In February 2019, the authors were separately assigned different topics to address through inductive PubMed searches: (1) pathophysiology of CDI; (2) chemistry and mechanism of action of bezlotoxumab; (3) pharmacology of bezlotoxumab; (4) efficacy of bezlotoxumab in phase 3 randomized controlled trials (RCTs); (5) bezlotoxumab in observational studies; and (6) safety of bezlotoxumab in clinical studies. Then, they were asked to prepare separated drafts related to their assigned research topic. Eventually, the drafts were merged into a complete manuscript to be reviewed and approved by all the authors.
Pathophysiology of CDI
After being ingested, the spores of C. difficile resist the gastric acid and pass through the stomach, ultimately reaching the gut. Once there, C. difficile can persist as spores or germinate into vegetative forms. Germination is dependent on sensing primary bile acids from the liver, recognized by the germinant receptor CspC, and is inhibited by secondary bile acids in the colon . In principle, while the “healthy” gut microbiota converts primary bile acids into secondary bile acids (which inhibit C. difficile germination), a disrupted microbiota following broad-spectrum antibiotic therapy, deficient of primary bile acid converters, may facilitate C. difficile germination and overgrowth. Once germinated, the vegetative forms of C. difficile are capable of producing toxins, the eventual mediators of the biologic damage (Fig. 1).
The pathophysiology of C. difficile relies mainly on the effects of toxin A and toxin B. These are two large proteins that contain a common multi-modular domain structure described as the ABCD model (A: biological activity; B: binding; C: cutting; D: delivery) . The crystal structure of toxin A and toxin B has recently been elucidated and reported [15, 16]. The toxins are encoded by the tcdA and tcdB genes, respectively, located within a region known as the pathogenicity locus or PaLoc, a chromosomally integrated DNA sequence. The PaLoc also contains three other genes: (1) tcdR, encoding an alternative RNA polymerase sigma factor that is responsible for tcdA and tcdB expression; (2) tcdE, encoding a putative holin-like protein necessary for the extracellular release of both toxins; and (3) tcdC, which negatively regulates TcdA and TcdB synthesis . PaLoc can be horizontally transferred to non-pathogenic strains characterized by the lack of tcdA and tcdB, converting them into a pathogenic strains producer .
Toxin A and toxin B bind to receptors on the surface of target cells. The main candidate receptors are glycosphingolipids containing the Galβ1-4GlcNAc motif for toxin A, and chondroitin sulfate proteoglycan 4, poliovirus receptor-like 3, and Wnt receptor frizzled proteins for toxin B . After surface binding, the toxins are internalized through a receptor-mediated endocytosis and hence translocate into the cytosol through a pore-forming mechanism. Once in the cytosol, the toxins undergo an inositol hexakisphosphate-dependent autocatalytic cleavage, with the consequent release of the glucosyltransferase domain (region A), which finally targets Rho proteins. Members of the Rho family of guanosine triphosphatases are hence inactivated, thereby producing cytopathic effects, cytotoxic effects, induction of the programmed cell death, and activation of the inflammasome. Overall, this leads to colonocyte death, loss of intestinal barrier function, and development of neutrophilic colitis [17, 20]. Some bacterial strains may produce a binary toxin called C. difficile transferase. Binary toxin causes depolymerization of F-actin and rearrangement of the actin cytoskeleton, thereby disturbing the dynamic balance between actin and microtubules in target cells . The pathogenic role of the binary toxin is still debated, but several studies have reported an association between binary toxin production and worse outcomes . It has still not been fully elucidated why C. difficile disposes of two similar toxins to exert its pathogenic effects. However, it now seems clear that toxin B, apart from being several-fold more potent than toxin A, is the one more strongly related to CDI pathogenesis . Finally, in addition to the well-known toxin-mediated effects on the gut, attention has recently also been given to the possible extra-intestinal effects of toxins and toxaemia, that are likely implied in systemic manifestations of the disease. For example, cardiotoxic effects of toxins have been described in animal models .
Other factors that significantly contribute to pathogenesis of CDI are: (1) flagellar expression , that is variable among C. difficile strains and contributes to colonization efficiency; (2) the expression of type IV pili  that interact with the intestinal epithelium contributing to C. difficile aggregation and biofilm formation; and (3) the combined action of proteins, such as the adhesin fibronectin-binding protein A, cell wall proteins (e.g. Cwp84), Sl-layer protein A, and its modifying protease Cwp84, which contributes to C. difficile adherence, which have a role in biofilm formation, ensuring an “ecological niche” to the bacterium .
