The Ecology and Pathobiology of Clostridium difficile Infections: An Interdisciplinary Challenge

Summary

Clostridium difficile is a well recognized pathogen of humans and animals. Although C. difficile was first identified over 70 years ago, much remains unknown in regards to the primary source of human acquisition and its pathobiology. These deficits in our knowledge have been intensified by dramatic increases in both the frequency and severity of disease in humans over the last decade. The changes in C. difficile epidemiology might be due to the emergence of a hypervirulent stain of C. difficile, aging of the population, altered risk of developing infection with newer medications, and/or increased exposure to C. difficile outside of hospitals. In recent years there have been numerous reports documenting C. difficile contamination of various foods, and reports of similarities between strains that infect animals and strains that infect humans as well. The purposes of this review are to highlight the many challenges to diagnosing, treating, and preventing C. difficile infection in humans, and to stress that collaboration between human and veterinary researchers is needed to control this pathogen.

Introduction

Clostridium difficile is an anaerobic organism that was first described in the 1930’s, when it was termed Bacillus difficilis because of difficulties in culturing this bacterium in vitro. B. difficilis was initially believed to be part of the normal intestinal flora of newborns (Hall & O'Toole, 1935). However, in 1978 Bartlett, et al. implicated this organism (by then known as C. difficile) in pseudomembranous colitis, a disorder often associated with antibiotic use (Bartlett et al., 1978). Prior to that report, antibiotic-associated colitis was thought to be caused by Staphylococcus aureus (Oeding & Austarheim, 1954, Prohaska, 1959). Since that time, C. difficile infection has evolved from the role of a nuisance complication of antimicrobial therapy to being one of the most rapidly rising and feared nosocomial pathogens worldwide (Zilberberg et al., 2008).

The incidence and severity of CDI are increasing in North America and Europe (Figure 1) (Barbut et al., 2007,McDonald et al., 2006, Redelings et al., 2007, Gravel et al., 2009). Recent outbreaks of CDI have been associated with increased severity, leading to more colectomies and attributable deaths (Figure 2) (Redelings et al., 2007,Pepin et al., 2005b, McEllistrem et al., 2005, Pepin et al., 2004, McDonald et al., 2005, Dallal et al., 2002, Loo et al., 2005). These increases in CDI incidence and severity have been associated a new, hypervirulent strain of C. difficile, commonly referred to as the epidemic strain. Because of the numerous methods available for molecular typing of C. difficile, this strain has been given several different names based on the type of typing performed: NAP1, 027, and BI (McDonald et al., 2005). This epidemic strain has a mutation in an important toxin production down regulatory gene, the tcdC gene that renders this gene non-functional. As a result, this strain is able to produce up to 16 times more toxin A and 23 times more toxin B in vitro than what have historically been the most common strains of C. difficile (Warny et al., 2005). Other potential virulence factors of the epidemic strain include presence of the genes for binary toxin and high-grade fluoroquinolone resistance.

Figure 1.

Increasing incidence of CDI, mid 1990s to mid 2000s, United States (from (McDonald et al., 2006) with permission):

Figure 2.

CDI Mortality rates, United States, expressed as fatalities per million, 1999–20004 (from (Redelings et al., 2007), with permission).

CDI also causes significant morbidity and mortality in endemic, and community settings, and might be responsible for as many as 20,000 deaths and costs as much as $3.2 billion per year in US acute care facilities alone (Campbell et al., 2009, Dubberke et al., 2008a, O'Brien et al., 2007). There are concerns that CDI incidence in the community is increasing, and an expanding number of studies report that C. difficile contaminates food (Weese, 2010, Wilcox et al., 2008). This pathogen is clearly a target for pre-ingestion control and prevention. The purpose of this review is to provide an overview of the many clinical challenges posed by C. difficile and the illnesses it causes in humans, and to prompt collaboration between veterinary and human microbiologists and ecologists and human clinical investigators.

The Organism

C. difficile is a Gram-positive bacillus that forms oval, subterminal spores. It is a strict anaerobe that forms non-hemolytic, rhizoid colonies on blood agar plates and has a characteristic “horse barn” odor. C. difficile is motile and is catalase, indole and urease negative. It is also negative for lipase and lecithinase on egg-yolk agar. C. difficile can be cultured and isolated from clinical specimens using well established methodology (Clabots et al., 1989), but culture is rarely used to diagnose CDI, as discussed below. When employed, alcohol shock increases the yield of recovering C. difficile from stool compared to direct plating onto selective media (Clabots et al., 1989). In this technique, equivalent volumes of stool and 95% ethanol are incubated and gently mixed. After washing, the specimen is inoculated onto pre-reduced cycloserine-cefoxitin-fructose agar with taurocholate (TCCFA) agar and incubated at 37°C in an anaerobic environment for up to one week. To optimize C. difficile recovery from rectal swabs, rectal swab specimens can be first inoculated into TCCFA broth (Arroyo et al., 2005). Identification of candidate C. difficile is made using commercial systems, such as the RapidID™ ANA II system or by traditional biochemical methodologies, such as a positive reaction with the Pro disk and a negative spot indole reaction.

