History of enteritis necroticans

Enteritis necroticans (EN) first received attention after World War II when it emerged in Germany and other European countries and was called Darmbrand or ‘fire bowels’. This disease was notable for severe abdominal pain, nausea and vomiting and was often fatal. Clostridium perfringens (previously, C. welchii) isolates were recovered from necrotic intestinal tissue and the disease was thought to be related to toxins produced by these clostridia [5]. Seven of these Darmbrand-associated strains were recovered from storage cultures and characterized by modern genotypic and phenotypic analyses [6]. Five of the strains were confirmed as type C and carried genes for beta toxin (cpb) and enterotoxin (cpe) on large plasmids. All seven produced highly heat-resistant spores and shared spore heat-resistant mechanisms with C. perfringens strains that cause human food poisoning. There was evidence supporting food-borne transmission, but malnutrition and other factors may have also been involved in the pathogenesis of Darmbrand [7]. The disease disappeared in the 1950s possibly because of the overall improvement in sanitation and general nutrition of the population. Unlike cases in PNG; however, the cases in Germany affected primarily people 30–50 years of age [7].

The first case of pigbel in PNG was reported in 1961 and was thought to be caused by the consumption of contaminated pork. The pig in Papua New Guinean culture is of social and dietary importance. Pigs were kept in close contact to humans and were used for ceremonial practices where they were eaten to celebrate rite of passages—the most notable being called “Pig Feasts”. The exact feast preparation varies by PNG region [7], but standard preparation includes butchering it publicly (oftentimes large numbers of pigs), shortly roasting the pig over an open fire to singe off the hair, and then cooking it over several hours in ground pits heated with hot stones. Samples of soil from the highland villages in which these rituals take place have identified C. perfringens type C by fluorescent antibody staining [8]. C. perfringens type C isolates were recovered in the intestines of patients in PNG exhibiting pigbel [9] suggesting these cultural practices compounded with an otherwise low-protein diet apart from pork at ritual feasts provide a route of infection and predisposition for pigbel [10]. It was also postulated that consuming sweet potato, a major staple of the traditional diet among highlanders, played a role in the disease. Sweet potatoes contain trypsin inhibitors that could potentially suppress pancreatic enzyme activity that aids in breakdown of the beta toxin leading to the proliferation of the toxin and subsequent necrosis of the jejunal and ileal mucosa [2]. Evidence from multiple sources supports the role of trypsin inhibitors in the pathogenesis of this disease including experiments where guinea pigs were challenged intragastrically with C. perfringens type C and raw soybean flour developed pigbel-like lesions but not when challenged with the pathogen alone or when the soybean flour was autoclaved destroying the trypsin inhibitors [11]. Trypsin inhibition is a particularly important risk factor given the central role of the trypsin-sensitive beta toxin in the pathogenesis of EN. However, host factors other than a diet replete with trypsin inhibitors may also contribute to susceptibility including malnutrition and chronic diseases such as diabetes, particularly outside of PNG.

From time of the first reported cases in early 1960s to the 1980s, pigbel reached epidemic significance in the highlands of PNG and was cited as the second leading cause of death in children over the age of 1 [11]. Children aged 2–10 years old were more at risk of infection, while males were affected more than females. Among those, 16–20 aged females were more likely to contract the disease compared to men, possibly explained by marriage ceremonies involving pig feasting where the bride consumes large amounts of pork to encourage fertility [7].

Pigbel is clinically characterized by acute abdominal pain and is often accompanied with bloody diarrhea, vomiting blood [12], and bloating of the stomach due to upper abdominal distension—where the name Pigbel (pain in belly associated with ritualistic pig-feasting) was coined [9]. Death of affected patients can occur within 24 h after onset of clinical signs due to enteric necrosis progressing to peritonitis and/or septic shock [9,11]. Macroscopic intestinal lesions are characterized by segmental hemorrhage and necrosis of mainly the jejunum. Lesions can progress to a full thickness necrosis in small intestinal segments, and also extend to the large intestine. Histopathological hallmarks in acute cases are deep mucosal necrosis with vascular thrombosis and necrosis accompanied by extensive hemorrhage. Subacute lesions are more often described as multifocal patchy necrosis of the small intestine.