Peripheral leucocytosis is common, especially in severe CDI episodes. Neutrophils are the primary cells that respond to C. difficile invasion, and neutrophil inflammation is the hallmark of CDI. C. difficile toxins (mainly toxin B) activate neutrophils through formyl peptide receptor-1, and generate bactericidal concentrations of reactive oxygen species . Neutrophils can also ingest complement or anti-C. difficile antibody-coated bacteria. However, although useful, these concerted mechanisms also need to be balanced, since they can also fuel tissue damage, and the boundary between “friend or foe” can be narrow . Finally, hypoalbuminemia and hypogammaglobulinemia may be implicated in the pathogenesis of CDI. Indeed, it has been recently been shown that human serum albumin is capable of binding C. difficile toxins, impairing their internalization into the host cells thus reducing the toxin-dependent glycosylation of Rho proteins . In clinical studies, hypoalbuminemia has been associated with mortality and recurrent CDI [31, 32]. With regard to hypogammaglobulinemia, humoral immunity is a major protective mechanism against CDI, and it has been demonstrated that lower antibody titres against toxins predisposes to disease development [33, 34].
Chemistry and Mechanism of Action of Bezlotoxumab
Bezlotoxumab (molecular weight 148.2 kDa) is a fully human IgG1 monoclonal antibody against C. difficile toxin B . It was developed using mice transgenic for human immunoglobulin genes, and exposed to various antigens and adjuvant for 6–12 weeks . Whole toxin A and toxin B toxoids and a recombinant C-terminal fragment of toxin B were used as immunogens and the splenic fusions were performed on mice with potent immune responses. Distinct toxin-reactive hybridomas were then screened according to in vitro and in vivo toxicity assays. Bezlotoxumab was the human monoclonal antibody derived from recombinant C-terminal toxin B fragment immunization .
Bezlotoxumab has been shown to bind and neutralize toxin B. Hernandez et al. assessed bezlotoxumab neutralization potency, as measured in a cell growth/survival assay with purified toxins from various C. difficile strains . The authors showed that bezlotoxumab is active against toxins from all C. difficile strains, although toxins of ribotypes 027 and 078 were bound with lower affinities resulting in lower neutralization potency. The precise mechanism for different affinities is unknown, but it has been speculated that in these ribotypes the Fab region of bezlotoxumab binds to a single epitope, while in other strains it binds two epitopes. Nevertheless, even in 027 and 078 ribotypes, nearly complete toxin neutralization was achieved at concentrations of antibody that were still below plasma concentrations measured in CDI patients, thus the lower affinity against toxins from hypervirulent strains is likely irrelevant .
In 2014, both the mechanism of action of bezlotoxumab and the toxin B epitopes involved in binding the monoclonal antibody were elucidated. Orth et al. demonstrated that bezlotoxumab binds to two epitopes existing in distinct regions within the N-terminal half of the combined repetitive oligopeptide (CROP) domain of toxin B, causing partial obstruction of two of the four putative carbohydrate-binding pockets involved in colonocytes binding . The stoichiometry of bezlotoxumab to toxin B combined repetitive oligopeptide domain is 1:1, suggesting a direct toxin neutralization mechanism, more than a system mediated by large immune complexes . This hypothesis was confirmed 1 year later by Yang et al. through experiments using multiple murine models of CDI . In addition, in 2017, Gupta and colleagues showed that bezlotoxumab binding to the toxin B CROP domain prevented the host receptor (chondroitin sulfate proteoglycan 4) in mammalian host cells from toxin binding .
The transport of bezlotoxumab from the basolateral to the luminal compartment of colonocytes take place through the paracellular path after toxin disruption of the epithelial cells and the intercellular junctions  (Fig. 1). Basically, this observation support the hypothesis that bezlotoxumab could be more effective in patients with severe CDI episodes , since an increased disruption of colonocytes may allow more monoclonal antibodies to reach the gut lumen.