C. difficile virulence factors

C. difficile toxins A and B

The principal C. difficile virulence factors are a pair of closely related large toxins known as C. difficile Toxins A (TcdA) and B (TcdB). The two large toxins are believed to account for most of the clinical manifestations of CDI. Toxin non-producing strains do not appear to cause disease in colonized individuals, regardless of recent antibiotic usage. Based on these observations, inhibition of Tcd toxin activity might be expected to prevent or resolve the clinical features of CDI.

The genes encoding TcdA and TcdB share 66% nucleotide sequence identity. TcdA and TcdB are large molecules (308 and 270 kD respectively) that also share extensive amino acid sequence homology (74%) within the enzymatic and substrate recognition domains (Reinert et al., 2005). The tcdA and tcdB genes appear to have arisen from a gene duplication event, are transcribed as single open reading frames, and are found within a large pathogenicity island (Jank et al., 2007). Recent in vivo studies demonstrated TcdB to be most crucial to virulence (Lyras et al., 2009).

The Tcd toxins are believed to bind to host cell receptors (but such receptors have not been identified with certainty), and are then internalized and transported into the cytoplasm where they are enzymatically active. Both of these toxins are glucosyltransferases that target the Rho family of GTPases (Jank et al., 2007). After they are glucosylated, these targets, which include RhoA, Rac1 and Cdc42, become inactive. This inactivation results in a variety of effects: disruptions in signaling cascades, arrest of cell cycle progression, and damage to cytoskeletal integrity (Jank et al., 2007). An immediate effect of this injury is the increased secretion of fluid and electrolytes from enterocytes and a decreased permeability barrier of the intestinal mucosa. Clinical and experimental evidence indicate that inflammation is an early response to toxin exposure. Within a few hours of toxin exposure, enterocytes become rounded and an inflammatory cascade characterized by release of inflammatory cytokines is activated (Ishida et al., 2004, Hippenstiel et al., 2000).

Some C. difficile strains also produce an actin-ADP-ribosylating toxin termed the C. difficile transferase (CDT) toxin (also called binary toxin) (Martin et al., 2008, Geric et al., 2004, Stubbs et al., 2000, Goncalves et al., 2004). Though CDT is frequently found in C. difficile strains associated with severe CDI, establishing the role of this toxin in pathogenesis remains elusive (Barbut et al., 2005, McDonald et al., 2005). A recent report suggests that CDT elicits protrusions from epithelial cells, and augments bacterial adherence (Schwan et al., 2009). It is plausible that this phenotype leads to increased host mucosal inflammation during symptomatic CDI in patients infected with CDT-expressing C. difficile.

Animal Models

The rabbit was used in the original characterization of B. difficilis by Hall and O’Toole (Hall & O'Toole, 1935). Culture supernatants injected intra-peritoneally caused seizures and death, with mouse lethal dose₅₀ testing near that for C. botulinum, but there was no description of intestinal pathology. In 1943, the guinea pig was used in studies putatively to show the value of penicillin for treatment of gas gangrene, but the animals developed large fluid-filled and hemorrhagic cecae (Hamre et al., 1943). It was also noted that stool from these animals caused cytopathic changes in Vero cells, but this was attributed to a virus (Green, 1974). The association between C. difficile and pseudomembranous colitis in the 1970’s brought clarity to the field, and prompted the need for new animal models to study this now recognized pathogen.

Bartlett, et al identified the etiology and pathology of clindamycin-associated enteritis as possibly due to C. difficile using the Syrian Golden hamster (Bartlett et al., 1977). The pathology of the Syrian Golden hamster infection model is generally regarded as closest to human antibiotic-associated enterocolitis. Recently, the key role of toxin B in C. difficile virulence was clarified using ΔtcdA and ΔtcdB genetic constructs in the hamster model (Lyras et al., 2009). In regard to treatment strategies, the hamster infection model has also been used to demonstrate protection from CDI using colonization with a nontoxigenic strain of C. difficile and response to antibiotics (Sambol et al., 2002, Anton et al., 2004). Unfortunately, hamster genome sequencing data, and hamster-specific serologic reagents to further investigate candidate cellular and immune targets that have been described in mouse pathogenicity models of CDI, are not available, so extensive use of this animal model is hindered.

