46 Pseudomembranous colitis and Clostridium difficile infection
RICHARD J. FARRELL, LORRAINE KYNE AND CIARAN P. KELLY
Introduction
Clostridium difficile was first identified in 1935 as a commensal organsim in the fecal flora of healthy neonates [1]. The organism was given its name because it grew very slowly in culture and was difficult to isolate. Although it produced cytotoxins, and was pathogenic for guinea-pigs and rabbits, the organism was considered part of the normal neona- tal gut flora that disappeared after weaning. C.
difficile remained a laboratory curiosity until 1978, when Bartlett and colleagues [2] identified it as the source of a cytotoxin found in the stool of patients with antibiotic-associated pseudomembranous coli- tis. Since that time the incidence of C. difficile infection has increased dramatically and the organ- ism is now recognized as the most frequent cause of nosocomial infectious diarrhea in developed coun- tries [3-5]. Incidence rates of nosocomial infection range from 0.1 to 30 per 1000 hospital admissions in non-epidemic settings [6-11]. Over 30% of high-risk patients, such as those admitted to acute-care general medical medical wards and receiving antibiotics, may be colonized with C. difficile [12, 13]. In commu- nity populations the reported prevalence of C. diffi- c/fc-associated diarrhea ranges from 8 to 12 per 100 000 person-years [14, 15].
As shown in Fig. 1, the sequence of events leading to C difficile diarrhea and colitis in susceptible individuals comprises disturbance of the normal colonic microflora, exposure to and colonization by C. difficile, toxin production and toxin-mediated intestinal injury and inflammation. Depending on host factors, especially the immune response to C.
difficile toxins, the outcome of colonization is either asymptomatic carriage or a spectrum of disease ranging from mild diarrhea to life-threatening pseudomembranous colitis.
During the two decades since its identification as a pathogen our understanding of the epidemiology.
Antibiotic Therapy
t
Alteration of colonic microflora Clostridium difficile exposure & colonization
Release oi Clostridium difficile toxins A & B
—y f—
Binding to enterocyte receptors Asymptomatic carriage
Colonic mucosal injury & acute inflammation
Diarrhea and colitis
Figure 1. Pathogenesis of C difficile diarrhea and colitis.
pathogenesis and management of disease caused by C difficile has increased dramatically. Yet despite this increased knowledge there has been no substan- tial decline in the frequency of hospital-acquired C.
difficile diarrhea and colitis.
Clostridium difficile
Clostridium difficile is a Gram-positive, obligate anaerobic rod that grows best in selective media containing cycloserine and cefoxitin and enriched with fructose and egg yolk [16]. This selective med- ium can detect as few as 2000 organisms in a stool sample. Individual strains of C difficile, identified by agglutinating antisera [17], or DNA fingerprinting [18], may diff'er with regard to virulence [19].
Hospital outbreaks have been attributed to toxin- producing epidemic strains [20]. The organism forms spores, allowing it to survive in harsh environments and withstand antibiotic therapy.
Stephan R. Targan, Fergus Shanahan andLoren C. Karp (eds.), Inflammatory Bowel Disease: From Bench to Bedside, 2nd Edition, 823-844.
© 2003 Kluwer Academic Publishers. Printed in Great Britain
Pathogenesis
C difficile diarrhea is a toxin-mediated disease.
Pathogenic strains of C. difficile produce two potent protein exotoxins, toxin A and toxin B. These toxins are encoded by two distinct genes in close proximity on the bacterial genome [21-23]. They are structu- rally similar and show 49% homology at the amino acid level [22].These high molecular weight proteins are believed to bind receptors on the luminal aspect of the colonic epithelium and are then transported into the cytoplasm. However, specific cell surface receptors for toxin A or toxin B have not yet been characterized. In rabbit ileum the brush-border ectoenzyme, sucrase-isomaltose, binds C. difficile toxin A and functions as a cell surface receptor [24].
Since this enzyme is not present in human colonic mucosa, other membrane surface glycoproteins pre- sumably serve as toxin receptors. Both toxins potently activate cell signaling molecules including N F - K B and MAP kinases in human monocytes, leading to the production and release of proinflam- matory cytokines including IL-ip,TNF-a, and IL-8.
These proinflammatory effects appear to precede toxin internalization and may be mediated by cell surface receptor binding [24, 25].
The aminoterminal regions of both toxins carry a series of repeated protein sequences that are believed to mediate toxin binding to the host cell membrane, while the carboxyterminal regions of both toxins possess similar glucosyltransferase activity. Once internalized, both toxins inactivate Rho proteins, a family of small GTP-binding proteins. The critical enzymatic action is the glycosylation of a specific, conserved threonine amino acid on Rho [26, 27]. The Rho protein targets of toxins A and B are rhoA, rac and cdc42, key cell signaling molecules that direct gene expression and are essential to maintain the actin cytoskeleton. Consequently, toxin-mediated Rho inactivation results in depolymerization of actin filaments, disruption of the cytoskeleton, cell round- ing and cell death [24, 28]. In contrast to cholera toxin or E. coli heat-stable toxin, C. difficile toxins have no effects on intracellular levels of cyclic AMP or GMP. However, a number of other bacterial toxins target Rho proteins in a similar manner [29].
For example the cytotoxins from C sordellii and C.
novyi add a glucose to Rho and toxins from Bacillus cereus and Staphylococcus aureus also modify Rho family proteins. Thus it appears that C difficile toxins and other structurally unrelated bacterial cytotoxins modify host cell structure and function by attacking
Rho family proteins that are vital for maintenance of normal cell architecture and function.
Toxin A is an inflammatory enterotoxin that induces fluid secretion, increased mucosal perme- ability and marked enteritis and colitis when injected into the intestinal lumen of animals [24]. Toxin A also possesses weak cytotoxic activity against cultured cells [30, 31 ]. Although toxin B is an extremely potent cytotoxin, it has no enterotoxic activity in animal intestine in vivo [28, 30, 32, 33]. This led to the widely held belief that toxin B did not participate in the pathogenesis of C. difficile diarrhea and colitis in humans. However, recent evidence appears to con- tradict this hypothesis and suggests that toxin B may indeed be pathogenic in humans. First, toxin A and toxin B both cause injury and electrophysiological changes in human colonic strips in vitro. Toxin B is 10 times more potent than toxin A in inducing these changes [34]. Second, there have been reports of the isolation of toxin A-negative/toxin B-positive strains of C difficile from patients with antibiotic-associated diarrhea and colitis [19, 35-38]. Toxin A-negative/
toxin B-positive strains accounted for 3% of clinical isolates referred for typing to the Public Health Laboratory Service Anaerobic Reference Unit in England and Wales [19].
Both toxins of C difficile bind to and damage human colonic epithelial cells [34]. C. difficile toxins produce colonic injury as a result of damage to the enterocyte cytoskeleton and disruption of tight junc- tion function [34, 39]. The toxins also cause a severe inflammatory reaction in the lamina propria with the formation of microulcerations of the colonic epithe- lium that are covered by an inflammatory pseudo- membrane. A characteristic of C difficile infection is the intense acute neutrophilic inflammation seen in pseudomembranous colitis patients and in animal models of the disease. In contrast to cholera toxin, which stimulates massive intestinal fluid secretion without a significant inflammatory response, C.
difficile toxin A stimulates fluid secretion accompa- nied by considerable mucosal edema, inflammatory cell infiltration and necrosis.
