Coronavirus disease 2019 (COVID-19) is a highly contagious infectious disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus, a novel RNA beta-coronavirus, with more than 663 million cases and 6.71 million deaths worldwide documented until 20 January 2023 (1). COVID-19 is mainly a respiratory illness, ranging from asymptomatic, mild-moderate, severe, and critical illness (2), especially affecting elderly subjects with underlying medical conditions (3).

After COVID-19, some patients may experience persistent symptoms or other conditions that are colloquially referred to as long-COVID. The Centers for Disease Control and Prevention have defined post-COVID conditions as new, returning, or ongoing symptoms that people experience ≥4 weeks after being infected with SARS-CoV-2 (4). The prevalence of these conditions varies widely from 5% to 80%, and the most frequently reported symptoms are fatigue, cough, shortness of breath, and chest pain (2, 5). Furthermore, half of the patients report persistent neurological symptoms at 6 months, the most frequent being “brain fog” and cognitive changes, described in up to one-third of subjects (6).

With regard to the gastrointestinal (GI) tract involvement, early reports from Wuhan showed that 2% to 10% of patients with acute COVID-19 had GI symptoms including nausea and diarrhea (7), but more recent metaanalyses reported a higher prevalence, up to 20% of patients (8). SARS-CoV-2 virus has been detected in anal swabs and stool samples in almost 50% of patients with COVID-19, suggesting that the digestive tract might be an extrapulmonary site for virus replication and activity (9), through ACE2 receptors binding with spike protein-S.

The human gut microbiota harbors up to 10¹⁴ resident microorganisms, including bacteria, archaea, viruses, fungi, and other eucaryotes, with bacteria being the most abundant microorganisms at the gut level. The most represented phyla at the gut level are Firmicutes, Bacteroidetes, Actinobacteria, Proteobacteria, Verrucomicrobia, and Fusobacteria (10). An increase in bacteria has been documented from duodenum to colon, with a decrease in facultative anaerobic Bacilli (Firmicutes) and Enterobacterales (Proteobacteria) taxa and an increase in obligate anaerobic bacteria, especially Bacteroidia (Bacteroidetes) and Clostridia (Firmicutes) classes (11, 12).

Gut microbiota is crucial for several functions, such as energy extraction from the diet, vitamin and short-chain fat acids (SCFAs) production, and immunomodulation, with the regulation of TH17 and T reg balance (13–15). A complex equilibrium exists among prebiotics, like microbiota accessible carbohydrates (MAC), probiotics, and postbiotics, like their products, SCFAs (16, 17), with involvement of several networks between gut microbiota and other body sites through axes (i.e., gut-lung, gut-liver, gut-brain axis), influencing processes in health and disease.

An unbalance of the crucial homeostasis between Firmicutes, Bacteroidetes, Actinobacteria, and Proteobacteria phyla ( Figure 1 ) is often associated with a change in the numbers of microbes and/or diversity of the microbiota; such a condition is defined as dysbiosis (18). Recently, a new definition of dysbiosis has been suggested, based on a model represented in several diseases, defined by the increase in facultative anaerobic bacteria, like Bacilli class and Enterobacterales order, and a parallel decrease in obligate anaerobic bacteria, such as propionate and butyrate-producing bacteria (BPBs) (11).

Gut microbiota dysbiosis can have a role in several disease models affecting the lung, brain, liver, and heart (19).

In the last decade, research on lung microbiota and its pathogenetic link to pulmonary conditions has significantly improved. Previously, the lung has been considered a sterile organ; however, numerous studies have demonstrated the presence of bacterial DNA in the lower respiratory tract in healthy individuals. The lung microbiota of healthy subjects is characterized by the presence of differentiated ecological niches belonging to Bacteroidetes, Firmicutes, and Proteobacteria phyla and Prevotella, Streptococcus, Veillonella, Fusobacterium, and Haemophilus genera (20). Its balance is the result of acquisition and clearance ( Figure 1 ). Many other factors contribute to this complex mechanism, such as the immune system (innate and adaptive immune recognition, secretory IgA), in addition to various exogenous components such as diet, environmental biodiversity, and drug treatments, in particular antibiotics (21).

