Recurrence of symptoms in Clostridium difficile infection—relapse or reinfection?

Recurrence of symptoms infection-relapse M. H. Wilcox*, Department

W. N. Fawley,

of Microbiology,

in Clostridium or reinfection? C. D. Settle

difficile

and A. Davidson

Unirlevsity of Leeds and The Geneva1 Infivmaql, Ileeds LS2 YJT, UK

Received 20 August 1997; manuscript accepted 17 Septemhev I997 Summary: \\‘e have fingerprinted (,‘hstvidium dif$ci/e isolates from patients Lvith symptomatic recurrences of infection, using random amplified pal>-morphic DNA (RAPD). The medical records of 55/79 patients xvere esamined, from whom multiple C. dif$cilc-positive faeces were received during hospitalization at least five days, but no more than two months, apart. In 20 of these cases s);mptoms either did not recur (i.e., absent for at least three days txt\veen episodes), or wwe explainable by other causes, such as lasatil c administration. Of the remaining 35 patients, 27 sets of C’. difjcile isolates (23 pairs and four triplicates) wcrc available for RAPD fingerprinting. Differing C’. d<[fici/c DS.4 fingerprints (at least three major hands difference) wcrc obtained for 1 S/27 patients, and hence at least 56% of the clinical recurrences of infection were in fact due to t-c-infection as opposed to relapse. Since TVVCfound that an endemic C’. difficile clone lvvas present in 1X out of 27 patients (67%) and accounted for 53% (31/5X) of all isolates, it is probahlc that the majority of symptomatic recurrences are in fact rc-infections, with either a different or the same C. difficile strain. \Ve conclude that more attention must he given to preventing the reinfection of C’. di’jcilr symptomatic patients. Isolation of symptomatic individuals is the preferred option for the protection of other patients, hut measures must bc taken to ensure that further strain acquisition by the index cases dots not occur. k’c,wcords:

r.

tiifficile; rc-infection;

relapse;

RAPD

Introduction d;fJ; ~1 . ‘I e infection is a major cause of morbidity, \vith reports in the elderly increasing approximately eightfold between 1990 and 1994 in England and Wales.’ Latest figures from the Communicable Disease Surveillance Centre indicate that the numbers of cases are continuing to rise with the provisional half-yearly reports for 1997 being the largest ever.? C. dif$ci/e is now endemic in many hospitals, and at least 17 patients died in a recent, large, nosocomial outbreak.’ The additional, identifiable, costs associated Lvith C. difficicile infection amount to more than E4000 per patient, Clostridium

94

M. H. Wilcox

et al.

the major factor being a mean increased duration of stay of 21 days.4 Patients appear to acquire C. difficile infection and as a result stay longer in hospital, rather than becoming infected because of an increased length of stay. Furthermore, patients who acquire C. difficile infection have an increased risk of dying before hospital discharge, although a direct cause and effect relationship is often not apparent.4 It is not uncommon to see patients with multiple clinical relapses of C. Such patients represent a therapeutic dilemma, pardifjkile infection.’ ticularly as to whether metronidazole or vancomycin should be prescribed for apparent treatment failures. While reported relapse rates for treatment with metronidazole or vancomycin are not significantly different (523% and 9-24%, respectively),’ it should be remembered that ‘relapse’ is normally a clinical definition. In a recent study it was found that repeat courses of antibiotic treatment were given to 37% of those patients requiring specific therapy for C. difficile infection.4 It was not known whether this high clinical failure rate was due to relapses or re-infections. Many hospitals are unable to isolate all symptomatic cases so increasing the risk of re-infection. Clearly, high clinical failure rates will exacerbate increases in the length of stay of patients with C. dif$ ci 1e infection. Two collections of case histories have shown that by using DNA fingerprinting, at least half of these socalled relapses are in fact re-infections with a different strain.6’7 However, these studies only examined 10 and 11 patients, and it is important to confirm or refute these findings with a larger series. Methods

Study design We retrospectively identified all hospitalized patients who had multiple C. difficile cytotoxin-positive faecal samples (>five days to two months apart) submitted to our laboratory between October 1995 and September 1996. Where possible hospital medical and nursing notes were obtained for scrutiny. The laboratory processes faecal specimens from six different hospitals. We only examine faecal samples for C. difficile cytotoxin on request in patients aged over two years, and do not process repeat specimens from patients previously positive within the last five days. Hospital policy requires that symptomatic C. difficile cytotoxin-positive patients are nursed in a side room until cessation of diarrhoea. C. difficile cytotoxin is detected by a microtitre tray method using Hep 2 cells with Clostridium sordellii antitoxin protected controls, and an approximate 1 in 50 final dilution of faeces in cell culture medium. All C. difficile cytotoxin positive faeces are stored at -20°C. C. difficile culture and identi3cation C. difJicile isolates were recovered from frozen faeces samples by culture on cycloserine, cefoxitin egg-yolk agar (Lab M, Bury, UK) for 48 h in an

