Hypothesis
Crohn’s disease: the cold chain hypothesis
Jean-Pierre Hugot, Corinne Alberti, Dominique Berrebi, Edouard Bingen, Jean-Pierre Cézard
Crohn’s disease is the result of an abnormal immune response of the gut mucosa triggered by one or more environmental risk factors in people with predisposing gene variations, including CARD15 mutations. Epidemiological data allow assessment of familial environmental risk factors related to western lifestyle, diet, bacteria, and domestic hygiene. All findings point to refrigeration as a potential risk factor for Crohn’s disease. Furthermore, cold-chain development paralleled the outbreak of Crohn’s disease during the 20th century. The cold chain hypothesis suggests that psychrotrophic bacteria such as Yersinia spp and Listeria spp contribute to the disease. These bacteria have been identified in Crohn’s disease lesions and we discuss their pathogenic properties with respect to our knowledge of the disease. From a molecular perspective, we postulate that the disease is a result of a defect in host recognition by pathogenic bacterial components that usually escape the immune response (eg, Yop molecules), which results in an excessive host response to these bacteria.
Crohn’s disease is characterised by a relapsing
disease has been positively associated with good standards
inflammatory process in the digestive tract.1 Since its first
of domestic hygiene (such as hot running water),9–11 and
description in 1913, the cause of Crohn’s disease is still
that environmental risk factors contribute to familial
largely unknown, although it is recognised to be a
aggregations of the disease.12 Because inflammation
complex trait that is the result of an interaction between
occurs in the digestive tract, antigens present in the gut
environmental and genetic risk factors. Although some
lumen, including components of food and intestinal
genetic factors are known,2,3 uncertainty still exists around
bacteria, have been implicated in the development of
environmental factors, except for cigarette smoking.4
Crohn’s disease. However, no specific dietary component
Knowledge of causative environmental factors is crucial
has been identified, suggesting that many foods contain
if disease mechanisms are to be understood and a specific
putative contaminants. Several infectious agents have
treatment developed. We postulate that Crohn’s disease is
been proposed, including species of Mycobacterium,
the result of an excessive response to pathogenic
Listeria, and Yersinia, and Escherichia coli, but none has
psychrotrophic bacteria in some genetically predisposed
been proven as a causative factor. However, the link
people. Such bacteria can exist and grow at temperatures
between bacterial components and Crohn’s disease was
between –1ºC and 10ºC, and we believe that the advent of
lent strong support by the discovery that mutations in
domestic refrigeration contributed to the outbreak of
CARD15, a gene involved in innate immunity,
Crohn’s disease in the 20th century.
In view of the observation that Crohn’s disease is linked
to a familial environmental risk factor related to modern
Crohn’s disease is common in Europe and North
western lifestyles, domestic hygiene, diet, and infectious
America, where incidence has risen in the second half of
agents, we propose a specific candidate for the
the 20th century.5 Thus, environmental factors, probably
development of Crohn’s disease: the refrigerator.
related to the modern occidental way of life, might have a
Domestic refrigeration began with the advent of ice
role in Crohn’s disease.6–8 From the first to the second half
containers in the 19th century. The first refrigerating
of the 20th century, many crucial changes occurred with
machines were built around 1875, and during the first
respect to food, housing, transport, leisure, and clothing.
part of the 20th century, refrigeration methods were more
Thus, to tease out the relevant causative risk factors for
widespread in the USA than in other countries.13 The first
Crohn’s disease from the mesh of inter-related variables is
domestic refrigerator was developed by Kelvinator in
1918 in the USA. It became more and more popular and
It is beyond the scope of this paper to discuss
in 1921, 5000 refrigerators were produced, with the yearly
epidemiological data in detail, but of note is that Crohn’s
production rate rising to 75 000 in 1925, 850 000 in1930, and 1 700 000 in 1935. By 1937, 49% of
In 1930s’ Europe, only wealthy families had
Laboratoire de Génétique des Maladies Inflammatoires de
refrigerators, and ownership became more common only
l’Intestin, Projet Avenir INSERM, Fondation Jean Dausset CEPH,
after World War II with improvements in living standards,
Paris, France (J P Hugot MD); Department of Paediatric
and with the supply of electricity to homes. Even in 1958,
Gastroenterology (J-P Hugot, Prof J-P Cézard), Unité d’Epidémiologie
only 10% of French and 12% of British families had a
Clinique, Department of Public Health (C Alberti MD), EA3102,
refrigerator—ownership was as low as 2% in Spain, the
Department of Pathology (D Berrebi MD), and EA3105, Department
USSR, and Japan. The European exception was Sweden,
of Microbiology (Prof E Bingen), Hôpital Robert Debré, Assistance
where at the same time just over half of families had a
refrigerator developed by Electrolux.
