Recent advances in the understanding of antibiotic resistance in Clostridium difficile infection

Abstract

Clostridium difficile epidemiology has changed in recent years, with the emergence of highly virulent types associated with severe infections, high rates of recurrences and mortality. Antibiotic resistance plays an important role in driving these epidemiological changes and the emergence of new types. While clindamycin resistance was driving historical endemic types, new types are associated with resistance to fluoroquinolones. Furthermore, resistance to multiple antibiotics is a common feature of the newly emergent strains and, in general, of many epidemic isolates. A reduced susceptibility to antibiotics used for C. difficile infection (CDI) treatment, in particular to metronidazole, has recently been described in several studies. Furthermore, an increased number of strains show resistance to rifamycins, used for the treatment of relapsing CDI. Several mechanisms of resistance have been identified in C. difficile, including acquisition of genetic elements and alterations of the antibiotic target sites. The C. difficile genome contains a plethora of mobile genetic elements, many of them involved in antibiotic resistance. Transfer of genetic elements among C. difficile strains or between C. difficile and other bacterial species can occur through different mechanisms that facilitate their spread. Investigations of the fitness cost in C. difficile indicate that both genetic elements and mutations in the molecular targets of antibiotics can be maintained regardless of the burden imposed on fitness, suggesting that resistances may persist in the C. difficile population also in absence of antibiotic selective pressure. The rapid evolution of antibiotic resistance and its composite nature complicate strategies in the treatment and prevention of CDI. The rapid identification of new phenotypic and genotypic traits, the implementation of effective antimicrobial stewardship and infection control programs, and the development of alternative therapies are needed to prevent and contain the spread of resistance and to ensure an efficacious therapy for CDI.

Introduction

Clostridium difficile is recognized as the major cause of healthcare antibiotic-associated diarrhea [To and Napolitano, 2014]. Antibiotics used for treating every kind of infection may potentially promote C. difficile infection (CDI). After antibiotic therapy, the protective intestinal microbiota is disrupted allowing ingested or resident C. difficile to colonize the gastrointestinal tract and infect the host. Antibiotic resistance enables C. difficile to grow in the presence of drugs, so strains resistant to multiple agents may have a selective advantage for their diffusion from the usage of these antibiotics.

An alarming increase in incidence of CDI has been observed across the United States, Canada and Europe over the last decade, with a significant financial burden on the healthcare system [Redelings et al. 2007; Burckhardt et al. 2008; Bauer et al. 2011; Gravel et al. 2009; Miller et al. 2011a; Dubberke and Olsen, 2012; Lessa et al. 2012]. These changes have been associated with the emergence of highly virulent (or hypervirulent) strains of C. difficile. The most prominent hypervirulent type is recognized as polymerase chain reaction (PCR) ribotype (RT) 027, North American pulsed field gel electrophoresis type I (NAPI) and restriction endonuclease analysis group B1, depending on the typing method used. Strains RT027, responsible for severe infections, characterized by a high rate of recurrences, mortality and refractory to traditional therapy, have spread globally over recent years [Pépin et al. 2004, 2005b; McDonald et al. 2005; Muto et al. 2005; Goorhuis et al. 2007; Clements et al. 2010]. C. difficile epidemiology is rapidly evolving. In recent years, CDI has emerged as a cause of diarrhea in the community, especially in populations previously considered at low risk, such as young individuals, healthy peripartum women, antibiotic-naive patients or those with no recent health care exposure [Kim et al. 2008; Pituch, 2009; Freeman et al. 2010]. In addition to RT027, a number of emergent highly virulent RTs, correlated to RT027 or not, have recently been identified [Cartman et al. 2010; Valiente et al. 2012]. Among these types, the hypervirulent RT078 has been recognized as a cause of infections in humans, in hospitals [Rupnik et al. 2008; Bauer et al. 2011], in the community [Limbago et al. 2009], and also in animals [Kee et al. 2007; Goorhuis et al. 2008].

Antibiotic resistance plays an important role in driving the current epidemic of CDI and the emergence of new types. A glaring example of this role is represented by the emergence and spread of C. difficile RT027, that has been correlated with the massive use of fluoroquinolones (FQs) and the acquisition of resistance to these antibiotics, a trait not present in historic strains of the same type [He et al. 2013].

C. difficile show a high capability to adapt to the environment through metabolic and genomic changes. Molecular investigations have demonstrated that C. difficile has a versatile genome content, with a wide range of mobile elements, many of them encoding for predicted antibiotic resistances [Sebaihia et al. 2006; He et al. 2010, 2013]. In addition to horizontal gene transfer, other mechanisms may contribute to antibiotic resistance in this pathogen, and recent studies support the multifactorial nature of this phenomenon.

The present paper will review phenotypic and genotypic traits of antibiotic resistance in C. difficile taking into consideration the most recent data.

Antibiotics promoting CDI

Certain antibiotics, such as cephalosporins (CFs), clindamycin (CLI) and, more recently, FQs, are known to carry a higher risk for CDI than others [Gerding, 2004; Slimings and Riley, 2014].

In the 1970s, CLI was recognized as the highest-risk agent for CDI [Bartlett et al. 1977]. Since then, many CDI outbreaks involving CLI-resistant C. difficile strains have been described [Samore et al. 1994, 1997; Johnson et al. 1999]. Antimicrobial stewardship policies were implemented to control the use of CLI in the US and Europe, therefore the attributable risk of CLI-associated diarrhea and CDI was reduced in the following years [Wiström et al. 2001]. CFs had become the antibiotics with the highest relative and attributable risk of CDI, between the 1980s and 1990s for their frequent use in hospitals [Bignardi, 1998]. A decrease in the numbers of patients with C. difficile has been observed in the hospitals that have curtailed the use of these antibiotics [de Lalla et al. 1989; Khan and Cheesbrough, 2003]. Currently, the risk of hospital-acquired CDI remains high for CFs and CLI, so their importance as promoting agents should not be minimized. The rise in the FQ-associated CDI has been concomitant with the increasing incidence of C. difficile RT027 in the early 2000s. Current strains RT027 show high-level resistance to FQs, never observed in historical isolates of the same RT [McDonald et al. 2005]. Although infection control procedures and antimicrobial stewardship have led to a significant reduction in the incidence of CDI by C. difficile RT027, this RT is still predominant in both Europe and the US [Muto et al. 2007; Lessa et al. 2015; Freeman et al. 2015].