Interactions between neuropeptides and inflam- matory mediators released from inflammatory cells of the intestinal lamina propria and from epithelial cells are also critical initiators of the toxin A-induced inflammatory process (Fig. 2). Pothoulakis and colleagues reported the release of the neuropeptides substance P (SP) and calcitonin gene-related peptide (CGRP) from sensory nerves, and degranulation of mast cells within 15 min of luminal application of
B. C.
vrtLus vT««^A Ef>ilh««aioeH<ii«truc«km
VILLUS Qf ^ ^ ^ ^ ^
Figure 2. Pathogenesis of inflammatory diarrhea caused by C. difficile toxin A. A: Toxin A binds to its brush-border receptor(s) on intestinal epithelial cells, causing release of cytokines from these cells which diffuse into the lamina propria and activate primary sensory afferent neurons whose cell bodies are present in the dorsal root ganglia (DRG). Activation of primary sensory neurons causes early release of substance P (SP) which stimulates mucosal mast cells and other resident immune cells, such as macrophages. B: Significant mucosal mast cell degranulatlon occurs early after toxin A administration, releasing several proinflammatory mediators, such as histamine, platelet-activating factor (PAF), leukotriene C4 (IT), and proteases (rat mast cell protease II). Activated intestinal lamina propria macrophages also release potent inflammatory mediators, such as macrophage inflammatory protein-2 (MIP-2), IT, tumor necrosis factor-a (TNF-a), and SP. These mediators directly stimulate fluid secretion from epithelial cells and also up-regulate expression of adhesion molecules on endothelial cells and polymorphonuclear neutrophils (PMN).
G: PMN subsequently enter into the intestinal mucosa and release more proinflammatory mediators, which act on epithelial cells, causing acute destruction and necrosis of villus enterocytes 2-3 h after toxin A exposure. Toxin A can also directly damage enterocytes by inactivating rho proteins and by damaging the enterocyte cytoskeleton. (Adapted from ref. 40, with permission of the publisher.)
toxin A in animal intestine [41]. This is followed by release of TNF-oc from macrophages and up-regula- tion of adhesion molecules on endothelial cells, allowing neutrophil attachment and invasion.
Pretreatment of rabbits with a monoclonal antibody directed against the neutrophil adhesion molecules CD 18 prevented neutrophil infiltration and substan- tially reduced toxin A-induced secretion and muco- sal injury [42]. The importance of sensory neuro- peptides in C difficile diarrhea is also demonstrated by a report that prevention of SP and CGRP release from sensory neurons by administration of specific SP or CGRP antagonists substantially inhibits toxin- A-mediated diarrhea and inflammation [43]. More- over, mice genetically deficient in the NK-1 (SP)
receptor are largely protected from the secretory and inflammatory changes induced by toxin A, and mast cell-deficient mice have markedly diminished responses to the toxin [44].
Recent work also suggests that the proinflamma- tory chemokine macrophage inflammatory protein-2 (MIP-2) plays a pivotal role in mediating the early interaction between sensory nerves and mast cells and macrophages of the intestinal lamina propria following luminal exposure to toxin A. Intestinal epithelial cells release MIP-2 within 15 min of exposure to toxin A, well before the onset of fluid secretion or inflammation [45]. Moreover, an anti- body to MIP-2 substantially inhibited intestinal secretion and inflammation in this model, support-
Table 1. Antimicrobial agents that predispose to C. difficile diarrhea and colitis
Frequently Infrequently Rarely or never
Ampicillin and amoxicillin Cephalosporins Clindamycin
Tetracyclines Sulfonamides
Macrolides (including erythromycin) Chloramphenicol
Trimethoprim Quinolones
Parenteral aminoglycosides Metronidazole
Bacitracin Vancomycin
Adapted from ref. 46, with permission of the publisher.
ing the view that release of this chemokine is critical for pathogenesis. These results suggest that inflam- matory mediators such as MIP-2 and IL-1 (3 released from enterocytes in response to toxin A activate sensory nerves in the subjacent lamina propria.
Sensory nerves then release proinflammatory neuro- peptides such as SP and CGRP, which in turn stimulate inflammatory cells leading to release of proinflammatory cytokines such as T N F - a and leukotrienes that elicit neutrophil recruitment via activation of adhesion molecules on vascular endothelial cells.
Predisposing factors
As shown in Table 1 almost all antibiotics have been associated with C dijficile diarrhea and colitis, including metronidazole and vancomycin [46-49]
However, the precise risks associated with individual agents are difficult to establish [3, 50, 51]. While the duration of antibiotic therapy, the number of differ- ent antibiotics used, and the route of administration significantly influence the risk of C difficile diarrhea [51-53]. pseudomembranous colitis associated with a single preoperative antibiotic dose has been reported [54]. A recent meta-analysis demonstrated that the 'big three'classes of antibiotics predisposing to C difficile diarrhea are clindamycin, cephalospor- ins, and ampicillin/amoxicillin [51]. While early work focused attention on the prominent role of clindamycin as an inducing agent, subsequent stu- dies have shown that cephalosporins are the most common agents implicated in C. difficile diarrhea, especially in nosocomially acquired cases [55].
AmpicilHn, amoxicillin, or amoxicillin-clavulanate (Augmentin) are also common causes, especially in outpatients. Less commonly implicated antibiotics include penicillins other than ampicillin/amoxicillin,
macrolides (erythromycin, clarithromycin, and azi- thromycin), tetracyclines, sulfonamides, trimetho- prim, chloramphenicol, and quinolones. Antibiotics that are rarely or never associated with C difficile infection include parenteral aminoglycosides, vanco- mycin, bacitracin, nitrofurantoin, or antimicrobial agents whose activity is restricted to fungi, mycobac- teria, parasites, or viruses. Antineoplastic agents that possess antibacterial properties, principally methotrexate, have occasionally been implicated.
Presumably these agents induce a sufficient disturbance of the intestinal microflora to allow colonization with C difficile [56].
Immunity and liost defense factors
The first line of defense again C difficile infection is the normal bowel microflora that inhibits growth of this pathogen in vitro and in vivo [57]. Indeed, normal adults not exposed to antibiotics are rarely infected with C difficile. While C. difficile is frequently cul- tured from the stools of healthy neonates it is seldom part of the normal colonic microflora in healthy children above age 2, or in adults. Colonization by C difficile follows alteration of the endogenous microflora by antibiotics, or cancer chemotherapy agents. The protective effect of the normal stable intestinal flora is frequently referred to as 'coloniza- tion resistance'. Disruption of this barrier by anti- biotics and subsequent infection with C difficile was originally demonstrated in animal models [58-60]. C.
difficile can colonize the intestine of 'germ-free' mice.
Wilson et al. demonstrated that inoculation of these animals with fecal flora from normal mice led to the disappearance of C. difficile, confirming the impor- tance of the normal flora in preventing colonization [60]. The phenomenon of 'colonization resistance' has also been demonstrated in vitro where the growth
of C. difficile is inhibited by emulsions of feces from healthy adults but not by sterile extracts [61]. The specific organism or group of organisms of the normal adult microflora that exclude C. difficile is not entirely clear, but anaerobic species including Bacteroides may be especially important. For exam- ple, treatment with lyophilized Bacteroides species can inhibit the growth of C. difficile in the stool of patients with chronic recurrent infection [62].
