Article Text
Statistics from Altmetric.com
Few would contest that advances in uncovering genetic risk factors for Crohn’s disease and ulcerative colitis over the past decade have changed the way we think about inflammatory bowel diseases (IBDs). Perhaps the most important message has been that much of the genetically determined risk lies in how the host interprets its microbial environment.1–3 However, the primacy of environmental factors was already evident from several sources; notably, studies of genetically identical twins showing a relatively low concordance rate for both Crohn’s disease (<50%) and ulcerative colitis (<10%), and the increased frequency of both disorders in many countries during a period too short to involve significant changes in the population gene pool.4 What are the environmental or lifestyle risk factors for IBD? How do they collude with genetic susceptibility?
The lesson of Helicobacter pylori and peptic ulcer disease was that the solution to some chronic disorders cannot be found by studying the human host alone. Rather, the answer may lie at the interface with the microbial environment. A more sobering lesson was the failure of conventional epidemiological studies to recognise that peptic ulcer disease is caused by a transmissible agent. How did disparate epidemiological observations miss this association and fail to guide medical scientists toward this conclusion? Could IBDs (or a subset thereof) be due to an infectious agent, waiting to be identified? Or is the relationship between host susceptibility and the microbial environment a more subtle one? While the complexity and heterogeneity of these diseases surely accounts for some of this dilemma, it seems reasonable to ask whether several decades of epidemiological studies have directed or distracted researchers seeking clues to the cause of IBD.
Some of the false leads and false promises of epidemiology have been highlighted elsewhere, with the most vigorous attacks in the area of “risk factor epidemiology”.5 The epidemiology of risk factors does not actually explain risk, seldom reveals anything about causal mechanisms, tends to generate contradictory findings and is particularly prone to publication bias, with positive associations in excess of null results.5 There have, of course, been successes; epidemiology can alert us to causal mechanisms as with smoking and lung cancer. Therefore, it would be foolhardy to abandon classical epidemiological approaches to IBD, but reasonable to contend that they be concentrated where the disease is emerging or changing, and to be designed with biological mechanisms in perspective. The delayed identification of H pylori is not only an epidemiological lesson but also a reminder that biologists can look but not see, even when the answer is before them in the tissues. Perhaps disconnected epidemiological and biological thinking ensured that neither knew what to look for. This should be avoided in IBD.
Our purpose here is to offer a perspective on some of the more consistent epidemiological observations in IBD and to attempt to reconcile them with current concepts of disease mechanisms. While much attention has been paid to comparative studies of different ethnogeographic regions, we will also focus on changes over time, because these are more likely to reveal clues to underlying disease mechanisms. We will address the evidence for the influence of dietary composition as one possible contribution to the changing epidemiology of IBD—in part, because this is the most frequent question patients pose to clinicians when seeking an explanation for the cause of their disease. Finally, we offer some general recommendations on the implications for future research directions.
POLARISING INFLUENCES
Many of the environmental or lifestyle risk factors associated with IBD are common to Crohn’s disease and ulcerative colitis, but smoking is a striking exception of a polarising influence. While the cessation of smoking is often associated with the onset of ulcerative colitis or the exacerbation of established disease, smoking represents one of the most consistently observed environmental influences on Crohn’s disease.6 7 However, the highest rates of smoking are in Asia and Africa (up to 65% of adult males), yet these regions have very low incidence rates of Crohn’s disease. Conversely, in Canada (27%) and Sweden (19%), which have some of lowest prevalence rates for smoking amongst adult males, the incidence rates for Crohn’s disease are amongst the highest in the world.8 Thus, although smoking has consistently been a reported risk factor associated with Crohn’s disease and a predictor of an aggressive disease course, it is unlikely to account for worldwide trends in disease incidence. Smoking may modulate outcomes in Crohn’s disease through effects either on vascular endothelium, on mucin production, or other toxic effects of smoking byproducts on gut mucosal homeostasis that are beyond the scope of this review. It seems apparent that the impact of smoking on disease phenotype expression is independent of the presence of NOD-2 mutations,9–11 and so countries with high smoking rates with potentially low rates of NOD-2 mutations should still be at risk for high rates of Crohn’s disease if smoking was causative. Hence, smoking represents an important contributory influence on established disease but, for environmental factors linked to causal mechanisms, one has to look elsewhere. Appendectomy is another factor with potentially opposing effects on Crohn’s disease versus ulcerative colitis. It seems to protect against future presentation with ulcerative colitis while either having no effect or predisposing to presentation with Crohn’s disease.12 Until the biological role of the appendix is unravelled, the impact of its presence or absence on development of either form of IBD will remain elusive.
