According to the Assessment of SpondyloArthritis (SpA) International Society (ASAS) and the European League against Rheumatism (1,2), SpAis a group of associated chronic systemic inflammatory immune-mediated rheumatic diseases affecting axial and peripheral joints and entheses. SpA includes different disease phenotypes: Ankylosing spondylitis (AS), reactive arthritis (ReA), arthritis/spondylitis with inflammatory bowel disease, psoriatic arthritis and undifferentiated (u)SpA (1,2). Extensive progress has been achieved in the understanding of SpA pathogenesis in the last two decades, including the complex interaction between environmental, immune, genetic and epigenetic factors (3–8). Furthermore, the gut and/or urogenital microbiota, representing one of the most important environmental factors, was revealed to have a much more complex configuration than previously thought, when studied by using modern high-throughput techniques, including genomic and metagenomic approaches (9–11). Infections with pathogens including Yersinia spp., Salmonella spp., Campylobacter spp., Shigella spp. and Chlamydia trachomatis have been reported at increasing rates worldwide (12,13), with a subsequent increase in the number of patients at risk that are prone to developing extra-intestinal complications. ReA was reported in 2–10% of patients with enteritis. Besides usual pathogens in the enteric and/or urinary tract (14), several other infectious agents were incriminated as triggering factors in SpA in a smaller percentage, including certain anaerobes (Porphyromonasgingivalis, Prevotellaintermedia) (15), Borrelia spp. (16), human immunodeficiency virus (17) and hepatitis C virus (18). Despite of remarkable scientific progress, the pathogenesis of SpAhas remained to be fully elucidated. It was demonstrated that genetic factors, including human leukocyte antigen (HLA) B27 (19–22), certain phenotypes of endoplasmic reticulum aminopeptidase-1 (19,21,22), interleukin (IL)-23 receptor (21,23) and tumour necrosis factor (TNF) gene polymorphisms (24) have an important role in the development of the disease and/or in persistence of bacterial components at different anatomic sites of the host. The persistence of bacterial antigens of Salmonella, Yersinia and Chlamydia trachomatis, including the presence of bacterial endotoxin and/or DNA in synovial fluid, was identified in patients with ReA (25). Differences in the serumantibody pattern amongst patients may suggest differences between triggering factors and various clinical forms of SpA (26,27). The aim of the present study was to improve the laboratory-based information in order to better understand the correlation between the disease activity/severity and the microbiological results/immune status against intestinal and/or urogenital pathogens.

The present study aimed to investigate whether the different patterns of clinical manifestation and disease progression in SpA may be explained by differences in initial infectious triggering factors and whether the overall dynamics of the microbial/host association in patients impacts their genetic predisposition to developing SpA. Several previous studies demonstrated that a fraction of patients initially diagnosed with ReA further developed AS. For instance, Herrlinger and Asmussen (34) reported that 5% of the patients with ReA triggered by Yersinia spp. developed AS and another 5% of patients with pre-existing AS progressed. Sairanen et al (35) reported that in patients presenting with ReA triggered by Shigella spp. 20 years earlier, 14% had AS and 32% had radiologically confirmed sacroiliitis. In addition, up to 20% of patients with ReA developed AS after 10–20 years, as reported by Leinsalo-Repo (36). In a 10-year observational study, Amor et al (37) reported that radiologically positive sacroiliitis was observed in ~30% of patients with Reiter's disease, usually triggered by urethritis.

The clearest indication of the association between the intestinal microbiota and joint damage in SpA is the triggering of ReA following previous gastrointestinal infections with bacteria including Salmonella, Shigella, Yersinia and Campylobacter (38).

In addition, it should be noted that bacterial components from previous infections may be recovered from the joints of patients who developed ReA (39–41). Hypothetically, the detection of intestinal pathogens associated with autoimmune diseases, including Salmonella spp., Yersinia spp., Campylobacter spp., Klebsiella pneumoniae, Shigella spp. and fungi, in different biological samples including saliva, synovial fluid or feces would aid in the identification of triggering factors and assessing the evolution of the specific immune status towards these bacterial pathogens. According to previous studies, triggering microorganisms may be cultivated from feces during a variable time interval following the initial infection (42,43). Serological methods including ELISA may also be useful to identify proof of previous infectious episodes in patients who tested negative by culture-based methods (43).

