Gut, 1979, 20, 467-475

Morphology of experimental antibiotic-associatedenterocolitis in the hamster: a model for humanpseudomembranous colitis and antibiotic-associateddiarrhoeaA. B. PRICE, H. E. LARSON, AND JULIE CROW

From Northwick Park Hospital and Clinical Research Centre, Harrow, Middlesex

SUMMARY The morphology of antibiotic-associated enterocolitis in the hamster is described andcompared with human antibiotic-associated pseudomembranous colitis. It is shown to be a caecaldisease with proliferative mucosal changes and in this respect unlike the human counterpart. Thebacteriology and toxicology, however, are identical. In addition, mucosal changes are described inanimals on antibiotics but without established enterocolitis. As a result we suggest that there may bea spectrum of human disease ranging from mild antibiotic-associated diarrhoea to establishedpseudomembranous colitis. Therefore, despite the morphological variation, the hamster remains a

good model for investigating the pathogenesis of pseudomembranous colitis and antibiotic-associatedenteropathy in general.

Rapid progress in the aetiology of antibiotic-associated pseudomembranous colitis (PMC) hasfollowed the intial finding by Larson et al. (1977) ofa specific toxin in the stools of affected individuals.The toxin was further identified by showing that itscytopathic effect on cell cultures was neutralised byClostridium sordellii antitoxin (Larson and Price,1977; Rifkin et al., 1977) though C. sordellii couldnot he isolated. Next Bartlett et al. (1978) isolatedtoxigenic C. difficile from patients with the disease.Many workers (George et al., 1978b; Larson et al.,1978) since then have confirmed these findings anddemonstrated cross-neutralisation between the toxinof C. difficile and C. sordellii antitoxin (George etal., 1978a). It seems established that proliferationof toxigenic strains of C. difficile in the humanbowel is one mechanism in the pathogenesis ofpseudomembranous colitis.

In parallel with the findings in humans similarresults have been produced in Syrian golden ham-sters. Small (1968) initially demonstrated that lin-comycin produced a fatal enterocolitis when givento hamsters. Because lincomycin and the relatedclindamycin have been the antibiotics at the centreof a spate of reports on antibiotic-associated pseudo-membranous colitis (Cohen et al., 1973; Scott et al.,

Received for publication 25 January 1979

1973), the hamster was the obvious choice as anexperimental animal model. C. difficile and its toxinhave been isolated from hamsters with enterocolitis(Rifkin et al., 1978; Bartlett et al., 1977b) and theorganism will reproduce the enterocolitis whenreinoculated into susceptible animals (Larson et al.,1978). In this paper we report on the morphology ofthe hamster disease and compare it with humanantibiotic-associated PMC, we also discuss thespectrum of pathology encompassed by the termantibiotic-associated diarrhoea.

Methods

Syrian hamsters, 1-2 months old, were supplied bythe Animal Division, National Institute for MedicalResearch, Mill Hill, London. Strains of C. diffcileisolated from both hamsters and human cases ofPMC and identified by standard methods, includinggas liquid chromatography of glucose fermentationproducts(Holdeman and Moore, 1975) were providedby P. Honour of the Division of Hospital Infection,C.R.C. All produced toxin in broth culture.

Enterocolitis was produced in the animals bytwo methods, either by injection of clindamycin,method A, or by oral administration of C. difficile,method B. The following procedures were used(Table 1).467

468

Table 1 Summary of treatments

Method Group Antibiotic Bacterialsuspension

A 1 Clindamycin i.p. Nil2 Saline i.p. Nil

B 1 Vancomycin p.o. C. difficile p.o.2 Vancomycin p.o. Sterile broth p.o.3 Nil C. diffcile p.o.

METHOD AGroup Al. Thirty animals were given a single intra-peritoneal injection of clindamycin (Upjohns Ltd),50 mg/kg.Group A2. Six animals were given 05 ml of nor-

mal saline intraperitoneally.

METHOD B

Group BI. Thirty-six animals were given vancomycin(Eli Lilly Co. Ltd), 3 mg/ml, in their drinking waterfor seven days, then three days of tap water only.They were then force fed 0 5 ml of a toxin-freesuspension of C. difficile. Twenty animals received ahuman strain isolated from a case of PMC and 16a hamster strain isolated from an animal who hadsuccumbed to clindamycin-induced disease.Group B2. Thirty-six animals were pretreated

with vancomycin as in group Bi but then given 0 5ml of sterile broth instead of C. difficile.Group B3. Another group of 12 hamsters were

given seven days tapwater and then fed 0 5 ml of atoxin-free suspension of C. difficile (six a hamsterstrain, six a human strain).