A toxoid vaccine targeting C. perfringens type C was introduced in 1979. A widespread vaccination program for children in the highlands led to a dramatic decrease in pigbel cases [13]. Pigbel was then thought to be no longer endemic to PNG and disappeared as it did in Europe in the 1950s. With production of the vaccine ceasing in 1992 and the remaining stock used or discarded the vaccination program also ceased. However, recent studies from 2006 to 2015 have documented confirmed pigbel cases in Highland hospitals and there are potentially more cases in areas where surveillance is more difficult [1,14].

Epidemiology of enteritis necroticans in Papua New Guinea post-vaccination program

Prior to vaccination in areas where pigbel prevalence was high in 1979, Goroka Hospital in the Eastern Highlands Province reported more than 100 cases annually of children admitted with pigbel while Mount Hagen Hospital in the Western Highlands Province reported 60–100 cases annually. Post-vaccine era numbers from 1993 to 2001 demonstrate a marked decrease of reported cases of pigbel with cases reported less than 5% of pre-vaccination numbers in Goroka and less than 10–20% of pre-vaccination numbers in Mount Hagen [15]. At the start of the immunization program in 1981, Tari Hospital (Enga Province) reported that the number of deaths related to pigbel among children under 5 years of age had dropped from 36 to 5 within 2 years of administration.

Cases of pigbel were still being diagnosed in PNG after the vaccination had ceased to be administered. From 1999 until the end in 2001, the PNG National Health Information System (NHIS) reported 105 pigbel cases from hospitals in the Southern Highlands, Simbu, Eastern Highlands, Western Highlands, and Enga Provinces [15].

In a study conducted in 2002, 119 children with acute abdominal pain were identified in 38 health centers and hospitals in the highlands over the 12-month period. Nine percent were confirmed cases of pigbel and 16% were identified as probable EN. Of the confirmed and possible cases, 50% were found to cluster along the border of the Western Highlands and Enga Provinces [15].

A surveillance study was piloted for pigbel based on a standardized clinical case definition and a beta-toxin immunoassay (EIA) for testing of intestinal fluid in suspected cases from 2012 to 2015 [1]. This proof of concept trial was conducted at one hospital in the Jiwaka Province [14]. Inclusion criteria for the study included patients admitted to the hospital with severe abdominal pain of less than 2 weeks duration. A standardized case definition was used to identify likely cases of pigbel and patient specimens including gastric aspirates, stool, and intestinal fluid were tested for the presence of C. perfringens beta toxin by immunoassay (EIA). Among 105 patients with acute abdominal pain, 48 met the case definition. The mean age was 5 years (IQR: 2–6) and 35% (12/34) tested positive for beta toxin. Four patients who did not meet the case definition were also positive by the beta-toxin EIA and four with classical pathologic lesions tested negative by the EIA. Five children died, four of whom met the case definition. The EIA has not been subsequently used for resource reasons.

Between 2013 and 2022, the Paediatric Hospital Reporting program, a nation-wide surveillance program for all childhood hospital admissions, recorded 113 cases of pigbel and 14 deaths, diagnosed based on the standardized and/or surgical clinical criteria. Annual cases are listed in Table 1. This reporting system involves 16–24 hospitals, but not all are able to report a complete set of pediatric data each year. Additional epidemiologic data such as association with pig feasts or specific food consumption are not available from this program, but there is speculation that many of these cases are sporadic and not associated with feasts or large gatherings.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Annual Pigbel cases and deaths reported to the Papua New Guinea Paediatric Hospital Reporting program: https://pngpaediatricsociety.org/reports/annual-child-morbidity-and-mortality-reports/.

https://doi.org/10.1371/journal.pntd.0012836.t001

In conclusion, there is an ongoing burden of pigbel in PNG. Beta-toxin EIA testing is helpful for disease confirmation; however, the test is insensitive and a negative test does not rule out pigbel.