Relapse

or

reinfection

of C.

difficile

95

anaerobic cabinet at 37°C. Isolates tvere recognized by their characteristic colonial morphology and odour, and in cases of doubt, RapID ANA II System (Atlanta, GA, USA) was used. fingerprinting DNA fingerprinting of C. difficile isolates was performed by random amplified polymorphic DNA (RAPD). DNA was extracted from one to two overnight colonies on pre-reduced blood agar plates by first incubating these with 25 ~1, lysoayme 100 ug/mL (S‘i g ma, Poole, UK) at 37°C on a rotary shaker for 10 min. A further incubation for 10 min at 37°C was carried out after the addition of 75 PL of 0.1 >t Tris pH 8.0 and 25 uL proteinase K 100 pg/mI. (Sigma), and the mixtures were then heated to 97°C in a thermal cycler for 5 min. The following DNA primer (AP3)X was used: 5’ ‘I’CA CGA TGC A 3’ (Oligonucleotide Synthesis Service, Institute of Pathology, University of Leeds, UK). The following vvere added to each 2S uL reaction volume: 0.75 PL AI’3 (70 pmoles), 2 ~1, of 1 rnN1 MgClz (BioLine, UK), 5 PL of buffer ( x 10 concentrate, BioLine), 8 /.tL of 1.25 rnbt dNTP mix (Pharmacia Biotech, Herts, IJK), 0.5 1tL RioExact taq polymerase (2 units/reaction, BioLine), and 5 ILL DNA extract. L1NA amplification was carried out in an Ericomp ‘I’winblock EasyCycler”“’ Systems, South(1, azer Laboratory ampton, UK) using the following cycling program: four cycles, each lasting 3 min at 94”C, 3 min at 35°C and 3 min at 72”C, and 40 cycles each lasting 1 min at 94”C, 1 min at 35°C and 2 min at 72°C. Amplified DNA vvas separated by agarose electrophoresis using Tris-borate EDTA TBE buffer PI-I 8.0 3% Metaphor”“’ gels (Flowgen, Staffs, UK) and a 4 h run at 1SOI’/ 1SOmA. DNA fingerprints were visualized, after ethidium bromide staining, with a UV transilluminator and photographed using an orange filter (Kodak Wratten No. 16) and Polaroid film. DNA

Definitions Diarrhoea due to C. difficile infection was defined as at least three episodes of loose stools per 24 h for at least 48 h with laboratory confirmed cytotoxinpositiv,e faeces. Recurrence of C. difficile diarrhoea was defined as the resumption of symptoms, after cessation for at least three days, with laboratory confirmed cytotoxin positive faeces.

Results

Data retrieval A total of 79 patients were identified from whom at least two C. difficile cytotoxin positive faecal specimens were received at least five days but no more than two months apart. Twenty-four sets of medical records either

96

M. H. Wilcox

et al.

could not be obtained despite repeated attempts, or were located at peripheral hospitals. Of the 55 for which records were available, in 20 cases (36%) diarrhoea either did not recur (i.e., at least three days between episodes), or symptoms were explainable by other causes such aslaxative use. For the remaining 35 patients with recurrences of symptoms attributable to C. difficile, 27 sets of isolates (23 pairs and four triplicates) were available for RAPD fingerprinting. In 8/35 cases either faeces had not been stored (insufficient amount after original laboratory diagnostic testing), or proved C. difficile culture negative. The 27 patients (age range 35-93, median 86 years) received a total of 46 courses of antibiotic therapy for C. dif$ciZe infection, of which 83% consisted of metronidazole and the remainder either vancomycin alone or in combination with metronidazole. DNA jingerprinting An endemic C. difficile clone was identified, accounting for 53% (3115%) of all isolates and was present in 18 out of 27 patients (67%). Different C. difficile DNA fi ngerprints (2 three major bands) were obtained for 15127 patients [Figure l(a)-(d)], and hence at least 56% of the clinical recurrences of infection were in fact due to re-infection as opposed to relapse. Discussion