Healthy neonates and infants have poor 'coloniza- tion resistance' because they have not yet developed a stable complex colonic microflora [57, 63]. Coloniza- tion rates with C difficile of 25-80% have been reported in infants and children up to the age of 2 years; however, despite the presence of toxin, they rarely develop C difficile-sissocmiQd diarrhea [3, 64].
Immaturity of their enterocytes with absence of toxin receptor expression is a possible mechanism for this clinical phenomenon [65]. Although cats, dogs, horses, and donkeys are colonized by C difficile, there is no evidence that animals serve as reservoirs for colonization of humans [21].
The humoral immune system provides a second line of defense against C. difficile. Immunization of laboratory animals against toxin A protects against a subsequent challenge with C. difficile [66, 67]. Infant hamsters who drink milk from mothers immunized against toxins A and B are also protected [66]. The fact that only one-third of C difficile carriers develop diarrhea [68] suggests that the host's ability to produce antitoxin antibodies may play a similar role in humans in modifying disease expression. Serum antibodies against C difficile toxins are present in the majority of the adult population [71]. Secretory IgA antitoxin is present in colonic secretions and can inhibit binding of toxin A to its specific brush-border receptor, providing a possible mechanism of immune protection [72]. High levels of serum and intestinal antitoxin antibodies may be associated with mild colitis or asymptomatic carriage of C difficile [69, 70, 73]. Conversely, a deficient antibody response may predispose to severe, prolonged, or recurrent C difficile coMs [14-161
In a recent prospective study of nosocomial C difficile infection 41% of 47 patients who acquired C.
difficile remained asymptomatic [13]. At the time of colonization, serum levels of IgG antibody against toxin A were three times higher in asymptomatic carriers compared to patients who developed C difficile diarrhea (Fig. 3). Multivariable analyses indicated that patients with a low serum level of IgG antitoxin A were 48 times more likely to develop C
CO
'E O <
|i
U) E
i»5
^ 2.5
2.0H
1.5
0) o5 if) c
\a 1.0
no
CD 0.5-
Carriers
P=0.06
y/ P=0.002
Patients with diarrhea P = 0.001 _ P=0.005
.^°>
- r
* % D O
-//-
S^
(i-^
Figure 3. Serum IgG antibody levels against toxin A in asymptomatic carriers and patients with C. difficile diarrhea during hospitalization. The median levels of IgG antibody against toxin A are shown for 28 patients in whom Clostridium difficile diarrhea developed, 19 asymptomatic carriers, and 187 patients without colonization at the time of hospitalization, at the time of colonization by C. difficile, 3 days after colonization, and at discharge. The median interval between admission and colonization was 3 days (range 3 - 33), and the median interval between the third day after colonization and discharge was 12 days (range 2-56). Serum levels of IgG antibody against toxin A were three times higher in asymptomatic carriers compared to patients who developed C. difficile diarrhea. The p-values refer to the comparison among the three groups (by the Kruskal-Wallis test).
(Reproduced from ref. 13, with permission of the publisher.)
difficile diarrhea compared to patients who had high antibody levels (/?< 0.001). Although no protective association was found for serum IgG antitoxin B levels, these were significantly correlated with IgG levels against toxin A in the asymptomatic carriers.
While we await controlled trials, open-label studies have demonstrated the efficacy of passive immuno- therapy using pooled human immune globulin (con- taining antitoxin A IgG antibody) in patients with recurrent or refractory C difficile diarrhea [77, 78].
A third protective factor is gastric acid, which reduces the number of viable spores [79]. Normal intestinal peristalsis is also important as a defense mechanism by eliminating C difficile and its toxins.
Conversely, antidiarrheal medications that reduce intestinal peristaltic activity may delay clearance of the organism and its toxins, and worsen the duration or intensity of illness.
Epidemiology
C. difficile infection occurs primarily in hospitalized patients, causing approximately 250 000-500 000 cases of diarrhea and colitis per year in the United States [12, 80], compared to only 20 000 cases per year in outpatients [14]. Carriage of C difficile is rare in healthy adults not taking antibiotics; intestinal carriage rates of 0% to 3%, have been reported in American and European populations [81-83]. It remains unclear whether carriage is a temporary or permanent state [19]. In contrast, the incidence of C difficile carriage is unusually high following admis- sion to hospital and treatment with antibiotics. In one study, 7% of patients admitted to an acute-care hospital had positive stool cultures for C difficile and another 2 1 % became colonized with C. difficile during their hospital stay [12]. At the time of dis- charge, 82% of hospital carriers were still excreting C difficile in their stools [12], accounting for the high rate of infection in nursing homes and chronic-care facilities. More recent hospital studies have reported similar C difficile carriage rates [10, 13].
C. difficile survives in the hospital environment as antibiotic-resistant spores that are ingested by patients. Infected patients, environmental surfaces, inanimate objects, and the hands of health-care workers are all potential sources of C difficile in the hospital setting [84]. In one study, patients sharing a room with a C difficile carrying room-mate acquired C. difficile more rapidly than patients who were in single rooms or with room-mates who were culture- negative (mean time to acquisition 3.2 days com- pared to 18.9 days respectively) [12]. The same group of investigators also cultured C difficile from the hands of 59% of hospital personnel caring for patients with positive C difficile cultures as well as from bedrails, commodes, toilets, floors, scales, call- buttons, windowsills, dust-mops, and in the rooms where these patients were nursed [12]. Thus, cross- infection may occur by patient-to-patient spread or through environmental contamination [12, 80, 8 5 - 87]. The spread of infection can be interrupted by careful hand washing after examining patients and by the use of disposable gloves [88].
Although asymptomatic carriers rarely go on to develop C. <i/^c//^-associated diarrhea [68], they can contaminate the hospital environment and serve as a reservoir of infection. McFarland ^^ a/, demonstrated that 29% of cultures taken from rooms of asympto- matic carriers were positive for C. difficile, whereas only 8%) of cultures from rooms of culture-negative
patients were positive [12]. Asymptomatic carriers have also been implicated as the source of strains of
C difficile that caused C. difficile-3.ssocmtQd diarrhea in other hospital inpatients [89]. In antibiotic-treated animals the infective dose of toxigenic C difficile may be as low as two organisms [81]. If human suscept- ibility is similar, control of C difficile infection in hospitals will continue to be a major challenge, as the organism is excreted in high numbers in liquid feces (up to lO"^ organisms/g) [85, 90]. Furthermore, C difficile can be cultured in a hospital room 40 days after discharge of an infected patient [73], and it is likely that spores of C difficile may persist for many months in hospital wards, as they are particularly resistant to oxygen, desiccation, and to many disinfectants [85].
While antibiotic exposure is the most important risk factor for C difficile infection, other risk factors include increasing age (after infancy), and severity of underlying disease. In England and Wales 75% of all reports of C difficile to the Public Health Laboratory Service Communicable Disease Surveillance Centre between 1992 and 1996, occurred in patients over 64 years of age [91]. Studies in the US have also demonstrated that increasing age is an independent risk factor for C difficile diarrhea [79, 92]. Indepen- dent of age, sicker patients are also more likely to acquire C difficile [\3]. In a recent study of antibiotic recipients, patients with severe underlying disease at the time of hospital admission were 8 times more likely to develop C difficile infection compared to patients who were less severely ill [13].
An increased incidence of C difficile infection in oncology and HIV patients appears to be related to specific risk factors among these groups of patients.
