BioProject accession ID: PRJNA1010693).

Analyses were performed with the R and QIIME2–2022 at a rarefaction level of 8,000 ASVs per sample. The α-diversity metrics assessed were (1) observed features, also called amplicon sequencing variants (ASVs); (2) Shannon diversity; and (3) Pielou’s Evenness. All statistical tests were conducted with a two-tailed alpha level of 0.05. Comparison between PEth categories and each α-diversity metric was done completed in R using the “glm” function for generalized linear modeling and adjusted covariates and BH. For β-diversity, analysis of principal coordinates was performed using the R vegan package (Oksanen et al., 2008) for Weighted and UnWeighted UniFrac (Lozupone & Knight, 2005). Statistical differences for PEth groups were calculated through pairwise PERMANOVA for each distance matrix using the adonis2 command with 10,000 permutations, adjusted for covariates and BH. Microbial measures of taxonomic relative abundance were aggregated at the phylum and genus levels. Differential relative abundances of taxa based on PEth category were assessed using “ancom-bc2” function (Lin & Peddada, 2020) with adjusting for covariates (prevalence filter 30%, p-values adjusted via BH).

Short Chain Fatty Acids

Fecal concentrations of short-chain fatty acids (SCFA) were analyzed by stable isotope gas chromatography–mass spectrometry using a method adapted from Lamarre et al. (2014). In brief, fecal samples were subjected to an alkylation procedure where the combined sample and alkylating reagent were vortexed for 1 minute and incubated at 60°C for 15 minutes. Following cooling and addition of n-hexane to allow for separation, 200 μL of the organic phase was transferred to glass and analyzed by gas chromatography–mass spectrometry. Results were quantified by reference to a standard curve and normalized to sample weight. Statistical analysis of SCFAs were conducted as described above for blood measures.

Results

Demographics of Participants

This study included a cohort of 19 participants that had a mean burn TBSA of 28.9% with a range of 1.1% to 73%. Variability was also observed in age, with a mean of 47.7 years with a range of 18–87 years. Compared to US national averages (SAMHSA, 2021), participants had increased prevalence of alcohol use disorder (36.8% this cohort vs. 11.3% US national average), substance use disorder (26.3% this cohort vs. 16.5% US national average), cannabis use (52.6% this cohort vs. 18.7% US national average), and cigarette use (47.4% this cohort vs. 15.6% US national average). Participants were grouped by PEth concentrations of low PEth (0–19 ng/mL, n = 12) and high PEth (20–200 ng/mL, n = 7). A summary of the metadata is shown in Table 1. The only statistically significant differences between low PEth and high PEth groups were related to alcohol.

Fecal Microbiota of Participants

Fecal microbiota were analyzed for all 19 burn injury participants. Overall, the most abundant phyla were Firmicutes (mean relative abundance 55.2%, standard deviation ± 11.8), Bacteroidota (27.9% ± 13.4%), and Verrucomicrobiota (5.6% ± 9.9%). The six most abundant known genera across for participants were Bacteroides (16.6% ± 12.5%), Akkermansia (5.6% ± 9.4%), Blautia (4.2% ± 3.2%), Parabacteroides (3.7% ± 5.6%), Alistipes (3.0% ± 2.4%), and Escherichia-Shigella (2.8% ± 8.9%).

PEth Levels Correlated to Differences in Gut Microbiota

Participants grouped into low PEth and high PEth showed differences in the fecal microbiota in this study. At the phyla level, the high PEth compared to the low PEth group had increased abundance of Firmicutes (63.8% vs. 50.3%) plus lower abundance of Bacteroidota (21.6% vs. 31.6%) and Verrucomicrobiota (2.4% vs. 7.5%). (Figure 1A, Figure 1) The Firmicutes/Bacteroides ratio was significantly different between the two groups (p = 0.046, glm) with lower levels in low PEth group (1.0 ± 1.9) compared to high PEth group (4.6 ± 5.3). (Figure 1B).

Figure 1.

(A) Differences in the relative abundances of the top six most abundant phyla between PEth groups, (B) Firmicutes to Bacteroides ratio between PEth groups. Box plots of % relative abundance are shown in each group, (C) Six most abundant phyla by PEth group and participant. For (A) and (B) median is solid line, box is interquartile between of 25^th - 75^th percentile, whiskers are 1.5 times interquartile. Based on differential abundant testing, none of these phyla were significantly different based on PEth groups.