Correspondence to: Dr Jean-Pierre Hugot, Service de
Although difficult to define precisely, population-based
Gastroentérologie, Hopital Robert Debré, 48 Boulevard Sérurier,
data suggest that the increase in prevalence of Crohn’s
disease was in the 1940s or before in the USA,8 in the 1950s
(e-mail: [email protected])
or before in Sweden,14,15 in the 1960s in the UK,6,7 and later
THE LANCET • Vol 362 • December 13, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
in southern Europe. These examples show temporal and
bacterial detection. In accordance with this point of view,
geographical coincidences between the development of the
yersinia was not a predominant microorganism in biopsy
refrigerator and the outbreak of Crohn’s disease. Of course,
samples from ten patients with the disease,21 lending
the cold theory for the development of Crohn’s disease
support to the theory of a chronic bacterial infestation that is
cannot be limited to the domestic refrigerator, which is the
last part of the cold chain. The cold chain is much more
Because data from microbiological studies are compatible
complex: of the 520 kg of food eaten every year per person
with the cold chain hypothesis, we have further developed
in France, 320 kg are, at some time, conserved at a low
the theory of chronic infestation by psychrotrophic bacteria.
temperature. Furthermore, external and complex factorscould affect the quality of the cold chain including machine
maintenance procedures, food conservation habits, and the
Crohn’s disease lesions occur predominantly where there is
a high density of lymphoid follicles, in the small bowel
We are aware that the suggested association between the
where the follicles are grouped to form Peyer patches, and
cold chain and Crohn’s disease needs confirmation by more
in the colon where they are isolated. In the small intestine,
detailed analyses before being retained. Additionally, even
lesions are more frequent in the distal ileum where Peyer
if firmly established, temporal and geographical
patches are most common. In addition to this spatial
coincidences cannot prove a causal relation by themselves,
relationship, a temporal relationship has also been
and we cannot rule out the possibility that the cold chain is
suggested by Van Kruiningen and colleagues.26 Peyer
a confounding risk factor that occurred in parrallel with the
patches develop from birth to age 10–15 years, and then
undergo involution. Of note is that the age-dependentincidence curve for Crohn’s disease is roughly parallel to
the number of Peyer patches with a peak in the third decade
If our hypothesis is valid, it points to the existence of
of life. A delay of 5–10 years is then seen between the
bacteria capable of surviving or developing at low
development of Peyer patches and the occurrence of
symptoms. Such a delay is compatible with chronic
psychrotrophic bacteria, have optimum growth at
infection. Conversely, the ileal location of Crohn’s disease
temperatures higher than 30ºC but are able to grow, at a
is less frequent in elderly people after the involution curve
slower rate, at temperatures between –1ºC and 10ºC—ie,
of Peyer patches.27 In the colon, by contrast with the small
the temperature inside refrigerators.
intestine, the number of lymphoid follicles is less subject to
The most frequently encountered psychrotrophic
variation. This observation can account for the fact that
colonic Crohn’s disease can occur at any age. Furthermore,
monocytogenes, Yersinia enterocolitica, Clostridium botulinum
aphthoid ulcerations (thought to be the first lesions in
and Bacillus cereus.16 The pathogenicity of B cereus and
development of Crohn’s disease) are centred on lymphoid
C botulinum is mainly related to toxin production. To our
knowledge, these bacteria have not been studied in Crohn’s
Together, these data suggest that lesions initially develop
disease. L monocytogenes is a gram-positive bacillus that
on lymphoid follicles and accord with a theory of specific
induces acute diarrhoea. It has been identified in Crohn’s
infection. Many infectious agents exhibit a specific tropism
disease lesions by immunohistochemical analysis, but not by
for M cells including Salmonella spp, Shigella spp,
PCR.17–21 Y enterocolitica is an ubiquitous gram-negative
M paratuberculosis, E coli, Vibrio cholerae, Campylobacter
bacillus that has been detected in the digestive tract of many
jejuni, reovirus, poliovirus, HIV, and the psychrotrophic
wild and domestic animals. In addition to Y enterocolitica,
bacteria Y enterocolitica and L monocytogenes.
