Infection, inflammation, and the irritable bowel syndrome

1. Introduction 

Infectious diarrhea is one of the commonest afflictions of mankind. Worldwide most of the burden, about 1 billion cases a year, is seen in children <5 years old [1], the vast majority in the developing world in communities where access to clean water and adequate sanitation is restricted. Here a child can expect to have 6–7 episodes per year compared to 1–2 in the developed world [2]. Following recovery from an episode of gastroenteritis (GE) the vast majority of healthy adults and children develop some degree of immunity to the organism responsible and return to normal functioning. However 7–31% develop post-infectious irritable bowel syndrome (PI-IBS). The proportion of unselected IBS that is post-infectious varies from 6 to 17% in the USA and Europe [3] but whether this differs in the developing world is unknown, though previous enteric infection is a known risk factor for IBS in Southern China [4].

This review will compare the epidemiology of infectious diarrhea in the developing and developed world and the link between mucosal inflammation and the development of IBS symptoms. The available evidence suggests that the acquisition of immunity in early childhood reduces the severity of subsequent gastroenteritis in adulthood. Since these are known risk factor for developing PI-IBS we hypothesize that this may underlie some of the regional differences in the incidence of both infection and IBS.

2. Epidemiology of IBS worldwide 

There are three key questions. Firstly is the incidence of IBS less in the developing world, secondly is the incidence increasing with the adoption of a western urban life style and finally is the disease itself different? The answer to all three is probably yes though interpretation of cross-cultural surveys is fraught with problems relating to the imprecise translation of questions into different cultures. Initial reports from small uncontrolled studies suggested that IBS was very uncommon and predominantly affected a subpopulation who pursued a “western life style” [5]. More recent and robust work gives a range of values for prevalence from very low in Iran and India with just 5.8 and 4.2% respectively [6], [7], to values in developed Asian countries that are generally lower but not dissimilar to those seen in the west [4], [8], [9]. The key factors associated with rapid westernization that underlie this increase in numbers is unclear but could include the effect of improved hygiene, increased overcrowding, stress and changes in diet. The best evidence comes from studies in which the same populations have been studied over a number of years as has been done in Singapore where after a decade of steady industrial growth the prevalence of IBS has risen from 2.3% [10] to 8.6% [8]. While female preponderance is the norm the recent Indian Society of Gastroenterology Task Force Survey of IBS reported a male/female ratio of 3/2 even in patients who had not sought medical help, thus controlling for the increased use of medical services by men [7]. This may reflect an overlap with dyspepsia with its male predominance since over half the Indian IBS patients reported epigastric pain, while the overlap with dyspepsia in the USA is only approximately 1/3 [11].

Finally there is a most important fourth question, why should these differences occur? It is clear that major differences in the epidemiology of gut infection exist between the west and the developing world. This is illustrated by Campylobacter jejuni enteritis, which causes a shorter, less severe illness in childhood than in adulthood, which is when most Europeans and North Americans are infected. The greater degree of inflammation which adults experience may increase the risk of developing subsequent PI-IBS which might partly account for the higher prevalence of IBS in the westernized nations.

3. Infectious diarrhea 

Worldwide the average number of episodes of infection annually per person is 3 [1]. A poorly nourished child living in cramped conditions without access to sewerage and running water will have 8 or more infections in the first year of life, most frequently with enteric bacteria and parasites [12] whereas a child in better sanitary conditions would have less infections and these would be more likely to be viral in origin. Even in England an estimated 1 in 5 people per year have an episode of diarrhea in the community adding up to 9.4 million cases in total a year, largely unreported since only 1 in 30 present to their doctor [13]. It seems here that viral infections predominate in the very young, with bacterial infection particularly Campylobacter spp. being most common in adolescence and early adulthood. PCR analysis of stool in the same study showed that Norovirus and Rotavirus were the commonest pathogens detected across all age groups. Campylobacter spp. were most commonly found in age group 30–39 (16% compared to 6.7% of those aged 1–4) [14].

Early studies of Campylobacter spp. enteritis in the UK described a double peak in incidence, the first in early childhood between ages 1–4 years and the second peak aged 15–24 [15]. By contrast in developing countries, infections occurs earlier in life with Campylobacter spp. the most commonly isolated bacterial pathogen from children [16] under the age of 2, with a rapidly falling incidence of symptomatic infection as immunity is acquired so that the disease is unimportant in the adult population [17].