The most abundant genera in the low PEth group were Bacteroides (19.5% ± 13.1%), Akkermansia (7.5% ± 11. 5%), Parabacteroides (4.7% ± 6.8%), Blautia (4.2% ± 3.6%), Escherichia-Shigella (4.0% ± 11.2%), and Alistipes (2.7% ± 2.4%). The high PEth group shared only four of the six most abundant genera with the low PEth group, and had a different order of abundance. The order of abundance of top genera in the seven participants of the high PEth group were Bacteroides (11.6% ± 10.3%), Subdoligranulum (5.0% ± 8.6%), Phascolarctobacterium (4.7% ± 11.2%), Blautia (4.0% ± 2.5%), Alistipes (3.5% ± 2.6%), and Akkermansia (2.3% ± 5.9%) (Figure 2 and Supplemental Figure 1).

Figure 2.

Differences in the relative abundances of the top nine most abundant genera between Peth groups. Box plots of % relative abundance are shown in each group (median is solid line, box is interquartile between of 25^th - 75^th percentile, whiskers are 1.5 times interquartile). Based on differential abundant testing, none of these genera were significantly different based on PEth groups.

Differential abundance testing was conducted between low and high PEth groups using ANCOM-BC2. When accounting for multiple testing, no significant phyla or genera were observed between low and high PEth groups. Prior to correction, Sutterella (p =0.02, ANCOMBC-2) was observed as different between groups of high PEth (mean 0.95% relative abundant, present in 86% of samples) and low PEth (mean 0.58% relative abundant, present in 42% of samples). Additional details on differential abundance is provided in Supplemental Figure 2.

Alpha diversity between the two groups was not significantly different in measures of Observed Features (p = 0.69, Wilcoxon). Shannon Diversity Index (p = 0.53, Wilcoxon), or Pielou’s Evenness (p = 0.72, Wilcoxon). (Figures 3A–C) Community composition approached statistically significant values for the two groups for weighted UniFrac (p = 0.106, PERMANOVA), but not for unweighted UniFrac (p = 0.536, PERMANOVA), indicating relative abundance of taxa was an important factor between the groups of low and high PEth. (Figures 3D and 3E) Microbial compositions for weighted UniFrac were not different based on covariates of age (p = 0.658, PERMANOVA), gender (p = 0.561, PERMANOVA), alcohol use disorder (p = 0.733, PERMANOVA), substance use disorder (p = 0.969, PERMANOVA), or total burn surface area (p = 0.729, PERMANOVA).

Figure 3.

Differences in PEth groups for (A) Observed Features, (B) Shannon Diversity Index, (C) Pielous’s Evenness, (D) Weighted Unifrac, (E) Unweighted UniFrac. (Statistical differences in A-C from generalized linear model, ellipse in D-E created through stat_ellipse command with level of 0.8).

Clinical outcomes were measured against total days spent in the ICU and total days spent in the hospital in regards to PEth levels. The microbial communities between the PEth groups were not significantly different in weighted UniFrac analysis for ICU days (p = 0.147, PERMANOVA) or hospital days (p=0.913). No significant genus was observed for the PEth groups and ICU days. Genus Eubacterium Hallii Group (p_FDR < 0.001, negative correlation), Sutterella (p_FDR < 0.001, positive correlation), and Anaerotruncus (p_FDR =0.009, negative correlation) were significantly correlated to PEth groups and total days in the hospital. Clinically, a negative correlation were taxa that had higher abundance at time of admission of participants that spent less time in the hospital (i.e. potentially beneficial).

