Chronic dirrhea (Henry J. Binder, M.D.

http://content.nejm.org/cgi/content/full/355/3/236... Volume 355:236-239

20, 2006)

July

Diarrhea can be classified in several ways. It is both a symptom and a sign. As a symptom, it is whatever the patient says it is: a decrease in consistency, an increase in the number or volume of bowel movements, or any combination thereof. As a sign, diarrhea is an increase in stool weight (or volume) of more than 200 g (or ml) per 24 hours in a person on a Western diet. The distinction between chronic and acute diarrhea is arbitrary: it is determined by duration. Diarrhea is generally considered acute when it lasts less than two or three weeks. Such diarrhea is frequently caused by an infectious agent, is usually self-limiting, and often resolves without treatment. Diarrhea is categorized as chronic when it lasts more than two or three weeks. Chronic diarrhea has multiple causes. During the past three decades, studies have delineated several ion-transport mechanisms that may be disturbed in one or more diarrheal disorders. The identification of five specific congenital diarrheal disorders (see table) has confirmed the importance of certain iontransport mechanisms in health and disease. The report by Wang et al. in this issue of the Journal (pages 270280) implicates a new cause of congenital diarrheal disorders: the absence of enteric hormones. Diarrhea is often classified as osmotic or secretory.1 It is considered osmotic if luminal substances are responsible for the induction of the fluid secretion, and it is considered secretory if endogenous substances (often referred to as "secretagogues") induce fluid secretion that persists even when the patient is fasting. Textbooks frequently state that one can distinguish osmotic from secretory diarrhea by determining the electrolyte concentration in the stool: if there is an osmotic gap that is, a substantial difference between the stool osmolality and twice the concentrations of sodium and potassium in the stool the diarrhea is osmotic, and if not, it is secretory. The assessment of stool electrolytes, however, is more useful for teaching purposes than at the bedside, because data on what constitutes a substantial osmotic gap are scant. Moreover, many diarrheal disorders, such as celiac sprue, have more than one pathogenic mechanism, some of which may be secretory and others of which may be osmotic. I find it more helpful to assess the effect of a fast on stool output: when diarrhea ceases with fasting, a dietary nutrient is likely to be the cause; if diarrhea persists unabatedly with fasting, a dietary nutrient is not likely to be the cause. In the former case, malabsorption of carbohydrates, fats, or both is probably to blame. The most common cause of carbohydrate malabsorption is lactose intolerance, secondary to primary lactase deficiency or lactase nonpersistence. In the absence of lactase deficiency, there are several uncommon defects in carbohydrate absorption including sucraseisomaltase deficiency that may be responsible. In the absence of carbohydrate malabsorption in a patient with osmotic diarrhea, it is essential to determine whether steatorrhea is present. Although diarrhea alone may be responsible for an increase in fat excretion of up to 11 g per day (normally, less than 7 g of fat per day is excreted by persons consuming 100 g of fat per day), greater rates of fat malabsorption can be explained only by one or more defects in fat digestion and absorption, which consists of

five well-characterized steps.2 Successful therapy will depend on the identification of the specific defect or defects in fat digestion and absorption. For example, the administration of pancreatic enzymes will reduce steatorrhea in patients with chronic pancreatitis or cystic fibrosis whose steatorrhea results from the reduced hydrolysis of dietary lipids by pancreatic lipase. Central to almost all diarrheal disorders is the induction of fluid and electrolyte secretion in one or more segments of the small intestine, the large intestine, or both. In secretory diarrhea, secretagogues affect ion transport in the intestine both by stimulating chloride secretion through the activation of the cystic fibrosis transmembrane regulator and by inhibiting sodium and chloride absorption. In steatorrhea, diarrhea is caused by the induction of fluid and electrolyte secretion in the colon by nonabsorbed fatty acids. What are the mechanisms underlying the diarrhea of the patients described by Wang et al.? These patients have mutant neurogenin-3, a transcription factor that is pivotal to the development of the pancreatic beta cell and, it would now seem, the enteroendocrine cell. The salient observations include the facts that diarrhea ceased when the patients fasted and was induced when anything other than water was ingested; that the mucosa of the small intestine was essentially normal, except for the absence of enteroendocrine cells; and that there was no inflammatory component. Although diarrhea was induced by orally administered glucose, it is not known whether a glucose-tolerance test would have disclosed normal glucose absorption. The absence of intestinal endocrine cells and enteric hormones is likely to be central to the diarrhea in these patients; enteric hormones may increase sodium and chloride absorption or induce anion secretion. Hitherto, there has been little to implicate enteric hormones in causing diarrhea: a single patient with autoimmune polyglandular syndrome type I had a defect in cholecystokinin that presumably led to reduced pancreatic-enzyme secretion and steatorrhea.3 A similar mechanism, however, could not explain the diarrhea of the patients described by Wang et al., which persisted even in the absence of dietary fat. Overall intestinal function involves the coordinated interaction of intestinal ion transport and motor activity, which are functionally linked through the enteric nervous system. Enteric hormones would presumably affect ion transport through a paracrine mechanism, either directly by regulating absorptive processes, secretory processes, or both in intestinal epithelial cells (see figure) or indirectly, through the enteric nervous system, which is regulated by several enteric peptides and neurotransmitters. The mechanisms that are responsible for diarrhea in patients with mutant neurogenin-3 remain unclear; their delineation should shed welcome light on the physiology of enteric hormones.

Amebiasis (http://content.nejm.org/cgi/content/full/348/16/156 5..... Volume 348:1565-1573 April 17, 2003 )

Rashidul Haque, M.B., Ph.D., Christopher D. Huston, M.D., Molly Hughes, M.D., Ph.D., Eric Houpt, M.D., and William A. Petri, Jr., M.D., Ph.D.
Diarrheal diseases continue to be major causes of morbidity and mortality in children in developing countries. For example, in Bangladesh 1 in 30 children dies of diarrhea or dysentery by his or her fifth birthday.1 In developed countries the microorganisms that cause diarrheal disease remain of concern because of their potential use as bioterrorist agents. Bacillary dysentery is most commonly caused by microorganisms belonging to the genus shigella, whereas amebic dysentery is caused by the protozoan parasite Entamoeba histolytica. The annual number of shigella infections throughout the world is believed to be approximately 164 million.2 Estimates of E. histolytica infections have primarily been based on examinations of stool for ova and parasites, but these tests are insensitive and cannot differentiate E. histolytica from morphologically identical species that are nonpathogenic, such as E. dispar and E. moshkovskii.3,4 Specific and sensitive means to detect E. histolytica in stool are now available and include antigen detection and the polymerase chain reaction (PCR).5 Two recent studies in developing countries used these modern diagnostic tests. A three-year study in Dhaka, Bangladesh, showed that preschool children had a 2.2 percent frequency of amebic dysentery, as compared with a 5.3 percent frequency of shigella dysentery6 (and unpublished data). The annual incidence of amebic liver abscess was reported to be 21 cases per 100,000 inhabitants in Hue City, Vietnam.7 In the United States, where fecal oral transmission is unusual, amebiasis is most commonly seen in immigrants from and travelers to developing countries. The disease is more severe in the very young and old and in patients receiving corticosteroids.

Entamoeba histolytica
Molecular phylogeny places entamoeba on one of the lowermost branches of the eukaryotic tree, closest to dictyostelium. Although the organism was originally thought to lack mitochondria, nuclear-encoded mitochondrial genes and a remnant organelle have now been identified.8,9 Unusual features of entamoeba include polyploid chromosomes that vary in length; multiple origins of DNA replication; abundant, repetitive DNA; closely spaced genes that largely lack introns; a novel GAAC element controlling the expression of messenger RNA; and unique endocytic pathways.10,11,12,13

Pathogenesis
Ingestion of the quadrinucleate cyst of E. histolytica from fecally contaminated food or water initiates infection (Figure 1). This is a daily occurrence among the poor in developing countries and is a threat to inhabitants of developed countries, as the epidemic linked to contaminated municipal water supplies in Tbilisi, Republic of Georgia,

demonstrates.14 Excystation in the intestinal lumen produces trophozoites that use the galactose and N-acetyl-D-galactosamine (Gal/GalNAc)specific lectin to adhere to colonic mucins and thereby colonize the large intestine.15 The reproduction of trophozoites has no sexual cycle, and the overall population of E. histolytica appears to be clonal.16 Aggregation of amebae in the mucin layer most likely triggers encystation by means of the Gal/GalNAc-specific lectin.17 Cysts excreted in stool perpetuate the life cycle by further fecaloral spread. Colitis results when the trophozoite penetrates the intestinal mucous layer, which otherwise acts as a barrier to invasion by inhibiting amebic adherence to the underlying epithelium and by slowing trophozoite motility.18 Invasion is mediated by the killing of epithelial cells, neutrophils, and lymphocytes by trophozoites, which occurs only after the parasite lectin engages host N-acetyl-D-galactosamine on O-linked cell-surface oligosaccharides.15 The interaction of the lectin with glycoconjugates is stereospecific and multivalent.19 The identity of the high-affinity intestinal epithelial-cell receptor is unknown. Secretion by the ameba of amoebapore, a 5-kD pore-forming protein, may contribute to killing.20 Activation of human caspase 3, a distal effector molecule in the apoptotic pathway, occurs rapidly after amebic contact, and caspases are required for cell killing in vitro and for the formation of amebic liver abscesses in vivo.21,22 Interaction of the parasite with the intestinal epithelium causes an inflammatory response marked by the activation of nuclear factor B and the secretion of lymphokines.23,24 The development of this epithelial response may depend on trophozoite virulence factors such as cysteine proteinase and leads to intestinal abnormalities through neutrophil-mediated damage. Neutrophils can also be protective, however, in that activation of neutrophils or macrophages by tumor necrosis factor or interferon kills amebae in vitro and limits the size of amebic liver abscesses.25 In contrast to the intense inflammatory response typical of early invasive amebiasis, inflammation surrounding well-established colonic ulcers and liver abscesses is minimal, given the degree of tissue damage.26 During chronic infection, E. histolytica evades the host immune response in several ways. The Gal/GalNAc-specific lectin has sequence similarity and antigenic cross-reactivity to CD59, a human leukocyte antigen that prevents the assembly of the complement C5bC9 membrane attack complex.27 Amebic cysteine proteinases rapidly degrade the complement anaphylatoxins C3a and C5a.28 The cysteine proteinases also degrade secretory IgA and serum IgG, possibly protecting amebae from opsonization. Finally, amebae appear to suppress both the macrophage respiratory burst and antigen presentation by class II major-histocompatibility-complex (MHC) molecules.

Immunity
Immunity to infection with E. histolytica is associated with a mucosal IgA response against the carbohydrate-recognition domain of the Gal/GalNAc lectin. Over a one-year period, children with this response had 86 percent fewer new infections than children without this response.6 Cell-mediated responses have been described in patients with amebic liver abscess, characterized by lymphocyte proliferation and lymphokine secretion that is amebi-cidal in vitro.29 One study found that in patients with liver abscess, the prevalence of the class II MHC haplotype HLA-DR3 is increased by a factor of more than

three, suggesting a role of CD4+ T-cell function in the outcome of the disease.30 It is noteworthy, however, that the acquired immunodeficiency syndrome pandemic has not led to increases in invasive amebiasis, although asymptomatic intestinal colonization is undoubtedly common.31 In fact, in the murine model of amebic colitis, the depletion of CD4+ T cells decreases the severity of the disease.32

Intestinal Amebiasis
Infection with E. histolytica may be asymptomatic or may cause dysentery or extraintestinal disease (Figure 2 and Figure 3). Asymptomatic infection should be treated because of its potential to progress to invasive disease. Patients with amebic colitis typically present with a several-week history of cramping abdominal pain, weight loss, and watery or bloody diarrhea. The insidious onset and variable signs and symptoms make diagnosis difficult, with fever and grossly bloody stool absent in most cases.33,34,35 The differential diagnosis of a diarrheal illness with occult or grossly bloody stools should include infection with shigella, salmonella, campylobacter, and enteroinvasive and enterohemorrhagic Escherichia coli. Noninfectious causes include inflammatory bowel disease, ischemic colitis, diverticulitis, and arteriovenous malformation. Unusual manifestations of amebic colitis include acute necrotizing colitis, toxic megacolon, ameboma (Figure 3B), and perianal ulceration with potential formation of a fistula. Acute necrotizing colitis is rare (occurring in less than 0.5 percent of cases) and is associated with a mortality rate of more than 40 percent.35 Patients with acute necrotizing colitis typically appear very ill, with fever, bloody mucoid diarrhea, abdominal pain with rebound tenderness, and signs of peritoneal irritation. Surgical intervention is indicated if there is bowel perforation or if the patient has no response to antiamebic therapy. Toxic megacolon is rare (occurring in approximately 0.5 percent of cases) and is typically associated with the use of corticosteroids. Early recognition and surgical intervention are important, since patients with toxic megacolon usually have no response to antiamebic therapy alone. Ameboma results from the formation of annular colonic granulation tissue at a single site or multiple sites, usually in the cecum or ascending colon. An ameboma may mimic carcinoma of the colon (Figure 3B). In developing countries, intestinal amebiasis is most commonly diagnosed by identifying cysts or motile trophozoites on a saline wet mount of a stool specimen (Figure 2E, Figure 2F, and Figure 2H). The drawbacks of this method include its low sensitivity and false positive results owing to the presence of E. dispar or E. moshkovskii infection. The diagnosis should ideally be based on the detection in stool of E. histolyticaspecific antigen or DNA and by the presence of antiamebic antibodies in serum (Table 1). Field studies that directly compared PCR with stool culture or antigen-detection tests for the diagnosis of E. histolytica infection suggest that these three methods perform equally well.5,36 An important aid to antigen-detection and PCR-based tests is the detection of serum antibodies against amebae, which are present in 70 to more than 90 percent of patients with symptomatic E. histolytica infection.37,38,39,40 A drawback of current serologic tests is that patients remain positive for years after infection, making it difficult to distinguish new from past infection in regions of the world where the seroprevalence is high. Examination of colonic mucosal biopsy specimens and exudates can reveal a wide range of histopathological findings associated with amebic colitis, including diffuse,

nonspecific mucosal thickening with or without ulceration and, in rare cases, the presence of amebae in the mucinous exudate; focal ulcerations (Figure 2B) with or without amebae in a diffusely inflamed mucosal layer; classic flask-shaped lesions (Figure 2C) with ulceration extending through the mucosa and muscularis mucosa into the submucosa; and necrosis and perforation of the intestinal wall.41 Staining with periodic acidSchiff or immunoperoxidase and antilectin antibodies aids in the visualization of amebae.42