Low intensity of chemotherapy, reflecting a lower frequency of neutropenia, lack of parenteral vanco- mycin use and hospitalization within the previous 2 months were independently predictive of C difficile colitis in hospitalized oncology patients [9]). A CD4"^
cell count less than 50/mm'', in addition to clinda- mycin and penicillin, use were independent factors significantly associated with C difficile colitis among HIV-infected patients [94].
Other reported risk factors for C difficile infection include the presence of a nasogastric tube, non- surgical gastrointestinal procedures, acid antisecre- tory medications, intensive-care unit stay, and duration of hospital stay [51]. The strengths of the associations of these risk factors with C. difficile vary from study to study. Because these factors are often markers of disease severity and/or older age, their
association with C. difficile may lose its significance after controlling for these confounding variables [13, 79, 92].
Pathology
When the human colon is exposed to C difficile toxins, loss of actin filaments leads to cell rounding and shedding of cells from the basement membrane into the lumen, leaving a shallow ulcer on the mucosal surface. Serum proteins, mucus, and inflam- matory cells flow outward from the ulcer, creating the characteristic colonic pseudomembrane. The spewing forth of the inflammatory exudate from the mucosal microulceration produces the typical 'vol- cano' or 'summit' lesion of C difficile colitis (Fig. 4).
On gross or sigmoidoscopic inspection of the colonic or rectal mucosa, pseudomembranes appear as yellow or off*-white raised plaques 0.2-2.0 cm in diameter scattered over a fairly normal-appearing intervening mucosa. Edema and hyperemia of the fufl thickness of the bowel wall are common, and this is reflected by the typical radiographic appearance of 'thumbprinting', or massive wall thickening on computerized tomography scanning of patients with pseudomembranous colitis.
The patchy distribution of the pseudomembranes is probably related to a toxin dose-response effect.
For example, when human colonic mucosal strips in vitro were exposed to different concentrations of toxin B, cellular damage was very patchy at low concentrations but, as the toxin concentration was raised, the area of damage increased until it was nearly confluent [34]. Similarly, some patients with early pseudomembranous colitis have only scattered lesions on the colonic mucosa, while others exhibit a confluent pseudomembrane covering the entire mucosa.
The pathologic features of pseudomembranous colitis (PMC) have been classified into three distinct types [95]: in type 1 PMC, the mildest form, the major inflammatory changes are confined in the superficial epithelium and immediately subjacent lamina propria. Typical pseudomembranes and sum- mit lesions are present, and crypt abscesses are occasionally noted. Type 2 PMC is characterized by more severe disruption of glands and marked mucin secretion, and more intensive inflammation of basal lamina. Type 3 PMC is characterized by severe, intense necrosis of the full thickness of the mucosa with a confluent pseudomembrane. In practice.
Figure 4. Endoscopic-biopsy specimen from a patient with pseudomembranous colitis and a 'summit' or Volcano* lesion (hematoxylin and eosin, x 55). Focal ulceration of the colonic mucosa (lower arrow) is evident, with exudation of a pseudomembrane (upper arrow) made up of inflammatory cells, fibrin, and necrotic debris. The adjoining mucosa is intact. (Reproduced from ref. 3, with permission of the publisher.)
colonic histology is often normal in mild cases and may reveal only type 1 changes in the majority of cases, while the classical type 3 pseudomembranous colonic changes are seen in only a minority of patients.
Clinical features
Infection with C difficile can produce a wide spec- trum of clinical manifestations ranging from the asymptomatic carrier state in infants and adults to fulminant colitis with megacolon or perforation.
Asymptomatic carrier state
Asymptomatic carriage of C difficile is common in hospitalized patients. Several large epidemiologic studies have demonstrated that 10-16% of patients in hospital may be carriers of the organism [12, 13, 96]. Despite the fact that over 50% of C difficile isolates from these patients are toxigenic, they do not appear to be at an increased risk of developing symptomatic disease [6, 27]. The basis for this varia- bility in response is not entirely clear but, as men- tioned above, the host immune response appears to be more important than bacterial virulence factors.
Other important host response factors may include toxin receptor density and the presence or absence of the normal barrier flora.
Antibiotic-associated diarrtiea
Mild diarrhea is quite common during treatment with antibiotics, but it is related to C. difficile in only 20% of cases. Most antibiotic-associated diarrhea is related to an osmotic effect of unabsorbed carbohy- drate [97]. In normal individuals unabsorbed dietary carbohydrate delivered to the large intestine under- goes fermentation by the microflora to short-chain fatty acids, hydrogen, methane, and other metabo- lites. However, during antibiotic therapy this normal fermentation process is interrupted, allowing accu- mulation of carbohydrates that bind water and cause diarrhea. Diarrhea is watery, containing mucus but not blood. Sigmoidoscopic examination reveals normal colonic mucosa or mild edema or hyperemia of the rectum. Obvious colitis or pseudomembrane formation does not occur. Systemic symptoms are absent and diarrhea stops when antibiotics are discontinued in the majority of patients.
C. difficile diarrhea without pseudomembrane formation
This is the most common cHnical manifestation of C difficile infection. The incubation period for diarrhea after colonization is not known, but is likely to be less than a week, with a median time of onset of approxi- mately 2 days [10, 12, 13,80]. C c//y^"r/7c^ diarrhea is a more serious illness than simple antibiotic-asso- ciated diarrhea. C difficile diarrhea is typically watery and foul-smelling [96]. Mucus or occult blood may be present but visible blood is rare [98]. Some patients may present with fever, leukocytosis, and crampy abdominal pain [96]. Extraintestinal mani- festations of C difficile infection such as septic arthritis, bacteremia or splenic abscess may occur, but are extremely rare [99-102], while asymmetrical arthropathy affecting large, weight-bearing joints is more common [103]. Fecal leukocytes may be present in the stools but are not a reliable indicator of C difficile colitis, as they were absent in 72% of toxin-positive stools in one study [104]. Sigmoido- scopy may reveal a non-specific diffuse or patchy erythematous colitis without pseudomembranes.
Pseudomembranous coiitis (PIVIC)
This entity is the classic manifestation of full blown C difficile colitis and is accompanied by similar, but often more severe, symptoms than observed in C difficile diarrhea. Sigmoidoscopic examination
Figure 5. Colonoscopy view of pseudomembranous colitis resulting from refractory C. difficile infection. In the lower part of the figure coalescing pseudomembranes are visible (raised, adherent yellow plaques on the colonic mucosa that vary in size from 2 to 20 mm). In the upper part of the figure there is non-specific erythema of the colonic mucosa, with Isolated pseudomembranes visible.
reveals the classic pseudomembranes, raised yellow plaques ranging from 2 to 20 mm in diameter scattered over the colorectal mucosa (Fig. 5). In severely ill patients, white blood cell counts of 20 000 or greater and hypoalbuminemia of 3.0 g/dl or lower may be observed. Most patients with PMC have involvement of the rectosigmoid area, but as many as one-third of patients have pseudomem- branes limited to the more proximal colon [105].
There have been a few reported cases of pseudomem- brane formation involving the small intestine [106].