C. difficile resistance patterns

Rates of antibiotic resistance vary considerably in the different studies, probably depending on the geographic regions and local or national antibiotic policy. Data extrapolated from 30 studies published between 2012 and 2015 (Table 1) indicate that resistance to CLI and CFs is very common in C. difficile clinical isolates (55% and 51%, respectively), as the resistance to erythromycin (ERY) and FQs (47%). Most of the strains tested for susceptibility to second-generation CFs [cefotetan (CTT) and cefoxitin (FOX)] were resistant (79%), whereas resistance to third-generation CFs [ceftriaxone (CRO) and cefotaxime (CTX)] is present in a lower number of isolates (38%). Similarly, resistance to ciprofloxacin (CIP), a second-generation FQ, is very common in C. difficile (99% of the strains tested), while resistance to moxifloxacin (MXF) and gatifloxacin (GAT), fourth-generation FQs, have been detected in 34% of the strains analyzed for these antibiotics.

Table 1.

Antimicrobial susceptibility of C. difficile clinical isolates from 30 studies published between 2012 and 2015.

Year of isolation	Country	Number of clinical isolates	Percentage of clinical isolates resistant to												Predominant C. difficile types (method different from PCR-	Susceptibility method	References
			CFs						FQs						ribotyping)
			CTT	FOX	CRO	CTX	CLI	ERY	CIP	MXF	GAT	RIF	MTZ	VAN
2000–2009	Germany	34						76	88	68					001, 027	Etest	Reil et al. [2012]
2002–2005	Japan	73			93.2		87.7	87.7	100				0.0	0.0		Etest	Oka et al. [2012]
2006–2007	Canada	432			35.4		50.5			45.1			0.0	0.0	NAP1, 2, 4, 6 (PFGE)	AD	Karlowsky et al. [2012]
2006–2008	Switzerland	86			30.4		52.3		98.8				0.0	1.2		Etest	Büchler et al. [2014]
2008–2009	US and Canada	316					41.5			38		7.9	0.0		002, 017, 027, 053, 078,104, 106	Etest	Tenover et al. [2012]
2009	France	224					34.8	19,2		8			0.0	0.0	014/020/077, 078/126, 015, 002, 005, 027	Etest	Eckert et al. [2013]
2009	Spain	154					74	49	100	43		24	0.0	0.0		Etest	Rodríguez-Pardo et al. [2013]
2000–2009	Korea	120	19				81	80	100	42					001, 018, 017, 014/020	AD	Lee et al. [2014]
2008–2010	South Korea	131					67.9			62.6		19.1	0.0	0.0	018, 017, 001	Etest	Kim et al. [2012]
2005–2010	Taiwan	403					73.5			17.9			0.0	0.5		AD	Liao et al. [2012]
2007–2010	Texas	271		96.3					98.5				13.3	0.0		Etest	Norman et al. [2014]
2008–2010	Poland	17					100	100	100	100	100	0.0	0.0	0.0	027, 176	Etest	Obuch-Woszczatynski et al. [2014]
2008–2010	Hungary	88								19.7		14.9	0.0	0.0		Etest	Eitel et al. [2015]
2008–2010	Hungary	200					29.5	31		21.5		11.5	0.0			Etest	Terhes et al. [2014]
2009–2010	Poland	10										80.0	0.0	0.0	046, 017	Etest	Obuch-Woszczatynski et al. [2013]
2010	US	46					8.3			78.0			0.0		027,001, 014/020,005	Etest	Zhou et al. [2014]
2010–2011	Croatia	23					39	30	96	22	26		0.0	0.0	001, 014	Etest	Novak et al. [2014]
2007–2011	Spain	196						18.0		20.0			0.0	0.0	014, 078, 001	Etest	Weber et al. [2013]
2008–2011	Slovenia	92	1.1	98.9			42.4	13.0	100	11.9		2.2	0.0	0.0	002, 014/020, 012, 029, 046	BD	Pirš et al. [2013]
2010–2011	Iran	75				100	89.3	57.3					5.3	8.0		AD	Goudarzi et al. [2013]
2010–2012	China	60		86.67			73.3			30.0		1.7	0.0	0.0	tr017, tr065, tr014, tr012 (TRST)	AD	Dong et al. [2013]
2011–2012	US	317					19			21			0.0	0.0	078, PA01, PA04	Etest	Varshney et al. [2014]
2011–2012	Europe	953					49.62			39.99		13.40	0.11	0.87	027, 001/072, 078,014	AI	Freeman et al. [2015]
2012	Poland	83					27.7	85.5	100	83.1		18.0	0.0	0.0	027, 176	Etest	Lachowicz et al. [2015]
2011–2013	Japan	159					69	70		62	71	1.2	0.0	0.0	018, 369, 014, 002	Etest	Senoh et al. [2015]
2012–2013	Japan	130			49		59		99				0.0	0.0	ST2, ST17, ST8 (MLST)	AD	Kuwata et al. [2015]
2013–2014	Australia	440			18.2		84.3			3.4			0.0	0.0	014, 002	AD	Knight et al. [2015]
2014	Israel	208								66			18	47	027	Etest	Adler et al. [2015]
2014–2015	Czech Republic	20					10	100	100	100		65	0.0	0.0	176	Etest	Krutova et al. [2015]
–	Zimbabwe	23				100	100	43.5	100				0.0	0.0		DD	Simango and Ulandi [2014]

CFs, cephalosporins; FQs, fluoroquinolones; CTT, cefotetan; FOX, cefoxitin; CRO, ceftriaxone; CTX, cefotaxime; CLI, clindamycin; ERY, erythromycin; CIP, ciprofloxacin; MXF, moxifloxacin; GAT, gatifloxacin; RIF, rifampin; MTZ, metronidazole; VAN, vancomycin; PCR, polymerase chain reaction; PFGE, pulsed field gel electrophoresis; TRST, tandem repeat sequence typing; MLST, multi locus sequence typing; AD, agar dilution; BD, broth dilution; AI, agar incorporation; DD, disk diffusion.