Amebic Liver Abscess

Amebic liver abscess is 10 times as common in men as in women and is a rare disease in children.37,43,44 Approximately 80 percent of patients with amebic liver abscess present with symptoms that develop relatively quickly (typically within two to four weeks), including fever, cough, and a constant, dull, aching abdominal pain in the right upper quadrant or epigastrium. Involvement of the diaphragmatic surface of the liver may lead to right-sided pleural pain or referred shoulder pain. Associated gastrointestinal symptoms, which occur in 10 to 35 percent of patients, include nausea, vomiting, abdominal cramping, abdominal distention, diarrhea, and constipation. Hepatomegaly with point tenderness over the liver, below the ribs, or in the intercostal spaces is a typical finding.43,44 Laboratory studies may reveal a mild-to-moderate leukocytosis and anemia. Patients with acute amebic liver abscess tend to have a normal alkaline phosphatase level and an elevated alanine aminotransferase level; the opposite is true of patients with chronic disease.44 Ultrasonography, abdominal computed tomography, and magnetic resonance imaging are all excellent for detecting liver lesions (usually single lesions in the right lobe) but are not specific for amebic liver abscess (Figure 3). The differential diagnosis of a liver mass should include pyogenic liver abscess, necrotic hepatoma, and echinococcal cyst (usually an incidental finding that is not the cause of fever and abdominal pain). Patients with amebic liver abscess are more likely than patients with pyogenic liver abscess to be male, to be younger than 50 years of age, to have immigrated from or traveled to a country where the disease is endemic, and not to have jaundice, biliary disease, or diabetes mellitus. Less than half of patients with amebic liver abscess have parasites detected in their stool by antigen detection. Helpful clues to the diagnosis include the presence of epidemiologic risk factors for amebiasis and the presence of serum antiamebic antibodies (present in 70 to 80 percent of patients at the time of presentation) (Table 1). Preliminary studies indicate that the detection of serum amebic antigens is a sensitive, noninvasive means of diagnosis.5 Occasionally, aspiration of the abscess is required to rule out a pyogenic abscess. Amebae are visualized in the abscess fluid in a minority of patients with amebic liver abscess. Complications of amebic liver abscess may arise from rupture of the abscess with extension into the peritoneum, pleural cavity, or pericardium. Extrahepatic amebic abscesses have occasionally been described in the lung, brain, and skin and presumably result from hematogenous spread.

Therapy
Therapy for invasive infection differs from therapy for noninvasive infection. Noninvasive infections may be treated with paromomycin. Nitroimidazoles, particularly metronidazole, are the mainstay of therapy for invasive amebiasis45 (Table 2). Nitroimidazoles with longer half-lives (namely, tinidazole, secnidazole, and ornidazole) are better tolerated and allow shorter periods of treatment but are not available in the United States. Approximately 90 percent of patients who present with mild-to-moderate amebic dysentery have a response to nitroimidazole therapy. In the rare case of fulminant amebic colitis, it is prudent to add broad-spectrum antibiotics to treat intestinal bacteria that may spill into the peritoneum; surgical intervention is occasionally required for acute abdomen, gastrointestinal bleeding, or toxic megacolon. Parasites persist in the intestine in as many as 40 to 60 percent of patients who receive nitroimidazole. Therefore, nitroimidazole treatment should be followed with paromomycin or the second-line agent diloxanide furoate to cure luminal infection. Metronidazole and paromomycin should not be given at the same time, since the diarrhea that is a common side effect of paromomycin may make it difficult to assess the patient's response to therapy.46,47,48

Inflammatory Bowel Disease

Clara Abraham, M.D., and Judy H. Cho, M.D.
http://content.nejm.org/cgi/content/full/361/21/2066 volume 361:2066-2078 November 19, 2009

The idiopathic inflammatory bowel diseases comprise two types of chronic intestinal disorders: Crohn's disease and ulcerative colitis. Accumulating evidence suggests that inflammatory bowel disease results from an inappropriate inflammatory response to intestinal microbes in a genetically susceptible host. Genetic studies highlight the

importance of hostmicrobe interactions in the pathogenesis of these diseases.1,2,3,4,5,6,7,8,9,10,11,12,13 Prominent among these genetic findings are genomic regions containing nucleotide oligomerization domain 2 (NOD2),14 autophagy genes,4,7,8,10 and components of the interleukin-23type 17 helper T-cell (Th17) pathway.2 The NOD2 protein is an intracellular sensor of bacterial peptidoglycan, and autophagy enables cells to regulate and degrade diverse intracellular components, including pathogens.15 The autophagy gene, ATG16L1, has been associated with Crohn's disease but not, thus far, with ulcerative colitis. The interleukin-23Th17 pathway mediates microbial defense and intestinal inflammation.16,17 Multiple genes regulating this pathway have been associated with both Crohn's disease and ulcerative colitis. This review summarizes recent progress in studies of intestinal immunity and genetics in inflammatory bowel disease. Inflammatory bowel disease affects approximately 1.4 million Americans, and its peak onset is in persons 15 to 30 years of age.18 Crohn's disease generally involves the ileum and colon, but it can affect any region of the intestine, often discontinuously. Ulcerative colitis involves the rectum and may affect part of the colon or the entire colon (pancolitis) in an uninterrupted pattern. In Crohn's disease the inflammation is often transmural, whereas in ulcerative colitis the inflammation is typically confined to the mucosa. Crohn's disease can be associated with intestinal granulomas, strictures, and fistulas, but these are not typical findings in ulcerative colitis. Cigarette smoking affects these two diseases differently: smokers are at increased risk for Crohn's disease and tend to have more severe disease, whereas former smokers and nonsmokers are at greater risk for ulcerative colitis. Patients with inflammatory bowel disease are at risk for primary sclerosing cholangitis, ankylosing spondylitis, and psoriasis.19 Familial clustering of cases and twin studies have established a role for genetic factors, which are likely to play a more prominent role in Crohn's disease than in ulcerative colitis.14 The observation that cases of both these diseases can occur within the same family suggests that some of the genes may be common to both disorders. As with other complex genetic disorders, inflammatory bowel disease entails the interaction of genetic and nongenetic factors. Changes in diet, antibiotic use, and intestinal colonization (e.g., the eradication of intestinal helminths) have probably contributed to the increased prevalence of inflammatory bowel disease during the past century.20,21 Our current knowledge of inflammatory bowel disease is based on a combination of gene association studies, clinical investigations, and laboratory experiments in mice. In this review, we first describe homeostasis of the intestinal immune system in health and then focus on advances in our understanding of how genetic alterations in this system contribute to the development of inflammatory bowel disease.

The Intestinal Immune System

The Intestinal Microbiome and Inflammatory Bowel Disease The intestinal microbiome consists of the microorganisms that inhabit the gut. The intraluminal microbiota affects the development of the intestinal immune system, supplies key nutrients, and modulates energy metabolism.22 The intestinal microbiota is acquired at birth but changes rapidly during the first year of life. In adults, each person's unique

population of fecal microbiota is fairly stable over time, but fluctuations occur in response to environmental and developmental factors and in disease.21,23,24 Hostmicrobiome interactions can be mutually beneficial or can be deleterious, inciting intestinal inflammation. Observations in patients with inflammatory bowel disease and in animal models point to the role of bacteria in such inflammation. For example, antibiotics are effective in some patients with inflammatory bowel disease, and most mouse models of colitis require intestinal bacteria for inflammation to occur.25 Bacteria that can adhere to and invade the intestinal mucosa may be particularly important, as in the case of Escherichia coli.26 Although a number of specific pathogens have been incriminated in the development of inflammatory bowel disease, none have been confirmed as causal; rather, microbial antigens that are normally present in the intestinal lumen seem to drive inflammation in the gut. As compared with control subjects, patients with Crohn's disease and those with ulcerative colitis have depletion and reduced diversity of members of the mucosa-associated phyla Firmicutes and Bacteroidetes.21,27 Whether these alterations contribute to the disease or merely reflect secondary changes caused by the inflammation is not known. The Intestinal Epithelium The intestinal epithelium at the interface between the intestinal microbiome and the lymphoid tissue associated with the gastrointestinal system plays a critical role in shaping the mucosal immune response. Intestinal epithelial cells are a physical barrier against excessive entry of bacteria and other antigens from the intestinal lumen into the circulation. An intact mucosal barrier depends on intercellular junctions, which help to seal the space between adjacent epithelial cells (the paracellular space), and tight junctions, which are the key elements of the seal.28 In inflammatory bowel disease, the paracellular space has increased permeability, and the regulation of tight junctions is defective.28 These abnormalities may be due to a primary defect in barrier function or may be an outcome of inflammation.28,29,30,31,32 Additional defenses against bacterial invasion consist of specialized epithelial cells, including goblet cells and Paneth cells. Goblet cells regulate the production of mucus and factors that contribute to epithelial repair and regulation of inflammation.33,34 Paneth cells secrete antimicrobial peptides such as -defensins. Intestinal mucus overlies the epithelium, thereby limiting contact between bacteria and epithelial cells. Epithelial regeneration and repair serve to control and ultimately resolve the inflammatory response to injury. In inflammatory bowel disease, however, the inflammatory response often results in continued epithelial injury, which causes erosions, ulcerations, and a decrease in the production of defensin.35,36 The result is increased exposure to intestinal microbiota and amplification of the inflammatory response. In mouse models of inflammatory bowel disease, several types of epithelial dysfunction can cause intestinal inflammation. These include defects in epithelial-cell development or proliferation, barrier function, cell-matrix adhesion, endoplasmic reticulum stress, and epithelial restitution after injury.28,37,38 Prostaglandin E receptor 4 (EP4) contributes to mucosal repair and barrier function; in mice that are deficient in EP4, the colitis that develops in response to chemical injury is more severe than that in wild-type animals.39

Polymorphisms in proximity to the gene encoding EP4 (PTGER4) were recently implicated in Crohn's disease in humans6 (Table 1). In mice in which the gene encoding MUC2, a major mucin component, has been deleted, intestinal inflammation develops41; moreover, a variant in a genomic region that includes the gene encoding MUC19 has been associated with Crohn's disease.2 Endoplasmic reticulum stress (a cellular response to various environmental changes) is increased in inflamed intestinal epithelial cells,42 and deletion from murine intestinal epithelium of the XBP1 gene, which encodes X-box binding protein 1, a key component of the endoplasmic reticulum stress response, results in inflammation in the small intestine.38 The Inflammatory Response in Inflammatory Bowel Disease The intestinal lamina propria contains a complex population of immune cells that balance the requirement for immune tolerance of luminal microbiota with the need to defend against pathogens, the excessive entry of luminal microbiota, or both (Figure 1 and Figure 2A). The hallmark of active inflammatory bowel disease is a pronounced infiltration into the lamina propria of innate immune cells (neutrophils, macrophages, dendritic cells, and natural killer T cells) and adaptive immune cells (B cells and T cells). Increased numbers and activation of these cells in the intestinal mucosa elevate local levels of tumor necrosis factor (TNF- ), interleukin-1, interferon- , and cytokines of the interleukin-23Th17 pathway (Figure 2B). The initial immune response to intestinal microbiota is tightly regulated, and this regulation determines whether immune tolerance or a defensive inflammatory response ensues. Disturbance of the balance of these responses can lead to inflammatory bowel disease: in mouse models, perturbation of the proteins essential to immune function can incite intestinal inflammation.25 In experimental colitis, some intestinal lymphocytes respond to microbial antigens, but the extent to which specific intestinal microbial antigens drive intestinal lymphocytes in inflammatory bowel disease is unknown. Innate Immune Recognition The innate arm of the immune system provides an initial, rapid response to microbes. Cells of the innate system display receptors that recognize general microbial patterns (pattern-recognition receptors), in contrast to antigen-specific recognition by receptors of the adaptive immune system. The intestinal epithelial layer expresses various types of innate immune receptors (Figure 1) that mediate defenses against luminal microbiota but also condition epithelial and antigen-presenting cells for inducing the tolerance mechanisms that maintain immune homeostasis in the intestine.43,44,45,46 The expression of plasma-membrane toll-like receptors either intracellularly or basolaterally and the down-regulation of the expression and responses of pattern-recognition receptors limit activation of intestinal epithelial cells by luminal microbes.45,47,48,49,50 Continual sampling of intestinal microbiota is important in regulating the intestinal immune response. In laboratory animals, microbial sampling occurs by translocation of microbes across epithelial cells and the M cells of the epithelium of Peyer's patches, by immunoglobulins,51 and by dendritic cells52,53 (Figure 1). Activated antigen-presenting cells, notably dendritic cells, then present peptide antigens to T cells in secondary