It is well known that a significant proportion of patients with C. d$$ciZe infection have symptomatic recurrences. For example, we recently found that repeat courses of antibiotic treatment were required by 37% of patients needing specific therapy for C. difficile infection.” This increases the costs of treatment, the risks of cross-infection and the duration of hospitalization. Two small studies (of 10 and 11 patients)hz7 have shown that approximately half of such clinical recurrences are in fact re-infections with a different strain, although these reports included some outpatients and readmissions. We sought to confirm these findings using RAPD fingerprinting of C. dif$ciZe strains isolated from a larger series of patients with clinical recurrences. There is considerable pressure to explore new and unproved treatment options for individuals with multiple symptomatic recurrences.“‘” It is therefore desirable to document the incidence of relapse (possible treatment failure) as opposed to re-infection. Clearly, episodes of reinfection point to inadequacies in infection control procedures. It remains possible that some patients may have originally harboured more than one C. difficile strain, although this appears to be uncommon. DNA fingerprinting of multiple C. dif$ciZe colonies by ourselves (unpublished data) and others7 from individual specimens has failed to yield more than one strain. In the present study, we used RAPD to fingerprint multiple isolates, as this technique has been found to be highly discriminatory for C. difficile.8~“-‘3 While in some hands there are problems with the reproducibility of banding patterns produced by RAPD, for the

Relapse 1 2 3 a ba baba

4

19

5L678 bab

20

or reinfection 9 ab

ababab

21 L

22

23

10 b

of C.

difficile

10 11 12 13 c a babababc

14

24

26

25

L 15 16 ababcabab

L

27

17

18

N15

Figure 1. The figures sho\v the RXPD fingerprint profiles of pairs (lanes a, b) and triplicates (lanes a, b, c) of successive C. difficile isolates from patients l-10 (a), patients IO-18 (h), patients 19-23 (c), and patients 21-27 plus a repeat of patient 15 (d). h‘uccessive isolates from patients 2-7, O-1 2, 15, 18, and 22-24 x\ere considered to he distinct (i.e., differed h! at least three major bands). An endemic strain is seen in lanes la&b, St?, bb, 7h, 8a&b, ‘)a, lha,b&c, 17a&b, 18b, 19a&b, 22a, 23b, 2-tb&c, 2Sa&h, 26a,b&c, and 27a&h. Lane ‘1’ is a negative control. The ‘I,’ lanes are molecular size markers.

purposes of this type of study it is an ideal technique given its speed, cost and relative simplicity.” Furthermore, despite several protocol changes \ve bvere unable to obtain discrete DNA bands for our endemic C. difficile clone of isolates using pulsed-field gel electrophoresis of Smn I digests. This lack of typeability of C. dg$cile h as been noted elsewhere and is believed to be due to endonuclease-mediated digestion of DNA fragments~x’l”,15,1” JVe have compared the use of the primer used in the present study (AP3)X with two others”,” for RAPD fingerprinting of C. dijjicile, and have found the former to be the most discriminatory (data not shown). We have extended the findings of earlier reports”,7 by documenting that

98

M. H. Wilcox

et al.

at least 56% of clinical recurrences of infection are in fact due to re-infection as opposed to relapse. This is a disturbingly high proportion of cases, particularly given the observation that there is a predominant DNA clone of strains causing C. difficile infection in our patients. Such an observation could be expected to reduce the proportion of cases of re-infection with different C. dif$ciZe strains from the original pathogens. Similar endemic C. dif$cile clones have recently been identified in other UK hospita1s.‘7 Furthermore, it is not possible to determine using this approach whether some of the cases of apparent relapse (indistinguishable C. difj’icile strains between symptomatic episodes) are indeed re-infections with the original strain; for example, reacquisition of a strain from the hospital environment. Taking these factors into account it is probable that the great majority of symptomatic recurrences are due to C. difjcile re-infection as opposed to relapse. We specifically excluded patients in whom symptoms either did not actually cease temporarily (i.e., for at least three days) or where laxative administration was temporarily associated with a brief recurrence of loose stools. It is known that approximately a quarter of patients whose symptoms of C. difficile resolve still excrete the bacterium in faeces.““’ Hence, if we had not excluded the former patient categories from our analysis then the proportion of individuals with differing strains would probably have been less than the recorded figure of 56%. A recent study of risk factors for recurrences of C. difficile infection in 60 patients, included individuals whose symptoms had abated for only one day.20 Clearly, such an approach is likely to over-diagnose recurrences of infection. Also, the same study only fingerprinted sequential isolates from three cases, and will therefore inevitably have determined risk factors for a mixture of C. difficile relapses and re-infections.20 The implications of re-asserting that re-infections account for a large proportion of the observed recurrences are first that there is little rationale for using new or experimental treatment regimens, such as brewer’s yeast,’ for re-infections as opposed to relapses. Also, there is a need to reduce the use of oral vancomycin for the treatment of C. difficile infection, given its selective pressure for vancomycin-resistant enterococci,2’ and such data further weaken the justification for its prescription as a second-line agent to metronidazole. Our findings imply that the explanation for the frequent success of the tapering/pulse dose vancomycin regimen that is sometimes used for patients with multiple symptomatic recurrences,22 is that the prolonged antibiotic course merely prevents reacquisition of C. difficile, during the susceptible period when the patient’s bowel flora remains ineffectual against bacterial colonization. The original belief was that the regimen acted by allowing antibiotic resistant spores to germinate, and then vegetative bacteria are killed by subsequent vancomycin doses. Instead, more attention must be given to preventing the re-infection of patients, i.e., reinforcement of infection control procedures. Unfortunately,