The most common antibiotic susceptibility methods used for C. difficile are the agar dilution (AD) and the epsilometer test (Etest), a commercially available gradient diffusion system for quantitative antibiotic susceptibility testing. The Clinical and Laboratory Standards Institute (CLSI) indicates AD as the reference method for C. difficile and underlines that other techniques may be used as long as equivalence to the reference methods is established [Clinical and Laboratory Standards Institute, 2012]. The disadvantages of AD approach are the laborious, time-consuming steps required to prepare testing plates, particularly when the number of compounds to be tested is high or when only a limited number of bacteria are to be analyzed. For these reasons, most laboratories use Etest for routine (Table 1).

Mechanisms of resistance

Cephalosporins

Resistance to CFs is still uncharacterized in C. difficile, although most of strains are resistant to these antibiotics (Table 1). C. difficile overgrowth seems to occur after CFs therapy, as reported in several studies [Ambrose et al. 1985; de Lalla et al. 1989; Impallomeni et al. 1995]. C. difficile is often described as ‘constitutively resistant’ to CFs, but the variable minimum inhibitory concentration (MIC) values to the different CFs suggest that resistance to these antibiotics may be strain-dependent. β-Lactam antibiotic resistance is caused mainly by two mechanisms: antibiotic-degrading enzymes, β-lactamases, and modification of target sites, penicillin-binding proteins (PBPs). Genome analysis of C. difficile 630 (accession number AM180355.1) shows a number of coding sequences (CDSs) potentially involved in this resistance (Table 2). These CDSs strains are present also in other C. difficile strains, with identity ranging between 73% and 100%. Genomic comparison and functional analysis of C. difficile strains showing different phenotypes will be necessary to clarify the role of these potential β-lactam interacting genes.

Table 2.

C. difficile 630 CDSs potentially involved in resistance to cephalosporins.

Locus-tag in C. difficile 630	Product
CD630_03440	Putative β-lactamase-like protein
CD630_04580	Putative β-lactamase
CD630_04640	Putative β-lactamase-like hydrolase
CD630_04700	β-lactamase-inducing penicillin-binding protein
CD630_04710	Penicillinase transcriptional regulator
CD630_05150	D-alanyl-D-alanine carboxypeptidase, S11 peptidase family
CD630_05270	Putative β-lactamase-like hydrolase
CD630_05480	Putative penicillin-binding peptidase
CD630_06550	Putative β-lactamase-like protein
CD630_07810	Putative penicillin-binding protein
CD630_08290	Putative metallo-β-lactamase superfamily protein
CD630_08950	Metallo-β-lactamase superfamily exported protein
CD630_11480	Putative penicillin-binding protein
CD630_12290	Peptidoglycan glycosyltransferase
CD630_12910	Penicillin-binding protein
CD630_13740	Putative β-lactamase-inhibitor protein II
CD630_14690	Putative cell surface protein; putative penicillin-binding protein cwp20
CD630_16270	D-alanyl-D-alanine carboxypeptidase (penicillin-binding protein)
CD630_18020	Putative hydrolase, metallo-β-lactamase superfamily
CD630_21410	Serine-type D-Ala-D-Ala carboxypeptidase
CD630_24980	Putative sporulation-specific penicillin-binding protein
CD630_26560	Stage V sporulation protein D (sporulation-specific penicillin-binding protein)
CD630_27420	Putative hydrolase β-lactamase-like
CD630_31960	Putative penicillin-binding protein
CD630_36510	Putative metallo-β-lactamase-like hydrolase

MLS_B antibiotics

Ribosomal methylation is the most widespread mechanism of resistance to the antibiotics of the macrolide–lincosamide–streptograminB (MLS_B) family in C. difficile (Table 3). Erythromycin ribosomal methylases (erm) genes of class B commonly mediate resistance to these antibiotics, even if other erm genes have rarely been detected in C. difficile isolates [Roberts et al. 1994; Spigaglia et al. 2005; Schmidt et al. 2007]. Tn5398 is a mobilizable nonconjugative element of 9.6 kb length containing two copies of ermB genes [Farrow et al. 2001]. This element is able to transfer in vitro from C. difficile to Staphylococcus aureus and to Bacillus subtilis [Hächler et al. 1987; Mullany et al. 1995]. Since Tn5398 does not have genes encoding a recombinase, other conjugative transposons present in the donor genome could provide integration/excision functions to transfer the element to the recipient strain [Mullany et al. 2015]. The element could integrate into the recipient chromosome either by homologous recombination or by using a site-specific recombinase of the recipient. Another possibility is that a portion of the donor genome containing Tn5398 is transferred to the recipient and integrates by homologous recombination [Wasels et al. 2015b].

Table 3.

Genetic elements involved in C. difficile antibiotic resistance.

Antibiotic	Resistance mechanism	Genetic element	Gene	References
MLSB	Ribosomal methylation	Tn5398 and Tn5398-like	ermB	Farrow et al. [2001]; Brouwer et al. [2011]; Spigaglia et al. [2005]; Spigaglia et al. [2011]
		Tn6194	ermB	Wasels et al. [2013]; He et al. [2010, 2013]
		Tn6215	ermB	Goh et al. [2013]; Wasels et al. [2015]
		Tn6218	ermAB/cfr	Dingle et al. [2014]
Tetracycline	Ribosomal protection	Tn5397	tetM	Roberts et al. [2001, 2011]
		Tn916-like	tetM	Sebaihia et al. [2006]; Brouwer et al. [2011, 2012]; Spigaglia et al. [2005, 2007]
		Tn6164	tet44	Corver et al. [2012]
Chloramphenicol	Chloramphenicol acetyltransferase	Tn4453a and Tn4453b	catD	Wren et al. [1988, 1989]

