The host immune response to Clostridium difficile

Clostridium difficile is the most common cause of nosocomial bacterial diarrhoea in the Western world. Diarrhoea and colitis are caused by the actions of toxins A and B released by pathogenic strains of C. difficile. Adaptive immune responses to these toxins influence the outcomes of C. difficile infection (CDI). Symptomless carriers of toxinogenic C. difficile and those with a single episode of CDI without recurrence show more robust antitoxin immune responses than those with symptomatic and recurrent disease. Immune-based approaches to CDI therapy and prevention have been developed using active vaccination or passive immunotherapy targeting C. difficile toxins. Innate immune responses to C. difficile and its toxins are also central to the pathophysiology of CDI. An acute intestinal inflammatory response with prominent neutrophil infiltration and associated tissue injury is characteristic of CDI. Furthermore, inhibiting this acute inflammatory response can protect against the intestinal injury that results from exposure to C. difficile toxins in animal models. Studies examining host risk factors for CDI have led to validated clinical prediction tools for risk of primary and of recurrent disease. Risk factors associated with severe CDI with poor clinical outcomes have also been identified and include marked elevation of the peripheral white blood cell count and elevated creatinine. However, further work is needed in this area to guide the clinical application of new approaches to disease prevention and treatment including new antimicrobials as well as passive and active immunization.


Introduction
Clostridium difficile infection (CDI) usually occurs as a complication of nosocomial antibiotic use and is mediated by toxins A and B released by pathogenic strains of the bacterium.The host response to CDI varies from symptomless carriage, to mild self-limited diarrhoea, to recurrent CDI, to severe, life-threatening, colitis and, in increasing numbers of cases, to death (Kelly & LaMont, 2008).There is now considerable evidence to demonstrate that host immune and inflammatory responses contribute in large part to these different disease outcomes.Age, underlying and co-morbid disease, drug therapy and adaptive and innate immune responses all influence CDI severity and course.This review will examine several aspects of the host response to CDI, with an emphasis on the adaptive immune response, as this has been a focus of recent advances in our understanding of disease pathogenesis and is a very active area for new drug development.The potentially crucial role played by innate immune responses in CDI has drawn relatively little attention from clinical researchers but may prove to be an important target for adjuncts to antibiotics, especially in severe disease.Finally, we will discuss clinical prediction tools for CDI as a practical means to harness our advancing knowledge of host factors that determine disease outcomes.
The ability to accurately identify those at highest risk for severe or recurrent disease will allow us to target current and emerging interventions to prevent and more effectively treat CDI.

Natural adaptive immunity to C. difficile and its toxins
Antibodies to C. difficile in the general population Many healthy older children and adults (~60 %) have detectable serum IgG and IgA antibodies to C. difficile toxins A and B, even in the absence of C. difficile colonization or active infection (Kelly et al., 1992;Nakamura et al., 1981;Sa ´nchez-Hurtado et al., 2008;Viscidi et al., 1983).It is likely that antibody production is stimulated in infancy and perpetuated through adult life by environmental exposure to C. difficile itself or to other clostridial species, such as Clostridium sordellii, which possesses crossreacting antigens (Viscidi et al., 1983).

Antibodies to C. difficile and asymptomatic carriage
Early studies suggested that immune responses to C. difficile and its toxins played a role in disease presentation and course (Aronsson et al., 1985;Johnson et al., 1992;IP: 54.70.40.11On: Thu, 25 Oct 2018 15:15:08 Mulligan et al., 1993;Warny et al., 1994).We subsequently tested this hypothesis in a prospective study of 271 hospitalized antibiotic recipients in whom we measured serum and faecal antibody responses to C. difficile toxins A, B and non-toxin antigens over time (Kyne et al., 2000).We found that at the time of colonization, asymptomatic carriers of C. difficile had significantly higher serum IgG antibody levels to toxin A compared with colonized patients who later developed diarrhoea.Importantly, serum IgG antitoxin A levels were also significantly correlated with IgG levels against toxin B and non-toxin antigens, which were also higher in asymptomatic carriers; however, this difference was not statistically significant, possibly due to small sample sizes.Serum antibody levels were not affected by age, co-morbidity or severity of underlying disease; however, the effect of higher serum IgG antitoxin A levels in protecting against diarrhoea was less marked in patients who were severely ill.This moderation of the effects of protective immunity in severely ill patients may have implications for future immunotherapy-based treatments, which may be less effective in very ill patients.Faecal antibodies were not associated with protection against symptoms in this study but were in a previous study (Kyne et al., 2000;Warny et al., 1994).
More recently, Sa ´nchez-Hurtado et al. (2008) found that serum IgG and IgM antibody levels against crude and purified toxin A were similar in colonized patients with and without symptoms and matched controls.However, the authors acknowledge that they were unable to establish the time of colonization and that antibody levels may change over time, as has been demonstrated (Kyne et al., 2000).