THE EPIDEMIOLOGY OF CHANGE—TRADING RISK FOR LIFESTYLE
The abrupt increase in frequency of IBD when there is a transition from “developing” to “developed” nation status is one of the most compelling lines of evidence linking changing epidemiology with changing lifestyles or environmental influences in modern society. The consistency of the evidence linking these changing events deserves scrutiny before attempting a mechanistic explanation.
As IBD emerges in developing countries, ulcerative colitis appears first, followed at a variable interval by a rising trend in Crohn’s disease. This is true not only for entire countries or global regions, but also for societal sectors living in underdeveloped conditions within highly developed countries; examples of the latter being the First Nations of Canada13 and the Maoris of New Zealand.14
Data from developing nations such as South Korea,15 China,16 17 India,18 19 Iran,20 Lebanon,21 Thailand22 and the French West Indies23 reveal a clear pattern of greater incidence rates for ulcerative colitis over Crohn’s disease. Despite reservations regarding the comprehensiveness of data collection and variability in access to healthcare in many of these countries, the incidence rates for IBD are lower than in developed countries but appear to have increased over the past two decades. It is noteworthy that amidst Manitoba’s high rates of IBD are the First Nations Aboriginal community (comprising 10% of the entire population). This community is largely located in the relatively sparsely populated north of Manitoba, often with living conditions akin to those of the developing world, without flush toileting and with poor housing. Another substantial First Nations community exists within the city of Winnipeg. All of these communities, rural and urban, have been shown to have significantly lower rates of Crohn’s disease and ulcerative colitis than the non-First Nations Manitoban population.24 However, the First Nations population has approximately fourfold the rate of ulcerative colitis compared with Crohn’s disease. This mirrors the relative incidence rates of IBD from the mid-20th century in developed countries and from the developing world at present. See figs 1 and 2 for recent incidence rate studies worldwide showing the emergence of IBD in Spain to mirror rates in northern Europe as well as the emergence of IBD in eastern European countries with rates that somewhat lag behind western Europe. The highest rates of Crohn’s disease are evident in Manitoba and New Zealand, a finding which may dispel somewhat the north–south gradient of disease incidence rate.25
What happens after migration from a low-incidence country to one of high incidence? In general, the incidence of immune-mediated diseases such as IBD does not change amongst adult immigrants, but their children when raised in developed countries seem to acquire the same susceptibility as that of the native population. Thus, immigrants from India to Leicestershire and Vancouver were reported to have incidence rates of ulcerative colitis comparable with or even higher than those of native Britons and Canadians, respectively,26 27 but age at time of migration appears to be critical. Migrant offspring reared in the adoptive country or those who migrate during childhood have the greatest risk of developing IBD. This is testimony to the relative importance of environmental factors over genetic susceptibility. It is also consistent with a pivotal role for early life events in the aetiology of IBD; hence, the relevance of studies in childhood.
Childhood studies confirm that Crohn’s disease is becoming the predominant form of IBD in developed countries (fig 3). Reports have consistently shown increasing rates for Crohn’s disease; whereas, in contrast to adults, the incidence rates for ulcerative colitis in children are static or in decline.28–36 This might suggest that the environmental factors contributing to Crohn’s disease persist and might be more identifiable in children. Additional advantages to studies in children include the shorter previous lives, less varied eating and lifestyles, and limited geographic mobility.37 38 Their habits are also often carefully tracked by parents and caregivers, making survey data more reliable than in adults relying on distant recall. Another unexplained trend has been the evolution of a male predominance of paediatric Crohn’s disease.29 31 33 36 39
LINKING LIFESTYLE WITH THE MICROBIOTA AND THENCE TO IMMUNITY
The molecular pathogenesis of IBD (reviewed in Strobert et al1, Xavier and Podolsky2 and Baumgart and Carding3) is beyond our scope here, but there is a consensus for three interacting elements: genetic susceptibility factors, particularly at the level of epithelial barrier function and at the level of innate and adaptive immunity to the microbiota; environmental drivers or modifiers, particularly the commensal bacteria; and dysregulated immune-mediated tissue damage. Disease heterogeneity arises, in part, because multiple susceptibility genes and modifying genes are involved and because of the variable composition of the microbiota.