ermB-containing elements different from Tn5398 have been found in the majority of C. difficile strains resistant to MLS_B. A total of 17 genetic organizations of these elements, denominated E1–E17, have been identified by a PCR-mapping method designed on the genetic organization of Tn5398 [Farrow et al. 2001; Spigaglia et al. 2005, 2011]. The most frequent genetic organization detected among European C. difficile clinical isolates in 2005 was the E4 [Spigaglia et al. 2011]. A recent analysis has demonstrated that elements E4 are related to Tn6194, a conjugative transposon firstly identified in C. difficile 2007855 [He et al. 2010, 2013; Wasels et al. 2013]. This element has a conjugative region related to that of Tn916, a large family of conjugative elements widely spread in Gram-positive and Gram-negative bacteria, and an accessory region that is related to Tn5398. Transfer of Tn6194 from C. difficile to Enterococcus faecalis has been demonstrated in vitro [Wasels et al. 2014].

Tn6215 is a peculiar element conferring resistance to MLS recently characterized in C. difficile CD80 [Goh et al. 2013]. Interestingly, this mobilizable transposon of about 13 kb in length can be transferred between C. difficile strains by a conjugation-like mechanism and by phage ΦC2 transduction. Furthermore, very recent data suggest that both Tn6215 and Tn5398 can be transferred to the C. difficile recipient strain C. difficile CD13 by a transformation-like mechanism [Wasels et al. 2015b].

It has been demonstrated that the in vitro maintenance of ermB-containing elements in C. difficile genome has a cost on the fitness of the bacterium [Wasels et al. 2013]. Nonetheless, these elements are widespread in the C. difficile population, suggesting that, regardless of the burden that an element imposes on fitness, other factors (i.e. the capability of transfer and the intrinsic genetic characteristics of the strains) are involved in the successful spreading of an element among C. difficile isolates.

Several C. difficile erm-negative strains resistant to both ERY and CLI or only to ERY have been identified [Spigaglia and Mastrantonio, 2004; Pituch, et al. 2006; Ratnayake et al. 2011; Spigaglia et al. 2011]. Alterations in the 23S rDNA or ribosomal proteins (L4 or L22) have been found in some of these strains, but the presence of the same changes in susceptible isolates excludes their role in resistance [Spigaglia et al. 2011]. Furthermore, resistant erm-negative strains treated with reserpine and carbonyl cyanide m-chlorophenyl hydrazone (CCCP), two pump inhibitors, do not show any reduction in MICs, suggesting that resistance is not mediated by efflux mechanisms [Spigaglia et al. 2011]. Tn6218 are nonconjugative Tn916-related elements recently identified in C. difficile strains [Dingle et al. 2014]. These elements carry multiple accessory genes conferring resistance to clinically relevant antibiotics, including cfr and ermAB genes. In particular, the multidrug resistance (MDR) gene cfr, present in a wide range of Gram-positive and Gram-negative species, could have a role in C. difficile resistance to MLS_B in the absence of erm genes.

Fluoroquinolones

Resistance to FQs in C. difficile is due to alterations in the quinolone-resistance determining region (QRDR) of either GyrA or GyrB, the DNA gyrase subunits [Ackermann et al. 2001, 2003; Dridi et al. 2002; Drudy et al. 2006, 2007]. Several amino acidic substitutions have been identified in both GyrA and/or GyrB (Table 4), but almost all C. difficile FQs-resistant strains show the substitution Thr82Ile in GyrA [Ackermann et al. 2001; Dridi et al. 2002; Spigaglia et al. 2008a, 2011; Kuwata et al. 2015]. In vitro experiments have demonstrated the MXF and levofloxacin (LE) exposure induce high frequency of selection for GyrA and GyrB drug-resistant mutants in previously susceptible strains [Spigaglia et al. 2009]. Therefore, it can be hypothesized that during the first stages of antibiotic treatment, when the concentration of drug in the intestine is still not inhibitory, a subpopulation of C. difficile may be able to acquire substitutions conferring resistance to FQs. Interestingly, a recent study has shown that Thr82Ile in GyrA has not detectable cost on the fitness of C. difficile, suggesting that this substitution can be maintained in the bacterial population even in the absence of antibiotic selective pressure [Wasels et al. 2015a].

Table 4.

Amino acid substitutions detected in C. difficile clinical isolates resistant to fluoroquinolones or rifamycins.