It is preferable to demonstrate an increasing titer of IgG antibodies or high persistent levels of IgA in patients with autoimmune diseases, given the fact that IgA usually has a shorter half-life than IgG. The presence of IgG autoantibodies confirms a previous contact with the corresponding bacteria (41).

Several studies have suggested that Yersinia, Salmonella and Chlamydia trachomatis are the most common ReA-triggering bacteria, and that these bacteria are also present in patients with undifferentiated arthritis (44–48). Therefore, an increase in the titer of specific bacterial antibodies is considered to be an important criterion for identifying any triggering bacteria for these autoimmune diseases. In the present study, the principal results obtained by re-assessing the microbiological and immunological status and disease activity/functional scores of the 27 patients with SpA at different time-points following the onset of disease, generally confirmed the previously described results; however, they also provided additional information that will be discussed hereinafter.

The distribution of patients of the present study by sex and the presence of HLA-B27 were similar to those in previously reported studies (19–22). Though obligate enteric pathogens were only identified in 1 patient with SpA and in none of the subjects of the control group, the present study observed slight differences regarding the presence of opportunistic Klebsiella species, which were only present in the SpA group (12.0% of feces samples). In addition, the distribution of Escherichia coli isolates by phylogenetic groups exhibited slight differences from that reported in the previous literature, with B1 present only in the SpA group, and the E group only being present in the control group. The patients and control subjects with the B2 phylogenetic group of Escherichia coli in stool samples also exhibited the same phylogenetic group in their urine samples and reported frequently occurring symptoms, including urinary tract infections. Previous data also support the dominance and the etiological role of Klebsiella spp, which was isolated from the stool samples of patients with AS (49,50). A recent study has reported that Escherichia coli strains of phylogenetic groups B2 and D are able to produce virulence factors, which may disturb the immune response, while the phylogenetic group B2 may impede the regulatory mechanism of toll-like receptor signaling (51). A genetic molecular investigation was used in the present study for the detection of Clostridium spp. and Aeromonas spp. genes, as these bacterial genera have been implemented in the etiology of SpA (52). It should be noted that by using the PCR method, the presence of Salmonella spp. in the feces of the HLA-B27 positive patients who had been diagnosed with Salmonella reissen by traditional methods (culture) was not detected. This may be due to the presence of PCR inhibitors in the feces, or due to the inability to detect the cultivated species by the nucleotide probe used in the reaction kits. This result supports the requirement to combine classical and molecular methods for microbiological investigations in patients with SpA.

In the present study, several patients who had initially been diagnosed with ReA following an intestinal infection developed AS, whereas other patients, who had already been diagnosed with AS and who were also positive for anti-bacterial antibodies exhibited disease progression, which confirmed the results of a previously published study (37). A total of 18 of the 27 patients with SpA tested positive for at least 1 bacterial specific antibody, while the concomitant presence of 2 or 3 different bacterial-specific antibodies was identified in 9 out of the 27 patients. In the present study, over half of the patients (n=16) were positive for Yersinia antibodies, of which 6 were positive for IgG and IgA antibodies, and 7 tested positive for other types of antibodies concomitantly. There was a marked difference between the number of Yersinia-positive antibodies between the patients and the control group.

The positivity for IgA-specific bacterial antibodies was betterreflected by the ASDAS_CRP and BASFI than by the BASDAI. However, 2 of the 6 patients with SpA had highly modified inflammatory score, with one of these patients exhibiting high BASDAI (>4), BASFI (>6) and ASDAS_CRP (>2.1) values. The inflammatory markers were better associated with the presence of anti-Yersinia antibodies in the 16 HLA-B27-positive patients, and 8 out of the 16 patients had CRP values >3 times the upper limit of normal. All HLA-B27-positive patients exhibited high BASDAI, BASFI and ASDAS_CRP.