Mouse studies are not limited by these constraints, and various strains of mice have been used to study CDI pathology and transmission, immunology and signal transduction, vaccine efficacy, and probiotics (Gardiner et al., 2009, Ishida et al., 2004, Anton et al., 2004, Chen et al., 2008, Adams et al., 2007, Warny et al., 2000, Ghose et al., 2007). However, murine intestinal pathology differs from that observed in severe human CDI. Also, no model addresses the severe systemic response seen in humans with fulminant CDI. A recent notable exception is the gnotobiotic piglet model, which demonstrated systemic toxemia and circulating Interleukin-8 production with severe but not mild disease (Steele et al., 2010). This suggests that complications of severe CDI such as multiple organ dysfunction syndrome (Dobson et al., 2003), abdominal compartment syndrome (Shaikh et al., 2008) and adult respiratory distress syndrome (Jacob et al., 2004) might be due to toxemia and abnormal modulatory responses. The above hypothesis is supported by an embryonic zebrafish model that demonstrated cardiotoxicity of C. difficile toxin B (Hamm et al., 2006).

Clinical disease in humans

CDI presents a broad spectrum of clinical manifestations. The spectrum includes an asymptomatic carrier state, colitis with or without pseudomembranes (and colitis might or might not be manifest as bloody diarrhea), and, in its most feared incarnation, fulminant colitis with megacolon or perforation. Recent antibiotic exposure is identified in >90% of hospitalized patients who develop CDI. However, most diarrheas associated with antibiotics are independent of CDI (Bartlett, 2002). It is not at all clear why there is such as spectrum of CDI, but it is possibly explained by host in addition to bacterial factors. Such factors might include variable cellular susceptibility to the toxins made by C. difficile, host response to C. difficile and/or its toxins, underlying status of the health of the host and the constituents of the ambient, non-C. difficile, intestinal microbiota, and of circulating antibodies against bacterial toxins (Jiang et al., 2006, Janvilisri et al., 2009, Louie et al., 2009, Kyne et al., 2000).

Asymptomatic colonization with C. difficile is common (Marciniak et al., 2006, Riggs et al., 2007). It is quite likely that this subset of C. difficile-infected individuals serves as this organism’s reservoir. However, there are no data in support of treating asymptomatic (or post-symptomatic) carriers of C. difficile with antibiotics in attempts to clear this organism from the gut (Johnson et al., 1992).

Symptoms of CDI most commonly start about a week after antibiotic therapy starts. However, diarrhea can develop even after antibiotic treatment has ceased. This interval can complicate the diagnosis of CDI, because a clinician might not be aware of this previous risk factor. C. difficile can rarely affect the small bowel of humans, typically in patients post colectomy (Malkan et al., 2009, Causey et al., 2009). Additional (and also uncommon) non-colonic complications of C. difficile infection include extraintestinal dissemination (Libby & Bearman, 2009, Gregg & Alexander, 2009), and reactive arthritis (Birnbaum et al., 2008).

Most patients symptomatically infected with C. difficile have nonbloody diarrhea. Additional symptoms include abdominal pain (usually crampy), malaise, nausea, vomiting, dehydration and fever. Laboratory tests commonly demonstrate elevated white blood cell counts and neutrophilic predominance. Indeed, the sudden occurrence of leukocytosis in hospitalized patients who might be at risk for CDI (most particularly those receiving antibiotics or having recently been administered these agents) should prompt consideration of testing for C. difficile infection (Bulusu et al., 2000).

Pseudomembranous colitis is the entity most often associated with C. difficile infection. However, this disorder occurs in less than 50% of people demonstrated to have CDI (Olson et al., 1994, Bouza et al., 2005). Pseudomembranous colitis can present several weeks after antibiotics have been stopped (Pear et al., 1994). In addition to the symptoms described above for C. difficile associated non-pseudomembranous colitis diarrhea, pseudomembranous colitis often is accompanied by fever, chills, and tenesmus. Physical findings are non-specific and include diffuse abdominal tenderness and distention. Hematochezia is unusual in pseudomembranous colitis, and fecal leukocytes are demonstrated in only about half of the patients. However, as with non-pseudomembranous colitis, peripheral leukocytosis is common. Three leukocytosis patterns are recognized: a sudden increase in the white blood count simultaneous with, or preceding symptoms of C. difficile colitis, or exacerbation of already existing leukocytosis (Bulusu et al., 2000).

C. difficile can also cause fulminant colitis, manifest as transmural inflammation of the colon, and fulminant colitis has many serious associated complications. Fulminant colitis represents well under 10% of symptomatic CDI, but it might be more common in those infected with the recently described hypervirulent epidemic strain (McDonald et al., 2005, Hubert et al., 2007, Miller et al., 2009, Loo et al., 2005). Fulminant colitis can present de novo, or evolve from a milder infection. Severe leukocytosis (over 20,000 cells/μl) and hypoalbuminemia can accompany fulminant C. difficile-associated colitis (Lamontagne et al., 2007) Paradoxically, diarrhea can decrease as severe colitis evolves because of colonic dilation and ileus. Colectomy might be necessary for impending or established colonic perforation (Gash et al., 2010).

Relapsing CDI

Up to 30% of patients develop symptoms after apparently successful initial treatment of C. difficile infections regardless of which antibiotic was chosen as initial therapy (Bouza et al., 2005). Infection could be from endogenous strain recurrence, or acquisition of a different strain (Barbut et al., 2000).