Antibiotic	Target	Amino acid substitution	References
Fluoroquinolones	DNA gyrase subunits	GyrA
		Thr82Ile	Ackermann et al. [2001]; Dridi et al. [2002]; Spigaglia et al. [2008]; Kuwata et al. [2015]
		Thr82Val	Ackermann et al. [2001]; Dridi et al. [2002]; Spigaglia et al. [2008]; Kuwata et al. [2015]; Liao et al. [2012]
		Asp71Val	Dridi et al. [2002]; Walkty et al. [2010]; Liao et al. [2012]
		Ala118Thr	Dridi et al. [2002]
		Val43Asp	Carman et al. [2009]
		Asp81Asn	Huang et al. [2009]; Liao et al. [2012]
		Ala384Asp	Mac Aogáin et al. [2015]
		GyrB
		Asp426Asn	Dridi et al. [2002]; Spigaglia et al. [2008]; Liao et al. [2012]
		Asp426Val	Spigaglia et al. [2008]
		Arg447Lys	Walkty et al. [2010]; Liao et al. [2012]
		Glu466Val	Liao et al. [2012]
		Arg377Gly	Liao et al. [2012]
		Ser416Ala
		GyrA/GyrB
		Thr82Ile/Ser416Ala	Spigaglia et al. [2008]; Liao et al. [2012]
		Thr82Ile/Ser366Ala	Huang et al. [2009]; Kuwata et al. [2015]
		Thr82Val/Asp426Val	Huang et al. [2009]; Liao et al. [2012]
		Thr82Ile/Ser366Ala and Asp426Val	Walkty et al. [2010]; Kuwata et al. [2015]
		Thr82Ile/Asp426Asn	Walkty et al. [2010]; Kuwata et al. [2015]
		Thr82Ile/Leu444Phe	Walkty et al. [2010]
		Thr82Ile/Asp426Val	Spigaglia et al. [2011]
		Thr82Ala/Ser366Ala and Gln434Lys	Kuwata et al. [2015]
Rifamycins	RNA polymerase	RpoB
		His502Tyr	O'Connor et al. [2008]; Pecavar et al. [2012]
		His502Arg	O'Connor et al. [2008]
		Arg505Lys	O'Connor et al. [2008]; Curry et al. [2009]; Miller et al. [2011]; Spigaglia et al. [2011]; Pecavar et al. [2012]
		His502Asn	Miller et al. [2011]; Pecavar et al. [2012]
		Asp492Asn	Pecavar et al. [2012]
		Asp492Val	Pecavar et al. [2012]
		His502Leu	Pecavar et al. [2012]
		Ser550Phe	Pecavar et al. [2012]
		Ser550Tyr	Pecavar et al. [2012]
		Ser448Thr and Arg505Lys	O'Connor et al. [2008]; Curry et al. [2009]
		Asp492Asn and Arg505Lys	O'Connor et al. [2008]
		His502Asn and Arg505Lys	O'Connor et al. [2008]; Curry et al. [2009]; Miller et al. [2011]; Spigaglia et al. [2011]; Pecavar et al. [2012]
		Arg505Lys and Ile548Met	O'Connor et al. [2008]; Curry et al. [2009]; Pecavar et al. [2012]
		Ser498Thr and Arg505Lys	Curry et al. [2009]; Miller et al. [2011]
		Leu487Phe and His502Tyr	Pecavar et al. [2012]
		His502Tyr and Pro496Ser	Carman et al. [2009]

Antibiotics for treatment of CDI

Metronidazole

Metronidazole (MTZ) is a nitroaromatic prodrug that need the reduction of the 5-nitro group of the imidazole ring to become cytotoxic to bacterial cells [Goldman, 1982] and it is considered the first choice for mild to moderate CDI [Debast et al. 2014; Lyras and Cooper, 2015].

Although the percentage of C. difficile strains resistant to MTZ is, in general, low (Table 1), several studies have emphasized treatment failure after treatment with MTZ [Musher et al. 2005; Pépin et al. 2005a; Vardakas et al. 2012]. An elevated geometric mean of MICs to MTZ have recently been observed in RT027 (1.1–1.42 mg/l), RT001/072 (0.65 mg/l), RT106 (0.65 mg/l), RT356 (0.61 mg/l) and in the nontoxigenic RT010 (1.5 mg/l), compared with the values of the other RTs (0.13–0.41 mg/l) [Moura et al. 2013; Freeman et al. 2015]. Furthermore, a number of C. difficile strains with MICs >2 mg/l, the EUCAST epidemiological cut-off (ECOFF) for MTZ (see http://mic.eucast.org/Eucast2/), have been reported in both humans and animals, as shown in Table 5. In particular, a very recent study reports the spread of strains RT027 with reduced susceptibility to MTZ in Israel, where they cause severe infections and a wide outbreak in 2013 in Jerusalem [Adler et al. 2015].

Table 5.

Characteristics of C. difficile strains with reduced susceptibility to metronidazole from 16 studies published between 2012 and 2015.

Antibiotic	Origin	Year of isolation	Country	Number of strains	PCR- ribotypes	MIC value (mg/l)	MICs after strains manipulation (mg/l)	References
Metronidazole
	Human
		2002–2007	Taiwan	2		>2		Chia et al. [2013]
		2009	Canada	1	027	>32	8	Lynch et al. [2013]
		2007–2010	Texas	36		32		Norman et al. [2014]
		2010–2011	Iran	4		32–64		Goudarzi et al. [2013]
		2011–2012	Europe	1	106	8		Freeman et al. [2015]
		2014	Israel	38	027	>2		Adler et al. [2015]
	Animal
	swine	-	Spain	4	078	≥256		Peláez et al. [2013]
	dog	-	Sweden	1		64	2–8	Wetterwik et al. [2013]
	swine	2007–2010	Texas	21		32		Norman et al. [2014]
	zebra	-	Spain	1		≥256	< 2	Álvarez-Pérez et al. [2014]
	dog	2007–2013	Italy	14	010 078	3–≥256	0.25–32	Spigaglia et al. [2015]
Vancomycin
	Human
		2002–2007	Taiwan	5		>2		Chia et al. [2013]
		2010–2011	Iran	2		4		Goudarzi et al. [2013]
		2010–2012	China	2		8		Dong et al. [2013]
		2005–2010	Taiwan	2		4		Liao et al. [2012]
		2014	Israel	98	027	>2		Adler et al. [2015]

MIC, minimum inhibitory concentration; PCR, polymerase chain reaction.

Heteroresistance seems to be a preresistance stage in C. difficile, as part of the population acquires the capacity of growth in presence of an antibiotic [Falagas et al. 2008; Peláez et al. 2008]. In vitro analysis suggest that subinhibitory concentrations of MTZ could have a role in selecting and maintaining colonies with increased MICs [Peláez et al. 2008; Moura et al. 2013]. Since mean concentration of MTZ in the feces of patients ranges from 0.8 to 24.2 µg/g [Bolton and Culshaw, 1986] it has been hypothesized that the concentrations achieved in the colon could be insufficient for the treatment of CDI due to strains with higher MICs [Brazier et al. 2001; Baines et al. 2008; Moura et al. 2013]. Although the clinical relevance of reduced susceptibility to MTZ is still not completely understood, a recent study suggest a potential impact of decreased susceptibility to MTZ of strains RT027 on the pathophysiology of recurrent CDI [Richardson et al. 2015].