Chronic respiratory diseases are often characterized by an imbalance between microbial immigration and elimination in the lung. Moreover, the presence of chronic inflammation results in the alteration of physicochemical proprieties that facilitate the growth of select species in the microbial community, such as microorganisms from the Proteobacteria phylum, that are linked to a proinflammatory state (22). It is important to emphasize that lung and gut microbiota are in close communication with each other through the circulation of soluble metabolites (i.e., peptidoglycan or LPS) transported by the blood (21). These peptides are recognized by host cells that express pattern-recognition receptors (PRRs), such as Toll-like receptors (TLRs) and Nod-like receptors (NLRs). The interplay between lung and gut microbiota, defined as the gut-lung axis, has been demonstrated in different animal models (23–26).

Further studies are needed to better understand the complex gut-lung interplay and characterize the gut microbial metabolites (i.e., indole derivative, niacin, polyamines, urolithin, and pyruvic acid) that act as immunomodulants and might have a possible impact on respiratory health (27, 28).

Another captivating field of microbiota studies is related to its connection with the brain through the so-called gut-brain axis, which is thought to be a bidirectional system. On one side, there is the involvement of microbiota-derived metabolites on the blood–brain barrier like SCFAs, tryptophan, and linoleic acid metabolites as well as cytokines produced at the gut level; on the other side, the brain controls gut activity through the neuroendocrine and parasympathetic systems (i.e., regulation of intestinal permeability through the vagus nerve) (29). Such connections have been studied in animal models: physiological aging affects gut microbiota in mouse models through cognitive frailty (30).

Gut microbiota dysbiosis seems to play a role in several neurodegenerative and psychiatric disorders (31), as well as in other neurological conditions (32). For example, damage to the GI barrier is a possible pathological pattern for depression disorders; moreover, increased LPS and microbiota-cytokine production seems to be related to Alzheimer’s disease (29).

The relationship between gut microbiota and the brain could be deeper and more complex: alteration of the hypothalamic “master clock” could impact the diurnal environmental fluctuations and lead to dysbiosis-related metabolic disorders like obesity and/or diabetes (33). Furthermore, gut dysbiosis could determine sleep disturbances (sleep loss, alteration of circadian rhythm), eventually leading to fatigue (34). Following this hypothesis, the gut microbiota, which is mostly influenced by diet, could represent a link between the immune and endocrine systems through brain function and the host metabolism (35).

High-fat food intake can indeed damage the GI barrier, affecting both the “intestinal epithelial barrier” (characterized by the mucus layer and the epithelial cells) (36) and the “gut vascular barrier,” regulated by the expression of plasmalemma vesicle-associated protein-1 (PV1). This condition, known as “leaky gut,” can favor microbial translocation to the liver (37), leading to hepatic and systemic disease.

Finally, another example of the role of dysbiosis in disease has been studied in the cardiological setting, where the increased production of trimethylamine (and its metabolite-liver trimethylamine-N-oxide) by gut microbiota has been linked to the development of cardiovascular disease (29).

We identified 22 studies on gut microbiota in COVID-19 patients published in a 2-year window period between 03 January 2020 and 03 January 2022 ( Table 1A ).

To critically revise the studies, we first considered all the variables potentially influencing the final observations: study design, location, material source, microbial technology used, sample size, and patient characteristics—age, body mass index (BMI), gender, sexual behaviors, COVID-19 severity index, comorbidities, recent previous use of antibiotics/probiotics, diet, and lifestyle.

The cross-sectional study design was the most common. Less than half of studies (45%) had a longitudinal/prospective design, 20% of which focused on long-COVID-19.

The study location was a critical factor: most studies (19/22, 86%) were set in Asia (18 in China, one in South Korea), and three of 22 (14%) in Europe; no other geographic regions were represented.

Lifestyle and diet were not analyzed, even though both factors are crucial elements in shaping microbial core composition (32, 59, 60).

The material source was a fecal sample in 19/22 (86%) studies, while three of 22 (14%) were based on anal swab analysis. Most studies (12/22) used next-generation sequencing (NGS) technology through ribosomal-S16-DNA hypervariable region sequencing (V4 or V3–V4 regions preferred) to analyze microbiota; shotgun metagenomic sequencing was used in seven of 22 studies, whereas one study used multi-omics methodologies (55), one study nanopore technology (38), and another used quantitative PCR (39).

