Impact of the gut microbiome on immunological responses to COVID-19 vaccination in healthy controls and people living with HIV

The study addresses how baseline gut microbiota diversity and composition influence immunological responses to the BNT162b2 mRNA SARS-CoV-2 vaccine in healthy controls and people living with HIV (PLWH). Given that vaccine efficacy is affected by age, comorbidities, and immunodeficiencies, and that PLWH often show reduced vaccine immunogenicity, the authors hypothesized that gut microbiome features predict humoral (spike-specific IgG) and cellular (spike-specific CD4+ T-cell) responses after vaccination. The work aims to identify microbial markers linked to vaccine responsiveness and to explore whether microbiome modulation could enhance vaccine immunogenicity, particularly in PLWH.

The introduction summarizes extensive prior evidence linking gut microbiota to vaccine responses. Bacterial stimulation is necessary for proper mucosal immune development; dysbiosis is associated with immune imbalance (e.g., Th2 activation or Treg deficiency). Specific taxa such as Bifidobacterium can augment cellular immunity in the elderly and enhance NK/PMN function. Clostridium clusters XIVa/IV stimulate TGF-β for Treg activation; butyrate from clostridia promotes Treg differentiation and anti-inflammatory functions. Microbiota-derived adjuvants (e.g., flagellin, LPS) activate PRRs (e.g., TLR5), enhancing antibody responses; germ-free or antibiotic-treated mice show reduced vaccine responses that can be restored by flagellated E. coli. Nod2 recognition of commensals enhances mucosal adjuvant activity during oral vaccination (elevated IL-1β). Epidemiologic and interventional studies link microbiome composition to vaccine immunogenicity in infants (polio, BCG, tetanus, HBV, rotavirus). Known host risk factors (aging, obesity, comorbidities) reduce COVID-19 vaccine efficacy; elderly display lower spike antibody titers. Immunocompromised populations, including PLWH, have lower COVID-19 vaccine immunogenicity. Trials are investigating microbiome-targeting strategies (e.g., 5-ALA-phosphate; Bifidobacterium strains) to improve SARS-CoV-2 vaccine responses.

Design and cohort: Open-label, non-randomized clinical trial (COVAXID) at Karolinska University Hospital, Stockholm, Sweden (EudraCT 2021-000175-37). Ethical approval: Swedish Ethical Review Authority (ID 2021-00451); written informed consent obtained. The broader cohort included six groups; this analysis focused on healthy controls (HC) and PLWH. Baseline fecal samples were collected prior to vaccination. Participants: Initially HC (n=90) and PLWH (n=90). Exclusions: detectable baseline anti-spike antibodies, antibiotic treatment within 3 months before vaccination, or missing day 35 spike IgG data (excluded PLWH n=22; HC n=15). Final baseline characteristics table: HC n=75, PLWH n=68 (overall n=143). PLWH were on ART for a median 9–10.8 years; median CD4+ T-cell count 615 cells/mL; 86% with HIV RNA <50 copies/mL. Vaccination and sampling: Two doses of BNT162b2 (Pfizer-BioNTech) administered 3 weeks apart. Blood collected at baseline and day 35 (two weeks after second dose). Fecal samples collected at baseline in RNA/DNA shield. Immunological assays: Spike IgG measured using Elecsys Anti-SARS-CoV-2 S (Roche). Spike-specific CD4+ T-cell responses quantified at day 35 by activation-induced marker (AIM) assay (CD69 and CD40L upregulation). Total IgG quantified by routine diagnostic methods. In PLWH, CD4+/CD8+ T-cell counts and HIV viral load assessed (flow cytometry; Cobas Amplicor). Microbiome sequencing and processing: DNA extracted with ZymoBIOMICS DNA Kit. 16S rRNA gene sequencing on Illumina MiSeq. Quality control with FastQC; adapters/low-quality reads trimmed with Cutadapt. Denoising, read merging, chimera removal via DADA2 in QIIME2 using SILVA v132 for taxonomy. Microbiome analytics: Alpha diversity (Observed, Shannon, Simpson) via phyloseq; beta diversity using Bray-Curtis distances and NMDS; PERMANOVA (vegan Adonis). Relative abundance computed in QIIME2. Differential taxa by LEfSe; visualization with ggplot2 and GraPhlAn. DESeq2 used to model absolute abundance and estimate log fold changes associated with high vs low antibody levels, independent of age, gender, and disease group. Correlations by Spearman with FDR adjustment; network visualization in Cytoscape. Statistical analyses: Univariate and multivariate linear regression assessed relationships between spike IgG and baseline covariates (alpha diversity indices, age, gender, BMI, total IgG) with stepwise backward selection; multiple hypothesis testing controlled by FDR. MANOVA with post hoc Kruskal-Wallis used for multivariate patterns. Significance threshold p<0.05.