Detection of strains with reduced susceptibility to MTZ can be problematic. In fact, this resistance is often unstable and laboratory manipulation of strains frequently results in MIC decrease towards a susceptibility range [Peláez et al. 2008; Lynch et al. 2013]. Experimental methodology may affect the magnitude of measured metronidazole MICs for C. difficile. The overall data reported in a recent study suggest the agar incorporation method (AIM) [Freeman et al. 2005] as the method of choice to detect strains with reduced susceptibility to metronidazole compared with the Etest and the AD [Moura et al. 2013]. Differences in the media used (Schaedlers broth and Wilkins–Chalgren agar for AIM and Brucella broth/agar for both Etest and AD) and in the duration of the precultured period (24 h for AIM and 48 h for both Etest and AD) seems to affect MIC determination [Baines et al. 2008; Moura et al. 2013]. Metronidazole susceptibility breakpoint for C. difficile defined by the CLSI and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) are not equivalent: the first is defined as ≥32 mg/l, the second >2 mg/l [Clinical and Laboratory Standards Institute, 2015] (see also http://www.eucast.org/clinical_breakpoints/). Methodological differences and different interpretation categories may cause discrepancies in results, influencing therapeutic decision and comparison of data. For these reasons, international committees are currently cooperating with the intention of harmonizing susceptibility testing and international breakpoints.

In contrast to other pathogens, such as Helicobacter pylori and Bacteroides fragilis, nitroimidazole (nim) genes conferring resistance to MTZ [Gal and Brazier, 2004] have not been identified in C. difficile [Moura et al. 2014]. C. difficile mechanism of resistance to MTZ is still not completely understood. Data obtained in recent studies on strains RT027 and RT010 suggest that this resistance is a multifactorial process involving alterations in different metabolic pathways, such as activity of nitroreductases, iron uptake and DNA repair [Chong et al. 2014; Moura et al. 2014].

Furthermore, recent evidences obtained in vitro seem to demonstrate that sub-inhibitory concentrations of MTZ are able to enhance the biofilm production of strains RT010, susceptible or with reduced susceptibility to this antibiotic, suggesting a possible role of biofilm in C. difficile resistance to MTZ [Vuotto et al. 2015].

Vancomycin

Vancomycin, the first-line antibiotic for moderate to severe CDI [Debast et al. 2014; Jarrad et al. 2015], consists of a glycosylated hexapeptide chain and crosslinked aromatic rings by aryl ether bonds, with a poor absorption in the gastrointestinal tract [Yu and Sun, 2013]. Its mechanism of action results in inhibition of the biosynthesis of peptidoglycan, an essential component of the bacterial cell wall envelope [Perkins and Nieto, 1974]. Resistance to VAN has frequently observed in Enterococci and Staphylococci, but it is not so largely diffused in C. difficile (Table 1), although a number of strains with reduced susceptibility to VAN (MICs range >2–16 mg/l) have recently been described (Table 5).

The mechanism of resistance in C. difficile is still unclear. Several Tn1549-like elements have been found in C. difficile [Brouwer et al. 2011, 2012]. Differently from the original Tn1549 element described in E. faecalis, the Tn1549-like elements of C. difficile do not have a functional vanB operon. Recently, a vanG-like gene cluster, homologous to the cluster found in E. faecalis, have been described in a number of C. difficile isolates but, although this cluster is expressed, it is not able to promote resistance to VAN [Ammam et al. 2012, 2013]. A recent study has demonstrated that the amino acid change Pro108Leu in the MurG of VAN-resistant mutants is obtained in vitro [Leeds et al. 2014]. Since MurG converts lipid I to lipid II during the membrane-bound stage of peptidoglycan biosynthesis, this substitution may affect VAN activity. Biofilm formation could be also probably involved in VAN-resistance. C. difficile within biofilms have been found more resistant to high concentrations of VAN (20 mg/l) and biofilm formation seems to be induced in the presence of subinhibitory and inhibitory concentrations of the antibiotic [Dapa et al. 2013].

The clinical significance of reduced susceptibility to VAN remains to be determined, since the fecal concentration of this antibiotic is very high, ranging between 520 and 2200 mg/l [Young et al. 1985].

Rifamycins and fidaxomicin

An increased rate of treatment failure and recurrence of infection have been associated with MTZ and VAN treatment [Vardakas et al. 2012], therefore other therapy options for CDI have been proposed in the recent years.

Rifamycins, in particular rifaximin (RFX), have recently been prosed as ‘chaser therapy’ for the treatment of relapsing CDI [Oldfield et al. 2014], while fidaxomicin (FDX) is a bactericidal new narrow spectrum macrocyclic antibiotic that is used for the management of CDI with high risk for recurrences [Chaparro-Rojas and Mullane, 2013]. Both RIFs and FDX are inhibitors of bacterial transcription but they have different RNA polymerase (RNAP) target sites. FDX binds to the ‘switch region’ of RNAP, a target site that is adjacent to the RIF target but does not overlap [Mullane and Gorbach, 2011; Srivastava et al. 2011].

Susceptibility to rifampin (RIF) by either Etest or AD correlated completely with susceptibility to RFX [Miller et al. 2011b]. Thus, rifamycin class susceptibility in C. difficile can be assessed by testing susceptibility to RIF, a rifamycin that is related to RFX. Data extrapolated from recent studies (Table 1) indicate that 11% of C. difficile clinical isolates are resistant to RIF and the rate of overall resistance appears to be rising [Huang et al. 2013; Rodríguez-Pardo et al. 2013; Eitel et al. 2015; Terhes et al. 2014]. C. difficile clinical isolates resistant to RIF have been detected in 17/22 countries participating in a recent pan-European surveillance and, in particular, high percentages of resistance (between 57% and 64%) have been observed in Italy, Czech Republic, Denmark and Hungary [Freeman et al. 2015]. Prior exposure to RIFs has been reported to be a risk factor for RIF-resistant C. difficile [Curry et al. 2009; Miller et al. 2011b] and resistant C. difficile strains may emerge even during therapy [Johnson et al. 2009; Carman et al. 2012]. RIFs are commonly used as anti-tuberculosis (TB) agents. Interestingly, all strains belonging to the emergent RT046 isolated in Poland from patients affected by TB and with a prolonged RIF therapy have been found highly resistant to these antibiotics [Obuch-Woszczatynski et al. 2013]. Different SNPs within the encoding gene for the β-subunit of the RNA polymerase (rpoB) have been identified in strains resistant to RIF (Table 4). Among the amino acid substitutions identified, Arg505Lys is the most common particularly in strains RT027 [Miller et al. 2011; Spigaglia et al. 2011; Carman et al. 2012; Pecavar et al. 2012].

