Epidemiological explorations on Clostridium difficile Infection
Goorhuis, A.
Citation
Goorhuis, A. (2011, October 12). Epidemiological explorations on Clostridium difficile Infection. Retrieved from https://hdl.handle.net/1887/17925
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the
Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/17925
Note: To cite this publication please use the final published version (if applicable).
Chapter 1
General Introduction
Clostridium dif-icile Infection: Disease, Epidemiology, Typing, Diagnosis and Therapy
7
Introduction
Nosocomial diarrhea has been a long been regarded as a “nuisance” problem,
associated with prolonged hospital stay and use of antibiotics. Initially, antibiotic associated diarrhea was attributed to Staphylococcus aureus 1. The organism Clostridium dif-icile was Mirst discovered by Hall and O’Toole in 1935 2, but it was not until 1977 that Bartlett and colleagues identiMied C. dif-icile at the causative agent of “antibiotic associated pseudomembranous
colitis” 3. To date, Clostridium dif-icile infection (CDI) is the most important cause of nosocomial diarrhea and the emergence of severe disease associated with CDI has now become a problem of equal magnitude as methicillin resistant Staphylococcus aureus infections. Over the past 10 years, CDI has been increasing in incidence and severity, and is associated with an increased duration of hospitalization, costs, morbidity, and mortality among patients 4,5.
This thesis describes the emergence of outbreaks of severe CDI caused by certain virulent types of Clostridium dif-icile and in the Netherlands. In addition, type speciMic risk factors are investigated and the question is explored whether differences exist between CDI that occurs in outbreaks and CDI that occurs in non-‐epidemic settings. Another research question in this thesis concerns the usefulness of highly discriminatory molecular Mingerprinting techniques, to better understand the dynamics of CDI outbreaks.
Clinical disease
Clostridium dif-icile is a gram-‐positive, spore forming rod, that grows in a strict anaerobic environment. The bacterium is ubiquitous in the soil and is capable of causing disease among both humans and animals (also: chapter 5) 6-‐8.
The clinical spectrum of disease caused by CDI can range from mild diarrhea, deMined as 3 or more loose stools (taking the shape of the container) per 24-‐hour period, to fatal colitis
9,10. Severe CDI is deMined as shown in table 1 4.
9
CDI causing admission to a healthcare facility for treatment CDI causing admission to an intensive care unit for treatment
Surgery (colectomy) for toxic megacolon, perforation or refractory colitis
Death within 30 days after diagnosis if CDI is either the primary or a contributive cause
Table 1. DeMinitions for severe CDI. Adapted from Kuijper et al 4.
Severe disease is often accompanied by a typical endoscopic picture, known as
pseudomembanous colitis, where destruction of bowel anatomy with hemorrhage and deep ulceration causes the formation of pseudomembranes. As a result, patients suffer from dehydration and tremendous discomfort caused by abdominal pain, fever and nausea. The most serious disease entity, although rare, is the syndrome known as toxic megacolon, deMined by an acute dilatation of all or part of the colon to a diameter greater than 6 cm, accompanied by systemic toxicity; this syndrome is associated with a 33% mortality rate11. When CDI is complicated by toxic megacolon, surgery (colectomy) is required in 65–71% of cases 12.
! One of the most challenging aspects of caring for patients with CDI is the recurrence of disease after successful initial therapy is completed. A recurrence is deMined by an episode of CDI that occurs within 8 weeks following the onset of a previous episode 4. A recurrence can correspond to a relapse involving the same strain or to a re-‐infection with a different strain
13-‐16. Recurrence rates after treatment with metronidazole or vancomycin are similar (20.2%
and 18.4%, respectively) (Table 2). The use of either metronidazole or vancomycin impairs resistance to colonization, thereby facilitating recurrent infection. Treatment options for recurrent disease are discussed later.
T h e ne w e ngl a nd jou r na l o f m e dicine
n engl j med 359;18 www.nejm.org october 30, 2008
1936
of disease after successful initial therapy is com- pleted. Recurrence rates after treatment with metronidazole or vancomycin are similar (20.2%
and 18.4%, respectively) (Table 1). The use of ei- ther metronidazole or vancomycin impairs resis- tance to colonization, thereby facilitating recurrent infection, which typically occurs within 4 weeks after the completion of therapy. Antimicrobial resistance to vancomycin in patients with C. dif-
ficile infection has not been reported, and resis-tance to metronidazole is rare. Recurrence may result from reinfection with a different strain of
C. difficile or persistence of the strain responsiblefor the initial episode.
29Role of Host Immunity
The risk of recurrent C. difficile infection is in- creased in patients who have already had one re- currence, rising from about 20% after an initial episode to about 40% after a first recurrence and to more than 60% after two or more recurrenc- es.