Antibodies to C. difficile and recurrent disease
In patients with CDI, IgM levels against toxins A, B and non-toxin antigens, measured on the third day after the initial onset of diarrhoea, were significantly higher in patients who had a single episode of diarrhoea than in patients who later developed recurrent CDI (Kyne et al., 2001).Higher day 12 IgG antibodies to toxin A were also associated with protection against recurrence.Katchar et al. (2007) found that IgG2 and IgG3 subclasses of antitoxin A IgG antibodies were deficient in patients with recurrent CDI.
Taken together, these studies suggest that patients who become colonized with C. difficile who can boost an anamnestic systemic immune response to C. difficile toxins are less likely to develop symptoms.Likewise, symptomatic patients who can mount an immune response early in the course of their illness are less likely to have recurrent CDI.The immune responses to toxin A, B and non-toxin antigens are correlated.This observation is important as effective protection against initial symptoms or recurrence is likely to involve strategies aimed at inducing immunity to both toxins and to nontoxin antigens.

Antibody responses to non-toxin antigens
To further evaluate the role of the host immune response to C. difficile antigens, other than toxins A and B, several investigators have examined the role of antibody response to surface-layer proteins (SLPs) (Drudy et al., 2004;Sa ´nchez-Hurtado et al., 2008;Wright et al., 2008).SLPs have been shown to have in vitro and in vivo adhesive properties and to play a role in bacterial adhesion (Calabi et al., 2002).Therefore, they may play an important role in intestinal colonization and persistence of infection with C. difficile.Drudy et al. (2004) found that antibody levels to SLPs were similar in patients with CDI, asymptomatic carriers and controls.However, patients with recurrent CDI failed to mount an IgM immune response to SLPs compared to patients with a single episode of CDI.In another study, IgG but not IgM levels to SLPs were similar in cases and carriers, but higher compared to controls (Sa ´nchez-Hurtado et al., 2008).Outcomes in terms of recurrence were not examined in that study.Pe ´chine ´et al. (2005a) found that the bacterial flagellar proteins FliC and FliD, and the surface-associated proteins Cwp66 and Cwp84, were expressed during the course of infection and that antibodies against these proteins could be detected after CDI diagnosis and for at least 2 weeks.

Immunization against C. difficile
The results of the observational studies presented above demonstrate the relevance of antibodies both against toxin and against non-toxin antigens in influencing the outcomes of CDI.These studies have also provided scientific rationale for the development of antibody products and vaccines against C. difficile.The efficacy of active and passive immunity to C. difficile toxins in protection against disease has been explored in various small animal models of CDI and also in humans.Vaccines against surface proteins, aimed at preventing or reducing colonization, are also being developed and tested in animals.

Passive immunization against C. difficile toxins in animals
Oral and parenteral administration of antitoxin antibody formulations has been shown to protect against diarrhoea and death in hamsters and mice.Anti-C.difficile bovine immunoglobulin concentrate (BIC), prepared from the colostrum of cows immunized with culture filtrate toxoid, protected against diarrhoea and death in hamsters before and during clindamycin/C.difficile challenge (Kelly et al., 1996;Lyerly et al., 1991).Kink & Williams (1998) immunized hens, with various recombinant peptides spanning the entire C. difficile toxin proteins, for the production of egg IgY antibodies, which were administered orogastrically to hamsters.They found that neutralizing antibodies to both toxins A and B provided complete protection from diarrhoea and death in this model.Parenteral passive immunotherapy is also effective in animals.In one study, Corthier et al. (1991) administered a monoclonal antibody against the putative binding region of toxin A to gnotobiotic mice.The mice were protected against diarrhoea and death following challenge with toxinogenic C. difficile.More recently, Babcock et al. (2006) developed neutralizing, fully human monoclonal antibodies (HumAbs) directed against either toxin A or toxin B by immunizing mice transgenic for human immunoglobulin genes with inactivated toxins.Antitoxin A HumAbs reduced mortality in the hamster model; however combination intraperitoneal administration of HumAbs against both toxin A and B resulted in the best protection against mortality and recurrence.