Since genetic susceptibility factors within the population are relatively stable, the question for epidemiologists becomes one of reconciling the changing epidemiology of IBD during periods of socioeconomic transition with microbial-driven immune-mediated pathology. How does the microbiota influence the behaviour of the immune response and can the modern lifestyle of developed societies be linked with the microbiota? A biologically plausible explanation can be provided for the first question, and an affirmative answer for the second is likely and suited to future epidemiological testing.
Microbial education of the sense of danger
The increased incidence and prevalence rates for IBD in developing societies have been paralleled by similar trends for several other chronic disorders.40 Given that immune-mediated pathology is common to each of these conditions, the impact of the environment and changing lifestyles probably operates in some way at the level of the immune response.
The immune response is a sensory system—the sense of danger from microbes and other foreign antigens. In every respect it is the sixth sense, and displays the attributes of the other senses; it has an afferent or uptake limb, central processing of sensory information, an efferent or effector limb, and it displays learning and memory. As with other sensory systems, the immunosensory system is present at birth but requires developmental education in early life. This is achieved by interaction with the environment. Episodic exposure to infection during early life offers some educational benefit, but colonisation with commensal organisms offers a far greater educational experience for the developing immune system. This is so because of the enormous antigenic diversity within the colonised gut, which represents the most densely populated ecosystem on the planet.41 Comparative studies of germ-free and colonised animals have shown the importance of the microbiota not only as a regulatory influence on the developing immune response42 but also for the maintenance of immunological homeostasis.43
In common with the other senses, any disruption or deficit in the quality of educational development is likely to leave the host vulnerable in later life to risk of misperception or inappropriate responses to harmless environmental stimuli. Defective education of the developing immune system can be anticipated to create a risk for inappropriate immunological interpretation of antigenic challenge in later life. Has the modern lifestyle of developed, industrialised countries changed our microbial environment and its role in immune education? Is this where epidemiological studies need to be focused?
Can the modern lifestyle of developed societies be linked with the microbiota?
Some of the elements of a modern lifestyle in developed societies are shown in box 1. The relative contribution of any of these to the increased frequency of immune-mediated chronic diseases is uncertain, but in many cases an impact on the composition of the human microbiota is likely or already evident. In others, such as the role of obesity, an unexpected connection with the gut microbiota has recently emerged.41 44 In the case of antibiotic exposure, it has been recently reported that the consequences for the commensal microbiota may be more lasting than previously thought,45 and there is experimental evidence of antibiotic-induced dysbiosis in the gut with disturbed systemic immune responsiveness.46
Features of a modern lifestyle that may be linked with alterations in enteric microbiota
Improved sanitation
Decline in endemic parasitism
Life on concrete (less exposure to soil microbes)
Decline in H pylori
Increased antibiotic usage
Vaccination
Smaller family size
Less crowded living conditions
Refrigeration
Delayed exposure to mucosal infections
Sedentary lifestyle and obesity
Reduced consumption of fermented food products
Increased consumption of refined sugars and saturated fats
Perhaps the greatest influence on the microbiota is diet and, since colonisation commences at birth during delivery, the impact of maternal diet on the microbiota colonising the neonate may be particularly important. In this context, one commentator has described the changes in adult diet over the past century and its subsequent impact on neonatal microbiota as being the greatest dietary change since Neolithic times.47 New molecular approaches that are not dependent on the limitations of bacterial culture media have demonstrated a greater diversity to the gut ecosystem and suggest that it may be less stable than previously thought. Application of these approaches to human IBD has generated evidence consistent with the heterogeneity of Crohn’s disease and suggests disturbances of the gut microbiota in at least a subset of patients.48 In fact, several groups in both North America and Europe have independently reported an association of a novel adherent-invasive Escherichia coli species.49–51