Regarding patients’ characteristics, all studies included both men and women, but no studies considered sexual behavior, although its impact on microbiota core is known in several disease models (61, 62). Only one-third of studies (seven of 22) included BMI data, and control groups, when included, were often matched for BMI. Fifty percent of the subjects in the studies, 50% were aged 50 or younger.

The small sample size was a limit reported by several authors, with a total number of enrolled subjects below 40 in almost two-thirds of studies 13/21 (62%). The COVID-19 severity index was reported by most studies, with high heterogeneity in the works analyzed.

Scarce data were available on comorbidities and concomitant medications; hypertension was the most commonly reported, followed by diabetes.

No data were generally reported on COVID-19 vaccine status for subjects enrolled after the introduction of the vaccine; only one study investigated the microbiota changes in two groups of patients vaccinated with two different vaccines (58). During hospitalization, both antibiotics and/or antiretroviral treatments and probiotics were administered in several studies; however, these data were not critically investigated in most published studies.

After assessing the possible confounding factors, we compared the gut microbiota features according to two ecological measures, α-diversity and β-diversity, in association with relative abundance results.

In humans, α-diversity measures the level of diversity within individual samples; it includes several indexes gathered in two groups: richness indexes (Faith index, Observed and Chao-1 index) and evenness indexes (Shannon index, Peliou’s evenness, Simpson, and inverse Simpson indexes) (63, 64).

In parallel to other disease models, α-diversity at the gut level, more frequently described with richness indexes (like Chao-1), resulted in a global reduction in all COVID-19 patients compared to controls (see details in Table 1A ). An interesting study observed this reduction already in the acute phase of the disease (48). On the contrary, Yeoh et al. (43) did not report alterations in α-diversity indexes, even though they enrolled most COVID-19 patients with a mild or moderate severity index (90% of patients).

In a Korean longitudinal analysis performed on patients who were asymptomatic or affected by the mild disease, an increase in α-diversity (Peliou’s evenness) was observed in the recovered subgroups compared to infected patients (51). Interestingly, Xu et al. (46) observed a trend toward increased bacterial diversity from the early to late stages of COVID-19 in a 35-day longitudinal analysis of inpatients with mild disease. Furthermore, the same study described an interesting synchronous restoration of microbiota in both gut and upper airways, suggesting a possible role of the gut-lung axis.

Moreira-Rosario et al. (53) described a reduced α-diversity gradient trend (Shannon index) from mild to severe COVID-19 patients, and Chen et al. (48) showed how richness was not restored to a normal level even after 6 months in 30 COVID-19 patients (one-third with severe disease), although a trend toward healthy controls was noticed.

β-Diversity measures the level of diversity (or dissimilarity) between samples, mostly by using a Permanova analysis (65, 66). All the studies showed a difference between COVID-19 patients and controls, in general, and according to different severity index categories.

Mazzarelli et al. (44) have shown a difference in β-diversity among patients hospitalized in regular wards compared to ICU patients and hospitalized no-COVID-19 controls, although no data on prior antibiotic intake was gathered. Regarding this aspect, two studies (9, 43) compared microbiota composition in COVID-19 patient subgroups (with and without antibiotics) with healthy controls, confirming a separation among groups, with high heterogeneity revealed in the antibiotic subgroup.

Regarding relative abundance analysis, several studies described a significant reduction in Firmicutes members, especially for BPBs (both Lachnospiraceae and Ruminococcaeae families, mostly Faecalibacterium prausnitzii) in COVID-19 patients compared to no-COVID controls, while discordant data have been reported about Erysipelotrichaceae and Veillonellaceae taxa.

Conversely, several facultative anaerobic bacteria like members of the Bacilli class, resulted in increased growth, mostly in the Enterococcaceae family as well as Streptococcaceae and Lactobacillaceae ( Table 1 ). Contrasting data have been described regarding the Bacteroidetes phylum during COVID-19, with some works reporting an increase in Bacteroidetes phylum with a consequent reduction of the Firmicutes/Bacteroidetes ratio (53) as opposed to other studies reporting a reduction in taxa belonging to this phylum. Other factors, like diet and/or antibiotics, could play a role in these findings, highlighting the importance of assess for confounding factors when considering the study results.