Previous studies have demonstrated that in Yersinia-induced acute ReA, IgG antibodies associated with IgA antibodies may be detected in nearly 100% of the patients (53–55). The persistence of IgA antibodies in the serum of patients with Yersinia-induced arthritis was observed for 14–16 months following the initial diagnosis in 84% of patients (56), and high levels of IgA are directly correlated with the severity of the disease (56,57). In patients without arthritic symptoms, these antibodies disappeared within 5 months. In addition, anti-Yersinia IgG antibodies were more persistent in arthritic patients when compared with non-arthritic patients; however, not as consistently as IgA antibodies (56). This previous study also demonstrated that IgM antibodies persist for only 1–3 months following the onset of infection (54), and these antibody levels persisted for longer among patients with ReA (9–14 months) and patients with undifferentiated arthritis, when compared with those with uncomplicated enterocolitis (4 months) (58). In salmonellosis, all classes of IgA, IgG and IgM antibodies were observed to persist, unlike Yersinia-triggered ReA, where only IgA antibodies persisted, particularly in the IgA2 subclass (59).

Campylobacter species are able to induce gastrointestinal and systemic diseases in humans, and have not only been associated with acute disease, but also with a wide range of chronic inflammatory diseases (60). Patients with SpA may have IgG anti-Campylobacter antibodies in the same range as those of healthy subjects, and this may be due to poultry consumption in the general population (41). In the present study, 4 out of 5 patients who tested positive for anti-Campylobacter antibodies were also positive for other bacterial antibodies detected in the serum. However, no strains of Campylobacterspp. were detected in the analyzed feces.

High levels of Salmonella antibodies towards several Salmonella antigens have been reported previously (46,61). A study by Durán-Avelar et al (62) revealed that 71.4% of Salmonella IgG and IgA antibody-positive AS patients and 14.3% of control subjects, the proteic 30-kDa band of Salmonella typhimurium was recognized by using human anti-IgG western blot analysis. However, not all Salmonella species are associated with SpA. In addition to the previously mentioned p30 antigen of Salmonella typhimurium (63), it has been reported that the Salmonella typhi heat shock protein 60 antigen is also involved in the pathogenesis of SA (64). However, the association between the observed antibody levels and the mechanisms involved in the pathogenesis of AS have remained to be completely elucidated. In the sera from patients with SpA and healthy controls, all 4 IgG subclasses were revealed to be involved in p30 recognition of Salmonella typhimurium, while only the IgG3 antibody response to p30 was significantly different in patients with AS when compared with healthy subjects (63). The involvement of outer membrane protein H of Salmonella spp. in SpA pathogenesis was not supported by the presence of antibodies or the cellular immune response to this peptide (65). The association of Salmonella spp. with SpA is also supported by the presence of Salmonella spp. antigens in the synovial fluid of patients with SpA (66). However, the behavior of Salmonella typhimurium is influenced by the presence of HLA-B27 in transfected cells, leading to the increased production of interleukin (IL)-6, IL-8 and IL-10, and the attenuated production of TNF-α (67–69).

Previous studies have suggested that mucous membranes and their immune systems serve a pathological role in the development and maintenance of numerous auto-inflammatory diseases, including SpA (70–72). Salas-Cuestas et al (71) reported that there was a higher mean SgA concentration in patients with ReA and uSpA, independent of the HLA-B27 allele and disease duration. In the study of Granfors and Toivanen (72), there was a significant difference in the serum levels of SIgA, particularly anti-Yersinia enterocolitis in patients who developed arthritis following a confirmed infection. In the present study, the percentage of subjects with anti-Salmonella antibodies detected in serum did not differ between the controls and patient groups. This may be explained either by contact with Salmonella from food sources or by the fact that several subjects from the control group were laboratory workers who were repeatedly exposed to enteric pathogens.