30,31This dramatic escalation in the risk of re- current C. difficile infection is probably caused in part by the selection of patients without protec- tive immunity against C. difficile, which makes them vulnerable to repeated attacks. C. difficile infection develops in only half the hospitalized patients who become colonized with toxigenic C. difficile as a complication of antimicrobial therapy, where- as the remainder are symptomless carriers.
32Af- ter colonization, symptomless carriers manifest an early increase in serum IgG antibodies against toxin A, whereas patients in whom C. difficile in- fection develops do not have such increased lev-
els (Fig. 4A).
32During an initial episode of infec- tion, some patients manifest a primary immune response with an early rise in IgM antitoxin A, followed by an increase in IgG antitoxin (Fig. 4B).
33In one study, patients with the highest titers of serum IgG antitoxin at the end of antimicrobial therapy were at a decreased risk for subsequent recurrence by a factor of 44, as compared with those with lower antitoxin titers.
33Management of Recurrence General Considerations
First, the ultimate goal of treatment is to discon- tinue all antibiotics and allow the normal bowel microflora to restore itself. Early studies of anti- biotic-associated colitis (published before C. dif-
ficile was identified as the causative agent) report-ed complete recovery in most patients after the discontinuation of clindamycin.
34,35Second, not all patients in whom recurrent diarrhea develops when they stop taking metronidazole or vanco- mycin have recurrent C. difficile infection. Other conditions, such as postinfectious irritable bowel syndrome, microscopic colitis, and inflammato- ry bowel disease, may be responsible. Third, a positive toxin assay in a patient with minimal or no symptoms should not prompt treatment. Re- peated stool assays are not recommended after therapy, except in patients with moderate or se- vere diarrhea. Fourth, in patients with persistent diarrhea despite several weeks of treatment with metronidazole or vancomycin, another cause should be sought, since C. difficile is rarely if ever resistant to metronidazole or vancomycin.
Antibiotics and Probiotics
An approach to the management of recurrent
C. difficile infection is presented in Table 2.36Since antimicrobial resistance is not clinically prob- lematic, a first recurrence of C. difficile infection can be treated with the same agent used to treat the initial episode. There is no standard or prov- en therapy for multiple recurrences. However, in one study of 163 patients with recurrent infec- tion, regimens that incorporated tapering or pulsed administration of vancomycin resulted in significantly fewer recurrences, with rates of 31.0% (P = 0.01) for tapering and 14.3% (P = 0.02) for pulsed administration, as compared with the rate for all other metronidazole or vancomycin treatments combined (49.6%).
31Probiotics, such
Table 1. Treatment Failures and Recurrences of C. difficile Infection with Metronidazole and Vancomycin Therapy.*
Variable No. of
Studies Treatment Failure Recurrence no./total no. (%) Metronidazole
Year 2000 or before 4 18/718 (2.5) 48/715 (6.7) After 2000 5 275/1508 (18.2) 332/1162 (28.6) Combined periods 9 293/2226 (13.2) 380/1877 (20.2) Vancomycin
Year 2000 or before 11 22/637 (3.5) 112/624 (17.9)
After 2000 2 2/71 (2.8) 36/181 (19.9)
Combined periods 13 24/708 (3.4) 148/805 (18.4)
* Data are from Aslam et al.21 and Zar et al.24
The New England Journal of Medicine
Downloaded from nejm.org at LEIDS UNIVERSITY MEDISCH CENTRUM on January 12, 2011. For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.
Table 2. Treatment Failures and Recurrences of C. dif-icile Infection with Metronidazole and Vancomycin Therapy 17-‐19.
In The Netherlands, a national sentinel surveillance in 19 hospitals across the country is being conducted since since May 2009. Extrapolating the data of this surveillance to all
hospitals in The Netherlands, it is estimated that each year, more than 2700 hospitalized patients will develop CDI, of which 100 will succumb as a direct or indirect consequence of the infection.
In these estimations, the impact of CDI in other healthcare facilities than hospitals was not
included. Therefore, the true number of patients with CDI admitted to healthcare facilities will be higher.
Based on a recent study, current U.S. estimates suggest that CDI affects 7178 inpatients on any given day, and causes the deaths of about 300 patients per day 20. These staggering statistics have brought Clostridium dif-icile to the forefront as one of the major challenges that healthcare facilities are addressing today.