PK/PD of Bezlotoxumab
Currently, bezlotoxumab is approved for the prevention of rCDI in adult patients at high risk for rCDI. The product must be administered during the active CDI antibacterial treatment and is available as 1000 mg/40 mL single-dose vials. Reconstituted vials should be diluted in 0.9% sodium chloride or 5% dextrose to a final concentration between 1 and 10 mg/mL [41, 42]. The recommended dosage is based on the patient body weight, with 10 mg/kg intravenously over 60 min in a single administration [11, 41].
Like other intravenously administered monoclonal antibodies, bezlotoxumab possess a limited extravascular distribution . In patients with CDI receiving a single 10 mg/kg intravenous dose, bezlotoxumab mean volume of distribution was 7.33 L; the geometric mean AUC0–INF was 53,000 mcg per h/mL and the Cmax was 185 mcg/mL [41, 43]. Age, gender, ethnicity, and co-morbid conditions, which typically have only a limited effect on the exposure of therapeutic antibodies, are not expected to affect the exposure of bezlotoxumab [41, 44]. Moreover, no clinically meaningful differences in bezlotoxumab exposure have been observed in patients with renal or hepatic impairment, and therefore no dose adjustment is recommended for patients with renal or hepatic disease . Bezlotoxumab elimination occurs primarily by protein catabolism. The drug undergoes catabolism into smaller peptides, with a mean elimination half-life of 19 days . In a phase 2 randomized, double-blind, placebo-controlled trial on the efficacy of a combination of actoxumab (a monoclonal antibody against toxin A) plus bezlotoxumab in preventing rCDI, after the initial infusion CDI patients showed detectable serum levels of bezlotoxumab for 22 ± 13 days . Bezlotoxumab clearance increases with patient body weight, and the resulting exposure differences are addressed by the administration of a weight-based dose [41, 46].
Like other monoclonal antibodies, bezlotoxumab is eliminated via catabolic pathways, including proteolysis by the liver and reticuloendothelial system, target-mediated elimination, and non-specific endocytosis . Therefore, bezlotoxumab differs from other traditional drugs eliminated through non-catabolic pathways, i.e. liver enzyme-systems like the cytochrome P450 and renal and biliary excretion. Considering the bezlotoxumab elimination by protein catabolism, drug–drug interactions with traditional drugs are not expected. So far, there is no in vivo or in vitro evidence of any drug–drug interaction .
The clinical phase 3 trials, MODIFY I and II, randomized adult patients with recurrent CDI under anti-CDI antimicrobial treatment to receive the addition of actoxumab–bezlotoxumab versus placebo (for details on efficacy endpoints, see the next section) . These clinical trials provided data on pharmacokinetic sampling of a large, diverse population, and Yee and colleagues analysed these data, adopting a population pharmacokinetic modelling approach to assess covariate effects on bezlotoxumab pharmacokinetic . In total, bezlotoxumab concentrations from 1587 participants who received either bezlotoxumab alone or bezlotoxumab in combination with actoxumab were included in the population pharmacokinetic modelling analysis . The study confirmed that co-administration with actoxumab, age, ethnicity, hepatic function, ongoing anti-C. difficile antibiotic treatment, and concomitant proton pump inhibitor use do not significantly alter bezlotoxumab exposure .
Interestingly, Yee and colleagues also estimated albumin levels to positively correlate with bezlotoxumab exposure. The estimated bezlotoxumab exposures was up to 33% lower in patients with albumin levels of < 3.5 g/dL than in patients with normal albumin levels [AUC0–INF geometric mean ratio: 0.67; 90% confidence interval (CI): 0.65–0.69] . Several mechanisms have been proposed to explain the interaction between albumin levels and the clearance of monoclonal antibodies, including the protective effect from lysosomal degradation exerted by the neonatal Fc receptor (FcRn) [43, 48,49,50]. According to the ability of FcRn to rescue both albumin and immunoglobulins from early degradation, factors that affect the recycling capacity of FcRn, i.e. low albumin levels, may influence the pharmacokinetic of monoclonal antibodies, including bezlotoxumab . However, at present, there is no definite evidence that low albumin levels reduce bezlotoxumab exposure to a clinically meaningful extent, and no dose adjustments of bezlotoxumab are recommended in the presence of hypoalbuminemia .