4. Mechanisms of diarrhea 

Infectious diarrhea results from either an increase in fluid and electrolyte secretion, predominantly in the small intestine, or a decrease in absorption which can involve both the small and large bowel. During a diarrheal illness these two mechanisms frequently co-exist. Enterotoxins from Vibrio cholerae or enterotoxigenic E. coli induce profuse secretion while decreased intestinal absorption can be induced by mucosal injury caused by enteroinvasive organisms (e.g., Salmonella, Shigella, and Yersinia spp.). These invasive infections injure cells and excite an immune response and activate enteric nerves and mast cells resulting in an acute inflammatory infiltrate with the release of pro-inflammatory mediators and stimulation of secretion. Clinically the patient will have an acutely inflamed mucosa with ulceration and bleeding.

Viral infection on the other hand most often results in short-lived watery diarrhea associated with remarkably little tissue damage. Norovirus infection causes increased enterocyte turnover and apoptosis leading to a 50% reduction of villous surface area, weakening of tight junction, increased permeability, all associated with an increase in intraepithelial lymphocytes but not polymorphs [18].

As will be discussed later, the development of post-infective functional gut disorders depends on the extent and distribution of inflammation, duration and timing of infection (be it childhood, as a young adulthood or in older age) as well as psychological and immunological factors. The relative importance of these factors is most easily described by looking in detail at a single agent. C. jejuni which is ubiquitous worldwide and has been the subject of frequently studies.

5. Campylobacter jejuni 

This organism produces a range of toxins including cytolethal distending toxin [19], that first produces a secretory diarrhea in the small intestine in the early part of the illness after which there is invasion of the distal ileum and colon to produce an inflammatory ileocolitis, which can extend all the way to the rectum [20].

The disease is less severe in developing countries than in developed countries, with watery stool, fever, abdominal pain, vomiting and dehydration predominating as opposed to the severe abdominal pain, weight loss, fever and bloody stool that is seen more frequently in infections in the west [16]. Infants usually have milder disease with less fever and pain [21], which in some cases is due to immunity acquired during previous infection. The reasons for these differences between the developed and developing world are unclear. The organisms do not appear to be less virulent since travellers from Europe who acquire infection abroad have just as severe symptoms as those acquired at home [22]. The duration of diarrhea is greater in the developed world with more prolonged excretion of bacteria compared to developing world [23]. C jejuni infections in Sweden were in general only with a single serotype and were always symptomatic with re-infection being rare whereas in Mexico infections were frequently with mixed serotypes, and re-infection with new serotypes often resulted in no symptoms suggesting the development of immunity [24] which may also prevent bloody diarrhea [25].

6. Disturbances of gut microbiota in IBS 

The composition of the resident intestinal microbiota is highly variable between individuals but relatively stable for each individual [26], though IBS patients show a more unstable microbiota. This instability may be due to antibiotic therapy [27] or alterations in diet, both of which are commoner in IBS. Patients given antibiotics are 4 times more likely than untreated controls to report bowel symptoms 4 months later [28], and antibiotic use is a risk factor for developing IBS with an adjusted OR of 3.70 (1.80–7.60) [29]. Antibiotic use increases the incidence of post-infective functional diseases following both Salmonella enteritidis [30], [31] and travellers’ diarrhea, in whom antibiotic treatment gave a relative risk of developing PI-IBS of 4.1 (1.1–15.3) compared with those not receiving treatment [32].

During acute infectious diarrhea there is a decrease in anaerobes [33], [34]. Mice infected with Citrobacter rodentium or C. jejuni or subjected to a chemically induced colitis show significant reduction in the total numbers of microbiota, which is mainly due to activation of the host immune response and only to a much lesser degree by bacterial factors [35]. This loss of anaerobes is associated with a depletion in short chain fatty acids and an increase in the pH of the stool allowing overgrowth of other organisms which may contribute to disturbed bowel function.

7. Post-infectious irritable bowel syndrome (PI-IBS) 

The study of patients with PI-IBS has yielded many new insights for several reasons. Firstly the patients are a more homogenous group than unselected IBS, most having diarrhea with fewer psychological problems than unselected IBS [36]. Secondly the direction of causation is easier to ascertain as they represent a “natural experiment”, with subjects “randomised” to receive an infection, thus producing an unbiased study group. Finally onset of symptoms on a clearly defined date in a previously well patient provides an opportunity to examine the prior host and bacterial factors that predispose to developing IBS.