Fidaxomicin provides cure rates not inferior to VAN and is associated with a significantly lower rate of recurrence of CDI associated with strains non-RT027 [Louie et al. 2011]. Furthermore, it has a minimal impact on the composition of indigenous fecal microbiota, in particular Bacteroides species [Tannock et al. 2010; Louie et al. 2012], with a high local concentration in the gut and feces (1225.1 µg/g after 10 days of therapy) [Goldstein et al. 2012; Sears et al. 2012]. Reduced susceptibility to FDX is very rare and only one C. difficile clinical isolate with a MIC of 16 mg/l has been described [Goldstein et al. 2011]. In vitro analysis has demonstrated the presence of mutations in rpoB or CD22120, which encodes a homolog to the MDR-associated transcriptional regulator MarR, in C. difficile mutants resistant to FDX [Leeds et al. 2014]. Interestingly, FDX retains activity against RIF-resistant strains since mutations causing resistance to FDX arise in rpoB gene at distinct loci compared to those causing resistance to RIFs [Optimer Pharmaceuticals, Inc., 2011].

Other antibiotics

Tetracycline

Recent papers indicate that C. difficile resistance to tetracycline (TET) varies among countries, from 2.4% to 41.67%, and is not so widespread among C. difficile clinical isolates [Dong et al. 2013; Pirš et al. 2013; Lachowicz et al. 2015; Norman et al. 2014; Simango and Uladi, 2014; Zhou et al. 2014]. In this pathogen, resistance is commonly due to protection of the ribosomes from the action of antibiotic (Table 3). The most widespread tet class in C. difficile is tetM, usually found on conjugative Tn916-like elements [Spigaglia et al. 2005; Mullany et al. 2012; Dong et al. 2014]. The Tn916-like family is responsible for the spread of antibiotic resistance (usually referred to TET but also to MLS_B and other antibiotics) to many important pathogens [Roberts and Mullany, 2011]. In C. difficile, the best-known element of this family is Tn5397, a 21 kb element able to transfer between C. difficile and B. subtilis or E. faecalis in vitro [Mullany et al. 1990; Jasni et al. 2010]. Tn5397 differs from Tn916 for the presence of a group II intron and for a different excision/insertion module. In fact, Tn916 contains two genes, xisTn and intTn, encoding an excisionase and a tyrosine integrase, whereas Tn5397 has a tndX gene that encodes a large serine recombinase [Roberts et al. 2001]. Furthermore, Tn916 inserts into multiple regions of the C. difficile genome [Mullany et al. 2012], while Tn5397 inserts DNA predicted filamentation processes induced by cAMP (Fic) domain [Wang et al. 2006].

Tn916-like elements may have different genetic organizations and carry different tetM alleles [Spigaglia et al. 2005, 2006]. A peculiar Tn916-element, containing both tetM and ermB, has been detected in the clinical isolate cd1911 [Spigaglia et al. 2007]. This element is nonconjugative and probably originated from the combination of one or more plasmids and a Tn916-like element.

Although tetM is the predominant class in C. difficile, other tet genes have been identified. In particular, the copresence of both tetM and tetW have been described in C. difficile isolates from humans and animals [Spigaglia et al. 2008b; Fry et al. 2012].

Furthermore, other integrative mobile genetic elements probably have a role in resistance to TET. An interesting element of 106 kb, the Tn6164, has been identified in C. difficile strain M120, a RT078 isolate [Corver et al. 2012]. This transposon is composed by parts of other elements from different bacteria, particularly from Thermoanaerobacter sp. and Streptococcus pneumoniae. Even if M120 is susceptible to TET and streptomycin, Tn6164 contains tet(44) and ant(6)-Ib predicted to confer resistance to these antibiotics, respectively. An analysis of data from patients indicate that mortality was more common in patients infected with strains RT078 containing Tn6164 compared with those infected with strains without this element. Although preliminary, these data are indicative of a possible association between Tn6164 and higher virulence of strains RT078.

Chloramphenicol

Resistance to chloramphenicol (CHL) is not so common in C. difficile and only 3.7% of European clinical isolates have been found resistant to this antibiotic [Freeman et al. 2015]. C. difficile resistance to CHL is usually conferred by a catD gene, encoding for a chloramphenicol acetyltransferase [Wren et al. 1988, 1989] (Table 3). The catD gene is located on the transposons Tn4453a and Tn4453b, structurally and functionally related to the Clostridium perfringens mobilizable element Tn4451 [Lyras et al. 1998]. The conjugative transposon Tn6104, recently described, contains elements closely related to Tn4453ab and Tn4451 but instead of a catD gene it has genes predicted to encode for transcriptional regulator, a two-component regulatory system, an ABC transporter, three sigma factors and a putative toxin–antitoxin system [Brouwer et al. 2011]. The role of these genes is not clear and remains to be determined.