lymphoid organs of the gut, such as Peyer's patches, mesenteric lymph nodes, and isolated lymphoid follicles54,55 (Figure 1). This interaction initiates an adaptive immune response, after which memory lymphocytes develop. One characteristic of adaptive immunity is a rapid and robust response to subsequent challenge by antigen that is, immunologic memory. CD4+ T Cells Certain helper T cells (Th1, Th2, and Th17) and regulatory T cells (e.g., forkhead box P3 [Foxp3+]Treg), which are subgroups of CD4+ T cells, secrete characteristic types of cytokines (Figure 1). Regulation of these subgroups must be continually fine-tuned to maintain intestinal immune homeostasis.56,57 Effector subgroups (Th1, Th2, and Th17 cells) are critical for defenses against pathogens and excessive entry of luminal microbiota (Figure 2A), but expansion and overactivity of these cells relative to the regulatory CD4+ T cells can lead to intestinal inflammation (Figure 2B).25,57,58,59,60,61,62,63,64 Studies of inflammatory bowel disease in mice and humans implicate dysregulation of intestinal CD4+ T-cell subgroups in the pathogenesis of these diseases. In Crohn's disease, for example, there is increased production in the intestinal mucosa of the Th17 cytokine interleukin-17 and the Th1 cytokines interferon- and TNF- .60,65 In ulcerative colitis, by contrast, there is usually an increase in interleukin-17 and Th2 cytokines.60,65,66 The interleukin-23 pathway is central to the function of Th17 cells. Polymorphic variants of multiple genes involved in this pathway and in Th17 cell function have been associated with both Crohn's disease and ulcerative colitis.2 B Cells The role of B cells in inflammatory bowel disease has not been as extensively studied as that of T cells. Intestinal B cells produce IgA antibodies, which contribute to immune protection without provoking inflammation. In animal models of colitis, both antiinflammatory67 and proinflammatory68 roles of B cells have been described. The presence of circulating antimicrobial antibodies in patients with inflammatory bowel disease (e.g., anti-flagellin antibodies and antiSaccharomyces cerevisiae antibodies)69 but not in healthy controls indicates B-cell reactivity. Intestinal Vasculature and Leukocyte Migration The intestinal vasculature and endothelium regulate the entry of leukocytes into the gut and maintain an adequate blood flow. Entry of cells into intestinal tissues is modulated by adhesion molecules (selectins, integrins) and chemokines (secreted cell attractants). T cells that become activated in mesenteric lymph nodes and Peyer's patches become "guttropic" cells by expressing the integrin 47 and the chemokine receptor CCR970,71; this change requires retinoic acid.72 The relative specificity in cellular migration to the intestine is the basis for targeting these molecules for the treatment of inflammatory bowel disease. An accumulation of leukocytes in intestinal tissues is characteristic of inflammatory bowel disease. Leukocyte adherence and recruitment are increased in the microvessels in chronic disease,73 mediated in part by up-regulation of adhesion molecules on vascular

endothelial cells by TNF- and interleukin-1. Moreover, increased levels of tissue-specific and inflammatory chemokines enhance leukocyte migration.74 Abnormalities in microvascular function probably contribute to inflammation, ischemia, and impaired mucosal healing.73 Ischemia causes local tissue hypoxia, which in turn regulates factors that contribute to both intestinal injury and protection.75,76

Innate-Immune-Response Genes and Crohn's Disease

NOD2 and Crohn's Disease The importance of responses to intestinal bacteria in inflammatory bowel disease is highlighted by the association between Crohn's disease and the NOD2 gene77 (Table 1), which encodes an intracellular sensor of peptidoglycan, a component in bacterial cell walls.78,79 The association includes three NOD2 polymorphisms that change the amino acids in NOD2, each impairing responses to peptidoglycan. These three polymorphisms occur with increased frequency in persons of European ancestry but are not present in Asian patients and are significantly less frequent in African Americans with Crohn's disease.14 Approximately 30% of patients of European ancestry have at least one of the three polymorphisms. NOD2 carriers are more likely than noncarriers to have ileal involvement and complications related to fibrostenosis and to require intestinal resection.80 Heterozygosity for a polymorphism confers an increased risk of Crohn's disease (by a factor of 1.75 to 4), whereas homozygosity confers a much greater risk (by a factor of 11 to 27)81; these are the highest relative risks observed for any of the genes associated with this disease. NOD2 polymorphisms alone, however, are not sufficient to cause Crohn's disease, which is indicative of the complexities of a multifactorial disorder. Epithelial cells, Paneth cells, macrophages, dendritic cells, and endothelial cells all express NOD2.77 The activation of the NOD2 protein by bacterial peptidoglycan activates the nuclear factor B (NF- B) and mitogen-activated protein (MAP) kinase signaling pathways, which leads to the production of cytokines (e.g., TNF and interleukin-1) and antimicrobial peptides.77,82 Decreased secretion of proinflammatory cytokines and decreased activation of NF- B on acute stimulation of NOD2 with bacterial peptidoglycan components have been detected in NOD2 carriers.77 Intestinal inflammation does not develop in NOD2-deficient mice, as is the case with most human NOD2 risk-allele carriers.82 Normally, secretion of proinflammatory cytokines by intestinal antigenpresenting cells is minimal,55,83 yet bacterial killing occurs, implying that the intestinal immune system can defend against luminal microbiota while minimizing tissue injury.83 In contrast, the gut in inflammatory bowel disease contains an increased number of antigen-presenting cells that secrete proinflammatory cytokines.84 Various factors within the intestinal environment contribute to the down-regulation of proinflammatory cytokines by intestinal antigenpresenting cells. These include inhibitory cytokines (transforming growth factor [TGF-] and interleukin-10) and chronic stimulation through pattern-recognition receptors, such as chronic peptidoglycan stimulation through NOD2.85,86 The NOD2-mediated mechanisms that down-regulate proinflammatory cytokines during chronic NOD2 stimulation are defective in carriers of genetic variants that impair NOD2 function.85 How loss-of-function NOD2 polymorphisms increase susceptibility to Crohn's disease is unknown, but most likely this outcome reflects the myriad functions of NOD2 and the unique features of the intestinal environment.

Autophagy Genes and Crohn's Disease Associations with Crohn's disease have been established for ATG16L1 and immunityrelated GTPase M protein (IRGM)4,7,8,10 (Table 1), two genes involved in autophagy. Autophagy is a mechanism for clearing intracellular components, including organelles, apoptotic bodies, and microbes.15 In mice with low expression of ATG16L1, the morphologic features and gene expression of Paneth cells are abnormal.87 ATG16L1 carriers with Crohn's disease also have abnormal Paneth-cell morphology.87 In mice, ATG16L1 appears to regulate secretion of interleukin-1 and inhibit intestinal inflammation.88

Adaptive Immunity and Inflammatory Bowel Disease

Alterations in T-Cell Tolerance The inhibitory cytokines interleukin-10 and TGF- in Peyer's patches, mesenteric lymph nodes, and lamina propria are involved in T-cell tolerance in the intestine.25 Regulatory T cells can differentiate in Peyer's patches and mesenteric lymph nodes through the actions of TGF- and retinoic acid.89 Defects in the development and function of regulatory T cells, or alterations in the ability to respond to them, can result in intestinal inflammation in mice.56,90 The autophagy pathway contributes to T-cell tolerance at multiple levels, which suggests that polymorphisms of autophagy genes associated with Crohn's disease could increase a patient's susceptibility to intestinal inflammation through defects in Tcell tolerance.15,91 Moreover, there is a genetic association between the inhibitory cytokine interleukin-10 and ulcerative colitis.11 This association corresponds with animal models showing that interleukin-10 participates in down-regulating intestinal inflammation; for example, colitis and intestinal dysplasia develop spontaneously in interleukin-10 deficient mice.25 The central importance and sufficiency of defective interleukin-10 signaling in the mediation of intestinal inflammation are further highlighted by the finding that uncommon, recessive loss-of-function mutations in either the IL10RA or the IL10RB component of the interleukin-10 receptor result in Crohn's disease.92 Th17 Cells and Interleukin-23 Signaling in Inflammatory Bowel Disease Interleukin-23 signaling is mediated through the engagement of heterodimeric interleukin-23 (comprising p19 and p40 subunits) with its heterodimeric receptor (comprising interleukin-23R and interleukin-12RB1). The engagement activates the JAKSTAT (Janus-associated kinasesignal transducers and activators of transcription) signaling pathway, which regulates the transcription of several genes. The importance of interleukin-23 signaling in mediating inflammation has been shown in animal models.17,59,61,64,65,93 Moreover, reports of highly significant genetic associations between IL23R and inflammatory bowel disease,3 psoriasis,94 and ankylosing spondylitis95 indicate that inflammatory bowel disease shares genetic associations with certain other autoimmune diseases. Interleukin-23, secreted by macrophages and dendritic cells, may contribute to TR17 proliferation, survival, or both.96 Interleukin-23 also contributes to intestinal inflammation through Th17-independent pathways.93 Levels of interleukin-23 and Th17 cytokines are elevated in the colonic mucosa in both Crohn's disease and ulcerative colitis.58,60,63

Multiple independent associations within the IL23R gene region have been found in inflammatory bowel disease, most notably for an Arg381Gln variant of the gene: carriers of this uncommon glutamine allele are less likely to have inflammatory bowel disease, by a factor of 2 to 3, than persons who carry only the common arginine allele.3 The effect of the IL23R polymorphisms on interleukin-23R function has not been defined. In addition to IL23R, associations with Crohn's disease have been observed in genomic regions encompassing multiple genes involved in interleukin-23Th17 signaling2 (Figure 2 and Table 1).

Comparative Gene Association Studies in Crohn's Disease and Ulcerative Colitis

It is estimated that known genetic associations account for only about 20% of the genetic variance underlying susceptibility to inflammatory bowel disease; the remaining genetic factors have not been identified. Consistent with predictions inferred from epidemiologic studies, genomewide association studies have identified loci associated solely with Crohn's disease or ulcerative colitis and other loci associated with both disorders (Table 1). For example, multiple genes along the interleukin-23 signaling pathway notably IL23R, JAK2, STAT3, and p40 have been associated with both these diseases.2,11,40,97 Genomic associations with Crohn's disease alone include NOD2 and the autophagy gene, ATG16L1,2 but a number of well-powered studies have not shown associations between these genes and ulcerative colitis. Given the broad functional contributions of autophagy, a more general role in either Crohn's disease or ulcerative colitis may eventually be established. Significant associations between ulcerative colitis and a region on chromosome 12q15 encompassing the interferon- and interleukin-26 genes have been observed, but no such associations with Crohn's disease have thus far been reported.40 In the initial genomewide association studies in ulcerative colitis, a highly significant association was observed for a common polymorphism in a region on chromosome 1q32 containing the IL10 gene, with only a very modest association observed in Crohn's disease.11 However, the intriguing finding that complete, loss-of-function mutations in the interleukin-10 receptor92 result in a Crohn's diseasepredominant phenotype highlights the complex phenotypic overlap between these two diseases. Associations within the major histocompatibility complex class II region near HLA-DRA (alpha chain) are the most significant observed in genomewide association studies of ulcerative colitis.11,40,97 Distinct HLA-DRB1 (beta-chain) alleles have been associated with both diseases.97

Therapeutic Implications
Treatment of inflammatory bowel disease includes lifestyle alterations (e.g., smoking cessation for patients with Crohn's disease), medical management, and surgical interventions. A seminal advance was the introduction of treatment with an antiTNFmonoclonal antibody, which is particularly effective in Crohn's disease. The efficacy of this therapy alone probably reflects the pleiotropic effects of TNF; however, the therapy is often limited by a loss of efficacy, underscoring the need for novel therapies.

Anti-p40 monoclonal antibodies have been reported to be effective in psoriasis98 and Crohn's disease.99,100 The p40 cytokine subunit is common to both interleukin-23 and interleukin-12, and monoclonal antibodies against p40 inhibit both pathways. Selective blockade of interleukin-23 can be achieved by targeting the p19 subunit, and this approach has been reported to be effective in many,59,64 although not all,101 animal models of inflammatory bowel disease. Selective inhibition of interleukin-23 may, however, deregulate other, cross-regulated pathways and T-cell subgroups, with unintended consequences.100 Moreover, some Th17 cytokines may also have protective features; for example, interleukin-22 ameliorates disease in an animal model of colitis.102 A major question that remains to be resolved is whether selective interleukin-23 blockade will be more or less effective than combined interleukin-12interleukin-23 blockade in the treatment of inflammatory bowel disease. Other treatments under investigation include the infusion of interleukin-10producing T cells and the administration of interleukin-10producing bacteria.103,104 Certain bacterial components, commensal bacteria, and "probiotic" bacteria more generally are also being investigated.48,105 The increased levels of tissue-specific and inflammatory chemokines that enhance intestinal leukocyte migration74 are the basis for targeting these molecules in inflammatory bowel disease.104,106 With the development of new, potent antiinflammatory agents, one must consider balancing therapeutic benefit with side effects resulting from an increased risk of infection or reactivation of infections (JC virusinduced multifocal leukoencephalopathy in the case of natalizumab107 and tuberculosis in the case of antiTNF- monoclonal antibody108). Future progress in disease monitoring and therapy will depend on the development of a more refined and integrated understanding of the mechanisms mediating intestinal immune homeostasis.