While the majority of patients make a full recovery from their infectious diarrhea, about a quarter experience a persistent change in their bowel habit [37] and in 7–36% the change meets the Rome II criteria for IBS [38]. Post-infectious IBS (PI-IBS) usually follows bacterial infection with Salmonella spp. [39], C. jejuni [36], [40] or Shigella spp. [41], the highest reported incidence of PI-IBS, 36% being in the Walkerton outbreak, in which the infection was particularly severe, combining both C. jejuni and E. coli 0157 [42]. Another series with a high incidence (31%) included patients hospitalized for infectious disease who are likely to have the most serious infections [43], much higher than the 7% incidence in those studied in a similar UK community, only 10% of whom were ill enough to be hospitalized [37].

Evidence of IBS after infection with protozoa, worms and viruses is scant. A single study reports a 13.9% prevalence of new IBS symptoms after a large outbreak of infection with the Whip worm, Trichinella britov [44]. A similar study after an outbreak of presumed viral gastroenteritis at a gastroenterology conference suggests that IBS can occur after a viral infection, but that this is a more transient phenomenon. While 23.6% reported symptoms consistent with PI-IBS at 3 months versus 1 of 29 (3.4%) who remained well there was no difference between infected and uninfected at 6, 12, and 24 months [45].

A large waterborne outbreak of Giardia intestinalis infection in Norway resulted in 84 (7%) developing a post-infective functional gastrointestinal disorder, mostly IBS but around 2/3 also having functional dyspepsia, predominantly of the postprandial distress subtype [46].

There are a range of post-infective symptoms that seem to reflect the area of the gut that is worst affected by the original insult. Thus G. intestinalis being predominantly a duodenal infection is associated with post-infectious dyspepsia while Shigella spp. causes a predominantly distal colonic inflammation with diarrhea as the main post-infective symptom. Salmonella spp. and C. jejuni, affecting mainly the mid-gut are intermediate, producing both post-infective dyspepsia and IBS.

8. Severity of injury 

The duration of illness and severity are important risk factors in the development of PI-IBS. An illness lasting greater than 3 weeks confers a relative risk of 11.4 (95% confidence interval (CI), 2.2–58) compared to one lasting less than 7 days [37]. A smaller study after Shigella spp. enteritis gave a relative risk of 4.6 (2.1–9.9) for infections lasting >14 days compared to those with a duration of <8 days [47]. The toxicity of the infecting bacteria also plays a major role since patients whose cultured C. jejuni supernatants demonstrated toxicity to an in vitro cell line had a relative risk of developing persistently deranged bowel habits of 12.8 (95% (CI) 6.1–101) compared with those who had no toxin [40].

9. Immune response in childhood and old age 

The severity of injury is mediated not only by factors related to the infecting organism but also by the host's own immune response which develops in early life and declines in old age. However little is known about the incidence of PI-IBS in the paediatric population and whether it is different to the condition seen in adults. Functional bowel disorders are common in children, with IBS affecting 14% of high school and 6% of middle school patients in a US community study and are classified according to the main complaints made by parents or children rather than in an organ-specific way [48]. This makes comparisons with the adult population difficult however a single recent study reports a very high incidence of post-infectious symptoms in 88 children with positive bacterial stool culture results presenting to a single institution. These had a 36% prevalence of functional gastrointestinal disorders compared to 11% in age- and sex-matched healthy controls. This is much higher than most adult studies with the exception of the Walkerton outbreak [42]. Unlike adults, female gender is not a risk factor for PI-IBS in children suggesting the gender effect depends on hormonal and/or psychosocial factors rather than being genetic [49].

Despite uncertainty about PI-IBS in childhood we do know that age in adulthood does have an effect on the likelihood of developing PI-IBS. A meta-analysis indicates that patients who develop PI-IBS are slightly younger [50] and one study showed increasing age was protective with age >60 years giving a relative risk of PI-IBS of 0.36 (0.1–0.09) [37] though not all studies have shown this [30], [51].