Virulence
! Only toxigenic (toxin producing) forms of Clostridium dif-icile cause disease. Before the toxins can exert their effects, ingestion and germination of spores in the intestinal tract is required 21. The organism’s spores are very resilient to heat, desiccation, air, detergents and
11
alcohol and can remain viable in the hospital environment for weeks to years. Recently, differences in sporulation capacity of different C. dif-icile strains have been associated with virulence 22. Two recent studies by Lawley et al., gave more insight in the role of spores in the transmission of CDI 23,24. First, they described a novel protocol for the isolation of highly puriMied spores from cultures of a human-‐virulent C. dif-icile strain 24. The availability of puriMied spores facilitated, for the Mirst time, an estimate of the infectious dose of this
pathogen. Second, they described in a mouse model a highly contagious “super shedder-‐state”
of spores that was caused by antibiotic treatment of CDI and characterized by a dramatic reduction in the intestinal microbiota species diversity, C. dif-icile overgrowth, and excretion of high levels of spores 23. Stopping antibiotic treatment led to recovery of the intestinal
microbiota species diversity and suppression of C. dif-icile levels. Spore-‐mediated
transmission to immunocompetent mice treated with antibiotics resulted in self-‐limiting mucosal inMlammation of the large intestine. In contrast, transmission to mice whose innate immune responses were compromised, led to a severe intestinal disease that was often fatal.
Two structurally similar toxins, denoted A and B, are the main virulence determinants linked to CDI, and most pathogenic strains of Clostridium dif-icile produce both toxins 25,26. The role of these toxins in the pathogenesis of CDI has been well-‐described 26. Both toxin A and toxin B are pro-‐inMlammatory, cytotoxic and enterotoxic in the human colon 27,28. In a recent nature study, the importance of both toxins was demonstrated, because Clostridium dif-icile producing either one or both toxins showed cytotoxic activity in vitro that translated directly into virulence in vivo 29. The authors demonstrated this effect by creating a genetic knockout model, in which they neutralized the genes encoding for toxin A, toxin B, or both toxin genes.
In that way, they were able to show that in the situation where the genes for both toxin A and toxin B were neutralized, virulence was completely attenuated, whereas isogenic mutants of C. dif-icile, producing either toxin A or toxin B alone, caused fulminant disease in the hamster model of infection. The result of this study contradicted an earlier study that was published in
nature, that showed that toxin B, and not toxin A, was the determinant of virulence 30. A possible explanation, among others, for the contradicting result in that study, is the fact that different genotyping methods were used.
Toxins A and B are encoded by two genes, tcdA and tcdB, that map to a 19.6-‐kb pathogenicity locus (PaLoc, see Migure 1) containing additional regulatory genes 31. An important additional regulatory gene is tcdC, which is a putative negative regulator of the production of toxins A and B 32. Deletions in this gene could lead to an increased production of toxins A and B, due to lack of negative regulation.
Figure 1. Pathogenicity locus of Clostridium dif-icile 33.
tcdC lies downstream of tcdA and is transcribed in the opposite direction from the two toxin
genes, and TcdC is highly expressed in early exponential phase but declines as growth moves into the stationary phase 34. This decline in TcdC expression corresponds to increases in TcdA and TcdB, suggesting that TcdC may function as a negative regulator of toxin production.
tcdD is found upstream of tcdB and is coordinately expressed with both of the toxin genes.
TcdD is also homologous to TetR and BotR, which serve as positive regulators of tetanus and
13
botulinum toxin synthesis, respectively 35,36. It is assumed that TcdD is a major positive regulator of tcdA and tcdB expression.
The gene encoding TcdE is positioned between tcdB and tcdA and shows homology with holin proteins and thus has been speculated to facilitate the release of TcdA and TcdB through permeabilization of the C. dif-icile cell wall 37. In summary, toxin expression appears to be dependent on decreases in TcdC, TcdD-‐enhanced expression, and TcdE-‐mediated release from the cell.
Clostridium dif-icile isolates with varying genetic modiMications within the PaLoc have been
described 38,39. These include variant Clostridium dif-icile isolates that produce functional toxin proteins TcdA and TcdB, and toxin-‐variant isolates that fail to produce detectable toxins 38,40-‐42. Although originally thought to be non-‐pathogenic, clinically relevant toxin A-‐negative, toxin B-‐
positive strains of Clostridium dif-icile that cause diarrhea and colitis in humans have been isolated with increasing frequency worldwide (also chapter 3) 43-‐46.
Another proposed virulence factor is the binary toxin, which has been associated with increased disease severity and mortality 47. Genes for the binary toxin are located outside the PaLoc 48. The extent to which this toxin contributes to the pathogenicity of Clostridium dif-icile had been unknown until recently. However, in a recent study by Schwan et al., the effects of binary toxin on human colon carcinoma cells were studied 49. The authors observed that the toxin caused rearrangement of microtubules and the formation of long cellular protrusions.
The microtubule-‐based protrusions formed a dense meshwork at the cell surface, which wrapped and embedded Clostridia, thereby increasing the adherence of the pathogens. The authors also found that C. dif-icile colonization of the cecal content was signiMicantly decreased when binary toxin was functionally neutralized in the gut of mice. They concluded that the increased cell-‐surface adherence optimized the colonization by the pathogen.
Other possible virulence factors include surface layer proteins, cell wall proteins,
bacteriophages and other non-‐toxin factors; these are currently under intensive investigation
23,50-‐52.