Inflammatory Bowel Disease

Daniel K. Podolsky, M.D.
(http://content.nejm.org/cgi/content/full/347/6/417 ( volume 347:417-429 2002 ) August 8,

In the decades since the major forms of idiopathic inflammatory bowel disease were defined on the basis of clinical manifestations, investigators have been challenged to identify the fundamental pathophysiologic processes underlying these enigmatic disorders, and clinicians have struggled to provide effective therapy for the often dismaying clinical manifestations. Clinical experience has led to the generally accepted notion that Crohn's disease and ulcerative colitis are distinct, if not discrete, entities. However, whether these are fundamentally different diseases or part of a mechanistic continuum remains an unanswered question, with both conceptual and practical management implications.

Etiology and Pathophysiology

Inflammatory bowel disease is thought to result from inappropriate and ongoing activation of the mucosal immune system driven by the presence of normal luminal flora. This aberrant response is most likely facilitated by defects in both the barrier function of the intestinal epithelium and the mucosal immune system. Genetic Factors Several clinical observations suggest that genetic factors contribute to susceptibility to inflammatory bowel disease (Table 1). These include wide variations in the incidence and prevalence of Crohn's disease and ulcerative colitis among different populations, cosegregation of inflammatory bowel disease in rare kindreds with a variety of uncommon genetic disorders, and more directly, familial aggregation of inflammatory bowel disease. Multiple studies have suggested that first-degree relatives of an affected patient have a risk of inflammatory bowel disease that is 4 to 20 times as high as that among the background population; the absolute risk of inflammatory bowel disease is

approximately 7 percent among first-degree family members.1,2 A substantially higher rate of disease concordance has been observed in monozygotic twins than in dizygotic twins, especially in those with Crohn's disease.1 Collectively, these findings lend compelling support to the inference that susceptibility is inherited and that the genetic contribution to the development of disease is more important in Crohn's disease than in ulcerative colitis. However, the absence of simple mendelian inheritance suggests that multiple gene products contribute to a person's risk of inflammatory bowel disease. Over the past 15 years, a wide variety of candidate genes have been studied. Most associations identified between specific candidate genes, including majorhistocompatibility-complex loci, and inflammatory bowel disease have not been reproducible, have not shed new light on pathogenesis, or have not facilitated diagnosis.3,4,5 However, substantial progress has now been made with use of the less biased approach of genome-wide screening with microsatellite DNA markers. Screening of DNA from members of kindreds with multiple affected members identified an area of apparent linkage on chromosome 16 among kindreds with Crohn's disease but not those with ulcerative colitis.6 Although the relative risk associated with this putative locus, designated IBD1, was small, the association was replicated in all but one independent study. These studies also provided evidence that inflammatory bowel disease is linked to several other genomic regions.7,8,9,10 Most of these sites are equally linked to both major forms of inflammatory bowel disease, suggesting that Crohn's disease and ulcerative colitis share many common genetic and, therefore, mechanistic features. Detailed mapping of chromosome 16 has recently resulted in the identification of a gene responsible, at least in part, for this linkage.11,12 This gene encodes a cytoplasmic protein designated NOD2 (also referred to as CARD 15 [caspase activation and recruitment domain]), which is expressed in macrophages and may serve as a so-called patternrecognition receptor for bacterial lipopolysaccharide, perhaps regulating nuclear factorB activation and macrophage apoptosis. European and North American patients with Crohn's disease, including those without a family history of inflammatory bowel disease, are more likely to have variants of NOD2 than are persons without Crohn's disease. Paradoxically, these NOD2 variants appear to result in reduced macrophage activation of nuclear factor- B in response to lipopolysaccharide. Persons who are homozygous for variant NOD2 may have a 20-fold or more increase in susceptibility to Crohn's disease, with a particular predilection for ileal disease.13,14,15 Heterozygotes are also at increased risk. However, fewer than 20 percent of patients with Crohn's disease are homozygous for these NOD2 variants. A putative locus associated with early-onset Crohn's disease appears to be present on chromosome 5 in the vicinity of genes encoding a variety of cytokine receptors.16 Environmental Precipitants and Disease Cofactors Whatever part genetic loci play in conferring susceptibility to inflammatory bowel disease, studies of identical twins (in which "only" 45 percent of pairs of identical twins are concordant for Crohn's disease) make it clear that the development of disease depends on additional factors. Among myriad factors studied, the most consistent are the use of

nonsteroidal antiinflammatory drugs, which can lead to disease flares, possibly related to an altered intestinal barrier, and early appendectomy, which is associated with a reduced incidence of ulcerative colitis.17,18 Smoking may modify the phenotype; it protects against ulcerative colitis but increases the risk of Crohn's disease.19,20 Accumulating evidence suggests that the luminal flora is a requisite and perhaps central factor in the development of inflammatory bowel disease. This inference is supported by studies in murine models of colitis established through genetic manipulation and reinforced by a variety of clinical observations in patients (Table 2). For example, the development of "spontaneous" colitis in rats and mice appears to require the presence of luminal flora; colitis does not occur in any of several mutant strains when they are maintained in a germ-free environment, but it develops rapidly when these mice are colonized by commensal bacteria.21,22 Studies of murine models resonate with clinical experience. Broad-spectrum antibiotics and probiotics have proven clinical efficacy in specific subgroups of patients. Studies have demonstrated the presence of an increased number of surface-adherent and intracellular bacteria in the colonic epithelium of patients with inflammatory bowel disease.23,24 Such observations underscore the importance of further defining the mechanisms of interaction between the normal mucosa and luminal microflora and their alteration in association with inflammatory bowel disease. Immune Response and Inflammatory Pathways The aggregate effect of genetic, environmental, and other processes is the sustained activation of mucosal immune responses (Figure 1). It remains unclear whether the immune system is activated as a result of an intrinsic defect (either constitutive activation or the failure of down-regulatory mechanisms) or because of continued stimulation resulting from a change in the epithelial mucosal barrier.25,26 Pathogenesis of Inflammatory Bowel Disease. Normal epithelium, with its highly evolved tight junctions and products of goblet-cell populations, most notably trefoil peptides and mucin glycoproteins, provides an effective barrier against luminal agents. The integrity of the barrier may be compromised by genetic variations in key molecular determinants, a diminished reparative response to injury, or exogenous agents, such as nonsteroidal antiinflammatory drugs. Chronic, recurrent intestinal inflammation appears to result from stimulation of the mucosal immune system by products of commensal bacteria in the lumen. Antigens from dietary sources may also contribute. Stimulation may occur as a result of the penetration of bacterial products through the mucosal barrier, leading to their direct interaction with immune cells, especially dendritic cells and lymphocyte populations, to promote a classic adaptive immune response. Alternatively, bacterial products may stimulate the surface epithelium, possibly through receptors that are components of the innate immuneresponse system; the epithelium can, in turn, produce cytokines and chemokines that recruit and activate mucosal immune cells. Activation of classic antigen-presenting cells,

such as dendritic cells, or direct stimulation through pattern-recognition receptors promotes the differentiation of type 1 helper T cells (Th1) in patients with Crohn's disease (shown here) or, possibly, atypical type 2 helper T cells in patients with ulcerative colitis. The stereotypical products of Th1 promote a self-sustaining cycle of activation with macrophages. In addition to producing the key cytokines that stimulate Th1 (interleukin-12, interleukin-18, and macrophage migration inhibitor factor), macrophages produce a mix of inflammatory cytokines, including interleukin-1, interleukin-6, and most notably tumor necrosis factor, which target a broad variety of other types of cells. The latter include endothelial cells, which then facilitate the recruitment of leukocytes to the mucosa from the vascular space, as well as fibroblasts and epithelium, modulating their functional properties. Most important, these functions may be altered either by genetically determined variants, as exemplified by germ-line mutations in the gene encoding NOD2, the product of the IBD1 locus, in some patients with Crohn's disease, or by environmental factors.

Substantial progress has been made in characterizing immune-cell populations and inflammatory mediators in patients with inflammatory bowel disease and murine models.22,27 There is reasonable consensus that the mucosa of patients with established Crohn's disease is dominated by CD4+ lymphocytes with a type 1 helper-T-cell (Th1) phenotype, characterized by the production of interferon- and interleukin-2. In contrast, the mucosa in patients with ulcerative colitis may be dominated by CD4+ lymphocytes with an atypical type 2 helper-T-cell (Th2) phenotype, characterized by the production of transforming growth factor (TGF- ) and interleukin-5 but not interleukin-4.28 In murine models, the effects of the activation of Th1 cells may be enhanced by the concomitant decrease in subgroups of suppressor T cells, variously designated Th3 or Tr1, which produce the down-regulatory cytokines interleukin-10 and TGF- .29

In-depth characterization of murine lines suggests that the stereotypical Th1 cytokines activate macrophages, which in turn, produce interleukin-12, interleukin-18, and macrophage migration inhibitor factor and thus further stimulate Th1 in a self-sustaining cycle. Just as important, activated macrophages produce a potent mix of broadly active inflammatory cytokines, including tumor necrosis factor (TNF), interleukin-1, and interleukin-6. The activation of central immune-cell populations is eventually accompanied by the production of a wide variety of nonspecific mediators of inflammation (Figure 2). These include many other cytokines, chemokines, and growth factors as well as metabolites of arachidonic acid (e.g., prostaglandins and leukotrienes) and reactive oxygen metabolites such as nitric oxide.24,27 These mediators enhance the inflammatory process itself and tissue destruction, which eventuate in the clinical manifestations of disease. Recruitment of additional leukocytes from the vascular space to sites of disease activity is especially important in maintaining inflammation and depends on the expression of adhesion molecules in the local microvasculature and counterligands on the various leukocyte populations.30,31

Treatment
Treatment must begin with accurate diagnosis. The diagnosis of inflammatory bowel disease depends on the aggregate constellation of the clinical history, physical findings, and endoscopic, radiologic, and histologic features, as well as the results of routine laboratory tests (Table 2). Typically, these features permit a firm diagnosis of inflammatory bowel disease and distinction between ulcerative colitis and Crohn's disease. However, in as many as 10 percent of patients with inflammatory bowel disease that is limited to the colon, it may not be possible to distinguish ulcerative colitis from Crohn's disease, at least initially; thus, these patients are considered to have indeterminate colitis. Serologic markers, including perinuclear-staining antineutrophil cytoplasmic antibodies, present in up to 70 percent of patients with ulcerative colitis, and anti Saccharomyces cerevisiae antibodies, present in 50 percent or more of those with Crohn's disease.32,33,34,35 Although these markers may help in the differential diagnosis of the few patients in whom the nature of colitis cannot be determined according to the usual criteria, they are not recommended for routine diagnosis. The long-term management of inflammatory bowel disease must be multidimensional and governed by the type of disease and sites involved (Table 3). Adjunctive agents, such as antidiarrheal agents, that are directed primarily at relieving symptoms rather than controlling inflammation itself can be important. Elemental diets have also been advocated as a primary treatment for Crohn's disease, but patients' poor compliance often limits the practicality of this approach.35,36 Proper nutrition and attention to related secondary complications such as osteoporosis resulting from malabsorption, corticosteroid therapy, or both are also important. Supplemental nutrition is especially important in the treatment of growth failure in children. Surgery can have an important role in management, though a full consideration is beyond the scope of this review. Notwithstanding the importance of other treatments, an expanding number and variety of drugs that target the inflammatory processes broadly or selectively are usually effective in

controlling active disease in most patients and in sustaining symptomatic remission for prolonged periods in many. In general, most clinicians use a stepped approach to therapy in which more potent agents are added to the regimen if less active drugs fail to achieve an adequate response. 5-Aminosalicylic Acid The 5-aminosalicylatebased compounds have remained mainstays of treatment for patients with mild to moderately active ulcerative colitis and Crohn's disease since the recognition of the therapeutic efficacy of the prototypical agent sulfasalazine (see Supplementary Appendix 1, available with the full text of this article at http://www.nejm.org). Early studies demonstrated that 5-aminosalicylate was the functionally active moiety of this sulfapyridine congener. 5-Aminosalicylate may act by blocking the production of prostaglandins and leukotrienes, inhibiting bacterial peptide induced neutrophil chemotaxis and adenosine-induced secretion, scavenging reactive oxygen metabolites, and perhaps inhibiting the activation of nuclear factor- B. Over the past 15 years, a variety of newer 5-aminosalicylatebased compounds have become available.37,38,39,40,41,42,43 In general, a 5-aminosalicylatebased agent should be selected principally on the basis of disease location. For patients with distal colonic disease, a suppository or enema form will be most appropriate. Oral compounds in which 5-aminosalicylic acid (mesalamine) is conjugated to prevent absorption by the small bowel are appropriate for colonic disease. These include sulfasalazine, olsalazine (a 5aminosalicylic acid dimer), and the recently approved balsalazide. Topical formulations may be used alone or in combination with oral formulations for patients with left-sided colonic disease. Oral formulations in which 5-aminosalicylic acid is in a slow-release or pH-dependent matrix can deliver therapeutic concentrations to the more proximal small bowel or distal ileum, respectively, to expand the spectrum of patients with Crohn's disease who may be treated with this class of agent. Maintenance treatment with a 5aminosalicylic acid can be effective for sustaining remission in ulcerative colitis but is of questionable value in Crohn's disease. Corticosteroids Corticosteroids have been commonly used when 5-aminosalicylatebased compounds are inadequate. They presumably act through the same functional properties relevant to other inflammatory processes, though which among these may be especially important in inflammatory bowel disease has not been determined. Topical corticosteroids (hydrocortisone enemas or foam) can be used as an alternative to 5-aminosalicylatebased compounds for patients with ulcerative proctitis or distal ulcerative colitis. Oral prednisone or prednisolone is used for moderately severe ulcerative colitis or Crohn's disease, in doses ranging up to 60 mg per day for patients with the most severe disease. Intravenous administration is warranted for patients who are sufficiently ill to require hospitalization; the majority of such patients, even those with fulminant ulcerative colitis, will have a response within 7 to 10 days. Optimal administration schedules for either oral or intravenous corticosteroids have not been defined. Corticosteroids, irrespective of the route of administration, should be continued only as