Active immunization against C. difficile toxins in animals
Protection against lethal CDI by active immunization with C. difficile toxoids was shown in hamsters as early as the 1980s (Fernie et al., 1983;Kim et al., 1987;Libby et al., 1982).Torres et al. (1995) later found that combined parenteral and mucosal immunization with a formalininactivated culture-filtrate C. difficile vaccine provided complete protection against CDI in hamsters.Using a more purified toxoid preparation, Giannasca et al. (1999) demonstrated that parenteral administration induced high levels of serum toxin A and B neutralizing activity and protected hamsters against diarrhoea and death.Immunity could also be transferred to naı ¨ve hamsters using serum or ascites, demonstrating that circulating antitoxin antibodies mediated the effector functions in this model of CDI.
While these previous animal models examined parenteral and mucosal immunization routes, a more recent study assessed whether transcutaneous immunization (TCI) could induce immune responses against toxin A in mice (Ghose et al., 2007).In a novel approach to immunization against C. difficile, the investigators demonstrated that TCI with a toxoid derivative of toxin A and immunoadjuvant cholera toxin induced serum anti-C.difficile toxin A IgG responses.
Recombinant antigens have also been examined as potential vaccine candidates.Immunization with recombinant polypeptides from the cell-binding domain of toxin A partially protected against diarrhoea and death in hamsters (Lyerly et al., 1990).In a murine model, these recombinant polypeptides induced local and systemic neutralizing antibody responses after intra-nasal immunization along with adjuvant heat-labile toxin from Escherichia coli (Ward et al., 1999).Various attenuated live vectors have been investigated in an attempt to improve delivery of these recombinant polypeptides to the intestinal mucosa.Ryan et al. (1997) found that protective immunity against C. difficile toxin A (TcdA) was induced in rabbits by oral administration of the nontoxic repeated binding sequence of toxin A expressed as a fusion protein in an attenuated Vibrio cholerae vector strain.
A recent development in active immunization of animals has been the development of a DNA vaccine targeting the receptor-binding domain of TcdA (Gardiner et al., 2009).A synthetic gene was created encoding the receptorbinding domain of TcdA and optimized for expression in human cells.Parenteral administration of this DNA vaccine induced neutralizing antibodies to CDA and protected mice from death.
The need to target both toxin A and B for effective immunotherapy Many of the animal studies reviewed above have demonstrated that immunization against both toxins A and B, rather than against toxin A or toxin B alone, affords the best protection against CDI (Babcock et al., 2006;Fernie et al., 1983;Kim et al., 1984;Kink & Williams, 1998;Libby et al., 1982).Several investigators have focused on inducing protection against C. difficile toxin A alone because previous animal models had suggested that toxin B, although a potent cytotoxin, was not enterotoxic in animals (von Eichel-Streiber et al., 1996).In the hamster and murine model, toxin A, but not toxin B, was lethal when administered intragastrically (Lyerly et al., 1985).Purified toxin B did not cause any symptoms of disease unless intestinal damage was present or the toxin was coadministered with toxin A.
More recently there has been a paradigm shift in the understanding of the role of toxin B in CDI.Using isogenic tcdA and tcdB mutants of C. difficile, investigators have shown that both toxin A and toxin B contribute to disease pathogenesis in animal models (Kuehne et al., 2010;Lyras et al., 2009).Studies using human colon have shown that toxin B is a more potent enterotoxin than toxin A in human colon (Riegler et al., 1995;Savidge et al., 2003).Furthermore, human strains of toxin A-negative, toxin Bpositive C. difficile have emerged which have been associated with sporadic and epidemic CDI (Drudy et al., 2007).Taken together, these findings suggest that toxin B plays an important role in the pathogenesis of CDI, but do not disprove the role of toxin A in human disease.Thus, it is important that vaccines and passive antibody products that protect against and treat CDI in humans target both toxins.
Passive immunization against C. difficile toxins in humans (i) Oral passive immunization.There have been few studies examining the role of oral administration of anti-C.difficile antibodies for the prevention or treatment of CDI in humans.Despite promising results in animals and in a small human study, anti-C.difficile BIC has not been developed further for human use (Kelly et al., 1996;Warny et al., 1999).Using a similar approach, van Dissel et al.