CDI and its many Clinical Challenges

Challenges to Prevention of Human Disease

By the early 1990’s, several observations led investigators in the field to conclude that CDI was largely hospital acquired, and that the chief source of transmission was patients with diarrhea. The mode of transmission was believed to be health care workers in these settings, more precisely, health care workers who did not use gloves when handling feces from patients infected with C. difficile (Clabots et al., 1992, Samore et al., 1996, McFarland et al., 1989, Johnson et al., 1990). Infection control recommendations logically included isolation of patients with diarrhea in the hospital, and use of gloves when handling feces in the health care setting (Gerding et al., 1995, Fekety, 1997). However, only the wearing of gloves while handling stool earns an “A1” quality of evidence rating for prevention of CDI (Johnson et al., 1990), and we are left with the disturbingly increased incidence of these infections: (Barbut et al., 2007, Kuijper et al., 2006, McDonald et al., 2006, Paltansing et al., 2007, Redelings et al., 2007, Kyne et al., 2002, Valiquette et al., 2007, Muto et al., 2007, Loo et al., 2005). It is disheartening that few practical advances in CDI prevention have emerged in the three decades following its description.

It is possible that secular changes in CDI epidemiology might have rendered current infection control recommendations (i.e., those that focus on fomite avoidance in the hospital from symptomatic patients) less helpful. There is increasing recognition of outpatient CDI (Dubberke, 2009, Klein et al, 2006) and asymptomatic C. difficile carriage in the community (Dubberke et al., 2009a, Wilcox et al., 2008, Pituch, 2009, Limbago et al., 2009, Dubberke et al., 2009b). These patients could enter the hospital without symptoms but with C. difficile in their intestinal tract, which through unknown mechanisms causes disease later in hospitalization or is transmitted to other patients (Clabots et al., 1992). Disease control strategies that rely on limiting spread of this pathogen only from symptomatic patients would therefore not be effective if this was true.

Since the early 2000’s, alcohol-based hand hygiene products have been recommended as the primary form of hand hygiene in healthcare settings (Boyce et al., 2002). This is problematic, because C. difficile spores resist the bactericidal effects of alcohol, use of alcohol-based hand hygiene products is not associated with a reduction in C. difficile spores on hands, and contact with asymptomatic C. difficile carriers can contaminate the hands of unwitting healthcare workers (Riggs et al., 2007, Oughton et al., 2009). It is recommended to use soap and water after caring for a patient with CDI during a CDI outbreak (Dubberke et al., 2008b). However, asymptomatic carriers are not identified. This is problematic because gloves are not worn when caring for, and soap and water for hand hygiene is not used after contact with, asymptomatic carriers, thus potentially increasing their contribution to C. difficile transmission. It is notable that the increase in CDI incidence started in the early 2000’s, at the same time as the wide spread use of alcohol hand hygiene products (Figure 1). In addition, the increase in CDI onset outside of healthcare settings could indicate that there has been an increase in the number of asymptomatic carriers who develop symptomatic CDI, and also an increase in the number of asymptomatic carriers that are admitted to hospitals. The role that asymptomatic carriers play in the transmission of C. difficile in today’s healthcare settings, the proportion that become symptomatic, and the factors that influence the likelihood of developing CDI after C. difficile acquisition must be established.

While antibiotics play an enduring role as a risk factor for CDI, a wider range of antibiotics seem to pose greater risk than had been previously considered (i.e., fluoroquinolones), and use of gastric acid suppression agents are also emerging as risk factors (Dial et al., 2005, Dial et al., 2006, Planche et al., 2008). Clearly, we need to take a broader view of potential modes of acquisition, including the colonized host, the environment, or food, as methods through which hospitalized patients acquire symptomatic CDI.

CDI and challenges to its diagnosis in humans

We remain constrained in our ability to diagnose CDI (Bartlett, 2008b, Planche et al., 2008, Kvach et al., 2009). Stools are rarely cultured to find C. difficile, because culture is cumbersome with a protracted turn-around-time. In addition, after isolating C. difficile, further characterization needs to be performed because nonpathogenic as well as pathogenic C. difficile are equally recovered by culture, and >10% of hospitalized patients might be colonized with non-pathogenic C. difficile (Belmares et al., 2009). For these reasons non-culture diagnostic tests are often employed. The oldest is the cytotoxicity assay, and this test remains the clinical laboratory “gold standard.” This technique applies fecal filtrates to cultured fibroblasts, and seeks microscopic evidence of cell death or injury (which is also neutralized for purposes of specificity testing with antibodies to Clostridium sordellii toxin)(Bartlett, 2002). However, the minimum time from specimen processing to results for cytotoxicity assays is 24 hours, and time from original order for the test to results often takes several days. In addition, there can be variability between microscopist when interpreting the results. It should also be noted that C. difficile toxin is unstable and will degrade rapidly in clinical specimens left at room temperature following collection, so results are compromised if transport of the specimen is delayed.