long as needed to control acute inflammatory activity, since they have no proven maintenance benefit in the treatment of either ulcerative colitis or Crohn's disease. Optimal tapering schedules have not been defined and are generally guided by clinical experience.44 Although corticosteroids are frequently, though not invariably, effective, these effects must be balanced against their many and often serious side effects. In general, side effects correlate with the dose and duration of treatment. Some side effects may be partially offset by careful management (e.g., early administration of supplemental calcium, vitamin D, and bisphosphonates to reduce corticosteroid-induced osteoporosis and careful monitoring of blood pressure and blood glucose for corticosteroid-induced hypertension and diabetes). For some patients, the side effects may be partially circumvented by the use of budesonide, which has recently become available in the United States in a controlled-ileal-release formulation; an enema form is also available elsewhere.45,46,47 However, concern persists that patients who take budesonide for prolonged periods remain at risk for systemic side effects; although first-pass metabolism may eliminate more than 90 percent of the drug, budesonide has an affinity for the glucocorticoid receptor that is 50 to 100 times that of prednisone. Moreover, its use is limited primarily to patients with distal ileal and right-sided colonic disease, and its efficacy somewhat less than that of conventional corticosteroids. Immunosuppressive and Immunoregulatory Agents There is an expanding role for immunomodulatory drugs in the care of patients with inflammatory bowel disease. These agents are generally appropriate for patients in whom the dose of corticosteroids cannot be tapered or discontinued. All these agents (as well as corticosteroids) put patients at risk for opportunistic infections. Azathioprine and its active metabolite mercaptopurine have been the most extensively used immunosuppressive agents, despite early hesitation arising from concern about an increased risk of lymphoma and other side effects. Indeed, this issue remains unresolved, in part owing to uncertainty about the possibility of a small, underlying increase in the risk of lymphoma in patients with inflammatory bowel disease.48,49,50 The actual mechanisms of action responsible for the therapeutic effect of these drugs remain largely unknown but may include suppressing the generation of a specific and long-lived subgroup of T cells, which might account for the prolonged time needed to achieve a therapeutic response. Clinical studies have shown that these agents are efficacious when they are given in adequate doses in patients with ulcerative colitis or Crohn's disease, allowing a gradual decrease in the corticosteroids and prolonging remission.51,52,53 The onset of benefit typically takes several weeks and may require up to six months, so that these drugs are not useful in the control of acute disease activity and, conversely, should be used only when prolonged therapy is planned. Although no firm guidelines for initiating the use of these agents have been established, many clinicians consider their use if it has not been possible to wean patients from corticosteroids after six weeks.

Toxic effects, some quite serious, frequently limit the use of azathioprine and mercaptopurine. Dose-related suppression of bone marrow is uniformly observed, and the white-cell count must be monitored closely when therapy is initiated and periodically thereafter (typically every two to three months). Considerable progress has been made in delineating the pathways of metabolism of mercaptopurine, and these findings have highlighted the importance of genetic variants of the key enzyme thiopurine Smethyltransferase in determining susceptibility to life-threatening bone marrow suppression.54,55,56 Approximately 0.1 percent of patients are homozygous for a functionally null allele that leads to the accumulation of toxic metabolites derived from alternative degradative pathways. Neither blood counts nor metabolite concentrations reliably predict other toxic effects, which include pancreatitis and drug-induced hepatotoxicity. In both instances, these initially asymptomatic effects are usually reversible with discontinuation of the medication and rapidly recur with greater severity if treatment is resumed. Methotrexate is effective in the treatment of corticosteroid-dependent active Crohn's disease and in maintaining remission.57,58,59 Typically, the drug is administered as a weekly injection, either intramuscular (15 mg per week) or subcutaneous (25 mg per week), and the response becomes evident over a period of several weeks. The mechanism of action of this agent also remains mostly uncertain. Aside from the generic risks of immunosuppression, methotrexate may lead to interstitial pneumonitis, which is first manifested by nonproductive cough and dyspnea, and there is a dose-related risk of hepatic fibrosis. Cyclosporine can be effective in the treatment of patients with severe ulcerative colitis who are hospitalized and otherwise in need of urgent proctocolectomy.60,61,62 This agent is presumed to work by blocking lymphocyte activation. High doses can achieve at least short-term control of disease sufficient to allow up to 80 percent of hospitalized patients who have been unresponsive to corticosteroids to be switched to a tapering oral regimen of cyclosporine. Cyclosporine is usually added after an initial 7-to-10-day trial of intravenous high-dose corticosteroids. A recent study has shown essentially equivalent response rates in patients given either cyclosporine or corticosteroids as initial intravenous therapy after hospitalization.62 Although, as noted above, short-term control of disease will result from the addition of cyclosporine in a substantial number of patients, approximately 50 percent will undergo proctocolectomy within a year as a result of either the inability to taper medications or recurrent disease activity. Azathioprine or mercaptopurine may help sustain the cyclosporine-induced response. In general, cyclosporine should not be used in the treatment of patients with Crohn's disease, with the possible exception of patients with symptomatic and severe perianal or cutaneous fistulas.63 Recent trials have suggested that the related compound tacrolimus, as well as mycophenolate mofetil, is also effective in the treatment of patients with inflammatory bowel disease, though direct comparative studies with cyclosporine have not been carried out. Anti-TNF Therapy The availability of the prototypical anti-TNF agent infliximab has offered an important advance in therapy for patients with Crohn's disease.64 Infliximab's mechanism of action

is incompletely understood. Its effectiveness suggests that TNF, a product of activated macrophages, may have a pivotal role among the many regulatory peptides with altered expression in association with inflammatory bowel disease.65,66 This chimeric monoclonal antibody, composed of a complement-fixing "human" IgG1 constant region and a murinederived antigen-binding variable region, binds soluble TNF; however, its action is thought to depend in part on its ability to bind precursor cell-surface TNF, perhaps leading to monocyte apoptosis. Two pivotal trials demonstrated the efficacy of infliximab in patients with Crohn's disease.67,68 In the first, approximately two thirds of patients with moderately active Crohn's disease who received a single infusion of infliximab had a significant reduction in their score on a standard Crohn's Disease Activity Index; of these, approximately half (one third of the total) achieved actual clinical remission. The response was generally quite prompt (usually within two weeks). However, the durability of the response ranged from a few weeks to six months or more. This pattern of response was mirrored in the second reported study in which patients with perianal and cutaneous fistulas received three infusions over a period of six weeks. Since this drug was approved for use in the United States, the experience in treating patients in routine practice has largely resembled that observed in controlled studies.69,70,71 Important questions remain to be addressed to define the most appropriate use of infliximab. Recently completed trials suggest the effectiveness of serial administration of the drug to maintain the initial response, but this indication is not yet approved.72 Furthermore, although responses have been reported in small numbers of patients with a variety of other intestinal inflammatory conditions (e.g., Behet's syndrome), the usefulness of infliximab in patients with ulcerative colitis remains uncertain. The optimal timing of infliximab administration and the value of maintenance therapy must be determined in rigorous studies, for several reasons. First, patients may not remain responsive to infliximab therapy indefinitely, and in some, the duration of response after an infusion becomes progressively shorter. Second, some patients who received infliximab during trials and who resumed treatment after a prolonged hiatus had a serum sicknesslike reaction, suggesting that the drug should not be used intermittently. Collectively, these two observations necessitate judicious use of this agent in patients with a lifelong condition. Third, the therapy is quite costly in terms of both the cost of the drug itself and the logistics of infusing it (the average third-party reimbursement per dose administered to a 70-kg patient is more than $5,200). Fourth, though it is generally safe, serious complications can ensue. In addition to occasional hypersensitivity and infusion reactions, a number of deaths have been reported as a result of tuberculosis or sepsis. Complications have included the reactivation of tuberculosis with atypical features; thus, all patients should be screened for tuberculosis before beginning therapy. Although a lupus-like syndrome has been reported infrequently after treatment with infliximab, the clinical significance of the more frequently detected antinuclear antibodies and antichimeric monoclonal antibodies is uncertain. Finally, lymphoma has developed in a small number of patients receiving infliximab, though a causal relation has not been established.

A number of other agents that target TNF are in advanced stages of clinical evaluation. These include CDP571, a humanized monoclonal antibody in which only the small number of residues necessary to confer antigen specificity remain from the original murine antibody. The overall usefulness of CDP571 in clinical trials involving patients with active Crohn's disease appears to be generally similar to that observed with infliximab.73,74 These results contrast with the very limited efficacy of etanercept, a fusion protein composed of the ligand-binding domain of the soluble receptor for TNF and an IgG common region, in patients with Crohn's disease.75 There has been recent interest in the use of thalidomide in the treatment of Crohn's disease, because of its presumptive action in blocking the production of TNF through the inhibition of intracellular pathways.76 Two small pilot trials have suggested therapeutic effectiveness.77,78 However, given the infamous history of this agent as a teratogen, its use requires rigorous supervision, including confirmation of adequate contraception in patients of childbearing age. Antibiotics and Probiotics Empirical clinical experience has led to the recognition that antibiotics are useful in the treatment of subgroups of patients with Crohn's disease.79 In contrast, antibiotics have very limited use in patients with ulcerative colitis, suggesting a differential role of the luminal flora in the two forms of inflammatory bowel disease.80 Metronidazole can be effective in the treatment of patients with Crohn's disease who have perianal fistulas. However, high doses are usually necessary (up to 750 mg three times daily), and the side effects may be limiting, most notably neurotoxicity. Despite its lack of effectiveness in the treatment of ulcerative colitis, metronidazole can control colonic (but not smallbowel) Crohn's disease. Ciprofloxacin and clarithromycin have been advocated as alternatives to metronidazole, but only small trials have been reported. The effectiveness of nonspecific antibiotics and experimental evidence of the central role of the luminal flora as an essential cofactor for the development of disease in the susceptible host have provided an impetus to the development of alternative strategies to manipulate the intestinal flora for therapeutic benefit. Probiotics, the administration of "healthy" bacteria, appears to be one promising approach.81 Patients with pouchitis or active Crohn's disease who were treated with a mixture of commensal bacteria had a positive therapeutic response; these findings should prompt increased efforts to define the value of this approach, which could be free of systemic side effects.82 Investigational Agents Advances in the understanding of the pathophysiology of inflammatory bowel disease, even if they are still incomplete, have led to great interest in the evaluation of a variety of new therapeutic agents with novel actions (Table 4). The explication of the cytokine network in inflammatory bowel disease and the seemingly central role of macrophageproduced interleukin-12, as described above, have led to ongoing trials with antibodies against interleukin-12. Interleukin-10, a cytokine that generally down-regulates the activation of Th1 cells, was found to have minimal activity in Crohn's disease.83,84,85 Preliminary studies of interleukin-11, a cytokine that is thought to act by enhancing the

epithelial barrier as well as through the inhibition of inflammatory cytokines, showed an apparent benefit. Larger trials of interleukin-11 in patients with ulcerative colitis and Crohn's disease are in progress.86 Strategies that block the recruitment of leukocytes to the sites of mucosal inflammation have produced promising results for example, a large-scale study of patients with Crohn's disease who were treated with a monoclonal antibody directed against the 4 integrin subunit. This integrin is part of a heterodimer that functions as a homing molecule for mucosal lymphocytes.87 Studies in both patients with ulcerative colitis and those with Crohn's disease are also under way after the demonstration, in pilot studies, that a chimeric antibody specific for the 4 7 integrin heterodimer reduced disease activity in patients with moderate-to-severe ulcerative colitis.88 Increasing attention is being directed at efforts to block the intracellular signaling pathways central to the activation of lymphocytes and macrophages with the use of p38 antagonists and agents that target the transcription factor nuclear factor- B. A few patients have already received a p38-blocking small molecule, with apparent benefit. However, most of these agents are in a relatively early stage of development,89 and the overall safety of disrupting ubiquitous signaling mechanisms remains to be determined. In addition, a variety of more conventional agents, including growth hormone,90 transdermal nicotine,91 fish oil,92 and rosiglitazone,93 have been found to have some effect in trials of variable sizes (Table 4). Finally, several patients with Crohn's disease had a prolonged remission after undergoing allogeneic bone marrow transplantation for unrelated reasons; this effect was presumably the result of reconstitution with nonpathogenic T cells.94 However, this approach cannot be considered established therapy.

Conclusions
Considerable progress has been made recently in both defining the mechanisms underlying the development of inflammatory bowel disease and expanding the spectrum of effective therapies. It is reasonable to expect that, as the spectrum of genes that confer susceptibility to inflammatory bowel disease is identified, attention will be focused on the mechanisms through which they lead to ulcerative colitis and Crohn's disease. A definitive understanding will require the delineation of the nature of the interactions with environmental factors, especially microflora. Progress in achieving these goals should lead to more precise genetic-based diagnosis, including presymptomatic risk assessment in family members of affected patients. In the interim, progress in our understanding of facets of the pathophysiology of inflammatory bowel disease has led to the development of useful new therapies.