Cytokines Levels and Intestinal Permeability Measures Poorly Correlated to PEth Groups

Plasma levels of ten cytokines were measured in the present study. Low and high PEth groups were statistically significantly different for IL-1ra (p = 0.035, glm) and IL-6 (p = 0.03 glm) (Figure 4). Cytokines that were nearly significant included MCP-1 (p = 0.06, glm), SDF-1a (p = 0.06, glm), and IL-8 (p = 0.08, glm). PEth groups had significantly different genera in IL-6 (Sutteralla p_FDR < 0.001, negative correlation & Anaerotruncus p_FDR =0.02, negative correlation), IL-1ra (Anaerostipes p_FDR = 0.025, negative correlation & Coprococcus p_FDR = 0.032, positive correlation), SDF-1a (Anaerostipes p_FDR = 0.006, positive correlation & Subdoligranulum p_FDR = 0.046, negative correlation), IL-1B (Eubacterium Hallii Group p_FDR < 0.001, positive correlation, Sutterella p_FDR = 0.008, negative correlation), IL-8 (Eubacterium Coprostanoligenes Group p_FDR < 0.001 negative correlation, Anaerotruncus p_FDR =0.02, positive correlation), MCP-1 (Sutteralla p_FDR < 0.001 negative correlation, Anaerotruncus p_FDR =0.02, negative correlation), M-CSF (Eubacterium Hallii Group p_FDR < 0.001 positive correlation), and eotaxin (Eubacterium Hallii Group p_FDR < 0.001 negative correlation).

Figure 4.

Differences in PEth groups for (A) IL-1ra, (B) IL-1B (C) IL-6, (D) IL-8, (E) IL-18, (F) Eotaxin (CCL11), (G) SDF-1a, (H) MCP-1, (I) M-CSF, (J) TNF-A. (Statistical differences from generalized linear model)

Two intestinal permeability measures were compared between PEth groups, iFAPB and Zonulin. Neither measure was significantly different between the groups. (Supplemental Figure 3) One genus was associated with PEth groups and iFAB, Sutteralla (p_FDR < 0.001, negative correlation). Zonulin did not have significantly different genera.

Short Chain Fatty Acid Concentrations Not Correlated to PEth Groups

Short chain fatty acid concentrations were measured in the fecal samples, normalized by the weight of the sample. For all participants, the highest concentrations were noted for acetate (38.6 mmol/kg ± 17.7), propionate (12.2 mmol/kg ± 7.5), and then butyrate (1.5 mmol/kg ± 1.2). The three short chain fatty acid concentrations were not significantly different between low and high PEth groups. (Figure 5A–C) Seven differentially abundant genera were identified between PEth Groups and butyrate including Bacteroides. After corrections, only Ruminococcus Gauvreaulii Group remained significant (Figure 5D). Akkermanisia was one of four genera that was significant for acetate; however, after statistical corrections only NK4A14 group remained significant. Five genera were associated with propionate and PEth groups with none remaining significant after corrections. Akkermansia was negatively effect in to all three SCFA models.

Figure 5.

Differences in PEth groups for (A) acetate, (B) priopionate, and (C) butyrate (statistical differences between PEth groups reported from generalized linear model), and (D) Correlation between SCFAs and most abundant and/or significantly associated genera (effect size and significance from MaAslin2, *** = p< 0.001, ** =p< 0.01, * p<0.05).

Correlations between SCFAs and select metadata, cytokines, and intestinal permeability measures were assessed for the low and high PEth groups separately. (Figure 6A–B) The low PEth group had significant, positive correlations for butyrate to IL-18 and IL-6. The high PEth group had significant, positive correlations for butyrate to IL-6, IL-8, and MCP-1. A negative correlation was observed for butyrate and SDF-1a in the high PEth group. Acetate and propionate did not have any significant correlations to the measures investigated.

Figure 6.

Correlation matrix between metadata (blue label), short chain fatty acids (purple label), cytokines (black label), and intestinal permeability (brown label) for (A) low PEth group and (B) high PEth group. Correlations and p-values calculated based on Pearson Correlation Coefficient (*** = p< 0.001, ** =p< 0.01, * p<0.05).

Discussion

Variations in gut microbiota due to recent alcohol use were observed post-injury in the present study. A clinically meaningful measure of total days in the ICU was associated with microbial community structure based on low and high PEth groups. These overall findings are particularly notable considering this study did not limit participants by known factors that influence the gut microbiota (e.g. age, diet, substance use abuse, etc). In addition, PEth influences on the fecal microbiota, cytokines, and SCFAs were evident in this cohort despite a small cohort and variability in burn injury size.