A number of these were in postsurgical patients and included involvement of a defunctionalized limb of a jejunal-ileal bypass [107], an ileal conduit [108], or an end-ileostomy. Although abdominal CT scan in patients with PMC is not highly specific, it may reveal mucosal edema, thumbprinting, pancolitis, pericolonic inflammation, and pronounced thicken- ing of the colonic wall that may involve the entire colon, collections of fluid in the lower abdomen or pelvis as well as the characteristic 'accordion sign' of contrast trapped among the thickened folds [109- 111] (Fig. 6). A neutrocytic ascites with a low serum- to-ascites albumin gradient may occur in patients with hypoalbuminemia [112, 113]. Ascites may even be the presenting manifestation of PMC. A recent
Figure 6. Computerized tomograph of the abdomen in C. difficile coWWs. There is marked thickening of the colonic wall in the sigmoid colon (arrow). The accordion pattern is evident, produced by a series of broad edematous colonic haustral folds. (Reproduced from ref. 111, with permission of the publisher.)
radiologic review of typical sonographic appear- ances of common colonic diseases reported ascites in 64% of patients with PMC, compared to 24% of patients with diverticulitis, cancer, inflammatory or ischemic bowel disease [114].
Fulminant colitis
Fulminant colitis in C difficile infection occurs in approximately 3%) of patients but accounts for most of the serious complications including perforation, prolonged ileus, megacolon, and death [115].
Patients with fulminant colitis may complain of severe abdominal pain, diarrhea, and distension.
Some patients exhibit high fever, rigors, dehydra- tion, and marked leukocytosis. Diarrhea is usually prominent, but may be minimal in patients who
develop an ileus resulting in the pooling of secretions in the dilated, atonic colon. Hypoalbuminemia may also occur because of a severe protein-losing entero- pathy. An abdominal radiograph may reveal a dilated colon ( > 7 cm in its greatest diameter), consistent with toxic megacolon. Patients with mega- colon may also have dilated small intestine on plain abdominal radiographs with air-fluid levels mimick- ing an intestinal obstruction or ischemia (pseudo- obstruction) [64]. In some patients fulminant C.
difficile infection may present with signs and symp- toms of bowel perforation. Typically, these patients have abdominal rigidity, involuntary guarding, rebound tenderness, and reduced bowel sounds.
Abdominal radiographs may reveal the presence of free abdominal air.
C. difficile infection in patients witti chronic inflammatory bowel disease
Infection with C difficile may complicate the course of ulcerative colitis or Crohn's disease [116, 117].
Patients with inflammatory bowel disease (IBD) are often exposed to antibiotics and are frequently hos- pitalized, placing them at increased risk. Patients infected with C. difficile may develop diarrhea, abdominal pain, and low-grade fever mimicking a flare of their IBD. The diagnosis is established by identification of C difficile toxin in a stool sample.
Pseudomembrane formation is rare in this setting, probably because the colon is already involved by IBD. C difficile diarrhea and colitis responds promptly to appropriate therapy with metronidazole or vancomycin. Some patients have developed C.
difficile at the onset of their first attack of IBD, a situation that can lead to considerable diagnostic confusion. Infection with C difficile in patients with ulcerative colitis or Crohn's disease requires prompt diagnosis and management, since failure to diagnose the infection may lead to inappropriate treatment with corticosteroids or immunosuppressive agents.
Diagnosis
The diagnosis of C difficile diarrhea or colitis is based on a history of recent or current antibiotic therapy, development of diarrhea, or other evidence of acute colitis and demonstration of infection by toxigenic C. difficile, usually by detection of toxin A or toxin B in a stool sample [3, 118].
The diagnosis of C difficile diarrhea should be suspected in any patient with diarrhea who has received antibiotics within the previous 2 months and/or whose diarrhea began 72 h or more after hospitalization [118]. Recent studies have shown that up to 40% of patients with C. difficile diarrhea at tertiary referral centers are symptomatic on admis- sion to hospital [12, 13, 119]. Therefore, it is also prudent to consider the diagnosis in patients who present to hospital with antibiotic-associated diarrhea, especially if there is a history of recent discharge or transfer from another hospital or nursing home.
Testing of non-diarrheal stools for C difficile by culture or toxin assay is not recommended because many patients in hospital may be asymptomatic carriers of the organism [12, 13, 98]. Treatment of asymptomatic carriers is not recommended, as it may prolong the carrier state [120]. For the same
reason, 'test-of-cure' of C difficile in asymptomatic patients with recent episodes of C. difficile diarrhea is not indicated. The duration of stool carriage of this organism following an episode of C difficile diarrhea is unclear, but may persist for at least 3-6 weeks [121].
Fecal specimens
If C. difficile diarrhea is suspected, a freshly taken fecal sample should be submitted immediately to the laboratory in a clean watertight container, to test for the presence of toxigenic C difficile (usually by the detection of fecal toxin A or toxin B). Anaerobic culture or the use of transport media do not enhance recovery of the organism or its toxin, and therefore are not recommended [84]. Storage at ambient temperatures leads to possible denaturation of fecal toxin [ 19]. Therefore, samples should be tested imme- diately for toxin or refrigerated or frozen for future testing. As C. difficile readily forms spores, culture of the organism from stools should be largely unaffected by ambient storage.
Laboratory tests for C. difficiie
A variety of laboratory tests are available for the diagnosis of C difficile-'dssoc'vdiQd diarrhea (Table 2). Enzyme immunoassays to detect toxin antigens in the stool are increasingly used in clinical practice.
These tests have the advantages of being relatively inexpensive, quick, and highly specific. However, their sensitivity is not ideal, leading to frequent false-negative results. The tissue culture cytotoxicity assay is more sensitive, leading to greater diagnostic accuracy, but it is also more costly and time-consum- ing.
Tissue culture cytotoxicity assay
The 'gold standard' diagnostic test to identify C.
difficile toxins in the stool of patients with antibiotic- associated diarrhea is the tissue culture cytotoxicity assay [122-124]. By inactivating Rho proteins (see above) toxins A and B effect a disintegration of the actin cytoskeleton of mammalian cells leading to cell rounding. A suspension of diarrheal stool in phosphate-buffered saline is centrifuged and filtered.
The filtrate is then inoculated on to a monolayer of cultured cells (usually fibroblasts or Chinese hamster ovary (CHO) cells). The monolayer is examined after overnight incubation and again at 48 h for cell
Table 2. Stool tests for diagnosis of C. difficile infection
Test Detects Advantages Disadvantages
Cytotoxin assay
Enzyme immunoassay
Latex agglutination assay
Culture
Polymerase chain reaction
Toxin B
Toxin A or B
Bacterial enzyme- glutamate dehydrogenase Toxigenic and non-
toxigenic C. difficile
Toxin A or B genes in isolates or directly in feces
'Gold standard'; highly sensitive and specific Fast ( 2 - 6 h); easy to
perform; high specificity Fast; inexpensive;
easy to perform Sensitive; allows strain
typing in epidemics
High sensitivity and specificity
Requires tissue culture facility;
takes 2 4 - 4 8 h Not as sensitive as the
cytotoxin assay
Poor sensitivity and specificity
Requires aerobic culture;
not specific for toxin-producing bacteria; takes 2 - 5 days Requires expertise in molecular
diagnostic techniques Adapted from ref. 111, with permission of the publisher.
rounding [19]. The specificity of the test result is established by preincubating the sample with specific neutralizing antitoxin antibody [118].
The advantages of the cytotoxicity assay include its high sensitivity (67-100%) and specificity (85- 100%), if performed correctly [84]. However, the sensitivity of the test may be reduced by inactivation of toxins during transport and storage, by the age and type of cell line used, and by the dilution titer of the stool sample [19, 125-127]. Therefore a negative cytotoxicity test does not completely rule out C difficile as the cause of diarrhea. A positive stool cytotoxin test in a patient with antibiotic-associated diarrhea indicates that it is highly likely that C difficile is the cause of diarrhea. The main disadvan- tages of the cytotoxicity assay are that it is relatively expensive; it requires a cell culture facility; and it is slow, requiring incubation of the fecal filtrate for 24- 48 h.