Ingestion of Toxic Substances by Children

Michael Shannon, M.D., M.P.H.

http://content.nejm.org/cgi/content/full/342/3/186 Volume 342:186-191 2000

January 20,

Ingestion of a harmful substance is among the most common causes of injury to children less than six years of age. Fortunately, in most cases, the ingested agent has minimal or no clinically important toxic effect. Occasionally, however, such ingestion can be lifethreatening or can even result in death. Although preventive measures have been remarkably successful in reducing the frequency and severity of poisoning in children, poisoning continues to occur and requires strategies for treatment and prevention that are safe and effective. In this review, I discuss the management of poisoning in children, with an emphasis on current guidelines for treatment.

Epidemiology
According to the Toxic Exposure Surveillance System of the American Association of Poison Control Centers, 1.08 million instances of ingestion of a toxic substance by a child less than six years of age were reported to poison centers in 1998.1 However, the proportion of incidents reported to poison centers is estimated to be as low as 26 percent,2 suggesting a true incidence of more than 4 million poisonings annually. According to the Consumer Product Safety Commission, approximately 85,000 young children were treated for poisoning in emergency departments in the same year (Schroeder T: personal communication), for a projected incidence of 450 per 100,000 population. The toxic substances most commonly ingested by children are listed in Table 1. Substances that are most accessible to children, such as cosmetics and personal care products, cleaning products, analgesics, and cough and cold preparations, account for 58 percent of the products listed.1,3,4,5 lists the primary agents involved in deaths from poisoning among children from 1995 through 1998. Medications, both prescription drugs and over-the-counter drugs, were responsible for 52 percent of the deaths from poisoning. The distribution of responsible agents also reflects the common presence of such agents in the home rather than their inherent toxicity. On the basis of hazard-factor analysis (in which the number of episodes of major toxic effects or death is divided by the total number of reported cases of exposure to the substance, normalized to the rate of major toxic effects or death for all substances in the age group in question), the substances associated with the greatest risk of death to children include cocaine, anticonvulsant drugs, antidepressant drugs, and iron supplements

Treatment
The ability to reduce morbidity and mortality among children who have ingested a toxic substance depends on prompt, appropriate intervention. A telephone call by the parent to a physician or poison center is often the first step in obtaining treatment. In most circumstances, after the substance and its toxicity have been identified and the amount

ingested has been determined, treatment can be carried out at home without the need for referral to an emergency department. In children brought or referred to a health care facility, assessment and stabilization of vital signs are the initial steps of treatment. Rarely, emergency management involves the administration of an antidote (e.g., naloxone after an overdose of an opioid drug) or measures to enhance the elimination of toxins that have already been absorbed (e.g., multiple doses of activated charcoal after an overdose of carbamazepine). Laboratory analysis of serum or urine should be guided by the substance ingested, its anticipated degree of toxicity, and the value of measuring these concentrations; there is rarely a need for toxicologic screening tests in children, since the ingested substance is usually known.7 The term "gastrointestinal decontamination" includes interventions that are used to prevent the absorption of an ingested toxin. Gastrointestinal decontamination has three distinct components: gastric emptying, administration of an adsorbent agent, and catharsis. Gastric Emptying In principle, if the contents of the stomach can be rapidly and completely evacuated after the ingestion of a toxin, the toxin will not have the opportunity to enter the small intestine, the main site of absorption. Consequently, the clinical effects of the poisoning can be mitigated. Gastric evacuation can be accomplished either by chemically induced emesis or by mechanical removal with a lavage tube. Emesis is mediated in the central nervous system, either by direct stimulation of the chemoreceptor trigger zone in the area postrema or by gastric irritation mediated by vagal afferent fibers, with resulting stimulation of the vomiting center in the medulla.8,9 Agents that have been used in children in the past for this purpose include parenteral apomorphine, copper sulfate, fluid extract of ipecac, hydrogen peroxide, and ipecac syrup. Among these, ipecac syrup has proved to be the safest, most reliable emetic agent; the others are contraindicated for the treatment of poisoning. Ipecac syrup contains two active alkaloids, cephaline and emetine. Both stimulate gastric sensory receptors linked to the vomiting center in the brain. In addition, cephaline acts at the chemoreceptor trigger zone. The primary advantage of ipecac syrup is its ease of administration by parents or caretakers. Emesis usually begins about 20 minutes after ipecac syrup is given.10 Eighty-two percent of children will vomit after a single 15-ml dose, and 99 percent will vomit after two doses.11 The duration of vomiting is typically one to two hours. Guidelines for doses are presented in Table 3. Administration can be repeated if vomiting does not begin within 20 to 30 minutes. The administration of additional fluid after ipecac syrup does not appear to improve the efficacy of the agent.12 The use of ipecac syrup should be considered in cases in which children have ingested a potentially toxic substance in the preceding hour. However, the efficacy of ipecac syrup, even under these circumstances, has not been proved.10 Because uncontrolled vomiting will occur for at least one to two hours, ipecac syrup should not be administered after the ingestion of certain substances or under certain clinical conditions

Increasingly, both the safety and the efficacy of ipecac syrup have been questioned.10 Although the drug has an impressive safety profile, there have been reports of adverse effects, including prolonged vomiting, sedation, MalloryWeiss syndrome, gastric rupture, and fatal aspiration.13,14 The efficacy of ipecac syrup, measured in terms of the elimination of ingested toxins or improvement in outcomes, also appears to be limited. In studies in animals, ipecac syrup removes 10 to 60 percent of an ingested substance; in clinical studies, within one hour after ingestion, a mean of 30 percent of a toxin can be recovered.10 When administered 90 minutes or more after a toxic substance has been ingested, ipecac syrup has no identifiable benefit.15 Saincher et al. have suggested that ipecac syrup is no longer beneficial between 5 and 30 minutes after the ingestion of a toxic substance.16 Gastric lavage is an alternative method of removing the contents of the stomach. For lavage to be performed properly, competence of the gag reflex should first be confirmed. Once properly restrained, the child should be placed in a left lateral decubitus Trendelenburg's position in order to limit the movement of the gastric contents into the duodenum and minimize the risk of aspiration.17,18 A large-bore (24-to-32-French), singlelumen tube should be placed by an orogastric route. The proper placement of the tube is confirmed by the spontaneous or aspirated return of gastric contents or by auscultation of insufflated air when a stethoscope is placed over the stomach. After placement of the tube, room-temperature aliquots of 10 to 15 ml of saline per kilogram of body weight are instilled through the tube and then aspirated. This process continues until the aspirated contents are clear. Volumes as large as several liters may be necessary to produce a clear aspirate. As compared with induced emesis, gastric lavage has several advantages with respect to the treatment of poisoning. It can be performed promptly and completely, whereas emesis induced by ipecac syrup is associated with a delay in the onset of vomiting and a prolonged effect. The lavage tube also makes available a means of promptly administering an adsorbent to complete the decontamination process. However, the safety and efficacy of lavage have also been challenged. Although there have been conflicting reports about whether lavage is superior19,20 or inferior21 to emesis, recent data indicate that the two methods of gastric decontamination have similar efficacy. When performed one hour after the ingestion of a toxic substance, lavage retrieves less than 30 percent of the toxin.17 If not properly performed, gastric lavage has the potential complication of propelling toxins into the duodenum, thereby increasing the likelihood that the toxin will be absorbed.21 The greatest risk associated with gastric lavage is the inadvertent placement of the tube into the trachea or a mainstem bronchus.18,22 Other potential complications in children are esophageal injury, hypothermia, hyponatremia, and water intoxication.17,18 Gastric lavage is contraindicated if protective airway reflexes are absent or depressed and if a low-viscosity hydrocarbon or a corrosive agent has been ingested. Administration of Adsorbent Adsorbents bind toxins, reducing the amount of free agent available for absorption into the gastrointestinal mucosa. There are several adsorptive agents that are useful in the

treatment of poisoning: fuller's earth (for paraquat), potassium ferrocyanate (for thallium), milk (for fluoride), sodium polystyrene sulfonate (for lithium), and cholestyramine (for lindane). However, activated charcoal is the most broadly effective adsorbent available. The adsorptive capacity of activated charcoal is a function of its binding surface area, which ranges from 1000 to 2000 m2 per gram.23 The types of activated charcoal that have binding surface areas as large as 3000 m2 per gram bind a greater proportion of the toxin than those with smaller binding surface areas.24 Activated charcoal maintains its attachment to toxins through covalent binding and van der Waals forces.25 It does not appear that clinically significant desorption of toxins from charcoal occurs. The most frequently encountered substances for which adsorption to activated charcoal is clinically negligible are alcohols, hydrocarbons, metals, and minerals. On the other hand, multiple doses of activated charcoal are effective in enhancing the elimination of certain toxins that have already been absorbed, such as theophylline, phenobarbital, and carbamazepine.26 When administered within one hour after ingestion, activated charcoal can reduce the absorption of toxins by up to 75 percent.27 Optimal adsorption occurs when the ratio of charcoal to toxin is 10:1 or higher.17,28 However, a fixed dose of 1 g per kilogram is recommended (Table 3). Activated charcoal is administered as a slurry. Additives, such as chocolate or fruit syrup, make charcoal more palatable without reducing its efficacy.29 Because less than half of young children will voluntarily drink activated charcoal quickly enough for it to work optimally, placement of a nasogastric tube may be necessary for its prompt administration. The main hazards associated with the administration of activated charcoal are vomiting and aspiration, which can result in pneumothorax or empyema. Vomiting occurs in approximately 15 percent of patients to whom activated charcoal alone is administered.27 Although activated charcoal is often described as inert, data from experimental studies indicate that aspirated charcoal can produce pulmonary parenchymal injury or bronchiolitis obliterans.30,31 The instillation of charcoal into the lungs through the inadvertent placement of an orogastric or nasogastric tube into the trachea has had disastrous results, including death.3,32 Catharsis Cathartic agents are administered after poisoning has occurred, to increase gastrointestinal motility and hasten the expulsion of the toxin or the toxinadsorbent complex. There are three classes of cathartic drugs stimulants, osmotic agents, and bulk-forming agents but only osmotic agents (e.g., magnesium citrate and sorbitol) are used in cases of ingestion of toxic substances. Osmotic agents promote the retention of colonic fluid and activate constitutive, calcium-dependent nitric oxide synthase, which stimulates gastrointestinal motility through the release of cholecystokinins and alters gastrointestinal pH.33,34,35 Magnesium citrate in a 6 percent suspension is given in a dose of 4 ml per kilogram (Table 3); larger doses do not produce more rapid results.36 Doses of more than 0.5 g of sorbitol per kilogram will reliably produce diarrhea37; the recommended dose of sorbitol is 1 to 2 g per kilogram (Table 3). Sorbitol is not recommended for use in children less than one year of age.17 Ingestion of anticholinergic drugs does not lessen the cathartic efficacy of osmotic agents.38,39

Although generally safe, the use of cathartic agents in children has occasionally been associated with adverse effects. Administration of sorbitol to infants has resulted in hypernatremic dehydration and cardiovascular collapse.40,41 Cathartic agents also promote emesis. Magnesium-based cathartic agents have the potential to produce hypermagnesemia if given repeatedly or if given to children with renal disease.42 Whole-bowel irrigation has recently emerged as a means of gastrointestinal decontamination.43 In this procedure, large volumes of solution are administered enterally until the rectal effluent is clear. Polyethylene glycolelectrolyte lavage solution has been formulated to prevent extensive absorption or secretion of fluid across the gastrointestinal mucosa. Whole-bowel irrigation is safe in children; volumes as large as 44 liters have been administered without ill effects.44 Typical rates of administration are 500 to 1000 ml per hour, orally or by nasogastric tube (Table 3). The adverse effects associated with this procedure consist of vomiting, abdominal cramps, and bloating. The place of whole-bowel irrigation in the treatment of poisoning in children has not been well established. Data suggest that the most important role of whole-bowel irrigation is as an intervention for the removal of substances that are poorly adsorbed to activated charcoal for example, iron and lithium.45 Decisions about Management There are few data from experimental studies and no rigorous clinical investigations in children to indicate that any means of decontamination produces an important clinical benefit. However, substantial indirect evidence supports the continued use of decontamination. For example, the clinical efficacy of activated charcoal was first demonstrated in 1831, when Touery drank an ordinarily lethal dose of strychnine along with charcoal and suffered no ill effects.46 In cases of ingestion in which the anticipated outcome is mild toxic effects or none, gastrointestinal decontamination should not be performed, since its risks may outweigh any benefit. Similarly, if the interval between ingestion and intervention is so long that it is unlikely that the toxin is still in the gastrointestinal tract, no intervention is warranted. The decision to begin decontamination should be based on clinical need, estimated according to the highest amount of toxin that may have been ingested; decontamination should not be considered a requirement for every ingestion.10 In cases in which it is undertaken, decontamination should begin as soon as possible. If the child is outside a health care facility, the clinician should determine whether ipecac syrup should be given in the home or whether the child should be referred directly to a health care facility. Usually, this decision is best made in consultation with a toxicologist or regional poison center. In cases in which serious toxicity is possible, decontamination can be initiated at home with ipecac syrup if there are no contraindications to its use. The decision about whether to refer a child to a treatment facility is based on the need for clinical assessment, additional decontamination, or close monitoring. Rigorously performed studies have not found any substantial value associated with gastric emptying in the emergency department for the management of poisoning.10,47 In large,

prospective, randomized studies, the clinical outcomes of patients who underwent gastric evacuation before receiving activated charcoal were no better than the outcomes of those who received activated charcoal only.48,49,50 Gastric emptying may, in fact, increase the risk of adverse outcomes after decontamination.49,51 For children given ipecac syrup in the emergency department, as compared with children who do not undergo gastric emptying, the successful administration of activated charcoal is delayed and the stay in the emergency department is 20 percent longer.52 Administration of ipecac syrup in the emergency department is therefore not recommended.53,54 Gastric lavage has a similar lack of benefit and is not routinely recommended.17 In general, activated charcoal is the sole intervention needed to treat serious poisonings. A slurry consisting of activated charcoal and a flavoring agent should be given to the child. If it has not been swallowed within 20 minutes after ingestion of the toxin, activated charcoal should be administered through a nasogastric tube by trained personnel who are able to identify and treat any complications of the procedure. Administration of a cathartic agent as the sole intervention for gastrointestinal decontamination is ineffective. Moreover, the combination of cathartic drugs and activated charcoal does not provide a better clinical outcome than activated charcoal alone after a poisoning.33,38 Given their potential risks, cathartic agents should not be used routinely in the treatment of poisoning in children.38,54 These agents remain potentially valuable in the treatment of poisonings that require multiple doses of activated charcoal, because they may prevent inspissation of the charcoal. Although whole-bowel irrigation has few proved indications, its primary role is in the treatment of poisoning with iron supplements, modified-release pharmaceutical drugs, or illicit drugs (e.g., cocaine or heroin).55,56 New approaches to the treatment of poisoning in children have begun to focus on identifying the smallest dose of activated charcoal needed for effective decontamination. The availability of activated charcoal with high adsorbency, which theoretically permits the administration of volumes of charcoal that are small enough to eliminate the need for a nasogastric tube, may allow this antidote to be kept in the home for administration by parents or caretakers.