Epidemiology of enteritis necroticans outside of Papua New Guinea

Five reports of EN were identified in the literature since 2000 where C. perfringens type C was recovered from intestinal specimens or was confirmed present based on amplification of the cpb gene from involved tissue (Table 2). Four of these reports represented individual cases and the other report was a summary of an outbreak in eastern Sri Lanka where 42 patients were seen in a 6-month period in 2002 [16]. Two-thirds of the cases in this outbreak were younger than 10 years of age, laparotomy identified small intestinal lesions consistent with EN in 13 patients, and C. perfringens was recovered in 7, with C. perfringens type C confirmed by PCR typing of the isolates in 4. The four individual case reports gave more detailed epidemiologic and clinical findings and represented patients from the United States of America (USA) (two cases), Japan, and the United Kingdom (UK) [17–20]. These cases ranged in age from 12 to 66 years, three were male, and diabetes mellitus (DM) was a co-morbidity in three of the four. All four patients had consumed a food source prior to symptom onset that has been historically associated with other forms of food poisoning; Chitterlings, locally processed turkey sausage, raw seafood, and chicken. Three of the four patients recovered after surgical resection of the involved necrotic small intestinal segments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Enteritis necroticans (confirmed or presumed Clostridium perfringens type C associated) reported outside of Papua New Guinea since 2000.

https://doi.org/10.1371/journal.pntd.0012836.t002

In addition to the above-confirmed reports, seven additional reports of EN were identified where C. perfringens type C was suspected, but not confirmed by microbiological or molecular testing (Table 2). Five of these reports represented individual cases and two reports were case series from the Indian subcontinent. One of these case series described 18 laparotomy-confirmed cases of EN in children aged 6 months to 12 years over a 6-year period from New Delhi, India [21]. They documented patchy transmural mucosal necrosis of the jejunum (primarily) and ileum and the 10 patients that had complete resection of the involved intestine showed dramatic improvement following surgery. The authors also noted that many more cases were suspected clinically, but responded to conservative management. The second case series described segmental enteritis in 24 children over a 6-year period from Chittagong, Bangladesh [22]. All cases underwent laparotomy and showed predominantly jejunal necrotic lesions. The five individual case reports involved three males and two females, ranging in age 3–70 years, from the USA (two cases), Belgium, Switzerland, and Australia [23–25]. Potential predisposing factors included DM, cyclic neutropenia, and previous bowel surgery (two cases). One case, in a child with immunodeficiency, developed extensive hemorrhagic necrosis and pneumatosis of the entire small bowel one week after a bone marrow transplant (Fig 1). All cases involved necrotic sections of small intestine, and C. perfringens was identified in four of the five cases, but C. perfringens type C was not confirmed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Abdominal CT scan of a 15-year-old girl with abdominal sepsis 1 week after a bone marrow transplant.

Hemorrhagic necrosis of the small bowel is seen in addition to extensive pneumatosis intestinalis, ascites and the periportal pneumatosis in the liver. Clostridium perfringens was recovered from the blood and ascitic fluid.

https://doi.org/10.1371/journal.pntd.0012836.g001

In addition to clinical cases of EN, C. perfringens type C has been isolated from numerous clinical and environmental samples worldwide including stools of infants in Jordan [26], fresh water fish in China [27], farm animals in Saudi Arabia [28], horse fecal and soil samples from South Korea [29], and soil samples from the USA [30,31].

In summary, reports of EN from South Asia since 2000 describe cases similar to those reported in PNG and include primarily young children with similar frequency among males and females and with clinical presentations that varied in severity. In contrast, case reports of EN from the USA, Europe, and Japan describe a wide age range of affected patients who were predominately male (6 of 9 case reports). DM is a common predisposing risk factor (4 of 9 cases) and all of the confirmed cases outside of the Indian subcontinent describe a potentially contaminated meal eaten prior to symptom onset. The onset of symptoms after the meal in most cases (1–2 days) suggest infectious enteritis rather than ingestion of preformed toxin.