Epidemiology
Since 2000, the incidence of CDI has been increasing worldwide and several large outbreaks of CDI in Montreal starting in 2003 have marked a new era of Clostridium dif-icile investigations and research 53,54. The number of Clostridium dif-icile cases in North America and Europe has been expanding dramatically over the past several years, in part due to the emergence of novel strains and to prolonged outbreaks of disease, which facilitates the continuing spread of the organism (see also chapters 2, 3, 5 and 6) 55-‐59. As shown in Migure 2, data from U.S. show a signiMicant increase in CDI discharges since 2001 60,61.
!"#$%&'()!"#$%&'()*!('+*!,-(!!"#$%&'(')*+(',-'.'"/!'$$-%#'+*.!.#$*'$*!/0"1"23!4*(!563666!&-$4#+'7!.#$%&'()*$3!,(-8!599:!
+&(-;)&!<66=>!?(-8!599:!+-!<6653!+&*!('+*!-,!0"1"!4*(!563666!.#$%&'()*$!#@%(*'$*.!AB!C6!4*(%*@+!D*!+&*!('+*!-,!#@%(*'$*!
,(-8!<665!+-!<66=!D'$!%-@$#.*('A7B!$+**4*(E!9<!4*(%*@+>!F&;$!+&*!(*%*@+!$&'(4!(#$*!#@!0"1"!D'$!@-+!'++(#A;+'A7*!$-7*7B!+-!'@!
#@%(*'$*!#@!+&*!@;8A*(!-,!&-$4#+'7!.#$%&'()*$>!
Figure 2. Discharge rate for Clostridium dif-icile associated disease (CDAD), per 10,000 hospital discharges, from 1993 through 2005. The upper line represents the discharges where CDAD was one of the listed diagnoses. The lower line represents the discharges where CDAD was the principal dignosis. From 1993 to 2001, the rate of CDAD per 10,000
discharges increased by 60 percent while the rate of increase from 2001 to 2005 was considerably steeper: 92 percent. Thus the recent sharp rise in CDAD was not attributable solely to an increase in the number of hospital discharges.
Although CDI has traditionally been seen in elderly inpatients and those recently discharged from healthcare facilities, CDI can also occur in the community. According to the proposed deMinitions 4, CDI cases are classiMied as healthcare facility-‐onset if they are
15
diagnosed more than 48 hours after admission. They are deMined as community-‐onset if they are diagnosed in the community or within 48 hours after admission. Furthermore, community onset CDI cases are classiMied on the basis of the time since the last discharge: if within 4 weeks, CDI is considered to be healthcare facility-‐associated; if 4-‐12 weeks: indeterminate exposure; and if more than 12 weeks: community-‐associated. See Migure 3 for a schematic presentation of these deMinitions. A study carried out in the U.S. showed that only 42% of more than 1000 patients with CDI had onset of their infection in a healthcare facility 62. In fact, 34%
of the cases were acquired in the community and had no healthcare associated risk factors.
SEVERE CDAD CASE
This is a CDAD patient to whom any of the following criteria apply:
1. admission to a healthcare facility for treatment of community-associated CDAD;
2. admission to an intensive care unit for treat- ment of CDAD or its complication (e.g., for shock requiring vasopressor therapy);
3. surgery (colectomy) for toxic megacolon, per- foration or refractory colitis;
4. death within 30 days after diagnosis if CDAD is either the primary or a contributive cause.
OUTBREAK OF CDAD
An outbreak can be defined as the occurrence of two or more related cases of CDAD over a defined period agreed locally, taking account of the background rate [28].
ORIGIN (Fig. 2)
The proposed categories are based on information concerning the origin of CDAD (healthcare-asso- ciated or community-associated) and the onset of symptoms (within the context of healthcare or within the community).
Healthcare-associated case
This is a CDAD case patient with onset of symptoms at least 48 h (>48 h) following admis- sion to a healthcare facility (healthcare-onset, healthcare-associated) or with onset of symptoms in the community within 4 weeks following dis- charge from a healthcare facility (community- onset, healthcare-assocciated).
Community-associated case
This is a CDAD case patient with onset of symptoms while outside a healthcare facility, and without discharge from a healthcare facility
within the previous 12 weeks (community-onset, community-associated) or with onset of symp- toms within 48 h following admission to a health- care facility without residence in a healthcare facility within the previous 12 weeks (healthcare- onset, community-associated).
Unknown case
This is a CDAD case patient who was discharged from a healthcare facility 4–12 weeks before the onset of symptoms.
ONSET (Fig. 2) Healthcare onset
Symptoms start during a stay in a healthcare facility.
Community onset
Symptoms start in a community setting, outside healthcare facilities.
NECESSITY FOR INVESTIGATION AND REPORTING OF OUTBREAKS ON BOTH NATIONAL AND EUROPEAN LEVELS
The definitions proposed above may be used in implementing CDAD surveillance schemes in spe- cific populations. Depending upon the populations and the reasons for surveillance and reporting, all or some of these definitions may be appropriate.