Prevention
The reduction in the incidence of childhood poisonings in the past half-century has been dramatic. This reduction is largely the result of the combination of highly effective active and passive methods of intervention.57 Important passive interventions have included the federal regulation of products and product safety and the introduction of child-resistant containers for drugs and other dangerous household products. Child-resistant containers have been particularly effective in reducing the incidence of death from the ingestion of prescription drugs by children.58 Active interventions, which require a change in behavior by parents and caretakers, have included the safe storage of household products.57 Finally, poison centers, which were established nearly 50 years ago, will continue to have a vital role in the management of poisoning in children by effectively reducing unneeded visits to the emergency department59 and providing education about poisoning to the public.

Typhoid Fever
Christopher M. Parry, M.B., Tran Tinh Hien, M.D., Gordon Dougan, Ph.D., Nicholas J. White, M.D., D.Sc., and Jeremy J. Farrar, M.B., D.Phil.
http://content.nejm.org/cgi/content/full/347/22/1770 Volume 347:1770-1782 November 28, 2002 Typhoid fever is a systemic infection with the bacterium Salmonella enterica serotype typhi. This highly adapted, human-specific pathogen has evolved remarkable mechanisms for persistence in its host that help to ensure its survival and transmission. Typhoid fever was an important cause of illness and death in the overcrowded and unsanitary urban conditions of the United States and Europe in the 19th century.1 The provision of clean water and good sewage systems led to a dramatic decrease in the incidence of typhoid in these regions. Today most of the burden of the disease occurs in the developing world, where sanitary conditions remain poor. Reliable data from which to estimate the burden of disease in these areas are difficult to obtain, since many hospitals lack facilities for blood culture and up to 90 percent of patients with typhoid are treated as outpatients. Community-based studies have consistently shown higher levels of typhoid than public health figures suggest. Annual incidence rates of 198 per 100,000 in the Mekong Delta region of Vietnam2 and 980 per 100,000 in Delhi, India,3 have recently been reported. According to the best global estimates, there are at least 16 million new cases of typhoid fever each year, with 600,000 deaths.4 The introduction of chloramphenicol for the treatment of typhoid fever in 1948 transformed a severe, debilitating, and often fatal disease into a readily treatable condition.5 The emergence of resistance to chloramphenicol and other antimicrobial agents has been a major setback.6 We now face the very real prospect that untreatable typhoid fever will reemerge. Typhoid is usually contracted by ingestion of food or water contaminated by fecal or urinary carriers excreting S. enterica serotype typhi. It is a sporadic disease in developed countries that occurs mainly in returning travelers, with occasional point-source epidemics.7 In endemic areas, identified risk factors for disease include eating food prepared outside the home, such as ice cream or flavored iced drinks from street vendors,8,9 drinking contaminated water,10 having a close contact or relative with recent typhoid fever,8,11 poor housing with inadequate facilities for personal hygiene,12 and recent use of antimicrobial drugs.9

The Bacterium
S. enterica serotype typhi is a member of the family Enterobacteriaciae. The bacterium is serologically positive for lipopolysaccharide antigens O9 and O12, protein flagellar antigen Hd, and polysaccharide capsular antigen Vi. The Vi capsular antigen is largely restricted to S. enterica serotype typhi, although it is shared by some strains of S. enterica serotypes hirschfeldii (paratyphi C) and dublin, and Citrobacter freundii. A unique flagella type, Hj, is present in some S. enterica serotype typhi isolates from Indonesia.13

Phage typing, pulse-field gel electrophoresis, and ribotyping have shown that areas of endemic disease usually have many strains in circulation but that outbreaks are usually due to a restricted number of strains.14,15,16 The Genome Recently, the complete genome sequence was determined for a multidrug-resistant strain of S. enterica serotype typhi (CT18), which was isolated in 1993 from a child with typhoid fever in the Mekong Delta region of Vietnam.17 The CT18 genome harbors 4,809,037 base pairs with an estimated 4599 coding sequences. The genomes of S. enterica serotype typhi CT18, S. enterica serotype typhimurium LT2,18 and Escherichia coli19 are essentially collinear, despite the fact that E. coli and S. enterica diverged about 100 million years ago. Similar environmental requirements for these enteric bacteria presumably explain this conservation of gene order. Gene clusters unique to particular bacteria are likely to represent adaptations to particular environments or may contribute to pathogenicity. Unlike E. coli, S. enterica serotype typhi has several large insertions in its genome, termed salmonella pathogenicity islands, that are thought to be recent horizontal acquisitions and that encode genes important for survival in the host. In addition, there are multiple insertions of many smaller gene blocks and individual genes scattered in the genome that may potentially be involved in pathogenicity A striking feature of the S. enterica serotype typhi genome is the presence of 204 pseudogenes, more than half of which are inactivated by the introduction of a single frame-shift or stop codon, suggesting that they are of recent origin. A substantial number are predicted to be involved in housekeeping functions or in virulence or host interactions. This apparent inactivation of genes responsible for host interactions may explain why S. enterica serotype typhi, unlike other salmonella serotypes, is restricted to one host (i.e., humans) and suggests that S. enterica serotype typhi may have passed through a recent evolutionary bottleneck. S. enterica serotype typhi CT18 harbors two plasmids. The larger conjugative plasmid, pHCM1, is 218 kb in length and shares approximately 168 kb of DNA with the plasmid R27, with more than 99 percent sequence identity.20 R27 is an incH1 plasmid, first isolated in the 1960s from S. enterica, that is closely related to the chloramphenicolresistance plasmids detected in S. enterica serotype typhi in the 1970s.21 The pHCM1 plasmid encodes resistance to chloramphenicol (catI ), ampicillin (TEM-1, bla), trimethoprim (dhfr1b), sulfonamides (sul II), and streptomycin (strAB). The smaller plasmid, pHCM2, is 106.5 kb in length and is phenotypically cryptic, but it has striking homology with the pMT1 virulence-associated plasmid of Yersinia pestis. Pathogenesis The infectious dose of S. enterica serotype typhi in volunteers varies between 1000 and 1 million organisms.22 Vi-negative strains of S. enterica serotype typhi are less infectious and less virulent than Vi-positive strains. S. enterica serotype typhi must survive the gastric acid barrier to reach the small intestine, and a low gastric pH is an important defense mechanism. Achlorhydria as a result of aging, previous gastrectomy, or treatment with histamine H2-receptor antagonists, proton-pump inhibitors, or large amounts of

antacids lowers the infective dose. In the small intestine, the bacteria adhere to mucosal cells and then invade the mucosa. The M cells, specialized epithelial cells overlying Peyer's patches, are probably the site of the internalization of S. enterica serotype typhi and its transport to the underlying lymphoid tissue. After penetration, the invading microorganisms translocate to the intestinal lymphoid follicles and the draining mesenteric lymph nodes, and some pass on to the reticuloendothelial cells of the liver and spleen. Salmonella organisms are able to survive and multiply within the mononuclear phagocytic cells of the lymphoid follicles, liver, and spleen.23 At a critical point that is probably determined by the number of bacteria, their virulence, and the host response, bacteria are released from this sequestered intracellular habitat into the bloodstream. The incubation period is usually 7 to 14 days. In the bacteremic phase, the organism is widely disseminated. The most common sites of secondary infection are the liver, spleen, bone marrow, gallbladder, and Peyer's patches of the terminal ileum. Gallbladder invasion occurs either directly from the blood or by retrograde spread from the bile. Organisms excreted in the bile either reinvade the intestinal wall or are excreted in the feces. Counts of bacteria in patients with acute typhoid fever indicate a median concentration of 1 bacterium per milliliter of blood (about 66 percent of which are inside phagocytic cells) and about 10 bacteria per milliliter of bone marrow.24,25,26 Even though S. enterica serotype typhi produces a potent endotoxin, mortality from treated typhoid fever for patients at this stage is less than 1 percent. Studies have shown increased levels of circulating proinflammatory and antiinflammatory cytokines in patients with typhoid and a reduced capacity of whole blood to produce inflammatory cytokines in patients with severe disease.27,28,29 Typhoid induces systemic and local humoral and cellular immune responses, but these confer incomplete protection against relapse and reinfection. The interaction of host immunologic mediators and bacterial factors in infected tissue may contribute to the necrosis of Peyer's patches in severe disease.30 The evidence for an association between typhoid and infection with the human immunodeficiency virus (HIV) is conflicting,31,32 whereas there is a large increase in the incidence of non-typhi salmonella bacteremia in HIV infection. Major-histocompatibility-complex class II and class III alleles have been shown to be associated with typhoid fever in Vietnam. HLA-DRB1*0301/6/8, HLADQB1*0201-3, and TNFA*2(308) were found to be associated with susceptibility to typhoid fever, whereas HLA-DRB1*04, HLA-DQB1*0401/2, and TNFA*1(308) were associated with disease resistance.33 Polymorphisms in the genes encoding the naturalresistanceassociated macrophage protein were not associated with resistance to typhoid, in contrast to the importance of this allele in the murine model.34 Antimicrobial Resistance In 1948 chloramphenicol became the standard antibiotic for treating typhoid.5 Although resistance emerged within two years after its introduction, it was not until 1972 that chloramphenicol-resistant typhoid fever became a major problem.6 Outbreaks occurred in Mexico, India, Vietnam, Thailand, Korea, and Peru.6 Chloramphenicol resistance was associated with high-molecular-weight, self-transferable, IncHI plasmids. These S. enterica serotype typhi strains were also resistant to sulfonamides, tetracycline, and

streptomycin, but initially amoxicillin and trimethoprimsulfamethoxazole remained effective alternative drugs. Toward the end of the 1980s and the 1990s, S. enterica serotype typhi developed resistance simultaneously to all the drugs that were then used as first-line treatment (chloramphenicol, trimethoprim, sulfamethoxazole, and ampicillin).6 Outbreaks of infections with these strains occurred in India,35,36 Pakistan,37,38 Bangladesh,39 Vietnam,40 the Middle East,41 and Africa42 (Figure 2). These multidrugresistant strains also carried the 100,000-to-120,000-kD IncHI plasmids that encoded the resistance genes. Spread results from the clonal dissemination of individual multidrugresistant S. enterica serotype typhi strains or from transfer of the plasmid to multiple S. enterica serotype typhi strains.14,15,16 Resistance rarely emerges during the course of treatment.43 Multidrug-resistant S. enterica serotype typhi are still common in many areas of Asia, although in some areas strains that are fully susceptible to all first-line antibiotics have reemerged.44 There have been sporadic reports of high-level resistance to ceftriaxone (minimal inhibitory concentration [MIC], 64 mg per liter) in S. enterica serotype typhi and S. enterica serotype paratyphi A,45,46 although these strains are very rare. S. enterica serotype typhi strains with reduced susceptibility to fluoroquinolones have become a major problem in Asia.47,48,49,50 An outbreak of typhoid with such strains in Tajikistan in 1997 sickened 8000 people in a six-month period and caused 150 deaths.10 Although they were reported to be susceptible to fluoroquinolones, by disk testing with the use of recommended break points, these organisms were resistant to nalidixic acid and the MIC of fluoroquinolones for these strains was 10 times that for fully susceptible strains. This reduction in susceptibility results in a poor clinical response to treatment.48,49 Quinolone resistance is frequently mediated by single point mutations in the quinolone-resistance determining region of the gyrA gene, characteristically occurring at position 83 of the DNA gyrase enzyme (changing serine to phenylalanine) and position 87 (changing aspartate to tyrosine or glycine).47,48 Quinolones, such as nalidixic acid, are a group of synthetic compounds based on the 4quinolone nucleus. The introduction of fluorine at position 6 of the nucleus creates the fluoroquinolone group of compounds, which have substantially greater antimicrobial activity. In other Enterobacteriaciae, higher levels of quinolone resistance have been associated with additional mutations in the gyrA gene, mutations in other topoisomerase genes, or alterations in fluoroquinolone uptake. No such mutations have been reported yet in S. enterica serotype typhi, although there are sporadic reports of fully fluoroquinoloneresistant isolates.51 Because the clinical response to fluoroquinolones in patients infected with nalidixic acidresistant strains is greatly inferior to the response in those infected with nalidixic acidsusceptible strains, we believe that the break points for the classification of S. enterica serotype typhi strains according to their susceptibility to fluoroquinolones should be changed.50 A pragmatic solution would be to classify strains that are resistant to nalidixic acid but susceptible to fluoroquinolones according to current disk-testing criteria as resistant to quinolones or nonsusceptible to fluoroquinolones. All strains that have intermediate susceptibility or resistance to fluoroquinolones on disk testing (as defined by national guidelines) should be considered fluoroquinolone-resistant.