less virulent species of the Yersinia genus are also
Yersinia spp are able to produce an invasin, which allows
encountered in food, but their pathogenic role is less clear.
M cell penetration. This specific tropism for M cells makes
Yersinia species are commonly found in humans and in
yersinia pathogenic for the lymphoid follicles but not for
foods such as milk, beef, pork, chicken, sausages,
other parts of the mucosa. In mice infected with yersinia,
vesicles appear on the lymphoid follicles, which are
People with Crohn’s disease have been tested for the
progressively destroyed with micro-abscess formations.29
presence of Yersinia with conventional methods. Sera from
Inflammation of the gut is discontinuous and only 30–50%
patients were reactive to Y pseudotuberculosis more often
than were control sera in one study,23 but most studies
In human beings, Yersinia spp are able to induce a clinical
report negative results. More interestingly, with PCR-
ileitis or ileocolitis, and a mesenteric adenolymphitis, most
specific primers, Yersinia spp were detected by two
often in children and young adults. They might also induce
independent groups in 63% (13/19) and 31% (17/54) of
reactive arthritis and erythema nodosa. Histological analysis
people with Crohn’s disease, respectively.24,25 Intestinal
shows that intestine and lymph-node lesions can exhibit
samples were positive for Y enterocolitica, Y pseudo-
granuloma formations. All these findings are common to
tuberculosis, or both. The clinical and histological features of
Crohn’s disease and yersiniosis, explaining that these two
the yersinia-positive cases were indistinguishable from
disorders have to be classically differentiated during the
unlikely, since median follow-up in the two study groups
Additional features are relevant for Crohn’s disease. Like
was 9 and 10 years, respectively. Thus, the data clearly
Helicobacter pylori, yersinia produces a urease that is known
indicate that Yersinia spp can exist in lesions associated with
to be essential in gastritis formation, a finding which could
Crohn’s disease. Moreover, the large number of yersinia-
explain the high frequency of focally enhanced gastritis in
positive cases and the absence of specific clinical or
Crohn’s disease. Y enterolitica produces a heat-shock protein
histological features suggest that a large proportion of
that might cause diarrhoea and autoimmune colitis in
people with the disease have chronic contact with the
animal models.30 The bacteria has been implicated in
spondylarthropathies, which are known to be associated
However, Yersinia spp might be present at only very low
with Crohn’s disease. Finally, antibiotics active against
rates in Crohn’s disease lesions, in view of the negative
Yersinia spp such as ciprofloxacin have a limited effect in
results from studies that use conventional methods for
THE LANCET • Vol 362 • December 13, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
inhibition of the NF-B pathway in wild
CARD15 mutated patients, inducing a
variations have been reported for bothyersinia infections and relapses of
to predispose people to Crohn’sdisease.2,3
CARD15 encodes for a protein that isable to induce NF-B activation (andpotentially apoptosis), after recog-
inheritance at the CARD15 locus,which is characterised by a dosage
Links between Crohn’s disease, the cold chain, and psychrotrophic bacteria
effect of the Crohn’s disease muta-tions.2 However, the model does not
accord with the increased NF-B activation noted in
The three main mutations of the CARD15 gene associated
with Crohn’s disease in white people (G702R, R908W,
Of note is that CARD15 is also mutated in another
and 1007FS) have probably occurred only recently in
granulomatous condition characterised by an inflammation
human history. There were large outbreaks of plague in
of the eyes, joints, and skin but not the digestive tract—
Europe that occurred periodically between the 6th and the
Blau syndrome.38 In this mendelian dominant trait, a gain-
14th centuries.32 Thus, in the Middle Ages, CARD15
of-function model has been established with an excess of
mutations might have provided carriers with a selective
advantage during plague outbreaks and it could be
To relate the two disorders because of a common
postulated that mutation carriers have a more intense
activation of the CARD15 pathway that induces an
reaction against Yersinia spp.