The study of Salas-Cuestas et al (71) identified the presence of serum IgA anti-Chlamydia trachomatis antibodies levels higher than the reference value in 20% of patients with SpA. Conversely, none of the subjects in the control group had elevated levels of anti-Chlamydia trachomatis antibodies in serum SIgA or IgA. The same authors observed anassociation between serum inflammatory markers, including high CRP, and the serum levels of SIgA in uSpA and ReA (71). In 1996, Wendling et al (73) reported similar results regardingSIgA and CRP in patients with AS. A moderate correlation between the clinical indices of disease activity, i.e. BASDAI and ASDAS_CRP, and the serum concentration of SIgA was demonstrated. These positive correlations were stronger in the HLA-B27-positive subgroup and in patients with symptoms that had persisted for over 12 months. In addition, the strongest association was in the HLA-B27-positive subgroup. The abovementioned authors demonstrated that an increase in the serum concentration of SIgA translates into a better clinical control of the disease, arguing that the serum concentration of SIgA modulates the clinical status of patients with ReA and uSpA. A longer monitoring period would provide sufficient evidence to determine whether a high and persistent level of total IgA is able to modulate the progression of the disease to structural sacroiliac involvement or the development of extensive axial inflammation, as suggested by Franssen et al (74). The study by Cowling et al (75) confirmed the conclusions of several previous studies, demonstrating that serum IgA is increased in AS, and they reported that these high IgA values are predominantly associated with AS active phases, which were indicated by high CRP values. In the present study, 3 patients were detected to have Chlamydia trachomatis antibodies (IgA), 4 out of 5 patients were HLA-B27 positive, and these patients also exhibited other antibody types, had a high presence of inflammatory markers and a high BASFI score. Patients who tested positive for IgG anti-Chlamydia trachomatis antibodies were HLA-B27-positive and had high inflammatory markers with a CRP >3 times the upper limit of normal, high disease scores, including BASDAI, BASFI and ASDAS, and ocular manifestations; however, these patients were only positive for the IgG class of antibodies. In the healthy control group, 1 out of 26 subjects harbored anti-Chlamydia trachomatis antibodies, i.e. IgA and IgG. In addition, Chlamydia trachomatis was not detected in the urine of subjects from the patient or the control groups by genetic methods.

Different immunological studies performed by independent groups from 16 different countries revealed that antibodies against Klebsiella pneumoniae, but not against any other microorganisms, are significantly elevated among patients with SpA compared with those in patients with other diseases (hospital outpatients with fractures) or with healthy controls (76–80); however, these results have not been confirmed by any other studies. Smith et al (81) identified no correlation between the presence of Klebsiella in the fecal flora of patients with primary AS or in association with inflammatory bowel disease, ReA or psoriatic arthritis, and disease activity, CRP or ESR. In addition, the levels of Klebsiella antibodies were identified to be significantly higher in serum than in synovial fluid samples taken from patients with AS (82). The differences between these results may be explained by the differences between the collection and culture methods for feces and the disease status at the time-point of sampling. Furthermore, a Finnish study reported that high levels of anti-Klebsiella IgA antibodies in patients with AS were correlated with the degree of intestinal inflammation (80). It is well documented that there is a strong link between intestinal inflammation and AS (83), and an increased concentration of Klebsiella pneumoniae isolated from the feces of patients with AS with peripheral arthritis and ocular disease has been demonstrated previously (73). In addition, there is evidence of high levels of IgA against Klebsiella antigens in patients with active AS (53,83). The results of these studies linking Klebsiella, collagen and HLA-B27 to AS may explain various predominant clinical, genetic and immunological features present in patients with this disease (76). In the present study, paradoxically, the controls had a significantly higher prevalence of anti-Klebsiella antibodies than the patients with SpA. Of the 10 SpA patients positive for anti-Klebsiella pneumoniae antibodies, a high presence of inflammatory markers was identified in 4 patients, and the high BASDAI and BASFI values were correlated with the presence of inflammatory markers. In a 33-year-old patient from the present study, cultures dominated by Klebsiella pneumoniae were detected in the second sample of feces.

In the present study, 6 patients, being HLA-B27-positive, had concomitant ocular complications, and it was noted that in 3 of these patients, a Klebsiella pneumonia-dominant culture was isolated from feces. The ASDAS_CRP was >2.1, which is indicative of intestinal inflammation. Inflammation of the bowel may be a source of contamination and responsible for the transportation of bacterial components to various sterile sites in the body, including the joints and eyes. In the present study, the concomitant presence of 2 or 3 types of antibodies were detected in 3 out of 6 uveitis patients.

The involvement of bacterial antigens, particularly Yersinia spp., in thyroid autoimmunity was hypothesized as early as the 1970s. In 1976, Shenkman and Bottone (84) reported titers of >1:8 in 75% of 67 patients in the USAdiagnosed with a variety of thyroid disorders, in contrast with the low prevalence of antibodies in controls (>8%). The most prevalent antibodies developed against serotype 3 occurred in the highest titers, and were commonly identified in patients with Graves' disease (84).