Declining immune responsiveness may be important; older subjects have less immunocytes in their rectal mucosa [52] and lamina propria lymphocytes which are less responsive to antigen challenge [53], a feature which may reduce the chance of any ongoing unchecked low grade inflammation. Alternatively there may be a cohort effect and patients in their 60s and 70s who were exposed in their early years to many more infections in a less sanitized environment may continue to be protected by immunity they acquired as children. Recent European data showing a disappearance of the peak in campylobacter infection in the under 5s [14], [54] supports this by suggesting that the time of first infection may be becoming later and this acquisition of early immunity is no longer occurring.

10. Gender 

There is no evidence of any differences between the immune response of men and women to infection and rectal immunocyte numbers are not different [52], yet female gender is frequently reported risk factor for developing PI-IBS [36], [43], [55]. This may in part be due to confounding with anxiety and depression which is commoner in women, since when this was controlled for in multivariate analysis female gender was no longer a risk factor in two studies [36], [43]. However female gender did remain a factor in multivariate analysis in one recent study where there was relative risk of 2.36 (1.23–3.98) for gender versus 1.82 (1.05–1.22) for anxiety [51]. The gender effect is observed in other subtypes of IBS and recent evidence suggests that the difference may be in the brain response to pain since male and female patients with IBS show different activation patterns in response to rectal distension during PET scanning [56].

11. Smoking 

Patients who smoke are more likely to develop PI-IBS with an OR of 4.8 (1.5–15.2) [57]. Whether this is a direct effect of nicotine or more likely smoking is a marker for adverse psychological factors is unclear. However smokers report more anxiety [58] and panic attacks than non-smokers, features which are also commoner in IBS patients.

12. Psychological factors and coping styles 

Subjects with IBS have higher levels of anxiety, neuroticism and depression compared to controls without IBS. Furthermore in subjects without IBS high levels of illness behaviour, anxiety, sleep problems and somatic symptoms predict the subsequent development of IBS [59]. This vulnerability to develop IBS may depend on early learning since childhood exposure to parenting styles which reinforce illness behaviour in early life, is associated with the later development of functional bowel disorders [60]. The presence of hypochondriasis and neuroticism, which are thought to be enduring rather than acquired traits, increase the risk for PI-IBS, RR=2.0 (1.7–2.5) [55] while each standard deviation increase in depression also increases the risk 3.2-fold (1.8–8.2) [36] as do adverse life events in the 3 months preceding infection, RR=2.0 (1.7–2.4) [43]. Other psychological factors increasing this risk include perceived stress RR 1.10 (1.02–1.15) and negative illness beliefs RR 1.14 (1.03–1.27) [61].

The underlying mechanism whereby stress contributes to PI-IBS is unclear but ongoing stress can cause mast cell hyperplasia in rats [62] and may initiate a similar low grade inflammatory response in the human jejunum [63]. Stress induced by immersing the hand in ice-cold water causes jejunal biopsies to release more mast cell tryptase and prostaglandins [63]. Furthermore IBS patients have been shown to have an exaggerated response to infusion of cortisol releasing hormone (CRH) which correlated with elevated IL-6 [64]. However while anxiety and depression were the strongest predictors of developing chronic fatigue syndrome in a large group of patients who infected with either the Ebstein–Barr virus or C. jejuni, the nature of the precipitating infection (C. jejuni rather than EB virus) was the strongest predictor of subsequent IBS [51], reinforcing the important role that local gastrointestinal injury and inflammation has in the generation of gastrointestinal symptoms following infection.

13. Evidence for inflammation in PI-IBS 

Increased lymphocytes have been reported throughout the colon in unselected IBS with diarrhea [65], and in the rectum in PI-IBS [43], [36], [66]. This has been associated with increased mRNA for interleukin (IL)-1β [47], [67]. T lymphocyte numbers correlate with enterochromaffin cell numbers which are increased in man following C. jejuni infection and in mice following Trichinella spiralis infection. The mechanism underlying this correlation has been clarified by the recent report showing that EC hyperplasia is controlled by T lymphocytes, which activate IL-13 receptors found on EC cells [68]. T. spiralis infection leads to long lasting motor and sensory dysfunction [69] associated with increased EC numbers and reduced SERT expression[70]. The same model shows long term increases in mucosal 5HT content and spontaneous release of 5HT associated with increased afferent nerve response to distension which can be inhibited by the 5HT3 receptor antagonist, ondansetron [71]. Human studies show a similar increase in Enterochromaffin cell numbers (EC) [36], [66], [72] in PI-IBS compared to a control population. This has been associated with increased postprandial 5HT levels in platelet-poor plasma [73] whose secretory and prokinetic actions may contribute to the loose frequent stools seen in PI-IBS. EC numbers have also been linked to increased rectal sensitivity in D-IBS [74]. However EC cell changes have not always been reported and it may depend on the nature of the infection since in mice infection with T. spiralis increases EC cell numbers while C. rodentium infection reduces EC numbers [75]