(2005) developed an anti-C.difficile bovine immunoglobulin secretory IgA concentrate (40 % immune whey protein concentrate; Mucomilk) prepared from the milk of cows immunized against C. difficile and its toxins.Following successful studies in hamsters, the investigators found in an uncontrolled pilot study that Mucomilk was safe and well-tolerated in 16 patients with CDI (van Dissel et al., 2005).None of the patients treated experienced recurrent CDI.In a later randomized, double-blind study, Mucomilk was found to be as effective as metronidazole in reducing recurrent CDI (Mattila et al., 2008).Unfortunately, this study was terminated early because of a lack of funding.
(ii) Intravenous passive immunization.Normal pooled immunoglobulin preparations contain substantial amounts of both IgG antitoxin A and B and demonstrate toxin neutralizing activity in vitro.In an early case series of five children with relapsing CDI and low antitoxin antibody levels, intravenous immunoglobulin (IVIG) treatment resulted in resolution of their symptoms in association with an increase in antitoxin antibody levels (Leung et al., 1991).Since that time, IVIG has been used to treat many cases of severe or refractory CDI with mixed results (Abougergi et al., 2010;O'Horo & Safdar, 2009;Wilcox, 2004).Different doses of IVIG (150-400 mg kg 21 ) administered at varying frequencies (one to three doses), with or without adjunctive treatment with oral anti-C.difficile antibiotics, have produced varying results.Despite these many case reports and observational studies, there have been no randomized controlled clinical trials of IVIG for CDI.Likewise, there is no standardized algorithm for its use.Most recently, Abougergi et al. (2010) reported a high mortality rate (57 %) in 21 patients treated with IVIG for severe CDI.The authors found that higher Acute Physiology and Chronic Health Evaluation II (APACHE II) scores on the day of IVIG infusion were significantly associated with higher mortality.They suggest that IVIG may have a role to play in patients with severe CDI that is confined to the colon but that once extracolonic organ dysfunction and systemic inflammatory response syndrome develop it may be less beneficial.This is in keeping with our earlier observational studies of the immune response to C. difficile in humans, in which we also found that critically ill patients were less likely to be protected against symptomatic CDI despite having high antitoxin antibody levels compared to less severely ill patients with similar antibody levels (Kyne et al., 2000).
A recent major advance in passive immunotherapy for CDI has been the development of HumAbs against C. difficile toxins A and B for use in humans (Leav et al., 2010;Lowy et al., 2010).Following demonstration of efficacy in hamsters and safety in a phase I study in healthy volunteers, these HumAbs were tested in a large multicentre, randomized, double-blind, placebo-controlled trial in which the primary end point was recurrence of CDI (Babcock et al., 2006;Leav et al., 2010;Lowy et al., 2010).The HumAbs were administered intravenously as an adjuvant to standard antimicrobial therapy in patients with symptomatic CDI.HumAb treatment was associated with a 72 % relative reduction in recurrence rates compared to placebo but did not reduce the severity of infection or the duration of diarrhoea or hospitalization (Lowy et al., 2010).Patients with multiple recurrences had an 82 % relative reduction in recurrence compared to the placebo group.The anti-C.difficile HumAbs were as safe and as well-tolerated as placebo and had circulating halflives up to 26 days.Future studies are needed to determine whether monoclonal antibodies will be cost-effective for the treatment of initial cases of CDI.They may also be useful in patients with refractory or severe CDI or may be beneficial as prophylaxis in very high risk patients such as those with prolonged hospitalizations or in intensive care unit (ICU) settings.
These recent human studies on the use of HumAbs against C. difficile toxins are important for several reasons.First, they lend strong support to a cause and effect association between high serum antitoxin concentrations and protection against CDI.Second, they demonstrate the protective efficacy of passive immunization.Finally, they confirm the role of serum IgG antitoxin in protection against an infection that is predominantly confined to the large intestine.The mechanism by which serum antibodies protect against toxin-mediated disease in the colon is still not well understood.Antitoxin antibodies may be exuded through an inflamed colonic mucosa, as both toxins A and B increase the permeability of intestinal epithelial cells by disrupting tight junctions (Giannasca et al., 1999;Riegler et al., 1995).An alternative mechanism is the active transport of antitoxin antibodies via the FcRn receptor, which mediates bidirectional transport of IgG across the intestinal epithelium (Yoshida et al., 2006).
Several studies have suggested a role for secretory IgA in protection against C. difficile and its toxins (Kelly et al., 1992;Stubbe et al., 2000;Wada et al., 1980;Ward et al., 1999;Warny et al., 1994).However, the importance of secretory IgA versus systemic IgG responses is unclear and recent studies indicate that systemic IgG responses may be more important in determining clinical outcomes in CDI (Kyne et al., 2000(Kyne et al., , 2001;;Lowy et al., 2010).