Antigen tests (usually enzyme immunoassays for C. difficile toxins A and B or the C. difficile common antigen) fare variably well against cytotoxicity assays, but are typically less sensitive (Musher et al., 2007b). In addition, molecular assays detecting tcdB have become commercially available. Recent publications suggest that detection is improved with multistep algorithms that encompass culture, toxin detection, and gene amplification (Shin et al., 2009, Ticehurst et al., 2006, Fenner et al., 2008, Delmee et al., 2005, Reller et al., 2007) However, we cannot, with current data and technology, apply uniform diagnostic recommendations to all specimens in which we suspect CDI (Bartlett & Gerding, 2008). Ideally, future testing technology will be available at or near point of care, be economical, performed in a clinically relevant time frame (hours, at most) so that infection control and therapeutic action can be implemented rapidly, and be interpreted by personnel not requiring full microbiologic technical expertise.

Various typing methodologies have been described for characterizing and comparing C. difficile isolates, including PCR-ribotyping, pulse-field gel electrophoresis, multi-locus sequence typing, toxinotyping, each with different advantages and performance characteristics (Killgore et al., 2008). Typing of C. difficile is typically reserved for research or epidemiological purposes, with few laboratories routinely typing C. difficile isolates. Use of non-culture based diagnostics for CDI limit the number of isolates available for typing as well.

In recent years, new challenges to CDI diagnosis have emerged with the identification of the putatively hypervirulent epidemic strain (Loo et al., 2005, Warny et al., 2005, McDonald et al., 2005). This strain seems to have evolved into its current pathogenic genotype in the past two decades (Stabler et al., 2009), and further study is needed to determine whether the future diagnosis of CDI will need to consider the sub-type of infecting C. difficile and the presence or absence of C. difficile Toxin A and/or B (or the toxin genes), toxin expression regulatory gene deletions (tcdC), and/or the C. difficile binary toxin (CDT) (Rupnik et al., 2009).

On occasion the diagnosis of pseudomembranous colitis is made by sigmoidoscopy (though, if there is access to good clinical microbiology, it is rarely necessary to do so to make this diagnosis). Pseudomembranes are striking and their appearance is characteristic: there are yellow-white raised plaques that are up to 10 mm in diameter (Figure 4). These small plaques can also coalesce.

Figure 4.

Pseudomembranes in CDI (from (Bartlett, 2002) with permission from Massachusetts Medical Society).

The tissue of pseudomembranous colitis contains inflammatory cells, fibrin, and mucus, and on histopathology, there is a “volcano” lesion “erupting” from the mucosa (Figure 5).

Figure 5.

“Volcano” lesion (from (Tarr et al., 2009, Tarr et al., 2008) with permission).

Although sigmoidoscopy can be helpful, it is important to remember that pseudomembranes can be restricted to the proximal colon (Tedesco et al., 1982), or be completely missing.

These limitations of current diagnostic strategies for CDI obligate a coordinated approach to this infection with non-culture methodology, and a specimen collection that results in new and useful approaches.

Human CDI and challenges to its treatment

Since the late 1970s, metronidazole and vancomycin have been the pillars of CDI therapy. Metronidazole has been preferred as the first line therapy for mild CDI; it is well absorbed and tissue levels in the colon are easily achieved, no matter the route of administration (oral, intravenous, or rectal) (Bolton & Culshaw, 1986). Resistance is quite unusual (Barbut et al., 1999, Johnson et al., 2000, Hecht et al., 2007), but treatment failures are common (15–30%) for unclear reasons (Hu et al., 2008, Belmares et al., 2007, Musher et al., 2005). Metronidazole’s use is sometimes limited by nausea and vomiting, and, less commonly but more ominously, by peripheral neuropathy.

Vancomycin is a glycopeptide bacterial cell wall synthesis inhibitor that is often used in severe CDI (Bartlett, 2008a, Zar et al., 2007). Intravenously administered vancomycin does not achieve concentrations in the colon of such a magnitude to treat CDI. For this reason, orally administered vancomycin is preferred. Orally dosed absorbed vancomycin is minimally absorbed and is neither nephrotoxic nor ototoxic. However, oral administration of this antibiotic sometimes poses difficulties, because patients severely ill with CDI cannot take medications by mouth, or might have an ileus that precludes access of the oral vancomycin to the region of the gut with the highest C. difficile burden. There has also been understandable hesitancy to use vancomycin to treat CDI because of concern about selecting for vancomycin resistant enterococci (Al-Nassir et al., 2008, Shin et al., 2003), but in the setting of a severely ill patient, this concern should not deter use of vancomycin.