MDR in C. difficile

All of the most common RTs, including the hypervirulent RT027 and RT078, are associated to resistance (Table 1) and many of these strains are MDR. An analysis performed during the first European prospective survey of C. difficile infections, indicated that 55% of resistant clinical isolates were MDR in 2005 [Spigaglia et al. 2011]. Results from 13 studies published between 2012 and 2015, indicate that MDR patterns include resistance to CLI, FQs, ERY and CFs (Table 6). Interestingly, resistance to multiple antibiotics characterized recently emerged epidemic RTs. Resistance to ERY, MXF, CIP and RIF has been observed in strains RT176, a type closely related to RT027, recently circulating in Poland and the Czech Republic [Obuch-Woszczatynski et al. 2014; Krutova et al. 2015]. RT356 is a MDR type predominant in Italy, genetically related to RT018, another type common in this country [Spigaglia et al. 2010, 2015]. Resistant to CLI, ERY, MXF and RIF characterized almost all of the strains belonging to RT356 and RT018 in Italy [Spigaglia et al. 2015], while strains RT018 described in Korea and Japan show resistance only to CLI, ERY and MXF [Kim et al. 2012; Senoh et al. 2015]. The long use of RIFs in Italy, more than 20 years [Salix Pharmaceuticals, Ltd, 2003], could explain the spread of this resistance in Italian isolates RT018. Although not defined as hypervirulent, RT018 has peculiar virulent traits and strains RT018 have been demonstrated to be highly transmissible, with a transmission index tenfold higher compared with that of strains RT078 [Baldan et al. 2015]. Old age (≥65 years), severe pulmonary comorbidity, previous use of FQs, and infection by RT018 have been associated as significant risk factors for complicated infections [Bauer et al. 2011].

Table 6.

Antibiotic resistance patterns of MDR C. difficile clinical isolates from 13 studies published between 2012 and 2015.

Year of isolation	Country	Number of MDR clinical isolates (%)	C. difficile PCR-ribotypes associate to MDR	Pattern of resistance (number of clinical isolates)							References
2000–2009	Korea	94 (-)	001, 018, 017	CLI	ERY	CIP				(86)	Lee et al. [2014]
2008–2009	US and Canada	22 (27.5%)	027	CLI	MXF	RIF				(22)	Tenover et al. [2012]
2009–2010	Poland	7 (70%)	046	CLI	MXF	ERY	RIF			(7)	Obuch-Woszczatynski et al. [2013]
2008–2010	Poland	17 (100%)	176, 027	CLI	ERY	MXF	CIP	GAT		(17)	Obuch-Woszczatynski et al. [2014]
2012	Poland	71 (85.5%)	027, 176, 012, 046	ERY	MXF	CIP	RIF			(15)	Lachowicz et al. [2015]
				ERY	MXF	IMP				(21)
				CLI	ERY	CIP	IMP			(2)
2010–2011	Croatia	7 (30%)	001	CLI	ERY	CIP	LVX	GAT	MXF	(7)	Novak et al. [2014]
2010–2011	Iran	36(48%)		CLI	ERY	CTX	MTZ			(3)	Goudarzi et al. [2013]
				CLI	ERY	CTX	VAN			(2)
				CLI	ERY	CTX				(30)
				CLI	CTX	VAN				(1)
2010–2012	China	44 (73.3%)	tr017, tr065, tr014,tr012, tr46, tr039, trsh2	CLI	FOX	CIP				(14)	Dong et al. [2013]
				CIP	FOX	TET				(1)
				CLI	FOX	CIP	MXF			(4)
				CLI	FOX	CIP	RIF			(1)
				CLI	FOX	CIP	TET			(9)
				CLI	FOX	CIP	TET	MXF		(13)
				CLI	CIP	TET	MXF			(1)
				CLI	CIP	TET				(1)
2011–2013	Japan	96 (-)	018, 369	CLI	ERY	MXF	GAT			(86)	Senoh et al. [2015]
				CLI	ERY	GAT				(11)
2012–2013	Japan	51 (39.2%)	ST17, ST81	CLI	CIP	CRO				(51)	Kuwata et al. [2015]
2012–2013	Italy	61 (71%)	356, 018, 126, 027, 046	CLI	ERY	MXF	RIF			(48)	Spigaglia et al. [2015]
				CLI	ERY	MXF				(11)
				CLI	ERY	RIF				(2)
2014–2015	Czech Republic	13 (65)	176	ERY	MXF	CIP	RIF			(13)	Krutova et al. [2015]
–	Zimbabwe	23 (100)		CLI	ERY	CIP	CTX			(23)	Simango and Ulandi [2014]

MDR, multidrug resistance; CLI, clindamycin; ERY, erythromycin; CIP, ciprofloxacin; MXF, moxifloxacin; RIF, rifampin; GAT, gatifloxacin; IMP, imipenem; LVX, levofloxacin; CTX, cefotaxime; MTZ, metronidazole; VAN, vancomycin; FOX, cefoxitin.

Conclusions

CDI is a growing concern for global public health. CDI has become the most common healthcare-associated infection in US hospitals and C. difficile is recognized as the most frequent cause of hospital-acquired gastrointestinal infections in Europe. Paralleling the increased CDI incidence, an increased morbidity and mortality was also observed and associated with the emergence and spread of the C. difficile hypervirulent type RT027. CDI due to strains RT027 is characterized by an increased incidence and severity, by being refractory to traditional therapy, and by a greater risk of relapse. Recently, additional types, highly virulent, have been reported to cause severe infections with poor outcomes. C. difficile adaptable capability and genome plasticity has determined an increase of isolates resistant to multiple antibiotics. The majority of epidemic clinical isolates are currently characterized as MDR. In particular, a wide range of mobile elements and alterations of antibiotic targets mediate resistance to the MLSB antibiotics and FQs, which are significantly associated with CDI. Furthermore, a decreased susceptibility to first-line antibiotics for therapy, in particular MTZ, and to those used for CDI recurrences, such as RIFs, may have a role in the low rate of response to treatment reported over recent years. Recent studies support the maintenance of antibiotic resistances in C. difficile population, regardless of the burden imposed by the acquisition of genetic elements/mutations conferring resistance and the decrease of antibiotics pressure. These data may partially explain the persistence of ‘old’ resistances (such as resistance to CLI) and the rapid diffusion of ‘new’ resistances, such as resistance to FQs, in C. difficile strains. The situation becomes more complex considering that antibiotic resistance in this pathogen can be a multifactorial phenomenon, which can involve more than one mechanism and alterations in various components. The rapid evolution of antibiotic resistance and the several mechanisms involved emphasize the need for a careful monitoring of C. difficile population to identify new phenotypic and genotypic characteristics. Effective antimicrobial stewardship, implementation of infection control programs, and the development of alternative therapies are also necessary to prevent and contain the spread of resistance and to ensure an efficacious therapy for CDI.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.