Reduction in the Actinobacteria phylum, including the Bifidobacterium genus and Collinsella genus (recently associated with SARS-CoV-2-ACE2 binding inhibition), represents another significant finding in COVID-19 studies (67). The Bifidobacterium genus was found to be increased only in three studies (notably, in one study, a probiotic including this taxon was administered (56)), while the Collinsella genus resulted was increased in a few other studies (40, 45, 49); the reason for this last difference is not clear. Proteobacteria resulted increased in almost all studies performed on COVID-19 patients, although some authors have described an increase in Enterococcaceae/Enterobacteriaceae ratio (39), probably linked to the use of antibiotics. Finally, the Akkermansia genus (Verrucomicrobia), a propionate-producing bacterium genus with anti-inflammatory features, resulted in reduced COVID-19 (but not in all studies). To note, the severity of COVID-19 disease seems to emphasize differences in the relative abundance of gut microbiota, although most studies included asymptomatic/mild/moderate categories.

Few studies tried to investigate α-diversity alterations during long-COVID: in this setting, Zhuo et al. (52) reported a reduced Shannon index in a 15-patient cohort, followed up for 3 months with at least one persistent COVID-19 symptom. Coherently with these findings, in a 6-month follow-up, Liu et al. (57) have confirmed in long-COVID patients both a persistently reduced α-diversity (Shannon and Chao-1 indexes) and different gut microbiota clusters compared to controls. Notably, the subgroup who had COVID-19 at baseline without developing long-COVID did not show the same dysbiosis pattern. Reduced BPBs were reported in both COVID-19 subgroups compared to controls, but only in the long-COVID subgroup the microbial composition was different compared to controls at 6-month follow-up ( Table 1A ). Interestingly, the authors found no correlation between viral load in the gut and respiratory levels and long-COVID development at 6 months, nor did they find any effect of previous antibiotic intake. On the contrary, in the long-COVID subgroup, increased fecal relative abundance of opportunistic pathogens was positively associated with fatigue, respiratory and neuropsychiatric symptoms, while decreased other anti-inflammatory/BPB taxa was negatively correlated with long-COVID at 6 months. Coherently, Zhuo et al. (52) described both a negative correlation between some taxa (Faecalibacterium prausnitzii, Intestinimonas butyriproducens) and chronic respiratory symptoms as well as a positive correlation between Proteobacteria members and long-COVID symptoms.

Persistent dysbiosis in long-COVID and its pathogenic role still need to be studied in humans, while rodent and non-human primate animal models of COVID-19 already showed long-term changes in both lung and gut microbiome (82, 83). The influence of gut microbiota on neurological symptoms, via the gut-brain axis, has been investigated in the animal model since the early decades of the new millennium. In murine models, Bercik at al. suggested that gut microbiota could influence the behavior of mice (84). Recently, Carloni et al. identified a closing in the choroid plexus vascular barrier during gut inflammation, suggesting a link between intestinal inflammation and neurologic/psychiatric symptoms, like a deficit in short-term memory and anxiety-like behavior (85). Moreover, a recent review summarized three different arms of inflammation for the gut-brain axis in a non-COVID-19 setting, where the systemic humoral pathway, cellular immune pathway, and neuronal pathway are involved (86). By translating these inflammatory patterns to the long-COVID setting, where gut dysbiosis persists at least after 6 months of follow-up, we can conclude that this microbial imbalance plays a role in maintaining both a chronic inflammatory status at the gut level and favoring the development of neurological/neuropsychiatric symptoms, as seen in the animal models mentioned above. However, it is not clear which immunologic pathway is dominant during long-COVID. It is plausible that several factors could coexist in the same disease model: (a) reduction in BPBs leading the butyrate loss linked to neuropsychiatric disorders (87); (b) development of the cytokine release syndrome during COVID-19, in particular with increased kynurenine:tryptophan ratio, already linked to depression syndrome (88); and (c) changes in l-DOPA production, regulated by ACE2 activation at the gut level (89).

There is still a lack of evidence on the role of microbiota dysbiosis in respiratory symptoms during long-COVID. Shortness of breath, frequently experienced by subjects after recovery from primary SARS-CoV-2 infection, could represent a clinical manifestation of the fibrosis secondary to chronic inflammation of lung parenchyma, leading to reduced total lung capacity. Such a condition is already linked to gut dysbiosis in non-COVID patients, as described in a recent review (90).