Finally, the required bezlotoxumab gut lumen concentration to effectively inactivate toxin B is not yet known . It is nonetheless of note that a higher bezlotoxumab concentration was observed in a CDI animal model of intestinal lumen toxin-damaged hamster, in comparison to controls with normal intestinal lumen .
Efficacy of Bezlotoxumab in Phase 3 RCTs
MODIFY I and II were two multicentre, double-blind phase 3 RCTs. Adults (≥ 18 years old) with first episode or recurrent CDI and receiving 10–14 days of standard of care antibiotic therapy for CDI (metronidazole, vancomycin, or fidaxomicin) were enrolled. Patients treated with vancomycin or fidaxomicin could also receive intravenous metronidazole . Participants were assigned in a 1:1:1:1 ratio to receive placebo (0.9% saline), actoxumab 10 mg/kg alone (only in MODIFY I), actoxumab 10 mg/kg plus bezlotoxumab 10 mg/kg, or bezlotoxumab 10 mg/kg single dose, respectively. Enrolled patients received a single intravenous infusion of monoclonal antibody or placebo during the treatment period of standard of care for CDI. The primary endpoint of the two studies was the proportion of rCDI during 12 weeks of follow-up in the modified intent-to-treat (mITT) population. The two MODIFY RCTs were independent, and both were adequately powered to assess the primary efficacy endpoint. The design of MODIFY I was adaptive (enrollment in bezlotoxumab or actoxumab arms could be discontinued in the case of inferiority vs. the combined arm in an interim analysis). In fact, this allowed discontinuation of enrollment in the actoxumab arm . As reported above, the actoxumab arm was not included in MODIFY II.
Overall, of 2655 randomized patients, 2559 (96%) were included in the mITT population (1396 in MODIFY I and 1163 in MODIFY II). In MODIFY I, the proportion of patients developing rCDI was lower in the bezlotoxumab (17%, 67/386) than in the placebo arms (28%, 109/395), with an adjusted difference of − 10.1% (95% CI − 15.9 to − 4.3). The same result was observed in MODIFY II [16% (62/395) vs. 26% (97/378), with an adjusted difference of − 9.9%, 95% CI − 15.5 to − 4.3]. The proportion of rCDI was conversely similar when comparing bezlotoxumab plus actoxumab versus bezlotoxumab alone. Indeed, in MODIFY I, the adjusted difference was − 1.4% with 95% CI − 6.7 to 3.9 [15.9% (61/383) vs. 17.4% (67/386), respectively], whereas, in MODIFY II, it was − 0.8% with 95% CI − 5.9 to 4.2 [14.9% (58/390) vs. 15.7% (62/395), respectively]. Similar results were observed in the pooled analysis of the two trials, overall supporting the efficacy of bezlotoxumab for the prevention of rCDI, whereas actoxumab was not efficacious and did not provide any additional benefit when combined with bezlotoxumab. Of note, most rCDI (71%) occurred within 4 weeks. Another aspect worth noting is that 77% of participants had at least one risk factor for rCDI or for a CDI-related adverse outcome. In most of subgroups stratified according to such risk factors (e.g. ≥ 65 years of age, previous CDI episodes, immunocompromised status, severe CDI according to Zar score ≥ 2), , the protective effect of bezlotoxumab was confirmed, whereas the 95% CI crossed the zero in participants with CDI due to ribotype 027 and in those with CDI due to ribotypes 027, 078, or 244, although the direction of the effect was in favour of a protective effect of bezlotoxumab. Only in these latter two subgroups, was the protective effect of bezlotoxumab plus actoxumab possibly increased compared with that of bezlotoxumab alone, although the small subgroup samples preclude definite conclusions .