To account for the increasing frequency of immune-mediated disorders, such as Crohn’s disease and ulcerative colitis, the hygiene hypothesis is often invoked.40 At its first inception, the hypothesis attributed the rising frequency of immunoallergic disorders to a sanitised environment with reduced exposure to childhood infectious agents, thereby creating a permissive environment for immunological mischief. The concept has undergone various refinements. It is now held by some that it is not the absence of pathogenic infections that needs consideration; rather, it is the reduced colonisation with specific commensals such as lactobacilli and bifidobacteria.52 In addition, to account for rising frequencies of immunological disorders associated with both ends of the spectrum of cytokine production (ie, disorders characterised by either the so-called TH1 or TH2 cytokine profiles), the concept of microbial education of regulatory T cells (Treg) has been invoked.53 54
ONCE BITTEN….PURSUING TRANSMISSIBLE AGENTS IN IBD
Throughout the millennia, most chronic disorders in mankind have turned out to have an infectious basis. Hopefully epidemiologists can help resolve persisting concerns regarding the possibility of a transmissible agent in IBD. Of the different infectious agents considered as possible causes of Crohn’s disease, Mycobacterium avium subspecies paratuberculosis is the most durable.55 To date, the circumstantial epidemiological evidence is inconsistent or conflicting, and the microbiological evidence that it is a definite zoonosis is lacking.
M avium subsp. paratuberculosis causes Johne disease in cattle, a disease that is similar to Crohn’s disease in its distribution, morphology and phenotype. M avium subsp. paratuberculosis may enter the food chain from a variety of sources including faecal shedding from infected animals, and contaminated raw meat and raw milk. Humans eat beef from cattle and drink cow’s milk some of which may come from cows or at least herds infected with M avium subsp. paratuberculosis. The rise in infection rate among herds of cattle in the USA has been alarming. In one study from Texas, up to 43% of beef cattle herds had at least one animal seropositive for the infection.56 57 However, the rise in incidence of Johne disease among beef cattle raises the concern that cattle with subclinical disease may enter the human food chain. Johne disease also occurs in other animals, including sheep, goats, deer, bison, alpacas, rabbits, pigs, and in several wild species.58–61 However, the relative paucity of M avium subsp. paratuberculosis in Scandinavian cattle where the incidence rates of IBD are comparable with North American incident rates is at variance with a cause and effect relationship between the organism and human Crohn’s disease.62 63 The seemingly moderate rates of M avium subsp. paratuberculosis in Chilean cattle, comparable with rates evident in North American cattle, are at variance with the low rates of IBD in Chile.64 65
Epidemiologists are limited in the extent that they can further develop the M avium subsp. paratuberculosis hypothesis in IBD because the biologists need to catch up with reliable and reproducible culture or detection assays. In a case–control study, sources of mycobacterial exposure were assessed and ingestion of ground water, or water sources without a filtration system, was not significantly different in Crohn’s disease cases and controls.66 In the same study, the consumption of pasteurised milk was associated with a significantly reduced risk for Crohn’s disease. However, consumption of meat (kg/month) was associated with a significantly increased risk of Crohn’s disease. Hence, these authors concluded that there was no proof that M avium subsp. paratuberculosis causes Crohn’s disease because there were no associations with two key primary sources for the organism (water and pasteurised milk). The flaw with this type of study design is that it inquired into dietary intake of persons already diagnosed with Crohn’s disease, which raises questions of cause or effect.
In some but not all laboratories M avium subsp. paratuberculosis is found by molecular diagnostics with much greater frequency in Crohn’s disease than in controls.67 68 Circulating DNA of M avium sub paratuberculosis has been suggested to be increased in Crohn’s disease69 whereas a serological response to M avium subsp. paratuberculosis was no more common among Crohn’s disease than in ulcerative colitis or in healthy unaffected controls.70 Proponents of a mycobacterial cause of Crohn’s disease signal alarm over detection of DNA from M avium subsp. paratuberculosis in up to 13% of pasteurised milk samples tested, but it has seldom been cultured from appropriately processed milk.71–76
Molecular microbiological techniques that consistently and reproducibly identify human M avium subsp. paratuberculosis infections are urgently needed to define whether this organism has human relevance. Can M avium subsp. paratuberculosis be isolated in humans in a form that either triggers a direct immune response or that facilitates an aberrant response by macrophages and denditic cells to other commensal bacteria? Much like biologists could have identified the important association between H pylori and peptic ulcers earlier in the 20th century, it is time for biologists to determine whether M avium subsp. paratuberculosis could in fact cause or facilitate a form of IBD. Until biologists develop reliable detection systems of M avium subsp. paratuberculosis such that the infection can be proven in persons with IBD, further epidemiological studies are not likely to advance the hypothesis.