- Immunogenicity by group and age: HC had significantly higher spike IgG titers than PLWH at day 35 (p=0.0001). Younger individuals (18–39 years) had higher antibody titers than middle-aged (40–59; p=0.003) and older (>60; p<0.0001); age negatively correlated with spike IgG (p=0.04). In PLWH, spike IgG was not associated with CD4+ counts or CD4/CD8 ratio. - Alpha diversity and spike IgG: Participants were split at median spike IgG (1972 U/mL). High responders had lower alpha diversity (Observed p=0.05; Shannon p=0.016; Simpson p=0.01) and lower phylogenetic diversity (Faith’s PD p=0.02). Negative linear associations between alpha diversity and spike IgG were observed. Beta diversity showed limited overall shifts; in HC, NMDS2 scores differed between high and low responders (p=0.0002). - Taxa associated with spike IgG (whole cohort): High responders were enriched in Agathobacter (p=0.02), Lachnospira (p=0.03), and Lachnospiraceae FCS020 group (p=0.03). Low responders were enriched in Butyricimonas (p=0.02), Cloacibacillus (p=0.009), Intestinimonas (p=0.02), Ruminococcaceae DTU089 (p=0.006), and Paraprevotella (p=0.02). - Group-specific taxa with spike IgG: In HC high responders, Bacteroidetes increased (p<0.01) with higher Bacteroides and Sutterella and decreased Alloprevotella, Anaerofilum, Succinivibrio, Moryella, Negativibacillus, and certain Ruminococcaceae (all p<0.05). In PLWH high responders, Flavonifractor, Lachnospira, and Oscillibacter increased (p<0.05); low responders showed higher Butyricimonas and Paraprevotella (p<0.05). Low IgG across both groups was associated with higher Hydrogenoanaerobacterium, Methanobrevibacter, Cloacibacillus, and Ruminococcaceae DTU089. - Spike-specific CD4+ T-cell responses: Individuals with higher CD4+ AIM responses (median split 0.36%) had lower alpha diversity (Shannon p=0.045; Simpson p=0.025) with negative linear associations (Observed p=0.003; Shannon p=0.001; Simpson p=0.005). Beta diversity differed (p=0.01). Low CD4 responders had increased Firmicutes (p=0.005; LDA>4.5) and decreased Bacteroidetes (p=0.005; LDA>4). Low responders were enriched in Ruminococcaceae (p=0.02; LDA>4.5), Erysipelotrichaceae (p=0.04; LDA 3.5), Akkermansiaceae (p=0.03; LDA 3.5), Akkermansia (p=0.035), Fournierella (p=0.014), and Alistipes (p=0.029). High responders had increased Lachnospira (p=0.035). In HC, high responders had more Lactobacillaceae (p=0.01; LDA>3) and Lactobacillus (p=0.014) and fewer Akkermansiaceae (p=0.02; LDA>3.5). In PLWH low responders, Firmicutes increased (p=0.01; LDA>4.5), Bacteroidetes decreased (p=0.046), and Erysipelotrichaceae (p=0.01), Eggerthellaceae (p=0.006), Succinivibrionaceae (p=0.036), and Fournierella (p=0.03) were enriched. Marvinbryantia was higher in low CD4 responders in both groups (p<0.05). - Correlations with antibody levels: Positive correlations with Sutterella (p=0.01), Bifidobacterium (p=0.015), Bacteroides, Lachnospira, Lactobacillus; negative correlations with Escherichia-Shigella (p=0.015), Marvinbryantia (p=0.002), Ruminococcaceae DTU089 (p=0.018), Methanobrevibacter (p=0.028), Cloacibacillus (p=0.015). - Age-microbiome relationships: Younger adults (18–39) had lower diversity than older (>60) (Observed p<0.001; Shannon/Simpson p<0.0001) with positive association between age and alpha diversity (Observed p=0.0002; Shannon p=1e-5; Simpson p=0.0001). Beta diversity differed by age (p=0.03). Older individuals had higher abundance of certain Ruminococcaceae members (p<0.05), Butyricimonas (p=0.01), Ruminiclostridium (p=0.005), Hydrogenoanaerobacterium (p=0.005), Fournierella (p=0.009), Christensenellaceae R_7 group (p=0.007), Methanobrevibacter (p=0.007). Younger adults were enriched in Lachnospira (p=0.02), Bacteroides (p=0.02), Agathobacter (p=0.009). Bacteroidetes were higher in younger (p=0.001) and Firmicutes higher in older (p=0.006; LDA>4.5); Firmicutes/Bacteroidetes ratio highest in elderly. - Baseline total IgG: Higher total IgG associated with lower diversity. Differential genera included Anaerostipes (p<0.01), Fournierella (p<0.01), Mitsuokella (p<0.05), Lactobacillus (p<0.05). High total IgG showed increased Anaerostipes (p<0.01; LDA>3), Bacteroides, and Bifidobacterium. - Network and DESeq2: Spike IgG positively associated with Lachnospira, Faecalibacterium, Bifidobacterium; Agathobacter, Lactobacillus, Bacteroides, and Lachnospira correlated with both spike IgG and spike-specific CD4 responses. Age-associated microbes (Hydrogenoanaerobacterium, Methanobrevibacter, Ruminococcaceae DTU089, Butyricimonas) negatively correlated with both immune readouts. DESeq2 showed higher Bifidobacterium (p=0.03), Faecalibacterium (p=0.03), Blautia (p=0.03), Catenibacterium (p=0.03), Hungatella (p=0.005) in high IgG; Cloacibacillus was higher in low IgG (p=0.0001). - Regression analyses: Univariate associations with spike IgG were significant for Observed (p=0.05), Shannon (p=0.01), Simpson (p=0.01), and age (p=0.03). In multivariable models, Shannon (p=0.02) and Simpson (p=0.019) remained significant independent of age; Observed combined with age was significant (p=0.04). Alpha diversity also significantly impacted spike-specific CD4 responses independent of age, gender, and total IgG (Observed p=0.006; Shannon p=0.0009; Simpson p=0.004).