All animals given vancomycin were kept in exhaust-ventilated isolators equipped with air filters on bothinlets and extracts. Animals were inspected daily fordiarrhoea or other evidence of illness. During theexperiment animals were killed only if they wereseverely ill, except that two animals from group Biand 2 from group B2 were killed daily for four days(16 animals).

Necropsies were carried out on all animals and,as a routine, 5-10 cm of terminal ileum, the caecum,and 5-10cm of colon were pinned out on a cork boardand fixed for 24 hours in 10% formol-saline. Blocksfrom the ileum, caecum, and colon were taken andstandard paraffin embedded, 5 IL haematoxylin andeosin stained sections prepared. A Gram-stain fororganisms and a Martius Scarlet Blue stain todemonstrate fibrin were carried out on selectedcases. Representative areas of caecum were chosenfrom both diseased and clinically normal animals forelectron microscopic examination. They were fixedin 3% glutaraldehyde in cacodylate buffer andstandard Spurr resin embedded 1 ,u and ultra-thinsections cut. The latter were examined on a PhillipsEM 300 microscope.

A. B. Price H. E. Larson, and Julie Crow

The caecal contents of all animals coming tonecropsy were tested for the presence of toxin bymethods previously described (Larson and Price,1977).

Results

Figure 1 summarises the clinical outcome. Sixty-threeper cent of animals given clindamycin intraperi-toneally died or were moribund in the first week and17% survived for three weeks. No animals receivingvancomycin followed by C. difficile survived beyondone week. Sick animals became sluggish and usuallyremained hunched up in one corner of their cage.Diarrhoea was manifest by wet fur around the anus.However, some animals were found dead withoutshowing prior symptoms or diarrhoea. No deathsoccurred in any of the control groups (A2, B2, B3).

40-

'D

c

-U

0E 30-~0a,-u

20-aE

10-

0 -

E: BlCO AlE B2* A2[ B3

Surviving animalskilled I

1 2 3 4 Week

Vancomycin and C.cffficileClindamycin pVandomycin and sterile broth-Saline ip ControlWater and C. difficile animals

Fig. 1 The outcome of each experimental method.

NECROPSY FINDINGSIntestinal disease was the only abnormal finding.Animals with enterocolitis had similar pathologywhich was independent of the experimental method.Disease invariably involved the caecum, usually theterminal 2-5 cm of ileum and rarely the proximalcolon.The caecum showed two main patterns of disease

that did not comply with any one experimental

Morphology of experimental antibiotic-associated enterocoliuis in the hamster

method nor with the time of death. One pattern wasof gross distension with widespread haemorrhage(Fig. 2) the caecal contents being fluid and the wallpaper-thin. The second pattern showed a caecumwith normal or reduced dimensions but having athickened velvet texture to the mucosa. Haemor-rhage, if present, was focal or punctate. Ulcerationwas noted in isolated cases only, the ulcer floorbeing covered by a white slough. The gross ileallesions were those of serosal reddening and mucosalthickening as in the caecum. In the few animalswhich showed colonic disease serosal congestion wasthe only macroscopic finding.

The caecums of animals receiving oral vancomycinfollowed by sterile broth (B2) were also found to beabnormal. They were enlarged and contained semi-fluid faeces, although there was no gross muralthickening. Animals belonging to groups A2 and B3were normal.

HISTOPATHOLOGYAll the animals treated with clindamycin or van-comycin and C. difficile who were found dead ormoribund invariably had caecal disease. Thehistological appearances correlated with the twomacroscopic patterns noted at necropsy.

Fig. 2 This compares a normalcaecum on the left with, on theright, a dilated haemorrhagiccaecum from an animal treatedwith dlindamycin intraperitoneally... .......

...... ........... '..

~is. 0Fig. 3 Caecal mucosa from thediseased animal in Fig. 2

l showing diffuse intramucosalhaemorrhage.