For example, in the UK (http://www.hpa.org.uk/
cdr/archives/2005/cdr3405.pdf), the population can be restricted to patients over 65 years of age, regardless of the presence or absence of specific risk-factors (e.g., prior antimicrobial therapy).
Since the implementation of comprehensive and systematic surveillance systems at the national level in each of the European member states will require some time, countries should first develop early-warning and response capa-
time
Admission Discharge
Healthcare-onset Community-onset
48h 4 weeks 8 weeks
Community-associated Unknown
Healthcare-associated (*)
(*) : - may be community- or healthcare-associated, depending on case’s history.
- if healthcare-associated, may have been acquired in the same facility or imported from another.
Fig. 2. Relationship among epide- miological definitions.
14 Clinical Microbiology and Infection, Volume 12 Supplement 6, 2006
! 2006 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 12 (Suppl. 6), 2–18
Figure 3. Epidemiological deMinitions for healthcare vs. community onset of CDI and a healthcare association vs. a community association of CDI 4.
Recently, severe CDI was also reported among pregnant women 63. The fact that CDI is increasingly found in a population that has previously been regarded as low risk, could reMlect a change in epidemiology related to the emergence of novel strains of Clostridium dif-icile (chapters 2, 3, 5 and 6)53,64-‐66.
In May 2009, Jarvis et al. published a survey regarding Clostridium dif-icile infections, which showed increasing prevalence of CDI in the United States 20.
For this survey, infection control departments were asked to report laboratory data from patients in their facility that were tested for Clostridium dif-icile on a single day sometime during the period of May to August 2008. In total, data were collected from 648 hospitals in 47 states (i.e. 12.5% of all the hospitals in the United States). See also Migure 4.
The average number of facilities participating per state was 14.3. On average, more than 12 of every 1000 inpatients in the U.S. healthcare system was found to be infected with Clostridium dif-icile. In Europe, a recent (2010) study by Bauer et al. also reported an increase in
incidence, from 2.5 to 4.1 per 10.000 patient-‐days, compared to data from Barbut in 2007
67,68. In this last study, all cause and attributable mortality rates were strikingly high: 22% vs.
7%.
Our respondents reported 1443 C difficile-colonized or -infected patients in 110,550 inpatients (our denomin- ator being nearly 20% of inpatients on any 1 day). Thus, the overall C difficile prevalence in US health care facili- ties was 13.1 per 1000 inpatients. This is 6.5 to 20 times higher than any previous incidence estimates (using different methodologies). Our C difficile rate should be considered a minimum estimate because all patients with diarrhea are not tested for this pathogen and the most frequent test used by our respondents was the EIA, which has a limited sensitivity (73%-75% on a sin- gle test).13-15Given that very few facilities were culturing for C difficile, it is not surprising that none of the respon- dents reported detecting the NAP1 strain. Our data also show that studying the epidemiology of the NAP1 strain (or any other emerging C difficile strain) will be severely hampered by the failure to use culture as a method to detect C difficile in patients with diarrhea.15
When we asked our respondents to examine the CDC surveillance criteria for health care- versus community- associated CDI, the majority (73%) of reported CDI patients was classified as health care-associated (with
either hospital or community onset).9Given the findings that 54.4% of CDI patients were detected within 48 hours of admission, 47% had been hospitalized within 90 days of onset, and 35% had long-term care facility admission within 30 days of onset, this suggests that a large propor- tion of the patients with CDI acquired the pathogen dur- ing previous health care facility admissions.
The major risk factors for CDI are age, health care facility exposure, and antimicrobial exposure. We found that nearly 70% of our reported CDI patients were .60 years of age (52.2% were .70 years of age) and that nearly 80% had received antimicrobials within 30 days of CDI onset. Our survey also shows that ,50% of respondent facilities had antimicrobial stewardship programs. Such programs were more common at medical school- than nonmedical school- affiliated hospitals and involved a wide variety of inter- ventions. The CDI prevalence rate was significantly higher at facilities without an antimicrobial steward- ship program, but these data likely are confounded by type of antimicrobial stewardship intervention(s), which varied widely across facilities, patient case-mix
A L
(9.55)
A K
(N/A)
A Z
(12.8)
A R
(26.67)
C A (18.70)
C O(13.76)
NH (11.40)
MA (24.96)
RI (28.88)
CT (7.29)
NJ (17.67)
DE (N/A)
MD (10.70)
V T (6.00)
F L
(12.22)
G A (7.36)
H I(0/00)
ID(3.68)
IL
(16.15) IN(10.69)
IA (6.82)
K S(12.84) K Y(21.80)
L A
(5. 45)
ME
(23.81)
MI(22.72)
MN(7.21)
MS(6.01)
MO(10.20)
N E
(7.04)
N V(11.11)
N M(10.40)
N Y(19.40)
N C(6.17)
N D(9.51)
O H
(10.42)
O K
(7.02)
O R
(9.43)
P A
(12.31)
S C(14.15)
S D(19.00)
T N(14.45)
T X
(12.70)
U T (12.99)
V A (9.25)
W A
(2.57)
W V(10.00)
W I
(17.6)
W Y(N/A)
MT(12.22)
≥ 20 per 1,000.