Enzyme-linked innmunoassay tests for toxin A or toxins A andB
There are several commercially available enzyme immunoassays (EIA) for the detection of toxin A or toxins A and B of C difficile in stool specimens [122- 124, 128, 129]. Most of the tests that are used in clinical practice are designed to detect toxin A by using a monoclonal antibody that reacts with an epitope located on the aminoterminal region of the toxin A molecule. The advantages of these tests are that they are easier to perform than the cytotoxicity
test, they are relatively inexpensive, results may be available within 2-6 h, and they have high specificity (75-100%) [84]. Their main disadvantage is that they are less sensitive than the cytotoxicity test (63-99%o) [84]. In addition, if stool samples are tested for toxin A only, C. difficile diarrhea due to a toxin A-negative/
toxin B-positive strain will not be diagnosed [35]. For this reason commercial kits that detect both toxins A and B have a slight advantage over those that detect toxin A alone [129].
Latex agglutination assay
The latex agglutination test was designed to detect toxin A in the stool, but instead recognizes glutamate dehydrogenase, another bacterial protein present in nontoxigenic strains of C difficile and other non- pathogenic Clostridia [130-132]. Thus the test is nonspecific and is also not sufficiently sensitive (48- 59%o) for the diagnosis of C. difficile diarrhea [3].
C. difficile cuMe
The stool culture test for C. difficile is sensitive (89- 100%) but is not specific for toxin-producing strains of C difficile [3, 84]. In-vitro testing for toxin produc- tion by isolates cultured from toxin-negative stools may improve specificity, but this is not a routine laboratory procedure and is costly and time-consum- ing. An advantage of stool culture of C difficile is that it permits strain typing of individual isolates.
The latter facilitates recognition of outbreaks of nosocomial infection.
Polymerase chain reaction for detection of toxin A or toxin B
Polymerase chain reaction (PCR), with the use of specific primers based on the gene sequences of toxins A and B, has been used to detect toxigenic C difficile in clinical isolates [133-135]. Although this is a highly sensitive (100%) and specific (97-100%) test, it is laborious and requires initial culture of C.
difficile. PCR methods for the detection of toxin genes directly in feces have been developed recently [136-140]. Using a nested PCR assay for the detec- tion of toxin B in fecal specimens Alonso et al.
reported 99% concordance with the cytotoxicity assay and sensitivity, and specificity of 96.3% and 100%, respectively [141]. Application of PCR methods in the clinical laboratory will require exper- tise in molecular diagnostic techniques and may not prove to be any more rapid or less expensive than stool cytotoxicity assay.
Endoscopic diagnosis of C. difficile diarrhea and colitis
Sigmoidoscopy or colonoscopy are not indicated for most patients with C. difficile diarrhea [3, 118].
Endosocopy is helpful, however, in special situa- tions, such as when the diagnosis is in doubt or the clinical situation demands rapid diagnosis. In these situations limited flexible sigmoidoscopy or colono- scopy may be performed at the bedside. In severe PMC, because of the risk of perforation, only mini- mal amounts of air should be introduced. The pre- sence of pseudomembranes in the rectum or sigmoid colon is sufficient to make a presumptive diagnosis of C difficile colitis. It is important to note that the results of endoscopic examination may be normal in patients with mild diarrhea, or may demonstrate nonspecific colitis in moderate cases. The finding of rectal pseudomembranes in a patient with antibiotic- associated diarrhea is virtually pathognomonic for C difficile colitis. Some patients without any diag- nostic features in the rectosigmoid will have pseudo- membranes in the more proximal areas of the colon [105]. Other endoscopic findings include erythema, edema, friability, and nonspecific colitis with small ulcerations or erosions.
Treatment
IVIanagement of mild to moderately severe C.
difficile diarrhea and colitis
The first step in the management of C difficile diarrhea and colitis is to discontinue the precipitat- ing antibiotic(s) if possible [3, 118]. Diarrhea will resolve in approximately 15-25% of patients without specific anti-C difficile therapy [8, 142]. However, conservative management alone may not be indi- cated in patients who are severely ill, or who have multiple medical problems, as it is difficult to predict who will improve spontaneously and who will have ongoing diarrhea. If it is not possible to discontinue the precipitating antibiotic, because of other active infections, the patient's antibiotic regimen should be altered to make use of agents less likely to exacerbate C difficile diarrhea, for example, parental amino- glycosides, trimethoprim, and quinolones [143].
Anti-peristaltic agents such as diphenoxylate plus atropine (Lomotil), loperamide (Imodium) or narcotic analgesics should be avoided because they may delay clearance of toxin from the colon and thereby exacerbate toxin-induced colonic injury or precipitate ileus and toxic dilation [118, 144 147].
Specific therapy to eradicate C. difficile should be used in patients with initially severe symptoms and in patients whose symptoms persist despite disconti- nuation of antibiotic treatment. Currently, the most widely accepted antimicrobials for the treatment of C difficile diarrhea are metronidazole (250-500 mg four times a day for 10 days) and vancomycin (125 mg to 500 mg four times a day for 10 days).
Bacitracin, teicoplanin, fusidic acid, and colestipol have also been used to treat this condition, but have few if any advantages over conventional antimicro- bials. In a recent systematic review of the efficacy of different treatments of C difficile intestinal disease, none of these agents showed clear therapeutic super- iority in terms of response rates [147]. The advantages and disadvantages of specific therapeutic agents are discussed briefly below.
Metronidazole
Metronidazole is now the drug of first choice in the treatment of C. difficile diarrhea and colitis [118]. It is inexpensive ($0.50 per 250 mg tablet) and is highly effective for the treatment of this condition. A number of clinical studies have demonstrated that metronidazole therapy results in the resolution of diarrhea and colitis in the vast majority of patients
treated (98% overall) [8, 148]. In a prospective randomized clinical trial, metronidazole (250 mg four times a day for 10 days) was found to be as effective as vancomycin (500 mg four times daily for 10 days) in terms of response and relapse rates for the treatment of C difficile diarrhea [142].
Metronidazole, unlike vancomycin, is well absorbed when administered orally. Fecal concentra- tions are low or absent in healthy individuals or asymptomatic carriers of C. difficile, but higher concentrations are observed in patients with C difficile colitis. In patients with diarrhea, metronida- zole may be secreted through an inflamed intestinal mucosa or decreased intestinal transit times may result in decreased absorption [149, 150]. Intra- venous metronidazole (500 mg four times per day) may be used if patients cannot tolerate oral medica- tion because it is excreted into bile and can accumu- late in bactericidal levels in the inflamed colon [149].
Systemic side-effects may occur with the oral use of metronidazole [151]. However, this antibiotic is remarkably well tolerated. In one institution, in a 10- year period, over 600 patients received metronida- zole for the treatment of C difficile diarrhea; only 1%
of those treated experienced significant side-effects [8]. Adverse effects include nausea and vomiting, a metallic taste, peripheral neuropathy (with pro- longed therapy) and a disulfiram-like reaction with alcohol. Metronidazole may potentiate the antic- oagulant effects of warfarin, resulting in prolonga- tion of the prothromin time. Its use in pregnant and nursing women is cautioned because of the unknown effect of metronidazole on fetal organogenesis, and reports of tumorgenictiy in rodents. Its safety in children has not been documented. Although the majority of C difficile isolates are sensitive to metro- nidazole, occasional resistant isolates have been reported, as have occasional cases of C difficile diarrhea induced by metronidazole [45, 118, 152, 153].