inflammation with granulomas is tempting. Several
With time, Y pestis disappeared—probably because it
bacteria are usually able to inhibit NF-B activation
was replaced in Europe by less pathogenic strains such as
through virulence factors. We can, thus, postulate that
Y pseudotuberculosis (and later Y enterocolitica), which
Crohn’s disease mutations might induce not only a loss of
probably induced a cross-immunisation to Y pestis in
NF-B activation by the peptidoglycan, but also a loss of
rodents carrying the plague agent.32 However, the intense
NF-B inhibition in the presence of bacterial agents of
reaction against Yersinia spp developed by mutation
virulence. As a result, hosts carrying CARD15 mutations
carriers might have become inadequate for these less
should have a more intense reaction toward virulent
virulent strains and the intestinal site of the infection,
which is usually characterised by an immune tolerance.
The virulence factors of the Yersinia species are mainly
Y enterocolitica was isolated for the first time in 1934 in
carried by a plasmid (pYV) encoding for 12 secreted Yop
the USA, 2 years after the complete description of the
and Ysc proteins forming the secretory apparatus type III.
regional enteritis by Burrill B Crohn, Leon Ginzburg, and
Some of the Yop proteins are involved in the inhibition of
Gordon D Oppenheimer. In Europe, Y enterocolitica
phagocytosis and NF-B activation. NF-B inhibition has
infections increased after the 1960s—the same time as the
been shown via YopJ/P, which is able to interfere with IKK
Crohn’s disease outbreak.32 Because Crohn’s disease is a
proteins.39,40 However, the mechanism of inhibition is not
complex genetic trait, progressive population exposure to
completely understood, Yop proteins might also interact
Yersinia spp might have allowed more and more genetically
physically with CARD15 to induce an inhibition of the
predisposed people to be affected. Interestingly, the
NF B by an alternative pathway. This NF-B inhibition
present stabilisation of the disease incidence in North
could be due to YopJ/P or other molecules, including
America and in some European countries suggests that
YopM which is characterised, like CARD15, by leucine
almost all people who are genetically at risk are now
rich repeats (LRRs). A similar model can also be proposed
THE LANCET • Vol 362 • December 13, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
for other psychrotrophic strains such as Listeria spp that
14 Hellers G. Crohn’s disease in the Stockholm county 1955–74:
carry LRR proteins at their surface.
a study of epidemiology, results of surgical treatment, and long-term
Our hypothesis could be tested in experimental models in
prognosis. Acta Chir Scand 1979; 490 (suppl): 1–84.
15 Brahme F, Lindstrom C, Wenckert A. Crohn’s disease in a defined
which cells expressing the wild type or mutated CARD15
population: an epidemiological study of incidence, prevalence,
protein were analysed in the presence of Yop proteins.
mortality, and secular trends in the city of Malmo, Sweden. Gastroenterology 1975; 69: 342–51.
16 Champagne CP, Laing RR, Roy D, Assanta Mafu A. Psychrotrophs
We propose a link between Crohn’s disease, the cold chain,
in dairy products: their effects and their control. Crit Rev Food Sci Nutr 1994; 34: 1–30.
and chronic infestation of the digestive tract by
17 Liu Y, Van Kruiningen HJ, West AB, Cartun RW, Cortot A,
psychrotrophic bacteria. Our model provides a structured
Colombel JF. Immunocytochemical evidence of Listeria, Escherichia
and unified explanation, based on biological evidence and a
coli and Streptococcus antigens in Crohn’s Disease. Gastroenterology
network of statistical relationships (figure). By themselves,
1995; 108: 1396–1404.
each line of evidence is quite convincing, but their
18 Walmsley RS, Anthony A, Sim R, Pounder RE, Wakefield AJ.
Absence of Escherichia coli, Listeria monocytogenes and Klebsiellapneumoniae antigens within inflammatory bowel disease tissues.