15- <20 per 1,000.
12.5- <15 per 1,000.
10- <12.5 per 1, 000.
7- <10 per 1,000.
0- <7.0 per 1,000.
No Responses. *Rate per 1,000 in patients.
DC (N/A)
Fig 2. The APIC National C difficile Inpatient Survey: C difficile prevalence rates by state.
268 Jarvis et al. American Journal of Infection Control
May 2009
Figure 4. The APIC National C. dif-icile Inpatient Survey: C. dif-icile prevalence rates by state 20.
17
Given the morbidity and mortality associated with CDI, the impact of these infections is considerable. If the data from the survey by Jarvis et all would be extrapolated to all healthcare facilities in the United States, then there would be approximately 7178 CDI inpatients on any 1 day 20. The published literature provides an average cost for each CDI patient of approximately 4475 U.S. dollars per episode 5,69-‐72. Thus, the approximate cost for these 7178 patients would be 32 million U.S. dollars. The estimated excess length of stay is 5.6 days 71-‐73, so these 7178 CDI patients would have an average of 40.197 extra days of
hospitalization as a result of their CDI. Furthermore, several studies have estimated the mortality associated with CDI; the mean mortality in these studies is 4.2% 74,75. Thus, 301 of the estimated 7178 CDI patients who present on any 1 day as inpatients would be expected to die. These estimates indicate that the impact of CDI is enormous and well worth the cost of prevention.
Patients in healthcare institutions are most at risk of acquiring the organism, which becomes a component of their gut Mlora 10. In a prevalence study by Johnson et al. , it was found that >20% of patients who remained in the facility longer than one week and up to nine weeks, became colonized with Clostridium dif-icile 76. This study supported the view that patients acquire Clostridium dif-icile during their hospitalization, as opposed to the view that the organism is a small component in virtually everyone’s bowel microbiotia and colonizes larger parts of the colon, following disruption of the normal bowel Mlora caused by the use of antibiotics. In fact, some patients who are already colonized with their own non-‐toxinogenic strain before they enter healthcare may be protected from acquiring a pathogenic version of Clostridium dif-icile 77.
Typing and identiJication of new Clostridium dif-icile strains
Several typing methods exist for Clostridium dif-icile. Depending on the typing method that is used, different names can exist for the same strain. In a study by Killgore et al, the
seven most frequently used genotyping methods were described and compared 78. The names of these methods are multilocus variable-‐number tandem-‐repeat analysis (MLVA), ampliMied fragment length polymorphism (AFLP), surface layer protein A gene sequence typing
(slpAST), PCR-‐ribotyping, restriction endonuclease analysis (REA), multilocus sequence typing (MLST), and pulsed-‐Mield gel electrophoresis (PFGE). All these techniques were capable of detecting outbreak strains, but only REA and MLVA showed sufMicient discrimination to distinguish strains from different outbreaks. The current epidemic strain is known as 027/
NAP1/BI, which is based on three different typing systems: PCR-‐ribotype 027, PFGE type NAP1 and REA type BI.
In Europe, PCR-‐ribotyping is used a the standard typing method. This method is based on the fact that every bacterial strain contains several rRNA operons and that there is a strain-‐
dependent variation in the size and number of the 16S-‐23S intergenic spacer regions.
AmpliMication of these regions results in a variety of PCR-‐products whose size and number will vary among different strains. PCR-‐ribotyping has appeared a robust typing method, that is stable and reproducible 79-‐81.
! To study spread and epidemiology among strains with the same PCR-‐ribotype, such as in outbreak situations, a more discriminatory method is needed. A good method in such a situation is Multi Locus Variable-‐number of Tandem-‐repeat Analysis (MLVA). This method is based on the ampliMication of regions with short tandem repeats. The number of tandem repeats within these regions (or loci) can differ between strains and can therefore be used for typing. The availability of the complete sequence of the C. dif-icile genome of strain 630
provided the opportunity to identify these short tandem repeats 82. The MLVA developed by van den Berg et al., uses automated fragment analysis and multi-‐colored capillary
electrophoresis and proved to be a highly discriminatory method, able to discriminate among strains that had the same PCR-‐ribotype 83. In chapter 3, this method is used to investigate a large outbreak of CDI caused by PCR-‐ribotype 017.