Rifaximin (an oral nonabsorbable form of rifampin) demonstrates some efficacy in CDI (Johnson et al., 2007, Garey et al., 2008a, Garey et al., 2008b), but reports of rifampin resistant C. difficile suggest some limitation to rifaximin’s utility for treating CDI (O'Connor et al., 2008, Curry et al., 2009). Nitazoxanide, a new anti-parasitic/anti-bacterial agent (Musher et al., 2006, Musher et al., 2007a), ramoplanin (Freeman et al., 2005) and teicoplanin (Nelson, 2007) also show promise. Toxin binders (e.g., cholestyramine (Sinatra et al., 1976, Kreutzer & Milligan, 1978)) have been used in uncontrolled studies. Telovamer, a toxin-binding resin, showed promise in in vitro (Hinkson et al., 2008), animal models (Hinkson et al., 2008) and early human studies (Louie et al., 2006), but more recent studies in humans have been less promising (Weiss, 2009).

Intravenous immune globulin (Leung et al., 1991, Wilcox, 2004) monoclonal antibody therapy (Lowy et al., 2010) and therapeutic vaccines have (Sougioultzis et al., 2005) also been considered, because circulating antibodies to C. difficile toxins are associated with protection from severe CDI in hospital (Kotloff et al., 2001, Aboudola et al., 2003, Kyne et al., 2000, Kyne et al., 2001). None of these agents are approved by the FDA to treat CDI, however, and cannot now be considered for first-line therapy (Balagopal & Sears, 2007, Abougergi et al., 2010).

Recurrent CDI presents treatment challenges, and occurs after cessation of first line therapy in 20–30% of patients. It is important to remember that antibiotic resistance of C. difficile is rare, and that there is no justification to using higher doses of metronidazole or vancomycin to treat recurrences of symptomatic CDI. Recommendations regarding the treatment of recurrent CDI are informed by extensive experience (Leffler & Lamont, 2009), and limited data.

Microbial influences on CDI development in humans

Despite the use of antimicrobials to which C. difficile is rarely resistant, we still encounter a 5–10% mortality from symptomatic CDI in hospitalized patients. We need better knowledge of host and microbial ecology and pathobiology so as to develop new ways to treat, or ideally, prevent, CDI,

A large body of circumstantial evidence suggests that CDI occurs because of perturbations in enteric microbial ecology. First, and most compellingly, and as noted above, antibiotics remain strong and independent risk factors for CDI. A recent Canadian study vividly demonstrates this risk: one in every 67 adults who received a peri-operative (i.e., once or twice only) antibiotic developed symptomatic CDI (Carignan et al., 2008). Antibiotics profoundly alter the microbial composition of the human gut (Antonopoulos et al., 2009), but there is considerable inter-individual variation in the extent of this effect (Dethlefsen et al., 2008). Though controversial (Miller, 2009), probiotics might prevent some cases of CDI (McFarland, 2006). Fecal “transplants,” in attempts to alter the human microbial biomass, have been used in severe refractory CDI and to prevent CDI relapses (Nieuwdorp et al., 2008, Persky & Brandt, 2000, You et al., 2008, Tvede & Rask-Madsen, 1989, Aas et al., 2003).

CDI stools have less overall bacterial diversity than control stools, as assessed by 16S rRNA gene sequencing (Chang et al., 2008). However, this study compared only four subjects with an initial episode and three subjects with a recurrent episode of CDI (all of whom had taken antibiotics), and three controls who had not taken antibiotics. Thus, the significance of the diminished bacterial diversity cannot be determined because it could relate simply to antibiotic use. A similar study of 11 patients who develop CDI after outpatient use of antibiotics demonstrated (by 16S rRNA gene sequencing and temporal temperature gradient gel electrophoresis), some potentially “permissive” antecedent (to the antibiotics) bacteria that might facilitate colonization with, or expansion of, asymptomatically carried C. difficile (De La Cochetiere et al., 2008).

Veterinary and Foodborne Aspects of CDI

C. difficile is a demonstrated enteric pathogen to many different animal species, including horses, hares, pigs, nonhuman primates, dogs, cats, ostriches and laboratory animals such as Syrian hamsters, mice, rats, and guinea pigs (Keel & Songer, 2006). In horses, C. difficile causes acute colitis in mature horses treated with antibiotics and foals (Baverud, 2004), and has been described as a nosocomial pathogen in equine veterinary hospitals (e.g.,(Baverud et al., 1997, Madewell et al., 1995)/ C. difficile has been documented as a major cause of neonatal enteritis in piglets since 2000 (Songer & Anderson, 2006). The case definition of porcine CDI includes piglets of 1–7 days of age with a history of scouring since shortly after birth (Songer & Anderson, 2006). More recently, C. difficile has been implicated as a causative agent of diarrhea in calves (Hammitt et al., 2008).