However, much more testing is needed before our model
J Clin Pathol 1998; 51: 657–61.
can be retained or discarded. Furthermore, some intrinsic
19 Chen W, Li D, Paulus B, Wilson I, Chadwick VS. Detection of
limitations can be raised. For example, because our
Listeria monocytogenes by polymerase chain reaction in intestinalmucosal biopsies from patients with inflammatory bowel disease and
hypothesis takes into account only one genetic risk factor
controls. J Gastroenterol Hepatol 2000; 15: 1145–50.
and one environmental component from several expected
20 Chiba M, Fukushima T, Inoue S, Horie Y, Iizuka M, Masamune O.
factors, it cannot elucidate the clinical heterogeneity seen
Listeria monocytogenes in Crohn’s disease. Scand J Gastroenterol
between patients with Crohn’s disease.
1998; 33: 430–34.
The cold chain has produced many benefits for western
21 Swidsinski A, Ladhoff A, Pernthaler A, et al. Mucosal flora in
inflammatory bowel disease. Gastroenterology 2002; 122: 44–54.
societies, including the prevention of enteric infections,
22 Greenwood MH. Human carriage of Yersinia species and incidence
allowing more people access to a well-balanced diet, and
in foods. Contrib Microbiol Immunol 1995; 13: 74–76.
the economic development of agriculture and fishing.
23 Blaser MJ, Miller RA, Lacher J, Singleton JW. Patients with active
These advantages clearly outweigh the putative risks
Crohn’s disease have elevated serum antibodies to antigens of seven
discussed here and, in the absence of experimental
enteric bacterial pathogens. Gastroenterology 1984; 87: 888–94.
evidence, practical conclusions should not be drawn.
24 Kalinowski F, Wassmer A, Hofmann MA, et al. Prevalence of
enteropathogenic bacteria in surgically treated chronic inflammatory bowel disease. Hepatogastroenterology 1998; 45: 1552–58.
25 Lamps LW, Madhusudhan KT, Havens JM, et al. Pathogenic
Yersinia DNA is detected in bowel and mesenteric lymph nodes from patients with Crohn’s disease. Am J Surg Pathol 2003; 27:
We thank Stéphane Bonacorsi, Pierre Laurent-Puig,
26 Van Kruiningen HJ, Ganley LM, Freda BJ. The role of Peyer’s
Patricia Mariani-Kurkdjian, Corine Miceli-Richard, Robert Modigliani,
patches in the age-related incidence of Crohn’s disease.
Michel Peuchmaur, and Jessica Zucman-Rossi for comments and fruitful
J Clin Gastroenterol 1997; 24: 470–75.
discussions. This project was supported by the Institut National de laSanté et de la Recherche Médicale (INSERM).
27 Polito JM, Childs B, Mellits ED, Tokayer AZ, Harris ML,
Bayless TM. Crohn’s disease: influence of age at diagnosis on site and clinical type of disease. Gastroenterology 1996; 111: 580–86.
28 Fujimura Y, Kamoi R, Iida M. Pathogenesis of aphtoid ulcers in
Podolsky DK. Inflammatory bowel disease. N Engl J Med 2002; 347:
Crohn’s disease: correlative findings by magnifying colonoscopy,
electron microscopy, and immunochemistry. Gut 1996; 38:
Hugot JP, Chamaillard M, Zouali H, et al. Association of NOD2
leucine-rich repeat variants with susceptibility to Crohn’s disease.
29 Gleason TH, Patterson SD. The pathology of Yersinia enterocoliticaNature 2001, 411: 599–603.
ileocolitis. Am J Surg Pathol 1982; 6: 347–55.
Ogura Y, Bonen DK, Inohara N, et al. A frameshift mutation in
30 Sukegawa Y, Kamiya S, Yagita A, Sugamata M, Atomi Y. Induction
NOD2 associated with susceptibility to Crohn’s disease. Nature 2001,
of auto-immune colitis by Yersinia enterocolitica 60-kilodalton
411: 537-539.
Heat-Shock Protein. Scand J Gastroenterol 2000; 35: 1188–93.