19
The changing epidemiology of CDI may well be related to the rise of new, epidemic strains that have spread rapidly across major geographic areas during the last decade. The Mirst of these, called “J” strain (now known as PCR-‐ribotype 001/PFGE type NAP2/REA type J), was Mirst recognized in Europe, causing outbreaks in England and Wales, where PCR ribotype 001 was the most common strain among hospitalized patients, accounting for 57 percent of all isolates in one survey and was responsible for a large outbreak in northwest England
involving 175 patients and 17 deaths at one hospital 84,85. Later, this strain was also found in the United States 86. The strain was resistant to clindamycin, causing the organism to survive when this antibiotic was administered and to substitute the original bowel Mlora that was killed by clindamycin, thus causing disease. Use of clindamycin has for many years been regarded as one of the most important risk factors for CDI.
The global increase incidence of CDI since the Mirst outbreaks in Canada, the United States and the United Kingdom has been associated with the spread of the hyper virulent type 027/NAP1/BI strain, that has been the cause of severe nosocomial disease (see also chapter 2) 53,56,59,65,87-‐90.
In the period between October 2003 and June 2004, the type 027 strain was recognized in the UK at the Stoke Mandeville Hospital in an outbreak involving 174 cases and 19 (11%) deaths that were deMinitely or probably due to C. dif-icile. A second outbreak occurred between October 2004 and June 2005 in Stoke Mandeville hospital involving 160 new cases and 19 (12%) further deaths 58,59. The Healthcare Commission investigation concluded that the outbreaks were a consequence of a poor environment for patient care, poor practice in the control of infection, lack of facilities to isolate patients and insufMicient priority being given to the control of infection by senior managers.
In July 2005, the Mirst outbreak with the type 027 strain was recognized in The Netherlands, in a hospital in Harderwijk 91-‐93. The incidence of CDI in the hospital had increased from 4 cases per 10.000 patient admissions in 2004 to 83 per 10.000 in 2005, at the time that the
outbreak was recognized. Measures taken by the hospital included isolation of all patients with diarrhea, cohorting of all C. dif-icile-‐infected patients on a separate ward, banning of all Mluoroquinolone use and limiting the use of cephalosporins and clindamycin. In January 2006, the situation appeared to be under control. A second outbreak occurred in another hospital 30 km from the Mirst and was probably related to the Mirst outbreak through a transferred patient with CDI 91,92. This outbreak is described in chapter 6.
In response to the outbreaks in The Netherlands, the Center for Infectious Disease Control (CIb) at the National Institute for Public Health and the Environment (RIVM) in Bilthoven issued guidelines for infection control and treatment94. Furthermore, diagnostic facilities were increased and made accessible to all microbiology laboratories in the country. Subsequently, three hospitals and a nursing home in the western part of the country also reported an increase in the incidence of severe CDI. A cluster of 12 patients with CDI caused by type 027 was reported in July and August 2005 in a large teaching hospital in Amsterdam. One patient died due to CDI and two other patients developed severe complications 95. Another hospital in Amsterdam also reported an increase in severe cases of CDI in July 2005 among patients who were cared for in the department of geriatrics. By April 2006, type 027 was found to be the cause of an outbreak in 11 hospitals and was isolated from sporadic cases in Mive hospitals
92,96. Simultaneously, severe CDI caused by type 027 was reported in Belgium and France 97,98. Because the type 027 strain is resistant to Mluoroquinolones, use of this antibiotic has been described as an important risk factor for nosocomial CDI caused by this type and has been implicated in outbreaks of CDI in several reports (see also chapter 2) 66,99,100. Other virulence factors of the type 027/NAP1/BI strains are the presence of binary toxin genes and a base-‐pair deletion at position 117 in the regulatory gene tcdC. This deletion in the type 027/
NAP1/BI strain has been associated with a 16 to 20-‐fold increase in toxin production as compared to wild type strains of Clostridium dif-icile 101.
21
Another emerging strain of Clostridium dif-icile is the type 078/NAP7/BK strain (toxinotype V), which has been associated with both food animals and humans in Europe (chapter 5) and more recently in the United States 102,103. This strain is also binary toxin positive and contains a mutation (C184T) in the toxin regulatory tcdC gene (chapter 5).
Several different strains of C. dif-icile, including 078, have been isolated from retail meats in Canada and the United States, further complicating the epidemiology of CDI 104. However, to date there have been no outbreaks of CDI speciMically linked to the consumption of food.
An important and largely unanswered question is why certain types of Clostridium dif-icile cause severe disease across the globe. To gain more insight in the genetic basis for the
emergence of C. dif-icile as a human pathogen, He et al. used whole genome sequencing to analyze genetic variation and virulence of a diverse collection of thirty C. dif-icile isolates 105. Phylogenetic analysis demonstrated that C. dif-icile is a genetically diverse species, which has evolved within the last 1.1–85 million years. The authors observed that the disease-‐causing isolates had arisen from multiple lineages, suggesting that virulence evolved independently in the highly epidemic lineages.