Relapse

or

reinfection

of C.

difficile

99

patients with symptomatic C. difjkile infection are likely to remain susceptible to acquisition of further strains due to impaired colonization resistance secondary to disrupted gut flora. It is interesting to speculate that by isolating patients with C. difficile infection, Eve may be creating enclosed ‘C. dif$ciZe enriched’ environments.2”24 Such theorizing does not justify the abolition of isolation procedures for patients with C. diffirile infection. It is clear that endemic C. difJiciIe clones will be further disseminated in the hospital environment if symptomatic patients are not isolated. Given the relative resistance of C. dif$cile spores to conventional disinfection we must define and ensure compliance with effective procedures for environmental decontamination. Our standard cleaning protocols for side rooms include daily use of neutral detergent and a chlorine-releasing sanitizing powder. Interestingly, Samore et al.” recently found extreme variations in the extent of environmental contamination by C. difficilr in the rooms of symptomatic patients, but it is not known \vhy such differences exist; they used a cluaternary ammonium compound for routine en\Gronmental disinfection. Not surprisingly, the prevalence of hospital environmental contamination correlated significantly with the extent of hand carriage by personnel of C. dif$cilr.” It is also pertinent to ask why some individuals in particular suffer multiple symptomatic recurrences of C’. d$$cile infection. ‘I’h ele . is evidence of a poor immune response to C‘. difficile toxins in such patients.“’ We found that the median age of patients \vith recurrences \vas X3 years, and elsewhere it has been shown the pre\ralence of antibodies against C‘. difficile is reduced in some elder11 individuals.27 Our findings also complicate the documentation of the cpidcmiology of C’. dif$cile infection. Patients Ivith repeat C. difficile-positive results ma! be excluded from laboratory returns of cases of infection. Our data impl! that this is an erroneous practice and will result in an underestimation of the numbers of genuine (new) episodes of C’. dif$cile infection. ITinally, we have demonstrated that is prudent in cases of multiple symptomatic recurrence to fingerprint successive C. dzffi . ‘ci 1e isolates in order to determine appropriate therapeutic strategies. As many laboratories rely on tosin detection rather than culture for the diagnosis of c‘. difficile infection, WY recommend that toxin-positive specimens be stored frozen so that isolates may later be recovered if necessary for fingerprinting studies. \Ve thank I,ill,Industries the infection control staff

Limited for their

for an educational help.

grant

to support

this

study,

and

References 1. Djuretic people. 2. CDSC. 3. Cartmill

‘I‘, Ryan

MJ, Fleming DM, \Vall PC;. Infectious intestinal disease in elderI> Dis Rep CDR Rev 1996; 6: R107-R112. Commun Dis Rep C’DR Wkly 1997; 7: 250. TDI, Panigrahi H, Worsley MA, McCann DC, Nice CE, Keith E. Management

Commun

all

100

4 5. 6.

7. 8.

9. IO. 11. 12.

13.

14. 15.

16. 17.

IX. 19. 20.

21. 22. 23. 24. 25.

26.

27.