Active immunization with a C. difficile toxoid vaccine in humans
The first candidate vaccine against C. difficile that has been tested in humans is a toxoid vaccine containing formalininactivated purified toxins A and B. In healthy volunteers, intramuscular administration of this vaccine was found to be safe, well-tolerated and associated with serum antitoxin antibody levels significantly higher than those observed in natural infection (Aboudola et al., 2003;Kotloff et al., 2001).Vaccination was also successful in treating a small number of patients with recurrent CDI (Sougioultzis et al., 2005).Sanofi-Aventis are currently sponsoring phase II clinical trials of their C. difficile toxoid vaccine for the prevention of CDI (clinicaltrials.govNCT00772343 and NCT01230957).Lee et al. (2010) recently published an economic computer simulation model of the potential value of a C. difficile toxoid vaccine.They found that over a wide range of vaccine costs, vaccine efficacies and C. difficile risks the vaccine would be cost-effective for the prevention of CDI in high-risk patients and also for the prevention of recurrent CDI.In particular, they noted that, as C. difficile risk and vaccine efficacy increased, vaccination became economically dominant, having the potential to save society, third-party payers and hospitals money, while preventing morbidity and mortality.

Future vaccines to reduce or prevent colonization with C. difficile
As a C. difficile toxoid vaccine is unlikely to affect colonization, alternative vaccine candidates are being explored that may reduce or prevent this first step in the pathogenesis of CDI (Brun et al., 2008;Nı ´Eidhin et al., 2008;Pe ´chine ´et al., 2007)