Laboratory diagnosis
! The diagnosis of CDI is usually based on the clinical history in combination with laboratory tests. Various laboratory tests are currently available for the detection of
Clostridium dif-icile or its toxins 106. The diagnostic tests for Clostridium dif-icile can be divided in different groups (table 3).
Groups of diagnostic tests Tests per group
Clostridium dif-icile products -‐ Cell culture cytotoxicity assay -‐ Glutamate dehydrogenase (GDH) -‐ Aromatic fatty acids
-‐ Toxins A and/or B
Groups of diagnostic tests Tests per group Culture methods for the detection of
toxin-‐producing Clostridium dif-icile -‐ Toxigenic culture
Genetic tests -‐ 16S RNA
-‐ Toxin genes -‐ Genes for GDH
Table 3. Diagnostic tests for Clostridium dif-icile
The cell culture cytotoxicity assay (CCA) is still regarded as the reference standard for the detection of Clostridium dif-icile toxins 107. Culture followed by in vitro toxin detection of the isolated strains has been adopted by some investigators as a more sensitive reference
standard than CCA, although the clinical relevance of this so called toxigenic culture (TC) is not entirely clear 108. Because these two standard methods are time-‐consuming and require speciMic laboratory facilities and technical expertise, many laboratories have replaced these methods by enzyme immunoassays 109. These are rapid and easy-‐to-‐perform assays designed to detect Clostridium dif-icile toxins or the enzyme GDH (table), which is produced by both toxigenic and non-‐toxigenic Clostridium dif-icile strains. Another new development is the application of real-‐time PCR to detect the toxin genes of Clostridium dif-icile directly from stools 110-‐114.
Although these rapid and easy tests can be attractive alternatives to the time-‐consuming reference standards, they have been reported to have limited sensitivity and/or speciMicity.
This has given rise to many different testing protocols, including multiple sample submission or multiple testing of samples by different methods 115-‐119.
In 2009, Crobach et al published a diagnostic guideline, based on a systematic review, in which the available evidence concerning laboratory diagnosis of CDI was evaluated and recommendations to optimize CDI testing in patients suspected of CDI were formulated 120. In this review, the test characteristics were analyzed of 13 commercially available enzyme
23
immunoassays (EIA) detecting toxins A and/or B, 4 EIAs detecting Clostridium dif-icile GDH, and a real-‐time PCR for Clostridium dif-icile toxin B gene. In comparison with CCA, and assuming a prevalence of CDI of 5%, positive and negative predictive values varied between 028-‐0.77 and 0.12-‐0.65, respectively. In comparison with toxigenic culture, positive and negative predictive values varied between 0.98-‐1.00 and 0.97-‐1.00, respectively. Positive predictive values would only be acceptable, if the tests were performed in a population with a hypothetical CDI prevalence of 50 percent (ranging from 0.71 to 1.00). Therefore, to overcome the problem of a low positive predictive value, Crobach et al. proposed a two step approach, with a second test or a reference method in case of a positive Mirst test (Migure 5).
Because the negative predictive values of the tests were very acceptable at low prevalences, the rapid rests can serve as screening tests in an endemic situation, with the emphasis on a negative test result. When a negative test result is obtained, CDI can very reliably be excluded. In case of a positive test result however, a second conMirmatory test must be performed. This can either be a reference test such s CCA or TC, or a second rapid test. The Mirst approach has been described by Ticehurst et al, who applied a two-‐step algorithm in which specimens were Mirst tested for the presence of GDH antigen by an EIA and the positive results were conMirmed by CCA 119. Because only GDH-‐positive samples were tested by CCA, this approach resulted in a reduced CCA workload (by 75–80%) and costs. Gilligan et al.
demonstrated that this two-‐step algorithm has an enhanced ability (by 40%) to detect CDI compared with the results of an EIA detecting toxins A and B 121.
The second approach, in which all positive samples would be conMirmed by an additional rapid test, the higher prevalence in the tested population will result in acceptable positive predictive values. However, due to the higher CDI prevalence, negative predictive values will be less acceptable. Therefore, samples with an initial positive test result, but a negative second test result for CDI, require testing with a reference method as a third step, as was described by the group Hussain et al. 116-‐118. With this three-‐step approach, results of 85% of samples were
available on the day that the specimens were received and the need for CCA testing was even further reduced to 15%. Fenner and colleagues have also applied this three-‐step approach. In their laboratory, the results of 92% of samples were available within a turnaround time of 4 h;
only 8% of samples had to be tested by CCA 122.
The lack of conMidence in the tests for CDI detection has motivated some clinicians to submit multiple samples per patient 115. In an endemic situation, all rapid tests have high negative predictive values, which implies that repeat testing is not useful 123,124. By contrast, in an epidemic setting, with a higher prevalence of CDI, negative predictive values of rapid tests will be signiMicantly lower, which implies that repeat testing will detect additional CDI cases 125.
25