An Italian study published in 2017, which aimed to assess the molecular mimicry theory in autoimmune thyroid disease hypothesized 40 years ago, revealed that ‘in genetically pre-disposed persons and under particular conditions, molecular mimicry between microbial and human antigens can turn a defensive immune response into autoimmunity’ (85). The authors also performed an in silico search for amino acid sequence homologies between several microbial antigens and thyroid autoantigens. These Bioinformatics-based results support the overlap between bacterial protein antigens and known autoepitopes, and the occurrence of specific HLA-DR binding motifs, and may explain the previously suggested triggering role of Borrelia, Yersinia, Clostridium botulinum, Rickettsia prowazekii and Helicobacter pylori. New data provided by the abovementioned study suggests the potential pathogenic importance of Toxoplasma gondii, certain Bifidobacteria and Lactobacilli, Candida albicans, Treponema pallidum and the hepatitis C virus in autoimmune thyroid disease (82).

However, a Danish study performed in 2006, investigating whether a correlation exists between the presence of Yersinia-specific antibodies and thyroid antibodies in a twin population, advanced the opinion that ‘Yersinia infection does not confer an increased risk of thyroid antibodies and the genetic contribution in the acquisition of Yersinia infection is modest’ (86).

On the other hand, in a comprehensive PhD thesis defended in 2008 by Strieder (87) entitled ‘The Amsterdam autoimmune thyroid disease cohort’ demonstrated that despite the fact that genetic factors exert pivotal functions in the pathogenesis of autoimmune thyroid diseases, environmental risk factors, including low birth weight, iodine excess and deficiency, selenium deficiency, parity, oral contraceptive use, reproductive span, fetal microchimerism, stress, seasonal variation, allergy, smoking, radiation damage to the thyroid gland as well as viral and bacterial infections all serve pivotal roles in the development of autoimmune thyroid disorders.

In the present study, only 2 of the 16 anti-Yersinia antibody-positive patients were detected to be positive for ATPO and 2 for ATG antibodies. Of note, 2 of the patients in the control group who had anti-Yersinia antibodies also tested positive for ATPO antibodies.

A very complex approach was promoted in 2017 by Bliddal et al (88), who advanced the concept of ‘polyautoimmunity’, culminating in the hypothesis of a ‘generalized dysregulation of immune system’. They concluded that the complex mechanism may be described by a ‘Swiss cheese model’, involving coincident immunity holes made from specific HLA, immunoregulatory and thyroid-specific genes, as well as environmental and existential factors, which may contribute to autoimmunity (88). This novel concept may be useful for clinical practitioners with respect to reconsidering the significance of the concomitant presence of thyroid autoantibodies in SA and taking into consideration therapeutic schedules addressing the underlying immune dysregulation in the context of polyautoimmunity.

In the present study, the most notable data provided by the complex bacteriological phenotypic/genotypic and serological investigation indicated that, compared with the controls, patients with SpA had a markedly higher prevalence of phylogenetic group B1 Escherichia coli in stool cultures (16.0 vs. 0%), a higher prevalence of anti-Yersinia antibodies in the serum (81.5 vs. 42.3%) and a higher prevalence of anti-Campylobacter antibodies in the serum (25.9 vs.3.9%). IgA anti-Yersinia antibodies were significantly better represented in patients with SpA than in control subjects. In addition, there were more patients positive for anti-bacterial serum antibodies in the upper categories of SpA disease activity when compared with the group with inactive disease. In certain patients with SpA with high clinical scores, thyroid-specific autoantibodies were also identified.

The difference in age and gender composition betweenthe two groups of the present study (SpA and controls) is unlikely to have significantly influenced any of the factors investigated, since a previous studyindicated that important changes in the gut microbiota occur in subjects aged <20 and >70 years (elderly type) (86).

In conclusion, SpA are systemic inflammatory diseases with complex pathogeny. Intestinal microbiota alteration (in conjunction with genetic predisposition and abnormal immune activation) is one of the mechanisms proposed for the initiation and development of SpA. The detection of anti-bacterial antibodies combined with other complex specific laboratory investigations may provide valuable information for better diagnosis, monitoring of the disease activity or for the prediction of disease progression and treatment in SpA and should be used more extensively to monitor patients with SpA in association with their symptoms and treatment options. Further data are required in order to propose modification of the current laboratory evaluation and standard of care.