Vancomycin
Vancomycin has been successfully used for the treat- ment of C. difficile colitis since 1978 [154]. Its pharmacokinetic properties make vancomycin an ideal agent for the treatment of C difficile diarrhea [3], When given orally it is neither absorbed nor metabolized, and is excreted virtually unchanged, in high concentrations, in the feces. A number of controlled trials have confirmed the efficacy of vancomycin in the treatment of C difficile colitis [142,154-156]. Symptomatic improvement is usually
evident within 72 h of initiating therapy, and complete resolution of diarrhea and colitis occurs in the majority of patients (96% overall) by the end of a 10-day treatment course [148]. In one observational study of 122 patients treated with vancomycin at one institution the response rate, drug intolerance rate, and relapse rate were 99%, 1%, and 10% respectively [8].
Fekety et al. demonstrated that vancomycin at a dose of 125 mg four times a day is as effective as vancomycin 500 mg four times a day [157]. The lower dose is recommended for patients with mild to moderate colitis but the higher dose is recommended if the patient is critically ill, or has impending ileus, colonic dilation, or fulminant PMC. Vancomycin may be administered by mouth, nasogastric tube, or enema [8, 118]. It should not be given intravenously as effective luminal concentrations of the agent cannot be obtained via this route [158,159]. Systemic side-effects associated with the use of oral vanco- mycin are rare.
Despite the many advantages of therapy with vancomycin, it is now considered a second-line agent for the treatment of C difficile. There are two main factors discouraging the use of oral vancomycin;
first, it is expensive (a 10-day course may cost up to
$800) and second, its use may encourage the spread of vancomycin resistance amongst nosocomial bacteria [160]. Oral vancomycin therapy should be reserved for patients who are intolerant of or fail to respond to metronidazole, have severe PMC, are pregnant, or are under the age of 10 years [118]. Table 3 compares metronidazole and vancomycin therapy of C. difficile diarrhea.
Other antibacterial agents
Bacitracin (25 000 U four times daily for 7-10 days) has been studied in several clinical trials and is less effective than metronidazole or vancomycin for the treatment of C <i#c/fe diarrhea [148, 155, 162-164].
The overall response rate is only about 80% and the relapse rate ( > 30%) appears to be higher than with conventional therapy. In randomized therapeutic trials teicoplanin, 100 mg twice a day for 10 days, has been shown to be as effective as vancomycin for the treatment of Cdifficile diarrhea [165, 166]. It also appears to be associated with a lower relapse rate (approximately 7%) [166]. However, teicoplainin is relatively expensive and, like bacitracin, it is not readily available for oral administration in the United States. The efficacy of fusidic acid for the treatment of C difficile diarrhea has been tested in a
Table 3. Metronidazole and vancomycin for treatment of C. difficile diarrhea
Metronidazole Vancomycin
Dose Frequency Duration Route Response rate
Cost (10-day oral course) Disadvantages
250-500 mg t.i.d. or q.i.d.
10-14 days Oral or intravenous
> 96%
$20
Systemic side-effects; rare resistant strains of C. difficile
125-500 mg t.i.d. or q.i.d.
10-14 days Oral
> 96%
$800 Encourages g
vancomycin-
t.i.d., three times a day; q.i.d., four times a day Adapted from ref. 161, with permission of the publisher.
limited number of patients [166, 167]. Once again, it is less effective than metronidazole or vancomycin and is associated with a relapse rate of approximately 28% [166]. Treatment with the ion exchange resin, colestipol (10 g four times daily) is associated with a very low response rate (36'V^)) and is not recom- mended as primary therapy for C difficile diarrhea [156].
Management of severe pseudomembranous colitis
As with mild to moderate cases of C difficile diar- rhea, the first step in the management of severe PMC is to discontinue precipitating antibiotics if possible, and start therapy with metronidazole or vanco- mycin. Although there are no published data indicat- ing that vancomycin is superior to metronidazole for the treatment of severe C difficile colitis, vancomycin is recommended as a first-line agent if the patient is critically ill [118]. This recommendation is based on clinical observations that such patients appear to respond more rapidly to vancomycin than to metro- nidazole. Intravenous metronidazole should be given if oral medication is not tolerated. Intravenous vancomycin is not recommended, for the reasons mentioned above. For patients with ileus, vancomy- cin (500 mg every 6 h) may be administered via a nasogastric tube, with intermittent clamping [8]. For critically ill patients a combination of antibiotics administered by various routes may be indicated. In one series six of eight patients with severe ileus were successfully treated using a combination of vanco- mycin administered by nasograstric tube, intra- venous metronidazole, and vancomycin-retention
enemas (500 mg of vancomycin in 100 cc of normal saline administered every 6 h via a no. 18 Foley catheter inserted into the rectum); patients treated with this regimen responded within 5-17 days [8].
Intracecal infusion of vancomycin has been reported, but is not recommended because of the risks associated with placement of a narrow-bore tube over a guidewire at colonoscopy in patients with severe active colitis [ 168].
Passive immunization with immunoglobulin pro- ducts has been shown to be effective for patients with severe colitis, who do not respond to therapy with metronidazole or vancomycin [69, 70]. Patients with severe or prolonged C difficile diarrhea have low serum and fecal concentrations of antibody against C. difficile toxins [73, 75-77, 120]. Intravenous infu- sion of normal pooled human immunoglobulin (IVIG) increases serum IgG antitoxin levels and has been used successfully to treat a small number of patients with severe C difficile colitis [77, 78]. A vaccine, based on formalin-inactivated C difficile toxins, has recently been developed and tested in human subjects [169]. The C difficile toxoid vaccine may be used to stimulate antitoxin antibody responses in healthy volunteers and thereby produce a hyperimmune IVIG against C difficile to treat patients with severe or recurrent C difficile diarrhea and colitis.
The presence of extreme leukocytosis ( > 30 000), fever, hypotension, and metabolic acidosis despite medical therapy are danger signs in fulminant colitis, and may indicate the need for emergency laparotomy and colectomy in patients with impending or actual perforation [170, 171]. However, surgical interven- tion in this setting is also associated with a high mortality rate, making the decision to operate
difficult. Ramaswamy and colleagues determined that the following factors predicted increased mor- tality in severe C. difficile colitis: a low serum albumin on admission to hospital ( < 2 . 5 g/dl); a fall in albumin greater than 1.1 g/dl at the onset of symptoms; exposure to more than 3 antibiotics; and persistent toxin in the stools 7 days or longer after therapy [172]. If surgery is required, the operation of choice is a subtotal colectomy and ileostomy [110, 171, 173]. In a review of the literature between 1976 and 1994, Grundfest-Broniatowski et al found that subtotal colectomy and ileostomy was associated with a failure rate of 24% [174]. In comparison, nontherapeutic laparotomy, diverting stomas, and segmental resections were associated with failure rates of 77%, 75%o, and 40%, respectively [174].