Andus T, Gross V. Etiology and pathophysiology of Inflammatory
31 Arnold GL, Beaves MR, Prvjdun VO, Mook WJ. Preliminary study
Bowel Disease-environmental factors. Hepatogastroenterology 2000; 47:
of ciprofloxacin in active Crohn’s disease. Inflamm Bowel Dis 2002;
8: 10–15.
Mayberry JF, Rhodes J. Epidemiological aspects of Crohn’s Disease: a
32 Mollaret HH. Fifteen centuries of Yersiniosis.
review of the literature. Gut 1984; 25: 886–99. Contrib Microbiol Immunol 1995; 13: 1–4.
Rose JD, Roberts GM, Williams G, Mayberry JF, Rhodes J. Cardiff
33 Sonnenberg A, McCarty DJ, Jacobsen SJ. Geographic variation of
Crohn’s disease jubilee: the incidence over 50 years. Gut 1988; 29:
inflammatory bowel disease within the United States. Gastroenterology 1991; 100: 143–49.
Fellows IW, Freeman JG, Holmes GKT. Crohn’s disease in the city of
34 Shivananda S, Lennard-Jones J, Logan R, et al. Incidence of
Derby, 1951–85. Gut 1990; 31: 1262–65.
inflammatory bowel disease across Europe: is there a difference
Loftus EV, Silverstein MD, Sandborn WJ, Tremaine WJ,
between north and south? Results of the European collaborative
Harmsen WS, Zinsmeister AR. Crohn’s Disease in Olmsted County,
study on inflammatory bowel disease (EC-IBD). Gut 1996; 39:
Minnesota, 1940–93: incidence, prevalence, and survival. Gastroenterology 1998; 114: 1161–68.
35 Zeng L, Anderson FH. Seasonal change in the exacerbations of
Gent AE, Hellier MD, Grace RH, Swarbrick ET, Coggon D.
Crohn’s Disease. Scand J Gastroenetrol 1996; 31: 79–82.
Inflammatory bowel disease and domestic hygiene in infancy. Lancet
36 Chamaillard M, Philpott D, Girardin S, et al. Gene-environment
1994; 343: 766–67.
interaction modulated by allelic heterogeneity in inflammatory
10 McCormick P, Manning D. Chronic inflammatory bowel disease and
diseases. Proc Natl Acad Sci 2003; 100: 3455–60.
the “over clean” environment: rarity in the Irish traveller community.
37 Schreiber S, Nikolaus S, Hampe J. Activation of nuclear factor kappa
Ir Med J 2001; 94: 203–04.
B inflammatory bowel disease. Gut 1998; 42: 477–84.
11 Karlinger K, Györke T, Mako E , Mester A, Tarjan Z.
38 Miceli-Richard C, Lesage S, Rybojad M, et al. CARD15 mutations
The epidemiology and the pathogenesis of inflammatory bowel
in Blau syndrome. Nat Genet 2001; 29: 19–20.
disease. Eur J Radiol 2000; 35: 154–67.
39 Orth K, Xu Z, Mudgett MB, et al. Disruption of signalling by
12 Hugot JP, Cézard JP, C JF, Belaiche J et al. Clustering of Crohn’s
Yersinia effector YopJ, a ubiquitin-like protein protease. Science
disease within affected sibships. Eur J Hum Genet 2003; 11:
2000; 290: 1594–97.
40 Orth K, Palmer LE, Bao ZO, et al. Inhibition of the mitogen-
13 Thévenot R. Essai pour une histoire du froid artificiel dans le monde.
activated protein kinase kinase superfamily by a Yersinia effector.
Institut International du froid: Paris, 1978. Science 1999; 285: 1920–23.
THE LANCET • Vol 362 • December 13, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
Final report for IFAR project 2008 Proposal code: 267IFAWAN. Name of Applicant: Dr. Mohamed El-Sayed Mohamed El-Boushy Assistant Professor of Clinical Pathology Faculty of Veterinary Medicine, El Mansoura University, Egypt. 1. Proposed Work Program: The possible impact of common zoonotic and food borne bacteria on fish and human health with suggestive control measures The