C. difficile is also found in apparently healthy animals, including food animals such as poultry (Indra et al., 2009, Zidaric et al., 2008), swine (Indra et al., 2009, Avbersek et al., 2009, Norman et al., 2009), and cattle (Indra et al., 2009), and household pets (Weese et al., 2009d). As in humans, young animals have greater colonization rates than older animals (Keel & Songer, 2006). A longitudinal study in one swine operation reported a 50% colonization of suckling piglets, but only 8.4% in weaned pigs and 3.9% in grower-finisher pigs (Norman et al., 2009). Little is known of the epidemiology and genotype distribution of C. difficile in animal populations. Studies that have genotyped animal isolates suggest lower strain diversity in animal populations compare to human populations (Avbersek et al., 2009, Keel et al., 2007). Keel et al. (2007) ribotyped C. difficile from piglets, calves, horses and dogs. Porcine isolates comprised four PCR ribotypes; one (ribotype 078), a toxinotype V strain, predominated (83% from a total of 144 isolates). This was also the most common ribotype (94%) among 33 calf isolates but was rarely identified in dogs and horses (Keel et al., 2007). In humans, ribotype 078 has caused severe cases of CDI at the community level in young populations (Goorhuis et al., 2008a). The C. difficile type 078 isolates from humans and pigs were highly genetically related, which suggested that pigs may be a potential reservoir for this strain (Goorhuis et al., 2008a).

The increasing recognition of human C. difficile as a community-acquired infection has prompted examination of its potential sources at the community level, including potential foodborne origins.A well recognized risk factor for community-onset C. difficile infection is recent discharge from a hospital, a risk that extends to 12 weeks after discharge (McDonald et al., 2007). The median incubation period for C. difficile is 2 to 3 days and stays in the hospital are becoming shorter (Clabots et al., 1992). Therefore it is possible patients recently discharged from a hospital remain at increased risk for developing CDI, but are acquiring their pathogenic C. difficile in the community after discharge (Weese, 2010). In the mid-1990’s, C. difficile was identified in spoiled vacuum packed meat products (Broda et al., 1996). Since then, C. difficile has been found in both animal and vegetable food products. C. difficile spores have been isolated in raw meat for dogs (Weese et al., 2005), in meats prepared for human consumption (Rodriguez-Palacios et al., 2007), as well as in vegetables in South Wales (al Saif & Brazier, 1996) and ready to eat salads in Scotland (Bakri et al., 2009). The strains more commonly found in animal products are toxinotype V (Songer et al., 2009, Weese et al.). This is notable because toxinotype V strains of C. difficile cause proportionately more cases of community onset infections than hospital onset infections (Limbago et al., 2009). However, in a Canadian study, meat C. difficile isolates contained genes encoding TcdA, TcdB and CDT belonged to toxinotype III (Rodriguez-Palacios et al., 2007). Interestingly,, these strains have the tcdC deletion associated with hyperexpression of TcdA and TcdB. Follow-on studies in the United States (Arizona) and Canada have expanded the list of foods containing C. difficile (ground beef, turkey, pork, and sausages and braunschweiger), and have suggested possible seasonality in the prevalence of contaminated product, with higher prevalence (20 % of retail meat) in winter (Rodriguez-Palacios et al., 2009). The density of C. difficile contamination is low (<60 spores/g) (Weese et al., 2009c).

Household pets might be another source of community-acquired CDI, and human-pet transmission might be common. A recent study documented 10% of dogs and 21% of cats were colonized with C. difficile (Weese et al., 2009d). The most prevalent strain to colonize the pets was ribotype 001, the most common strain of C. difficile in the hospital setting prior to the emergence of the current epidemic strain. Having an owner identified as being immunocompromised was associated with almost an 8-fold increased association of the pet being colonized. This is notable because dogs that enter healthcare facilities (i.e. “pet therapy”) are 2.7 times more likely to be colonized with C. difficile compared to dogs that participate in animal assisted interventions that do not enter healthcare facilities (Lefebvre et al., 2009). Dogs that entered healthcare facilities that licked patients or accepted treats were more likely to be colonized with C. difficile compared to dogs who did not lick patients or accept treats as well (Lefebvre et al., 2009).

Studies have identified strains isolated in food that are genetically identical to strains isolated from humans with CDI as well as isolates from humans with CDI genetically identical to strains isolated from animals, which has suggested that food animals may be a source of food contamination with C. difficile (Goorhuis et al., 2008b, Jhung et al., 2008). Nevertheless, the prevalence of some important human strains, such as the ribotype 027/NAP1, appears to be disproportionately high in food compared to food animals (Weese, 2010). Therefore, it has been postulated that other sources such as slaughterhouse environments, the processing facility, or the hands of personnel manipulating meat may also contribute to the contamination of food with C. difficile spores (Weese, 2010). C. difficile spores are also isolated in other environments including households, soil, pets, water (al Saif & Brazier, 1996, Weese et al., 2009d), and therefore, the relative importance of food as a source of C. difficile remains unclear.

Table 2 reviews isolation rates and organism characterization from recent world-wide surveys, and is derived from a recent and thorough review of this topic (Weese, 2010). It should be noted, however, that isolation techniques for C. difficile from food is far from standardized, and differences in procedures might account for recovery rate differences between studies.