Several pre-planned/post hoc analyses of the MODIFY RCTs were conducted. An important necessary premise is that several had limited power, which may imply a non-negligible risk of type II error in some of them. In patients at high risk of rCDI (age ≥ 65 years, previous CDI episodes, immunocompromised status, severe CDI according to Zar score ≥ 2, and/or infection by ribotypes 027, 078, or 244), a post hoc analysis confirmed the protective effect of bezlotoxumab versus placebo in patients with a least one risk factor for rCDI, with the greater reduction in risk being observed in patients with at least 3 concomitant risk factors . An increased protective effect of bezlotoxumab in patients at higher risk of rCDI was also suggested in another analysis . In another study, participants in the MODIFY trials with sustained clinical cure at 12 weeks were shown not to develop any rCDI after other 9 months of follow-up (0/69, 0%) versus 2/65 (3%) and 1/34 (3%) in the bezlotoxumab plus actoxumab and placebo groups, respectively . Using whole-genome sequencing, Zeng and colleagues differentiated recurrences due to new infection by a different ribotype (50/259 evaluable patients, 19%) from recurrences due to relapse of infection by the same ribotype of the index CDI episode (198/259 evaluable patients, 76%) . Unknown categorization of the type of recurrence was reported in 11 cases. The authors found that the cumulative incidence of relapses (assessed by means of a competing risk model) was lower in patients receiving bezlotoxumab versus non-bezlotoxumab (actoxumab or placebo), . Compared with placebo, in another post hoc analysis the use of bezlotoxumab was also associated with reduced CDI-associated hospital readmissions in patients at high risk of rCDI [5.1% (27/530) vs. 11.2% (58/520), with difference − 6.1%, 95% CI − 9.5 to − 2.8] . In a cost-effectiveness model based on pooled data from the MODIFY trials, the administration of bezlotoxumab led to a gain of 0.12 quality-adjusted life-years (QALYs) compared with placebo, and seemed cost-effective in terms of the prevention of rCDI in the entire study population, showing an incremental cost-effectiveness ratio of US$19 824/QALY gained . Favourable results were also observed when adapting the cost-effectiveness model to a Spanish setting . In patients enrolled in the MODIFY trials and receiving placebo, endogenous serum antibodies against toxin B were protective against rCDI, whereas endogenous serum antibodies against toxin A were not, this being in line with the protective effect observed for bezlotoxumab but not for actoxumab . Staying on the topic of endogenous antibiotics, the immunogenicity potential of bezlotoxumab has been shown to be low, and no development of treatment-emergent anti-bezlotoxumab antibodies was observed in patients enrolled in registrative studies . With regard to the timing of bezlotoxumab administration, efficacy in preventing rCDI was not influenced by the time of administration with respect to the onset of antibiotic therapy (i.e. 0–2, 3–4 and ≥ 5 days after onset) .
In a post hoc analysis of 44 MODIFY I/II participants with inflammatory bowel disease, treatment with bezlotoxumab showed a trend toward a protective effect when compared with placebo, although the wide 95% CI does not allow for firm interpretations before the conducting of more powered studies on this topic [26.7% (4/15) vs. 53.8% (7/13), with difference − 27.2%, 95% CI − 57.9 to 9.6] . In 382 MODIFY I/II participants with cancer, the proportion of patients developing rCDI was lower in the bezlotoxumab (26/146, 17.8%) than the placebo arms (42/138, 30.4%), with an absolute difference of − 12.6%, 95% CI − 22.5 to − 2.7 . As shown in another analysis, the mean cumulative inpatient-days were lower in the bezlotoxumab (12.1 days) than the placebo arms (14.1 days), with a mean difference of − 2.1 days (95% CI − 3.7 to − 0.4) . An exploratory study investigated if human genetic variations are able to influence the effect of bezlotoxumab in patients enrolled in the MODIFY trials. The single nucleotide polymorphism rs2516513 and the human leukocyte antigen alleles HLA-DRB1*07:01 and HLA-DQA1*02:01, which are located in the extended major histocompatibility complex on chromosome 6, showed an association with a reduced risk of rCDI in patients treated with bezlotoxumab. Notably, the same was not observed in patients receiving placebo . Finally, the protective effect of bezlotoxumab was confirmed in a subgroup analysis of Japanese patients enrolled in the MODIFY trials .
Available meta-analyses also support the use of bezlotoxumab for preventing rCDI. In this regard, Madoff and colleagues evaluated 38 RCTs of different treatments [antibiotics, faecal microbiota transplantation (FMT), monoclonal antibodies, and various prebiotics and probiotics] for the prevention of rCDI . They observed a greater risk reduction with FMT or monoclonal antibodies therapy, although any extrapolation about the relative efficacy of the different interventions should be made with caution because of the very different comparators employed in the included studies. In a Bayesian network meta-analyses of RCTs, a similar protective effect of bezlotoxumab versus (indirect comparison) single or multiple FMT was suggested, although FMT was possibly associated with a higher rate of non-serious diarrhoea as an adverse event (no differences were noticed for other adverse events) .