Alcohol use modulates the gut microbiota, potentially due to consumption of additional sugars and ethanol, malnutrition in heavy alcohol users, inflammatory processes, or disruption of intestinal barrier function. In the first several days of burn injury, the pre-existing microbial community in a cohort with alcohol use was a dominate factor in community structure in the present study. A short time between the burn injury and our sampling day might have limited changes found in the gut due to injury or hospitalization. However, even with a mean fecal sample provided 3.5 days post injury, it is notable that alcohol influences were apparent. The microbial community can change within days (Schlomann & Parthasarathy, 2019) from perturbations that include diet (David et al., 2014), osmotic shock (Fukuyama et al., 2017; Tropini et al., 2018), and antibiotics (Dethlefsen & Relman, 2011; Jakobsson et al., 2010; Palleja et al., 2018). Lima et al (2021) conducted a pilot study of ten burn injury participants with sampling on days 0, 3, 7, 14, 21, and 28. That study observed a microbiotas of individuals in two clusters; (1) days 0 and 3, and (2) days 7, 14, 21 and 28, potentially a delayed and substantial impact from 100% of participants receiving antibiotics.

Identification of microbial taxa that are consistently associated with alcohol use in human studies is non-trivial. A global measure of gut dysbiosis, the Firmicutes to Bacteroides ratio, was increased in the high PEth group, although not significantly. The use Firmicutes to Bacteroides ratio in alcohol use disorder has had mixed findings (Day & Kumamoto, 2022), possibility due to its connection to comorbidities (e.g. obsesity, type II diabetes, age, and cardiovascular disease) (Magne et al., 2020). In a review, Day and Kumamoto (2022) examined alcohol use disorder and gut microbiota with generalized observations of taxa abundance from ten alcohol use disorder studies. Our results between low and high PEth groups observed the same general trends at the phylum level of increased Firmicutes and decreased Verrucomicrobia and Bacteroides in those that use alcohol in excess. Likewise at the genus level, the present study also replicated findings for alcohol use of increased Lactobacillus and decrease in both Akkermansia and Bacteroides. These results again highlight a persistent microbial community, associated with alcohol use, that endures at least several days post burn injury.

Specifically with regard to the taxa, we observed lower relative abundance of Akkermansia and Bacteriodes in the high PEth group. Akkermansia muciniphila is an intestinal commensal which promotes barrier function in part by enhancing mucus production (Grander et al., 2018). A disruption of intestinal permeability has been noted in studies of burn injuries (Earley et al., 2015; Peng, Yan, You, Wang, & Wang, 2004; Wrba, Palmer, Braun, & Huber-Lang, 2017; Ziegler, Smith, O’Dwyer, Demling, & Wilmore, 1988) and exacerbated in the ileum when alcohol was administered prior to the injury in an animal study (Choudhry, Fazal, Goto, Gamelli, & Sayeed, 2002; Zahs et al., 2012). In one mouse model, an aged burn group had a reduction in Akkermansia and a corresponding increase in mRNA expression of neuroinflammatory markers in the brain (Choy et al., 2023; Walrath et al., 2023). Severe disruption of the barrier function post burn injury increases systemic bacterial translocation (Ziegler et al., 1988), and could lead to sepsis (Adiliaghdam et al., 2020). In the present study, the relative abundance Akkermansia was not identified as a factor associated with reduced intestinal permeability through plasma iFABP and Zonulin levels, perhaps due to the burn injuries or other confounding factors (e.g. alcohol use, age, diet, etc). Akkermansia was negative correlated with all three measured SFCA concentrations, a known modulator of maintaining intestinal barrier (P. Liu et al., 2021). In this small scale study, it is unclear if pre-existing Akkermansia might provide a protective effect and improve long-term burn injury outcomes; however, it is a potential future research area as there has been increasing interest in utilizing Akkermansia in a probiotic role for other conditions (Naito, Uchiyama, & Takagi, 2018; Zhou, 2017). Bacteroides, in contrast to Akkermasia, is a genus that contains a large number of species and strains that can impart different effects (Carrow, Batachari, & Chu, 2020), including anti-inflammatory properties of some strains (Qu et al., 2022; C. Wang et al., 2022). Several burn injury researchers have hypothesized that improved healing depends on a balance of pro-inflammatory and anti-inflammatory mediators (Antonacci et al., 1984; Jeschke et al., 2020), an observation supported by animal models of burn injury (Bird & Kovacs, 2008; Chen et al., 2013; Fontanilla et al., 2000; Zahs, Bird, Ramirez, Choudhry, & Kovacs, 2013).