Clinical Features

The clinical manifestations and severity of typhoid fever vary with the patient population studied. Most patients who present to hospitals with typhoid fever are children or young adults from 5 to 25 years of age.1,52,53 However, community-based studies in areas of endemic disease indicate that many patients with typhoid, particularly children under five years of age, may have a nonspecific illness that is not recognized clinically as typhoid.2,3,54 Between 60 and 90 percent of people with typhoid do not receive medical attention or are treated as outpatients.2,3 After a person ingests S. enterica serotype typhi, an asymptomatic period follows that usually lasts 7 to 14 days (range, 3 to 60). The onset of bacteremia is marked by fever and malaise. Patients typically present to the hospital toward the end of the first week after the onset of symptoms with fever, influenza-like symptoms with chills (although rigors are rare), a dull frontal headache, malaise, anorexia, nausea, poorly localized abdominal discomfort, a dry cough, and myalgia, but with few physical signs.1,37,52,53,55 A coated tongue, tender abdomen, hepatomegaly, and splenomegaly are common. A relative bradycardia is considered common in typhoid, although in many geographic areas this has not been a consistent feature. Adults often have constipation, but in young children and in adults with HIV infection, diarrhea is more common.31,56 It is unusual for a patient hospitalized with typhoid to have no abdominal symptoms and normal bowel movements. Initially the fever is low grade, but it rises progressively, and by the second week it is often high and sustained (39 to 40C). A few rose spots, blanching erythematous maculopapular lesions approximately 2 to 4 mm in diameter, are reported in 5 to 30 percent of cases. They usually occur on the abdomen and chest and more rarely on the back, arms, and legs. These lesions are easily missed in dark-skinned patients. There may be a history of intermittent confusion, and many patients have a characteristic apathetic affect. Convulsions may occur in children under five years of age.55 The hemoglobin level, white-cell count, and platelet count are usually normal or reduced. Disseminated intravascular coagulation may be revealed by laboratory tests, but it is very rarely of clinical significance. The levels of liver enzymes are usually two to three times the upper limit of normal. Complications occur in 10 to 15 percent of patients and are particularly likely in patients who have been ill for more than two weeks. Many complications have been described (Table 1), of which gastrointestinal bleeding, intestinal perforation, and typhoid encephalopathy are the most important. Gastrointestinal bleeding is the most common, occurring in up to 10 percent of patients. It results from erosion of a necrotic Peyer's patch through the wall of an enteric vessel. In the majority of cases, the bleeding is slight and resolves without the need for blood transfusion, but in 2 percent of cases, bleeding is clinically significant and can be rapidly fatal if a large vessel is involved. Intestinal (usually ileal) perforation is the most serious complication, occurring in 1 to 3 percent of hospitalized patients.57,58 Perforation may be manifested by an acute abdomen or, more covertly, by simple worsening of abdominal pain, rising pulse, and falling blood pressure in an already sick patient. A reduced level of consciousness or encephalopathy, often accompanied by shock, is associated with high mortality.59,60,61 The patient is commonly apathetic although rousable. Patients can be severely agitated, delirious, or obtunded, but complete stupor or coma is infrequent. The incidence of these neuropsychiatric presentations varies among countries. It ranges from 10 to 40 percent among hospitalized

patients with typhoid in Indonesia59,60 and Papua New Guinea61 but is less than 2 percent in Pakistan37 and Vietnam.40 This geographic variation is unexplained. Typhoid fever during pregnancy may be complicated by miscarriage, although antimicrobial treatment has made this outcome less common.62 Vertical intrauterine transmission from an infected mother may lead to neonatal typhoid, a rare but severe and life-threatening illness.63 Relapse occurs in 5 to 10 percent of patients, usually two to three weeks after the resolution of fever. The relapse is usually milder than the original attack, and the S. enterica serotype typhi isolate from a patient in relapse usually has the same antibioticsusceptibility pattern as the isolate obtained from the patient during the original episode. Reinfection may also occur and can be distinguished from relapse by molecular typing.39,64 Up to 10 percent of convalescing patients with untreated typhoid excrete S. enterica serotype typhi in the feces for up to three months; 1 to 4 percent become longterm carriers, excreting the organism for more than one year. Up to 25 percent of longterm carriers have no history of typhoid. Chronic carriage is more common among women and the elderly and in patients with cholelithiasis.65 Most carriers are asymptomatic. Patients with an abnormal urinary tract, such as those who have schistosomiasis, may excrete the organism in the urine for long periods. The average case fatality rate is less than 1 percent, but the rate varies considerably among different regions of the world. Among hospitalized patients, the case fatality rate varies from less than 2 percent in Pakistan37 and Vietnam40 to 30 to 50 percent in some areas of Papua New Guinea61 and Indonesia.59,60 The case fatality rates are highest among children under one year of age and among the elderly.37,52,55 However, the most important contributor to a poor outcome is probably a delay in instituting effective antibiotic treatment.

Management
Diagnosis The absence of specific symptoms or signs makes the clinical diagnosis of typhoid difficult. In areas of endemic disease, a fever without evident cause that lasts more than one week should be considered typhoid until proved otherwise. Blood cultures are the standard diagnostic method; provided a large volume of blood is cultured (15 ml in adults), they are positive in 60 to 80 percent of patients with typhoid. Culture of bone marrow is more sensitive. The result is positive in 80 to 95 percent of patients with typhoid, even patients who have been taking antibiotics for several days, regardless of the duration of illness.26,66,67,68 Blood cultures are less sensitive than bone marrow cultures because of the lower numbers of microorganisms in blood as compared with bone marrow.25,26 The sensitivity of blood culture is higher in the first week of the illness, is reduced by prior use of antibiotics, and increases with the volume of blood cultured and the ratio of blood to broth. Cultures have also been made from the buffy coat of blood,69 streptokinase-treated blood clots,68 intestinal secretions (with the use of a duodenal string capsule),67 and skin snips of rose spots.66 The sensitivity of stool culture depends on the amount of feces cultured, and the positivity rate increases with the duration of the illness. Stool cultures are positive in 30 percent of patients with acute typhoid fever. For the

detection of carriers, several samples should be examined because of the irregular nature of shedding. The role of Widal's test is controversial, because the sensitivity, specificity, and predictive values of this widely used test vary considerably among geographic areas. The test detects agglutinating antibodies to the O and H antigens of S. enterica serotype typhi. Unfortunately, S. enterica serotype typhi shares these antigens with other salmonella serotypes and shares cross-reacting epitopes with other Enterobacteriaceae. Furthermore, patients with typhoid may mount no detectable antibody response or have no demonstrable rise in antibody titer. Despite this, some centers have found Widal's test helpful when it is used with locally determined cutoff points.70,71 A Vi agglutination reaction has been used to screen for S. enterica serotype typhi carriers. Its reported sensitivity is 70 to 80 percent, with a specificity of 80 to 95 percent.72 Newer serologic tests are being developed but do not yet perform well enough to ensure their widespread adoption.73,74 DNA probes and polymerase-chain-reaction protocols have been developed to detect S. enterica serotype typhi directly in the blood.75 The methods are not yet widely used and are impractical in many areas where typhoid is common. Typhoid must be distinguished from other endemic acute and subacute febrile illnesses. Malaria, deep abscesses, tuberculosis, amebic liver abscess, encephalitis, influenza, dengue, leptospirosis, infectious mononucleosis, endocarditis, brucellosis, typhus, visceral leishmaniasis, toxoplasmosis, lymphoproliferative disease, and connective-tissue diseases should be considered. For patients in countries where typhoid is not endemic, a travel history is crucial. Clinical algorithms have been developed but have not generally been validated. Treatment In areas of endemic disease, more than 60 to 90 percent of cases of typhoid fever are managed at home with antibiotics and bed rest. For hospitalized patients, effective antibiotics, good nursing care, adequate nutrition, careful attention to fluid and electrolyte balance, and prompt recognition and treatment of complications are necessary to avert death. There is strong evidence that the fluoroquinolones are the most effective drugs for the treatment of typhoid fever. In randomized, controlled trials involving patients infected by quinolone-susceptible S. enterica serotype typhi, these drugs have proved safe in all age groups and are rapidly effective even with short courses of treatment (three to seven days).76,77,78,79,80 The average fever-clearance time is less than four days, and the cure rates exceed 96 percent. Less than 2 percent of treated patients have persistent fecal carriage or relapse (Table 2 and Supplementary Appendix 1). The published data also suggest that the fluoroquinolones are more rapidly effective and are associated with lower rates of stool carriage than the traditional first-line drugs (chloramphenicol and trimethoprim sulfamethoxazole).76,77 Concern has been expressed about three main issues regarding the use of fluoroquinolones in the treatment of typhoid fever: the potential for toxic effects in children, the cost, and the potential emergence of resistance. In preclinical testing, the

fluoroquinolones damaged the articular cartilage of young beagles. There is now a considerable body of reassuring evidence from the long-term use of fluoroquinolones in children with cystic fibrosis and from the short-term use of fluoroquinolones to treat typhoid fever and of fluoroquinolones or nalidixic acid to treat bacillary dysentery in children.81,82,83,84 There has been no evidence of bone or joint toxicity, tendon rupture, or, in long-term follow-up, impairment of growth. The production of generic fluoroquinolones in Asia has reduced the price considerably. However, the emergence of quinolone resistance in areas where these drugs are inexpensive and readily available is likely to be the greatest limitation on their use. Fortunately, full fluoroquinolone resistance is still rare. In areas where quinolone-resistant strains are uncommon, the fluoroquinolones are the current treatment of choice for all age groups (Table 3). Short courses of treatment (three to five days) are particularly useful to contain epidemics. Among patients with quinoloneresistant S. enterica serotype typhi infection, the rate of treatment failure is higher for those treated for less than seven days than for those treated for a longer period.48 Treatment at the maximal recommended doses (e.g., 20 mg of ofloxacin per kilogram of body weight per day) for 7 to 10 days has been successful in 90 to 95 percent of patients with resistant infections. However, the fever-clearance times are long (seven days, on average), and the rate of fecal carriage during convalescence can be as high as 20 percent (unpublished data). Fluoroquinolones should be used at the maximal possible dose for a minimum of 10 to 14 days, and the patients should be carefully followed to determine whether they are excreting S. enterica serotype typhi in their feces. Unfortunately, quinolone-resistant strains are often also multidrug-resistant, and therefore the choice of drugs is limited to azithromycin or the cephalosporins, which are expensive. The third-generation cephalosporins (ceftriaxone, cefixime, cefotaxime, and cefoperazone) and azithromycin are also effective drugs for typhoid. In randomized, controlled trials of third-generation cephalosporins, principally ceftriaxone and cefixime, the fever-clearance times averaged one week and the rates of treatment failure were 5 to 10 percent.77,78,85,86 The relapse rates were 3 to 6 percent, and the fecal-carriage rates were less than 3 percent. Cure rates of 95 percent were achieved with five to seven days of treatment with azithromycin.79,80,86,87 Fever resolved in four to six days, and the rates of relapse and convalescent fecal carriage were less than 3 percent. Aztreonam and imipenem are potential third-line drugs.76,88 Chloramphenicol, amoxicillin, and trimethoprimsulfamethoxazole remain appropriate for the treatment of typhoid fever in areas of the world where the bacterium is still fully susceptible to these drugs and where the fluoroquinolones are not available or affordable.89 These drugs are inexpensive, widely available, and rarely associated with side effects. They produce relief of symptoms, with defervescence usually occurring within five to seven days; however, two to three weeks of treatment is required, and adherence to a four-times-daily regimen over this period may be low. An adult will often have to take more than 250 capsules of chloramphenicol during a course of treatment. Although the cure rate is approximately 95 percent, the relapse rate is 1 to 7 percent, and the rate of convalescent excretion is 2 to 10 percent.

There are few data on the treatment of pregnant women with typhoid. The beta-lactam antibiotics are considered safe.62 In addition, there have been several case reports of the successful use of fluoroquinolones.90 Although these drugs have generally been avoided because of concern about safety, the general consensus is that they are also safe.91 Most of the data from randomized, controlled trials come from patients treated in regions where disease is endemic. There are few data from such trials of treatment in patients living in regions where the disease is not endemic or in returning travelers. Knowledge of the antibiotic susceptibility of the infecting strain is crucial in determining which drug to use. If no culture is available, knowledge of the likely susceptibility from the available global data may be useful (Figure 2). Severe Typhoid The parenteral fluoroquinolones are probably the antibiotics of choice for severe infections, but there have been no randomized trials of such treatment.92 In severe typhoid, the fluoroquinolones are given for a minimum of 10 days (Table 4). Adults and children with severe typhoid characterized by delirium, obtundation, stupor, coma, or shock benefit from the prompt administration of dexamethasone. The mortality rate was reduced from over 50 percent to 10 percent in Indonesian adults and children who were given dexamethasone at an initial dose of 3 mg per kilogram by slow intravenous infusion over a period of 30 minutes, followed by 1 mg of dexamethasone per kilogram given at the same rate every 6 hours for eight additional doses. Hydrocortisone at a lower dose was not effective.59,60,61