Innate immune responses in CDI
The intracellular mechanism of action of C. difficile toxins A and B is to catalyse the glucosylation of a conserved threonine residue on specific Rho family members (Just et al., 1995).This leads to both activation and inactivation of Rho signalling and ultimately to cell death and a pronounced host inflammatory response (Genth et al., 2008;Sun et al., 2009).Pathogenic, toxin A-negative strains of C. difficile produce toxin B variants with atypical cytotoxic effects including cell rounding and the development of filopodia-like structures that have been linked to glucosylation of R-Ras and RhoA activation, respectively (Chaves-Olarte et al., 2003).It has been suggested that differences in the intracellular effects of toxin B variants produced by toxin A-negative strains (compared to toxin B from toxin A-positive strains) may be responsible for the ability of these strains to cause disease in humans.Innate inflammatory responses that are activated by the glucosyltransferase activity of C. difficile toxins A and B include the NF-kB and MAP kinase pathways (Chae et al., 2006;Chen et al., 2006;Jefferson et al., 1999;Kim et al., 2005;Warny et al., 2000).Monocytes and macrophages are activated and a variety of proinflammatory cytokines including IL-1b, TNF-a and IL-8 are released (Jefferson et al., 1999;Linevsky et al., 1997;Sun et al., 2009;Warny et al., 2000).
The intense acute inflammatory response appears to be a major factor in causing intestinal injury in CDI (Savidge et al., 2003).Acute colitis with pseudomembrane formation and a heavy neutrophil infiltration of the colonic epithelium and submucosa is characteristic of severe CDI.Furthermore, as discussed below, a marked increase in neutrophils in the peripheral blood is also a feature of severe CDI and is linked to prognosis (Cloud et al., 2009;Lamontagne et al., 2007;Loo et al., 2005;Pe ´pin et al., 2007Pe ´pin et al., , 2009)).In one study, neutrophil infiltration of the intestine was prevented in rabbits by using a monoclonal antibody to block CD18, a neutrophil cell surface receptor that mediates firm adhesion to the vascular endothelium, an essential step in neutrophil diapedesis (Kelly et al., 1994).Exposure to C. difficile toxin A caused severe inflammation and tissue destruction of control intestine whereas anti-CD18 pretreatment led to almost complete protection.Interestingly, mild changes remained in the intestinal epithelium of anti-CD18-treated animals with increased numbers of apoptotic epithelial cells.One explanation for these findings is that C. difficile toxins cause direct injury to the intestinal epithelium, which is associated with a robust host inflammatory response with neutrophil activation and recruitment leading in turn to greatly amplified intestinal injury.In fact, a wide range of anti-inflammatory agents can reduce intestinal injury in animal models of CDI (Anton et al., 2004;Chen et al., 2006;Cottrell et al., 2007;Kim et al., 2005Kim et al., , 2007;;Kokkotou et al., 2009;Pothoulakis et al., 1993;Warny et al., 2000).The use of combinations of antibiotic and anti-inflammatory agents to treat severe CDI in humans warrants greater attention and investigation, especially given the rising incidence of severe and fatal disease (McDonald et al., 2006;Redelings et al., 2007).
Clinical prediction tools for CDI: using host factors to predict disease outcomes As discussed above, there is a spectrum of host responses to CDI, ranging from symptomless carriage to fulminant and fatal colitis (Kelly & LaMont, 2008).As the treatment options for CDI expand to include probiotic therapy (possibly including non-toxinogenic C. difficile), newer antibiotics with lower recurrence rates, and immune-based therapies including passive antitoxin infusion and active vaccination, it becomes increasingly important to identify those individuals at greatest risk for recurrent, severe or fatal disease (O'Donoghue & Kyne, 2010;Venuto et al., 2010).In that way, new treatment approaches can be applied to greatest effect and at the lowest cost-to-benefit ratio.Thus, there is renewed interest in the development and validation of clinical prediction tools for CDI, both to guide clinical practice and to aid clinical research and drug development.

Predicting primary CDI
The single greatest clinical problem in CDI is the recent dramatic and sustained increase in disease incidence, which is now three times that seen in 2000 both in the US and in many parts of Europe (McDonald et al., 2006;Redelings et al., 2007).A method to reliably identify those patients at greatest risk for CDI will be valuable in applying prophylactic measures in the form of: (i) more sparing antibiotic use, (ii) use of antibiotics that are associated with lower risk for CDI, (iii) targeted and more extensive measures to reduce exposure to C. difficile and its spores and (iv) use of novel probiotics or other novel agents to protect against CDI (Cohen et al., 2010;Crobach et al., 2009).Identification of high-risk patients can also facilitate the early diagnosis and treatment of CDI when it occurs.
Risk factors for CDI have been described in many studies and include antibiotic use, admission to a health-care facility, prolonged and multiple antibiotic use, prolonged hospital stay, advanced age, increased co-morbidities, gastrointestinal surgery, nasogastric intubation and more (Cohen et al., 2010;Howell et al., 2010;Kelly & LaMont, 2008).However, few studies have attempted to predict the risk for CDI based on combinations of these individual risk factors.We described previously that individuals who receive antibiotics while in hospital are at increased risk for CDI if they have critical underlying disease (Kyne et al., 2002).In this study of elderly, hospitalized antibiotic recipients, those with severe or extremely severe underlying disease (modified Horn's Index of 3 or 4) had incidence rates of CDI of 8.7 % in the derivation cohort and 11 % in the validation cohort (Kyne et al., 2002).It is probable that clinical prediction algorithms that utilize a wider variety of risk factors can be more accurate in disease prediction but the development and testing of such complex prediction tools will require studies in very large at-risk populations.