Pathogenesis of enteritis necroticans

C. pefringens type C has early on been identified as the cause of EN due to the epidemiological evidence of its association with the disease in humans. The pathogen also causes similar enteritic disease in horses, sheep, goats and cattle, but the pig is the by far most frequently affected host species (Table 3). Much of our knowledge about the pathogenesis of human EN is extrapolated from studies on naturally occuring C. perfringens type C enteritis in animals, ex vivo laboratory work and research using experimental animal models. The pathogenesis of C. pefringens type C in pigs is reviewed elsewhere [3]. Here, we relate the most noteworthy parallels to the human disease.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Enteric disease in animals due to Clostridium perfringens type C.

https://doi.org/10.1371/journal.pntd.0012836.t003

Type C strains are defined by secretion of alpha- and beta-toxin [32], but they secrete a plethora of additional toxins and enzymes. Whereas alpha-toxin (a phospholipase) is produced by all C. perfringens strains, including apathogenic strains, beta-toxin is unique to type C and B strains. Using animal models and genetic modification of C. perfringens type C strains it was shown that beta-toxin was the essential virulence factor for induction of enteric lesions [33].

Generally, C. perfringens spores or vegetative bacteria have to be orally ingested and germinate to colonize the intestine (Fig 2). This is followed by a phase of rapid proliferation, either immediately after colonization or in the case where favorable nutritional conditions occur. During their exponential growth phase, pathogenic type C strains secrete large amounts of toxins as well as enzymes and produce metabolites which as a whole can have deleterious effects on the host intestinal barrier [34]. Currently, the initial steps leading to an alteration of the small intestinal epithelial barrier are incompletely understood. Epithelial cells are resistant to beta-toxin [35–37], thus other virulence factors and/or the presence of resident co-pathogens may be important for this. For example enterotoxin, another virulence factor of C. perfringens capable of causing damage to epithelial tight junctions, is encoded by several C. perfringens type C strains including the Darmbrand-associated strains mentioned previously [5,6]. Epithelial barrier damage most likely allows beta-toxin to diffuse into the intestinal mucosa and reach the vascular endothelium (Fig 3). Endothelial cells and potentially platelets and immune cells are targeted via the specific interaction of the toxin with its membrane receptor Platelet Endothelial Cell Adhesion Molecule 1 [38–40]. Beta-toxin forms oligomeric transmembrane pores in the plasma membrane of target cells leading to endothelial cell necrosis and acute vascular damage. The result is acute local hemorrhage and further hypoxic damage in the affected part of the small intestinal mucosa [35,41–43]. Upon exposure to such nutrients, C. perfringens type C proliferates even further and upregulates toxin secretion [44]. This leads into a vicious cycle of bacterial proliferation, toxin secretion, beta-toxin-induced vascular damage, hemorrhage and clostridial toxin and enzyme-induced tissue damage, ending up in rapidly progressive intestinal necrosis. Toxemia, resulting from resorption of beta and other toxins through the damaged intestinal barrier most likely contributes to the disease in later stages; however, systemic effects of specific toxins have so far not been conclusively demonstrated in naturally occurring disease.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Proposed pathogenesis of enteritis necroticans.

(1) Healthy intestine with microbiota; (2) colonization with Clostridium perfringens type C, beta-toxin degradation through trypsin; (3) proliferation and toxin secretion of C. perfringens type C, potentially due to disbalanced microbiota, uptake of large numbers of pathogen, trypsin inhibitors preventing beta-toxin degradation, or other factors that may facilitate proliferation and toxin elaboration; (4) initial epithelial barrier alteration (through additional clostridial virulence factors or other co-pathogens), diffusion of beta-toxin into lamina propria, local endothelial damage with plasma extravasation and small hemorrhages, lesions macroscopically most likely not yet visible; (5) increased nutrient supply for C. perfringens through vascular leakage, increased proliferation and upregulation of toxin secretion leads to acceleration of damage, lesions become macroscopically visible and extend rapidly; (6) vicious cycle of clostridial proliferation, toxin secretion, vascular damage and nutrient supply leads to rapidly progressing hemorrhagic transmural jejunitis with peritonitis, lesions potentially extend into distal small and also large intestine, death can occur at this stage; (7) in more protracted cases patchy lesions develop and there is marked inflammatory response in affected intestinal segment, lesions become macroscopically less hemorrhagic and more fibrinous, severe peritonitis can develop. Figure created with BioRender.com.

https://doi.org/10.1371/journal.pntd.0012836.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Clostridium perfringens beta-toxin targets endothelial cells of mucosal vasculature during early stages of lesion development in an experimental porcine infection model.