469

A. B. Price, H. E. Larson, and Julie Crow

The animals with caecal dilatation showed diffusemucosal haemorrhage. The caecal glands survivedin the midst of the haemorrhage (Fig. 3), though ..

in many areas the surface epithelium was lost.Submucosal vascular thrombi were an isolated alfinding in occasional animals.The majority of animals showed proliferative

changes of the caecal and ileal mucosa with accom-panying inflammation (Fig. 4). This was a diffuse *change in the caecum and accounted for the velvettexture noted macroscopically. It was accompaniedby vascular engorgement and variable amounts offocal haemorrhage. The caecal crypts were elongatedand the intercrypt surface epithelium showedflorid 'tufting' (Figs 4, 5), the cells having analmost syncytial arrangement. This was in markedcontrast to the neat smooth outline of normal caecalmucosa (Fig. 6). At times the proliferation was soflorid and villiform it became difficult to distinguishdiseased caecum from ileum. .

Degenerative changes were also present in thesurface cells. The individual enterocytes had losttheir brush border, were pale staining, frequentlyvacuolated, and had enlarged pale nuclei which wereno longer situated at the base of the cell. "The mucosal changes were accompanied by a C.

variable mixed inflammatory cell infiltrate in thelamina propria. Polymorphs were seen invadingthe surface and migrating into glands to form crypt.....abscesses and a plasma cell infiltrate was a constantfeature (Fig. 5). Inflammation was maximal in the Fig. 4 A caecum showing a greatly thickened mucosamucosa and superficial submucosa but did extend with irregularity of the luminal surface and lengtheninginto the muscle coat. Ulceration was noted in of the crypts. Inflammatory cells are present but are notoccasional cases and when present extended into the a conspicuous feature. Magnifications quoted in this andsubmucosa. A surface inflammatory slough covered the following figures are original magnifications. H andthese ulcers. E. x 90 (animal given vancomycin and C. difficile).~~~~~~~~~~~~

A0. AR ~~~~~~~~~~Fig.5 The caecum of ahamster with enterocolitisshowing the tufted appearance ofthe epithelium and loss ofsrface nuclear and cellular

regularity. A crypt abscess isseen in the centre. H and E,

Vb 225 (animal given clindamycin( r ~~~~~~~~~~~~~~~~~~ ~~itraperitoneally).

470

Morphology of experimental antibiotic-associated enterocolitis in the hamster

Fig. 6 Normal caecum. Theregular surface is a contrast withthat seen in animals with toxinpositive enterocolitis (Fig. 5)and that in animals treated withvancomycin followed by sterilebroth (Fig. 8). H and E, x 225.

Fig. 7 The caecum from one ofonly two animals that showedareas of ulceration with'structured' mucosal necrosis anda fibrinoid-like exudate in thelamina propria. These appearancesdo resemble those seen in humanpseudomembranous colitis butwere restricted to the two animals.H and E, x 225 (animal givenvancomycin and C. difficile).

In two cases ulceration and 'structured' mucosalnecrosis was noted (Fig. 7) with fibrinoid-likematerial in the lamina propria. The Gram stainsshowed mixed surface collections of gram positiveand negative organisms in all cases. Staining forfibrin showed that only an occasional vessel inisolated cases contained thrombus.The changes in the ileum were of a similar nature

to those in the caecum. With the increase in thicknessof the crypts the ratio of villus height to cryptdepth decreased and ileum came to resemble caecumin overall architecture. A similar proliferativepattern was noted in the proximal colon in the fewcases where it was involved.

Among surviving animals the histopathologywas normal except for the animals given vancomycinorally followed by sterile broth. In these the changeswere limited to the caecum. The basic normal un-dulant pattern was preserved but the surface in theintercrypt areas showed heaping up of cells formingbud-like outgrowths and the individual cells hadlost their nuclear polarity (Fig. 8). Their staining wasvariable and cytoplasmic vacuolation was common.The brush border was blurred and the nucleihyperchromatic and angular. An increase in in-flammatory cells in the lamina propria was oftenpresent. Having observed this change individualhamsters given oral vancomycin then broth (group

471

A. B. Price, H. E. Larson, and Julie Crow

Fig. 8 The caecum from ananimal treated with vancomycinfollowed by sterile broth. Thereis surface irregularity andtufting but only ofa minordegree compared with animalsthat had fully developedenterocolitis (Fig. 5). H and E,x 225.