COULD DIET HAVE A ROLE AFTER ALL?
Diet is probably the single most important factor influencing the composition and metabolic behaviour of the microbiota. Despite this, epidemiological investigations of dietary factors have generally been disappointing, mostly because they have been retrospective. Two recent Canadian studies have pursued dietary surveys in patients with IBD and controls.77 78 In one study of newly diagnosed children where recall bias may be less of an issue than in adults with the help of interviewed parents and with a shorter lifespan to review, it was reported that higher amounts of vegetables, fruits, fish, dietary fibre and long-chain ω-3 fatty acids significantly protected from Crohn’s disease.77 A higher ω-6/ω-3 fatty acids ratio was significantly associated with higher risks for Crohn’s disease. A retrospective population-based case–control survey of childhood lifestyle in adults found protection against Crohn’s disease in those who lived on a farm and who consumed unpasteurised milk, somewhat supporting the hygiene hypothesis.78 However, the factors remaining on multivariate analysis that protected against development of Crohn’s disease were exposure to pet cats prior to age 5 and larger families, both supporting the hygiene hypothesis.
Nonetheless, some of the most compelling data calling for a reconsideration of the role of diet in relation to the changing epidemiology of IBD have been generated in Japan. This is particularly noteworthy because the genetic epidemiology of IBD in Japan appears to differ from that in the West (eg, NOD-2 is not a risk factor for Crohn’s disease in the Japanese).79 It is also noteworthy because Japan may represent an exception to the pattern of epidemiological trends for IBD in developing versus developed societies. Japan is a well-developed country but has seen a notable increase in frequency of IBD within the past decade.80 The Japanese experience may relate more to dietary changes than to some of the other lifestyle trends seen in other countries (box 1).
Studies from Japan have previously shown an association between Crohn’s disease and consumption of total animal protein81 and intake of total animal fat—particularly ω-polyunsaturated fatty acids.82 In the first Japanese study, an assessment was made of national trends in consumption. Assessing national trends avoids the pitfalls of direct subject inquiry and the dilemma of choosing the most appropriate age of consumption about which to inquire. The univariate analysis showed that the increased incidence of Crohn’s disease was strongly correlated (p<0.001) with increased dietary intake of total fat (r = 0.919), animal fat (r = 0.880), ω-6 polyunsaturated fatty acids (r = 0.883), animal protein (r = 0.908), milk protein (r = 0.924) and the ratio of ω-6 to ω-3 fatty acid intake (r = 0.792). Crohn’s disease incidence was not correlated with intake of fish protein and was inversely correlated with intake of vegetable protein (r = −0.941, p<0.001). The multivariate analysis showed that increased intake of animal protein was the strongest independent factor, with a weaker second factor, an increased ratio of ω-6 to ω-3 polyunsaturated fatty acids.
In the later Japanese study, a semi-quantitative food frequency questionnaire was used to estimate pre-illness intakes of food groups and nutrients in a case–control study. Recall bias is obviously a potential problem in such a study. Ulcerative colitis was strongly associated (by a factor of two- to threefold) of a higher consumption of sweets and fats and oils. Crohn’s disease was associated (by a factor of two- to threefold) with consumption of fish and shellfish, the intake of total fat, monounsaturated fatty acids, polyunsaturated fatty acids, ω-3 and ω-6 fatty acids, and vitamin E. The intake of vitamin C significantly protected against ulcerative colitis. Elsewhere, citrus fruit consumption protected against development of both Crohn’s disease and ulcerative colitis.83 The study of dietary intake and the emergence of IBD in Japan may be very instructive. Japan, as opposed to other nations newly reporting rising incidence and/or prevalence rates of IBD, is a very developed nation. Hence hygiene and its accompanying alteration in responses to microbial flora may be less important than dietary changes that mirror western societies.