(A) Histopathology of an early jejunal lesions of necrotic enteritis induced by C. perfringens type C in a porcine ileal loop model showing hemorrhage in the tips of villi and epithelial cell sloughing. Magnification 400× (B) Immunohistochemistry for beta-toxin of a serial section from A. Note that beta-toxin signal (red) stains the vascular outline (endothelial cells) in the lamina propria. (Samples derived from study by Schumacher and colleagues [70]).

https://doi.org/10.1371/journal.pntd.0012836.g003

In case of human EN, it is not clear how and when during the disease process intestinal colonization occurs. C. perfringens spores can persist in the environment for years and the bacteria are believed to persist in the intestines of suitable hosts, such as pigs [45]. This can lead to long-lasting contamination of the environment and recurrent exposure of humans in endemic areas. This was most likely the case in the Highlands of PNG, where in the past, pigs were kept in very close contact to their human owners. The most prominent risk factor for humans in PNG in the past was the sudden consumption of pork meat following a traditional pig feast. Contamination of pork meat during the traditional slaughtering and cooking process, followed by distribution of the meat over a period of more than 1 day could have led to ingestion of large amounts of vegetative clostridia and/or spores. In addition, traditional pig feasting represented a sudden dietary change from a mainly vegetable to a meat-dominated diet, which could have affected the intestinal microbiota and favored clostridial overgrowth. Such sudden food changes are well-known risk factors for different clostridial enteric diseases in animals [46]. During pig feasting in PNG, newborn children still breast feeding were unlikely to be exposed to pork meat, but children 2–10 years of age were at highest risk to develop EN [7].

Another unproven factor might be the presence of resident co-pathogens that breach the upper intestinal epithelium, such as parasitic infection by Strongyloides or Ascaris. The adult forms of these parasites reside in the small intestine and were common among Khmer children with EN [47]. Compared to the disease in pigs, parasitic co-infections are not required for the development NE. However, Isospora suis invading and damaging small intestinal epithelial cells in newborn piglets, can initiate lesions that allow clostridia access to damaged tissue. Epithelial damage by coccidia (Eimeria sp.), are indeed an important predisposing factor in a very similar disease in chickens, caused by C. perfringens type G strains that produce NetB, a hemolysin beta-pore forming toxin related to beta-toxin [48]. In addition to providing a potential breach in the small intestinal epithelial barrier, adult parasites, such as Ascaris sp. are known to produce typsin inhibitors which are thought to facilitate their survival in a protease rich environment, but also may inhibit beta-toxin degredation [49].

The circumstances under which the more recent sporadic cases of EN occurred in PNG and elsewhere; however, are unclear. Association of sporadic cases with traditional pig feasts or sudden dietary changes is not consistently reported. With the currently available data, it seems likely that a common feature of EN is ingestion of contaminated food potentially in combination with dietary changes or other contributing factors facilitating C. perfringens type C proliferation and beta-toxin elaboration. Many human cases show pathological lesions resembling a more protracted course of the disease that is also observed in piglets dying of NE in the second or third week of life [50]. This could be due to either lower infectious doses ingested by the individual patient/animal, partial protection by the resident microbiota, or, a combination of both.

One important contributing factor in NE is that the presence of trypsin inhibitors in the small intestine. Beta-toxin is highly sensitive to degradation and inactivation by the pancreatic enzyme trypsin [51]. Thus, trypsin inhibitors can enhance stability and toxicity of beta-toxin [52]. In the highlands of PNG, sweet potato is the dietary staple and this is rich in trypsin inhibitors. One study showed that after 1 year of age PNG highlands children from areas of high risk for EN had significantly lower trypsin stool levels than controls including coastal village children, European or PNG children consuming “western” diets, and children under age 1 from high-risk area who were breast fed [53]. The association of NE with decreased trypsin activity is also paralleled in C. perfringens type C-induced enteritis in suckling piglets where sow colostrum contains high amounts of trypsin inhibitors [54].

DM has a striking association with EN cases outside of PNG and was documented in 4 of the 9 recent cases in our review. The reason for this association is not known, and while exocrine pancreatic dysfunction with trypsin/protease deficiency is a potential complication of type 1 DM, both types 1 and 2 DM appear to be associated [55]. DM is also associated with reduced gastric and small intestinal motility. A delay in small-intestinal transit time and overgrowth of bacteria in the proximal small bowel might also facilitate C. perfringens type C proliferation [56,57]. Further research is needed to elucidate the mechanism(s) whereby DM confers a risk for EN.