Table 2.

(from (Weese, 2010) with permission).

Prevalence of isolation and ribotype distribution of Clostridium difficile from food animals and retail meat

Country	Sample type	Prevalence (%)	Ribotype 027/ toxinotype III (%)	Ribotype 078/ toxinotype V (%)
Canada (Rodriguez-Palacios et al., 2006)	Calves	15	12	26
USA (Hammitt et al., 2008)	Calves	25	0	94
Canada (Costa et al., 2009)	Veal calves	49	0/1	65
Slovenia (Pirs et al., 2008)	Calves	1.8	0	0
Austria (Indra et al., 2009)	Cows	4.5	0	0
Slovenia (Zidaric et al., 2008)	Chickens	62	0	0
Austria (Indra et al., 2009)	Chickens	5	0	0
Zimbabwe (Simango & Mwakurudza, 2008)	Chickens	29	NT	NT
Slovenia (Pirs et al., 2008)	Piglets	52	0	0/77
USA (Yaeger et al., 2007)	Piglets	79	NT	NT
USA (Keel et al., 2007)	Piglets	NA	0	83
Austria (Indra et al., 2009)	Pigs	3.3	0	0/50
Canada (Weese et al., 2009b)	Piglets	95	0	94
Canada (Rodriguez-Palacios et al., 2007)	Beef, veal	20	0/67	0
USA (Songer et al., 2009)	Various	42	27	73
Canada (Rodriguez-Palacios et al., 2009)	Beef, veal	6.1	0/27	0
Slovenia (Pirs et al., 2008)	Piglets	52	0	0/77
USA (Yaeger et al., 2007)	Piglets	79	NT	NT
USA (Keel et al., 2007)	Piglets	NA	0	83
Austria (Indra et al., 2009)	Pigs	3.3	0	0/50
Canada (Weese et al., 2009b)	Piglets	95	0	94
Canada (Rodriguez-Palacios et al., 2007)	Beef, veal	20	0/67	0
USA (Songer et al., 2009)	Various	42	27	73
Canada (Rodriguez-Palacios et al., 2009)	Beef, veal	6.1	0/27	0
Canada (Metcalf et al., 2009)	Pork	1.8	43/57	0
Canada (Weese et al., 2009a)	Chicken	15	0	96
Canada (Weese et al., 2009b)	Pork Beef	12 12	7.1/14 7.1	71 86

NT, typing was not performed; NA, not applicable, as the study was an evaluation of previously collected isolates.

The above data make a case for the foodborne origin of human C. difficile strains. Systematic food and animal surveys will enable us to confirm or refute this postulated linkage.

Conclusion

CDI poses clinical and infection control problems to degrees not anticipated a decade ago. CDI’s implications compel us to include broad sets of disciplines to try to control this pathogen and the diseases it causes. Expertise from the fields of diagnostic microbiology, nursing, hospital management and infection control, pharmacology, microbial pathogenesis, genome biology, microbial ecology, microbial pathogenesis, and therapeutics development will be needed to help clinicians mitigate the effects of C. difficile once this organism gains access to human hosts. It appears likely that food plays a role in transmission and colonization of hosts by C. difficile, and, by implication, there will be an animal and/or environmental reservoir for this pathogen. We predict that the interface between veterinary microbiology and ecology, food technology and science, and human biology will be a fertile and worthwhile area for collaboration and research C. difficile.

Figure 3.

Resected colon from a patient with CDI fulminant colitis (from (Tarr et al., 2009, Tarr et al., 2008) with permission).

Table 1.

Treatment of Recurrent CDI (modified from (Leffler & Lamont, 2009) with permission).

Treatment of Recurrent C. difficile Infection
Initial recurrence
14-day course of oral metronidazole or vancomycin
Second Recurrence
Tapered pulse dose oral vancomycin
_ 125 mg 4 times daily for 1 week
_ 125 mg twice daily for 1 week
_ 125 mg daily for 1 week
_ 125 mg every other day for 1 week
_ 125 mg every third day for 2 weeks
Consider 1-month course of probiotics starting in the final 2 weeks of antibiotic therapy
Third or subsequent recurrence
Tapered pulse dose oral vancomycin (see above)
Followed by 14-day course of rifaximin, nitazoxanide, or toxin-binding resins
Consider 1-month course of probiotics starting in the final 2 weeks of antibiotic therapy
Consider intravenous immunoglobulin or fecal bacteriotherapy
Consider chronic low-dose suppressive therapy with oral vancomycin for elderly patients and those with multiple comorbidities

Three bullet points.

• Clostridium difficile infection frequency and severity are increasing.

• Diagnosing, treating and preventing C. difficile infection remain a challenge.

• More research is needed to define the pathobiology of C. difficile infection and to identify the major determinants and sources of C. difficile transmission and acquisition in the hospital and in the community