M. H. Wilcox

et al.

and control of a large outbreak of diarrhoea due to Clostridium difficile. J Hosp Infect 1994; 27: l-l 5. Wilcox MH, Cunniffe JG, Trundle C, Redpath C. Financial burden of hospitalacquired Clostridium dij$cile infection. J Hasp Inject 1996; 34: 23-30. Wilcox MH, Spencer RC. Clostridium dif$cile infection: responses, relapses and reinfections. J Hosp Inject 1992; 22: 85-92. Johnson S, Adelmann A, Clabots CR, Peterson LR, Gerding DN. Recurrences of Clostridium dif$cile diarrhea not caused by the original infecting organism. J Inject Dis 1989; 159: 340-343. O’Niell GL, Beaman MH, Riley TV. Relapse versus reinfection with Clostridium dijficile. Epidemiol Infect 1991; 107: 627-635. Chachaty E, Saulnier P, Martin A et al. Comparison of ribotyping, pulsed field gel electrophoresis and random amplified polymorphic DNA for typing Clostridium dijficile strains. FEMS Micro&o1 Lett 1994; 122: 61-68. Schellenberg D, Bonington A, Champion CM, Lancaster R, Webb S, Main J. Treatment of Clostridium dif$cile diarrhoea with brewer’s yeast. Lancet 1994; 343: 171-172. Tvede M, Rask-Madsen J. Bacteriotherapy for chronic relapsing Clostridium dij$cile diarrhoea in six patients. Lancet 1989; i: 1156-1160. Wilks M, Tabacqchali S. Typing of Clostridium difJicile by polymerase chain reaction with an arbitrary primer. J Hosp Infect 1994; 28: 231-234. Kilgore GE, Kato H. Use of arbitrary primer PCR to type Clostridium dijjicile and comparison of results with those by immunoblot typing. 3’ Clin Microbial 1994; 32: 1591-1593. Samore MH, Kristjansson M, Venkataraman L, Degirolami PC, Arbeit R. Comparison of arbitrarily-primed polymerase chain reaction, restriction enzyme analysis and pulsed field gel electrophresis for typing C. J Microbial Methods 1996; 25: 215-224. Power EGM. RAPD typing in microbiology-a technical review. J Hosp Inject 1996; 34: 247-65. Collier MC, Stock F, DeGirolami PC, Samore M, Cartwright CP. Comparison of PCRbased approaches to molecular epidemiologic analysis of Clostridium difficile. J Clin Microbiol 1996; 34: 1153-I 157. Prior SE, Hudson MJ, Minton NP. Analysis of Clostridium dijficile by pulsed field gel electrophoresis. PHLS Microbial Digest 1996; 13: 162-I 64. O’Neill GL, Brazier JS, Magee JT, Duerden BI. A comparison of PCR ribotyping and pyrolysis mass spectrometry for typing isolates of Clostridium difficile. Anaerobe 1996; 2: 211-21s. Bartlett JG. Treatment of antibiotic-associated pseudomembranous colitis. Kea Infect Dis 19X4; 6 (Suppl. 1): S235-S241. Fekety R, Silva J, Buggy B, Deery HG. Treatment of antibiotic associated colitis with vancomycin. J Antimicrob Chemother 1984; 14 (Suppl. D): 97-102. Fekety R, McFarland LV, Surawicz CM, Greenberg RN, Elmer GW, Mulligan ME. Recurrent Clostridium difficile diarrhea: characteristics of and risk factors for patients enrolled in a prospective, randomized, double-blinded trial. Clin Infect Dis 1997; 24: 324-333. Murray BE. What can we do about vancomycin-resistant enterococci? Clin Znject Dis 1995; 20: 1134-1136. Tedcsco FJ, Gordon D, Fortson WC. Approach to patients with multiple relapses of antibiotic associated pseudomembranous colitis. Am J Gastroenterol 1985; 80: 867-868. Sanderson P, Richardson D. Do patients with Clostridium difficile need to be isolated? J Hosp Inject 1997; 36: 157-I 58. Wilcox M, Settle C, Fawley W et al. Isolation of patients with Clostridium dij$ciZe infection. J Hasp Infect 1997 (in press). Samore MH, Venkataraman L, DeGirolami PC, Arbeit RD, Karchmer AW. Clinical and molecular epidemiology of sporadic and clustered cases of nosocomial Clostridium difficile diarrhea. Am J Med 1996; 100: 32-lO. Warny M, Vaerman JP, Avesani V, Delmee M. Human antibody response to Clostridium dijficile toxin A in relation to clinical course of infection. Infect Immun 1994; 62: 384-389. Bacon AE III, Fekety R. Immunoglobulin G directed against toxins A and B of Clostridium difficile in the general population and patients with antibiotic-associated diarrhea. Diagn Microbial Infect Dis 1994; 18: 205-209.