Other human clostridial diseases related to enteritis necroticans

Over the past two decades, it has become apparent that other bacteria can cause necrotizing enteric diseases in humans, particularly C. perfringens strains that do not produce beta-toxin. They are referred to as C. perfringens type A in the literature, but some were shown to produce enterotoxin and they would now be referred to as type F strains [32]. These C. perfringens strains have been implicated in several necrotizing enteric disease syndromes sometimes closely mimicking C. perfringens type C-induced EN [58–60]. Sobel and colleagues described four adult patients in North America with necrotizing enterocolitis, three of them died, in whom C. perfringens was isolated or where clostridial antigens were detected by immunohistochemistry in affected intestinal specimens [60]. Cases were complicated by portal or mesenteric thrombosis in three patients and the same enterotoxin-producing C. perfringens strain was recovered from one patient, his wife, and food from a restaurant they ate at. The site of necrosis was the colon or distal ileum in three; the entire small bowel and colon in one, and the jejunum only in one case.

Another syndrome involving necrotizing colitis among institutionalized adults with chronic constipation has been linked to food borne enterotoxin-producing C. perfringens strains [61,62]. Bos and colleagues described a cluster of seven cases, two of them died, among residents at a residential care facility for mental illness in Oklahoma [61]. The necrotizing lesions involved the transverse, descending or sigmoid colon and C. perfringens isolates were recovered that contained a chromosomal cpe gene supporting food borne origin of the infection. A Thanksgiving meal containing turkey was the suspected vehicle. Constipation due to anticholinergic effects of psychiatric medications and fecal impaction were suggested as risk factors. Another report involved 42 residents and 12 staff at a Louisiana state psychiatric hospital who developed acute gastrointestinal symptoms [62]. Three of the residents, all who received medications with anti-intestinal motility side effects, died. Necrotizing colitis was documented post-mortem in two cases. C. perfringens enterotoxin (CPE) was detected in the stool of 20 of 23 ill residents and a chicken dinner consumed the previous day was associated with illness. CPE+ C. perfringens strain was also recovered from samples of chicken.

In summary, despite the lack of beta toxin, other C. perfringens strains have been shown to cause disease similar to C. perfringens type C-induced EN. Although not proven, predisposing factors may also dictate the location of the lesions such as constipation leading to colonic necrosis in some patients with food borne C. perfringens strain infection and co-infection with parasites leading to small bowel disease in patients with C. perfringens type C.

Prevention of enteritis necroticans

To date, the most effective preventive method against EN caused by C. perfringens type C is vaccination. The human toxoid vaccine used from the late 1970s to 1990s in PNG was produced from formalin inactivated supernatants of C. perfringens type C strains and as such contained many secreted factors and exotoxins, including beta-toxin. As mentioned above, the vaccine campaign was remarkably successful, but discontinued after case numbers had dropped and remained low. The vaccine was administered mainly to the risk group of young children [13].

Similar vaccines are still widely and successfully used in veterinary medicine against different clostridial enteric diseases. (https://www.msdvetmanual.com/digestive-system/intestinal-diseases-in-pigs/clostridium-perfringens-type-c-enteritis-in-pigs#Treatment-and-Control_v3264050). Depending on the age of animals at risk, vaccines are used to induce protective immunity of the potentially affected animal itself (e.g., C. perfringens type D vaccines in sheep), or, in the case of pigs, to induce protective antibody levels in the colostrum of sows. These are then passively transferred to the piglets in the first days of life and provide sufficient maternal immunity to prevent disease outbreaks during the critical period of up to 3 weeks post-partum [63].

As C. perfringens vegetative forms or spores can be taken up by humans from the environment and contaminated food, standard hygiene measures and proper food processing are potentially the most important prophylactic measures against EN as well as the much more common enteric and other infections that lead to malnutrition, stunting, and chronic illnesses. Continued epidemiologic surveillance of EN and targeted investigations may clarify other risk factors such as sudden dietary changes, ingestion of contaminated foods, and the role of trypsin, and thus point to other preventive measures.