Mast cells are also increased in animal models of infection and are increased in the human terminal ileum after Shigella infection [47] and in rectal biopsies of PI-IBS patients in some studies but not all [66], [36]. They have also been reported increased in the terminal ileum of unselected IBS patients with diarrhea [76]. PI-IBS may well involve inflammatory changes in the small intestine as well as the colon since small intestinal permeability is increased [77], [78], a change which could be mediated by increased ileal mast cells.

Other markers of inflammation in IBS patients include increased peripheral blood mononuclear cells cytokines including IL-6, IL-8 [64], tumor necrosis factor (TNF)-alpha, IL-1beta, IL-6, as well as increased lipopolysaccharide-stimulated IL-6 levels when compared to healthy controls [79].

14. Causes of persistent mucosal inflammation in PI-IBS? 

Why should this inflammation persist in some and not others. As we have already discussed adverse life events, anxiety and depression may play a part however less psychological morbidity is seen in PI-IBS than IBS [72] indicating the presence of other factors which predispose to an exaggerated or prolonged inflammatory response.

These factors might be genetic since a larger proportion of IBS patients have the high producing heterozygous TNF-alpha G/A polymorphism at position-308 than controls [80]. Some PI-IBS patients were contained in this study but too few to examine as a subgroup. This study did not confirm an earlier finding of a decrease in the presumed immunoregulatory high IL-10 producing phenotype in IBS [81].

Although it is likely from animal work that infection does alter the gut microbiota there is no data on this in PI-IBS. There is some indirect evidence that altered microbiota may be important in IBS since fecal serine protease activity, which may be of bacterial origin, is increased in D-IBS [82]. This is of great interest because these proteases can increase visceral sensitivity in rats, acting via the protease activated receptor-2 (PAR-2) group of receptors found in the mucosa and enteric nerves [83].

Other sources of proteases in man are mast cells which are thought to be important in IBS since the number of activated mast cells in close proximity to afferent nerves have been shown to be correlated with severity of abdominal pain in IBS patients [85]. Furthermore mast cell mediators including histamine and serotonin released into the supernatant of IBS patients’ biopsies excite nociceptive visceral sensory nerves in rats [84].

Alteration in neurochemical coding of enteric nerves may also be important in increasing visceral sensitivity. This has been detected after inflammatory colitis [86] and Substance P and 5HT staining nerves were also increased after Shigella spp. infection [47]. Increased expression of transient receptor potential vanilloid (TRPV)1 receptors can sensitize afferent nerves, so it is of great interest that a recent study found an increase in TRPV1-immuno-reactive fibres related significantly to abdominal pain score in unselected IBS patients [87]. Whether this is also seen in PI-IBS remains to be determined.

15. Treatment 

The lack of an effective treatment for IBS of all types represents a substantial unmet need. Recent evidence of low grade mucosal inflammation in IBS has encouraged trials of anti-inflammatory therapies. A small randomised double blind placebo controlled trial of 3 weeks of prednisolone in PI-IBS showed no effect on symptoms though it did reduce lymphocyte counts [88]. Mesalazine has a number of postulated anti-inflammatory mechanisms and data in abstract form suggests benefit when used in patients with IBS, decreasing pain, mucosal proteolytic activity and altering stool microbiota [89]. A recent small randomised placebo controlled trial of Mesalazine suggested this could reduce mast cell numbers and improve symptoms, a finding which needs repeating with larger numbers [90]. Given the increase in 5HT availability and the effectiveness of 5HT3 receptor antagonists in animal studies and in unselected IBS-D patients [91] a trial of a 5HT3 receptor antagonist would also be logical.

Conflict of interest statement 

The authors declare that they have no conflict of interest as regards the present manuscript.