Management of recurrent C. difficile diarrhea Approximately 15-20%o of patients treated success- fully for C. difficile diarrhea will have recurrence of diarrhea in association with a positive stool test for C. difficile toxin [3,118]. Recurrence is manifested by the reappearance of diarrhea and other symptoms, usually within 1-2 weeks of stopping treatment with metronidazole or vancomycin. Symptomatic recur- rence is rarely due to treatment failure or antimicro- bial resistance to metronidazole or vancomycin. It may result from germination of C. difficile spores persisting in the colon despite treatment. Recent evidence now suggests that recurrence most likely results from reinfection with the same or a different strain of C difficile from the environment [175, 176].
Using DNA fingerprinting, Wilcox et al. demon- strated that 56% of clinical recurrences of C. difficile diarrhea were due to infection with a different strain of C. difficile [177]. It is worth emphasizing that therapy with metronidazole or vancomycin perpe- tuates disruption of the colonic microflora and there- fore predisposes to reinfection with C. difficile. An impaired host immune response to C difficile toxins may also increase the risk of recurrent C difficile diarrhea [3, 70].
Regardless of the mechanism of recurrence, treat- ment of this form of disease can be problematic.
Approaches to management include conservative therapy or treatment with specific anti-C. difficile antibiotics, the use of anion-binding resins, therapy with micro-organisms (probiotics), and immuno- globulin therapy. The basic principles of manage- ment involve: (a) treatment of C difficile diarrhea, and (b) reduction of the susceptibility of the indivi-
dual to C. difficile reinfection and/or C difficile toxin-mediated colonic injury.
Conservative therapy
As with initial episodes of C. difficile diarrhea, conservative management of recurrent diarrhea may be preferable to re-treatment with metronidazole or vancomycin. While diarrhea usually responds to these agents, they do little to eradicate C. difficile spores within the colon or in the environment. They also perpetuate the disturbance of the normal intest- inal flora and the associated loss of 'colonization resistance' [118]. In clinical practice, however, it is often impossible to withhold antibiotic therapy, as many patients with recurrent disease are elderly and infirm and are not able to tolerate diarrhea [178].
Even in healthier patients persisting or worsening diarrhea caused by recurrent C difficile infection are clear indications for active treatment.
Re-treatment with specific anti-C difficile antibiotics The most common therapy for recurrent C. difficile diarrhea is a second course of the same antibiotic used to treat the initial episode [118]. In one large observational study in the United States, 92% of p a t i e n t s with r e c u r r e n t C. difficile d i a r r h e a responded successfully to a single repeated course of therapy, usually with metronidazole or vancomycin [8]. In this study multiple recurrences were uncom- mon (8%); however, other investigators have found that patients with a history of recurrence have a high risk of further episodes of C difficile diarrhea after antibiotic therapy is discontinued [179, 180]. In a 1994 study, patients with at least two previous relapses had a subsequent relapse rate of 65% after standard therapy with metronidazole or vancomycin [179]. Fortunately there is no evidence to suggest that sequential episodes become progressively more severe or complicated [180].
A variety of treatment schedules have been sug- gested for patients with multiple recurrences of C difficile diarrhea (Table 4). One approach is to give a prolonged course of vancomycin (or metronidazole) using a decreasing dosage schedule followed by pulse therapy [181]. The unproven rationale for this treat- ment course is that pulse therapy with antibiotics allows C. difficile spores to vegetate on the off days and then be killed when the antibiotics are taken again [181]. A combination of vancomycin 125 mg four times a day and rifampicin 600 mg twice a day for 7 days was used successfully in a study of seven patients with relapsing disease [182]. However, this
Table 4. Approach to management of recurrent C. difMle colitis
First relapse Confirm diagnosis
Symptomatic treatment if symptoms are mild 10-14-day course of metronidazole or vancomycin Second relapse
Confirm diagnosis Vancomycin* taper
125 mg q6h for 7 days 1 2 5 m g q 1 2 h f o r 7 d a y s 125 mg q.d. for 7 days 125 mg q.o.d. for 7 days 125 mg every 3 days for 7 days Further relapse
Vancomycin in tapering dose as above plus cholestyramine 4 g b.i.d.
or
Vancomycin 125 mg q.i.d. and rifampicin 600 mg b.i.d. for 7 days or
Therapy with micro-organisms, e.g. Saccharomyces boulardii'm combination with metronidazole or vancomycin
or
Intravenous immunoglobulin
^Metronidazole may be substituted for vancomycin, although there are no published data regarding its efficacy in this treatment regimen.
q6h, every 6 h; q.d., every day; q.o.d., every other day; b.i.d., twice daily; q.i.d., four times daily.
Adapted from ref. 111, with permission of the publisher.
study was uncontrolled and there is no evidence that this combination of antibiotics has any unique activ- ity against C difficile [118].
Anion-binding resins
Cholestyramine (4 g three or four times daily for 1-2 weeks), an anion-exchange resin, binds C difficile toxins and may be used in conjunction with anti- biotics to treat frequent relapsers [183]. Because cholestyramine may bind vancomycin as well as toxins, it should be taken at least 2-3 h apart from the vancomycin [184].
Biotherapy
Biotherapy (therapy with micro-organisms or 'pro- biotics') is an attractive approach to the management of recurrent C. difficile diarrhea because it aims to restore the 'colonization resistance' of a normal colonic flora. Several agents and routes of adminis- tration have been evaluated, including a mixture of colonic bacteria in saline administered as a rectal
Table 5. Practice guidelines for prevention of C. difficile diarrhea
1. Limit the use of antimicrobial drugs
2. Wash hands between contact with all patients
3. Use enteric (stool) isolation precautions for patients with C. difficile diarrhea
4. Wear gloves when contacting patients with C. difficile diarrhea or their environment
5. Disinfect objects contaminated with C. difficile with sodium hypochlorite, alkaline glutaraldehyde, or ethylene oxide 6. Educate the medical, nursing, and other appropriate staff
members regarding the disease and its epidemiology Adapted from ref. 118, with permisssion of the publisher.
infusion, fresh feces administered as a rectal enema, Lactobacillus GG given as a concentrate in skim milk, oral administration of non-toxigenic C difficile, brewer's or baker's yeast {Saccharomyces cerevisiae) taken by mouth and Saccharomyces houlardii given in capsule form [60, 179, 185-189]. Unfortunately many of these studies have been small, open-labeled and uncontrolled.
S. houlardii is a non-pathogenic yeast that has been reported to reduce the incidence of antibiotic-asso- ciated diarrhea [190]. A randomized, double-blind, placebo-controlled trial involving 124 patients examined the efficacy o{S. houlardii (500 mg twice a day for 4 weeks) in combination with metronidazole or vancomycin in patients with C difficile diarrhea [179]. S. houlardii significantly reduced recurrences compared with placebo in patients with multiple episodes of C difficile diarrhea (recurrence rate 35%
versus 65%, p = 0.04), but not in those with an initial episode of C difficile diarrhea (recurrence rate 19%
versus 24%; p = 0.86) [179]. The preparation of S.
houlardii used in this trial is not currently approved for use in the US, but is available in other countries.
Immunoglobulin therapy
As mentioned earlier, there is now substantial evi- dence that the immune response to C. difficile toxins plays a major role in determining host susceptibility to disease [13, 69, 70]. Several investigators have found that serum antibody levels against C difficile toxins are low in patients with recurrent C. difficile diarrhea [15-77, 191]. In a study of six children with relapsing C. difficile colitis, Leung et al. found that these children had low serum levels of IgG antibody against toxin A [77]. Treatment with normal pooled intravenous gamma globulin, that contains IgG anti-