It is established that alcohol use promotes pro-inflammatory cytokine production (Adams, Conigrave, Lewohl, Haber, & Morley, 2020; Choy et al., 2023; Waldschmidt, Cook, & Kovacs, 2008). We observed non-standard correlations in cytokines that differed between low and high PEth groups. Overall, the high PEth group, compared to the low PEth group, had higher percent of negative correlations with cytokines, SFCAs, intestinal permeability, and clinical outcomes. Notably, IL-1ra, an anti-inflammatory cytokine (Dinarello, 2018), was suppressed in the high PEth group, indicating a potential disruption in the proinflammatory immune response to burn injury. Gut microbes might contribute to that response through the bi-directional signaling in the gut-brain-axis. For example, Sutterella had higher relative abundance in the high PEth group and was negatively correlated to IL-6. Sutterella has been associated with higher obesity, smoking, and depression (L. Liu et al., 2023; Manor et al., 2020). In the present study, a higher abundance of Sutterella was also correlated to longer time spent in the hospital.

Short chain fatty acid concentrations were similar between low PEth and high PEth groups. Previous research has reported mixed results in terms of SFCA and alcohol consumption. In one study, chronic alcohol users had higher combined SCFA levels than day workers but lower levels than night workers (Swanson et al., 2020). Another study reported lower levels of SFCAs for alcohol consumers versus healthy controls (Bjørkhaug et al., 2019) while Gonzalez-Zancada and colleagues (2020) observed higher levels of butyrate in beer consumers compared to abstainers. A lack of difference between SCFA concentrations was also observed in our earlier study on TBSA and age groups (Dyamenahalli et al., 2022). Increased SCFA concentrations are believed to provide anti-inflammatory properties that assist in supporting intestinal barrier integrity (Kelly et al., 2015). A lower concentration of SCFA has been noted for critically ill patients with sepsis (Valdés-Duque et al., 2020), the leading cause of death in burn patients (Nunez Lopez, Cambiaso-Daniel, Branski, Norbury, & Herndon, 2017). Therefore, SFCA might provide a universal therapeutic treatment either directly or through SCFA-producing bacteria for burn-injury with or without prior alcohol use.

Several limitations and strengths were present in this study. The overall participant size and small group size limited the power to detect statistically significant and biologically relevant associations with the two PEth groups. Additional studies are needed with larger sample size and broader range in PEth concentrations to confirm our preliminary findings. Moreover, PEth is a record of only several weeks and therefore not a historical record of alcohol use. The PEth groups also had significant differences for other measures of alcohol (e.g. blood alcohol content on admission and a clinical diagnosis of alcohol use disorder); therefore, similar microbiota results might be observed in other alcohol related analysis. A strength of this paper is the combination of 16S rRNA sequencing, short chain fatty acids, cytokines, and intestinal permeability measures to create an in-depth dataset for each participant. Finally, this is the first study to evaluate the combined effects of burn injury and alcohol consumption on the gut microbiota, SCFAs, and cytokines.

5. Conclusion

A recent history of alcohol use at the time of burn injury represents an additional negative confounding factor in the health outcomes of a patient. An understanding of PEth levels and microbiota at the initial intake of a burn injury patient might provide support for evidence-based interventions that can reduce the severity of the injury and duration of hospitalization. More research is warranted in the field of burn injury, alcohol, and gut microbiota to develop a better understanding of the factors associated with negative clinical outcomes, with a goal in the future of interventional trials.

Highlights.

• Recent history of alcohol use prior to burn injury has an impact on fecal microbiota

• Increased evidence of burn injured alcohol users having lower Akkermansia and Bacteroides

• Overall, short chain fatty acid and intestinal permeability marker concentrations were not correlated to alcohol use in burn patients

• Several genera were significantly different for butyrate between the two groups of burn patients who are alcohol users