Predicting severe CDI
In a landmark study, Zar et al. (2007) compared treatment of CDI for 10 days with oral metronidazole 250 mg four times daily to oral vancomycin 125 mg four times daily.Patients were prospectively stratified as having mild or severe CDI based on clinical criteria [severe defined as two points or greater; one point each for age over 60 years, temperature above 38.3u C (101 u F), serum albumin of less than 2.5 mg dl 21 , peripheral white blood cell count above 15 000 cells mm 23 ; two points for pseudomembranous colitis at colonoscopy or requirement for ICU care].In patients with mild CDI, responses to metronidazole (90 %) and to vancomycin (98 %) were similar (P50.36).In severe CDI, response rates to metronidazole (76 %) were significantly lower than to vancomycin (97 %, P50.02).These study findings indicate that vancomycin is more effective than metronidazole in severe CDI and have led to recommendations that oral vancomycin be used as firstline therapy in severe CDI (Cohen et al., 2010).
Although the study by Zar et al. (2007) showed that clinical criteria can predict response to therapy in CDI, the scoring system used was not formally evaluated or validated.Other investigators have also identified several clinical parameters that are associated with poor clinical outcomes in CDI, such as a requirement for ICU admission, colectomy or death (Cloud et al., 2009;Lamontagne et al., 2007;Loo et al., 2005;Pe ´pin et al., 2004Pe ´pin et al., , 2007Pe ´pin et al., , 2009)).Several studies have identified an elevated peripheral white blood cell count and a rising serum creatinine concentration as important predictors of poor outcomes in CDI (Cloud et al., 2009;Lamontagne et al., 2007;Loo et al., 2005;Pe ´pin et al., 2004Pe ´pin et al., , 2007Pe ´pin et al., , 2009)).Additional prospective studies are needed to develop and validate clinical prediction tools for severe CDI in order to guide choice of therapy and optimal timing of surgery (Cohen et al., 2010).
We recently developed and validated a prediction tool for recurrent CDI based on three clinical factors: (i) age greater than 65 years, (ii) severe or extremely severe underlying disease (modified Horn's Index 3 or 4) and (iii) continued antibiotic use (i.e.other than metronidazole or vancomycin) after CDI therapy (Hu et al., 2009).The clinical prediction rule discriminated between patients with and without recurrent CDI, with an area under the curve of the receiver operating characteristic curve of 0.80 (95 % CI: 0.67-0.92) in the validation cohort.The rule correctly classified 72 % of patients as having a single episode of CDI versus recurrence.Thus, a relatively simple and validated tool that is based on readily available clinical parameters can be used to predict recurrent CDI and to identify highrisk patients most likely to benefit from measures to prevent recurrence.
In another study, we found that serum IgG antibodies against C. difficile toxins A and B were higher in subjects with a single episode of CDI than in those with recurrent disease (Kyne et al., 2001).However, measuring serum antitoxin concentrations added little to the clinical parameters listed above in predicting recurrent CDI (Hu et al., 2009).

Conclusions
There are several notable unmet medical needs in CDI.The first and most fundamental is the alarming increase in the number of cases reported in recent years.The second is the marked worsening of disease severity which, compounded to the increased incidence, has led to sharp rises in CDIrelated mortality.The third is the ever-present problem of recurrence.Adaptive immune responses, especially against C. difficile toxins A and B, are now known to influence disease expression and course.Both active vaccination and passive immunotherapy approaches to CDI prevention and treatment show substantial clinical promise.Conversely, innate immune responses to CDI appear to be counterproductive and to exacerbate toxin-mediated intestinal injury.Agents to temper the innate immune response, used in conjunction with antibiotics, hold the potential to improve clinical outcomes in severe disease.Validated clinical prediction tools, based on host characteristics, will be useful to identify those at greatest risk for complications of CDI and therefore the best candidates for current and future prophylactic and therapeutic interventions.
. Pe ´chine ´et al. (2005a, b) have evaluated C. difficile surface proteins, including the flagellar cap protein FliD and Cwp84, as vaccine antigens in a human flora-associated mouse model.After rectal immunization with different antigen combinations and subsequent toxigenic C. difficile challenge, colonization levels of C. difficile were significantly lower in mice who received the antigen combinations than in mice who received control PBS.However, immunization did not prevent C. difficile colonization.Similarly, Nı ´Eidhin et al. (2008) found no significant benefit of active immunization of hamsters using SLP.Future multi-component vaccines may be developed which incorporate both neutralization of toxins and prevention or reduction in colonization by C. difficile through targeting surface proteins or other factors needed for colonization and persistence of infection.Preventing colonization may have the added benefit of reducing bacterial carriage, environmental contamination and spread thereby augmenting the beneficial effects of herd immunity to C. difficile.