Why might fats be important in the genesis of Crohn’s disease? Saturated fats increase the activation of nuclear factor-κB through Toll-like receptors, and this activation is suppressed by derivatives of ω-3 fatty acid.84 In addition, polyunsaturated fatty acids of the ω-6 series produce proinflammatory leucotriene B-4 (LTB-4) which enhances the inflammatory reaction. In contrast, polyunsaturated fatty acids of the ω-3 series produce anti-inflammatory mediators LTB-5 and prostaglandin E3 which suppress inflammation.85 It is difficult to reconcile the association of ω-3 fatty acids with the disease in Japan unless the ratio of ω-6 to ω-3 fatty acids is of paramount importance rather than simply the presence of ω-3 fatty acids.
Finally, vitamin D has increasingly been shown to have an important impact on health, and this may include the predisposition to IBD.86–88 1,25-Dihydroxy vitamin D3 regulates TH1 and dendritic cell function while inducing regulatory T cell function. This can result in a decrease in the TH1-driven autoimmune response.89 It has been shown that vitamin D supplementation is associated with increased circulating transforming growth factor β,90 but it is not known whether this translates to enhanced clinical outcomes.91 Vitamin D deficiency in IBD is well recognised,92 93 but is generally assumed to be consequential rather than causal to the disease. A paucity of vitamin D because of less access to sunlight year round, with reduced milk consumption in childhood, could be a contributory factor to the higher incidence rates evident in northern countries such as Canada, the UK and Scandinavia.25 94
CONCLUSIONS: IMPLICATIONS FOR FUTURE RESEARCH
Epidemiological surveys of westernised societies, while informative on some aspects of IBD, particularly the transition from ulcerative colitis to Crohn’s disease, have limited value without some rapprochement with disease mechanisms. Risk factor epidemiology in the absence of concurrent biological plausible support is more likely to yield conflicting information. Epidemiologists can, however, enhance our understanding of the aetiology and pathogenesis of IBD by addressing the clues offered by the changing epidemiology of these disorders in developing societies. To do otherwise may be a wasted opportunity. The populations ripe for study exist in developing nations, migrant groups and paediatric populations of the developed world. Further, populations who do not develop IBD or have very low incidence rates can be as instructive to study as those with high incidence rates. These epidemiological studies should be supported by longitudinal studies linked with some assessment of immunological, microbiological or other biological variables. For example, the new science of metagenomics is a molecular strategy for studying complex microbial ecosystems and promises to overcome the limitations of culture-dependent methods for studying the human microbiota.87 Several consortia around the world are underway to apply metagenomic methods to understanding the normal microbiota in health and disease in different populations.
Since circumstantial and direct evidence point to the importance of early life events in relation to the developing immune system and to the composition of the colonising microbiota, the impact of epidemiologists is likely to be most revealing if focused at early phases of the life spectrum. In addition to assessing the early phase of life, epidemiologists should assess the evolution of IBD in the developing world.
Pursuing epidemiological studies in inflammatory bowel disease (IBD)
The solution to some human disorders cannot be found by research focused exclusively on the human host, without attention to the interface with the microenvironment. This was the lesson of Helicobacter pylori and peptic ulceration.
Epidemiological studies in IBD should be linked to concepts of disease mechanisms
Considering peak incidence rates in the third decade of life and more aggressive disease being apparent at younger ages of presentation, early life events may hold the key to unlocking important aetiological clues
Changes in the risk of developing IBD with migration between developing and developed regions is a useful model for reconciling disparate lifestyle and environmental factors with causal mechanisms
Mechanisms underlying the changing epidemiology of IBD as societies undergo transition from developing to developed status also represent the most promising model for future study.
Acknowledgments
CB is supported in part by a Research Scientist Award from the Crohn’s and Colitis Foundation of Canada. FS has been supported in part by Science Foundation Ireland, the Health Research Board of Ireland, the Higher Education Authority and the European Union.
REFERENCES
Footnotes
Competing interests: None.