Bezlotoxumab in Observational Studies
So far, only a few observational studies on the use of bezlotoxumab in real life have been performed (see Table 1). A retrospective, multicentre case series was conducted in the US . Among 200 patients with CDI receiving bezlotoxumab in addition to oral antibiotic therapy, 15.9% developed rCDI within 90 days. The median age of patients was 77 years, and the median Charlson Comorbidity Index was 5. A higher risk of rCDI was observed in patients with ≥ 2 previous recurrences before receiving bezlotoxumab (hazard ratio 2.77, 95% CI 1.14–6.76, p = 0.025) . Another multicentre, retrospective case series was conducted in five university hospitals in Finland . The study included 46 CDI patients who received a standard dose of bezlotoxumab. Their mean age was 66 years (range 15–97 years). Based on a Zar score, 18/46 (39%) had severe CDI. In addition, 28/46 (61%) were immunocompromised. As many as 36/46 (78%) had ≥ 3 risk factors for rCDI. Notably, 28/46 (61%) received concomitant antibacterial treatment for infections other than CDI. Two patients died before 3 months of follow-up after bezlotoxumab infusion. At the end of the 3-month follow-up, 32/44 (73%) patients remained free of rCDI. A similar result was observed in immunocompromised patients (71%). In severe CDI cases, the proportion of patients with no rCDI at 3 months was 63% . Of note, eight were waiting for faecal microbiota transplantation but all remained free of recurrence and did not need the procedure. A prospective observational study assessing the impact of bezlotoxumab on rCDI rate in patients at high risk of recurrence is currently ongoing in five different hospitals in Spain (NCT04075422) .
Finally, successful prevention of rCDI with a combination of bezlotoxumab administration with FMT has been described in a patient with two previous FMT procedures that were unable to prevent rCDI .
Safety of Bezlotoxumab in Clinical Studies
In the MODIFY RCTs, infusion-specific reactions (i.e. adverse events occurring within 24 h of the infusion) were observed in 10.3%, 11.1%, 8.0%, and 7.6% of patients receiving bezlotoxumab, actoxumab, bezlotoxumab plus actoxumab, or placebo, respectively . The most frequent infusion-specific reactions were headache (2%), nausea (2%), fatigue (1%), pyrexia (1%), and dizziness (1%), with equally distributed rates across the study arms. Of note, there was a drug-related discontinuation of bezlotoxumab because of ventricular tachyarrhythmia occurring approximately 36 min after the start of bezlotoxumab infusion. Bezlotoxumab was discontinued and the adverse event resolved [11, 72].
During the 4 weeks after infusion, one or more adverse events, mostly gastrointestinal disorders, were registered in 61.7%, 67.2%, 58.6%, and 61.2% of patients receiving bezlotoxumab, actoxumab, bezlotoxumab plus actoxumab, or placebo, respectively. The rates of drug-related adverse events (with causality being assessed by the blinded investigator) were 7.5%, 7.2%, 6.4%, and 5.9% in patients receiving bezlotoxumab, actoxumab, bezlotoxumab plus actoxumab, or placebo, respectively. Serious drug-related adverse events were observed in 0.5%, 1.3%, 0.6%, and 0.3% in patients receiving bezlotoxumab, actoxumab, bezlotoxumab plus actoxumab, or placebo, respectively. Death during the 12 weeks after infusion (mostly related to infectious events, followed by cardiac disorders) occurred in 7.1%, 11.5%, 6.6%, and 7.6% of patients receiving bezlotoxumab, actoxumab, bezlotoxumab plus actoxumab, or placebo, respectively. Of note, the number of patients with baseline cardiac failure experiencing adverse events, severe adverse events, or death was higher in the bezlotoxumab than the actoxumab plus bezlotoxumab or placebo arms .
In the previously described multicentre, observational case series conducted in Finland, possible bezlotoxumab infusion-related adverse reactions were observed in two patients . One experienced startling sensations after the infusion, and the other one presented with fever the day after the infusion . No infusion-related reactions were observed in the US case series by Hengel and colleagues . Two deaths, conceivably unrelated to bezlotoxumab infusion, occurred during follow-up .