The findings support a robust association between baseline gut microbiome features and the immunogenicity of the BNT162b2 vaccine in both healthy individuals and PLWH. Contrary to the common assumption that higher microbial diversity is uniformly beneficial, lower alpha diversity at baseline correlated with stronger humoral (spike IgG) and cellular (spike-specific CD4+ T-cell) responses. Specific taxa enriched in high responders (e.g., Bacteroides, Lachnospira, Agathobacter, Bifidobacterium, Faecalibacterium) have recognized immunomodulatory roles and may promote effective vaccine responses, potentially through endogenous adjuvant activity and SCFA-mediated immune regulation. Conversely, taxa enriched in low responders (e.g., Cloacibacillus, Methanobrevibacter, Ruminococcaceae DTU089, Succinivibrio, Marvinbryantia) align with patterns linked to aging and inflammatory states and were negatively associated with vaccine responses. Age-related microbiome shifts (higher Firmicutes and certain archaeal/bacterial taxa in older adults, higher Bacteroidetes in younger adults) paralleled the observed decline in vaccine immunogenicity with age. Importantly, multivariable analyses indicated that microbial diversity measures (particularly Shannon and Simpson) were independently associated with spike IgG responses beyond age, and similar independent effects were seen for CD4 responses. Overall, the data suggest that specific compositional features, rather than global richness, may be critical determinants of vaccine responsiveness, and that microbiome-targeted strategies could be leveraged to enhance mRNA vaccine immunogenicity, including in PLWH.

This study provides initial evidence that baseline gut microbiome diversity and composition are associated with immunogenic responses to the BNT162b2 mRNA COVID-19 vaccine in healthy controls and PLWH. Lower alpha diversity and enrichment of taxa such as Bifidobacterium, Faecalibacterium, Bacteroides, Agathobacter, and Lachnospira were linked to higher spike IgG and CD4+ T-cell responses, while Cloacibacillus and other age-associated taxa correlated with poorer responses. These associations were partly independent of age. The findings highlight candidate microbial markers of vaccine responsiveness and support the concept that microbiome modulation (e.g., via probiotic or other interventions) could optimize SARS-CoV-2 vaccine immunogenicity. Future research should include longitudinal, mechanistic, and interventional studies to test causality, evaluate specific microbial strains or consortia, and determine whether targeted microbiome modulation improves vaccine responses across diverse populations, including PLWH.

The study was open-label and non-randomized, and analyses are observational, limiting causal inference. Microbiome samples were collected only at baseline, precluding assessment of temporal dynamics post-vaccination. 16S rRNA sequencing was used (rather than shotgun metagenomics or metatranscriptomics), limiting taxonomic resolution and functional inference. Subset sizes for some analyses (e.g., CD4+ T-cell responses n=90) were smaller than the overall cohort. Participants were from a single country and health system, which may affect generalizability. Individuals with recent antibiotic use were excluded, which may limit applicability to those populations.