-._4:4 ., u Wi-W.- ,1-.MWN: . .-. - 1

Fig. 9 An electron micrograph ofsurface caecal mucosa from an animal with enterocolitis treated with vancomycinand C. difficile. Individual cells and their microvilli show irregularity with vacuolar degeneration of the cytoplasmNo intracellular organisms can be seen. x 16 200.

472

no.44.

Morphology of experimental antibiotic-associated enterocolitis in the hamster

B2) were killed on alternate days for 16 days. Thesechanges were maximal immediately after thecourse of vancomycin and diminished during thenext 21 weeks.

ELECTRON MICROSCOPYElectron microscopic examination of the fullydeveloped proliferative caecal disease showed thatthe surface epithelial cells were of uneven height anddensity. The cell surfaces showed marked 'tufting'(Fig. 9) and the micro-villi were irregular and some-times absent. Degenerative changes were noted incytoplasmic organelles and large cytoplasmicvacuoles were frequent. Although micro-organismswere seen in the lumen intracellular organisms werenot a feature.

In the normal animals occasional vacuolated anddegenerating cells were seen but the cell surfaces weresmooth and the microvillous border was regular.Sections from an animal in group B2 (oral van-comycin followed by broth), did, however, show someirregularity of the cell surface and microvilli,although the changes were less marked than in thoseanimals with established enterocolitis.

TOXIN ASSAYSThe caecal contents of all animals inoculated byeither method that died or were moribund within thefirst two weeks contained toxin to a titre of at least1:1,000. Five animals from group Al survived threeweeks (Fig. 1) and when killed no toxin was found.The toxin assays were negative in all the otheranimals that survived three weeks.

SEQUENTIAL KILLING OF ANIMALSINOCULATED BY METHODS B I AND B2The results are summarised in Table 2. On day 0both sets of animals had similar histology but weretoxin negative. On day 1 one animal challenged withC. difficile had a tiny but definite macroscopiccaecal haemorrhage but caecal contents were stilltoxin negative. Microscopy was similar to thevancomycin-only animals but more widespreadand in particular the crypts were beginning tolengthen. By day 2 the animals given C. difficile weretoxin positive, showed caecal thickening, inflam-mation and vascular engorgement. By day 3 oneanimal was dead with florid enterocolitis involvingthe ileum and caecum.

Discussion

Unlike other workers (Onderdonk et al., 1977;Lusk et al., 1978), our initial attempts to studyclindamycin induced enterocolitis in hamsterswere hampered by failure to produce universal

Table 2 Summary of results in experimentinvolving sequential killing of animals in groups B] andB2

Groups Vancomycin 7 days thenBJ, B2

C. difficile Sterile broth

Morphology Toxin Morphology Toxinassay assay

Day0

2

3

Intercrypt surface )changes

Nega-Pronounced surface tivechanges. Crypts |enlarged. FocalhaemorrhageObvious proliferativecaecitisObvious proliferative IPositivecaecitis; 1 animaldead

Intercryptsurfacechanges Negative

J

disease (Fig. 1). We thought this happened becausesome animals were not colonised with C. difficile.To produce uniform colonisation we treated animalswith vancomycin to eliminate indigenous C. difficileand then inoculated this organism by mouth. Thistechnique resulted in a universally fatal outcome(Fig. 1). Pathology and toxin assays show that bothmethods, clindamycin intraperitoneally and van-comycin orally followed by C. difficile, produce anidentical enterocolitis. C. difficile inoculation withoutprior antibiotic treatment had no effect and animalsgiven vancomycin alone then kept free from anyclostridial exposure also survived and remainedtoxin negative (Fig. 1). However, to ensure theselatter conditions the animals had to be housed inisolators.

Several comparisons with human pseudo-membranous colitis are suggested. First, humanderived strains of C. difficile induce the same diseasein hamsters as the hamster derived strains. Second,the necessity for antibiotic pretreatment mimics mosthuman situations. Thirdly, the association andproperties of the toxin with established enterocolitisis similar in both (Bartlett et al., 1977b; Rifkin et al.,1978). However, unlike human antibiotic-associatedpseudomembranous colitis, enterocolitis in thehamster invariably occurred in the caecum, often inthe terminal ileum, but rarely in the colon. Sodifferent is the physiological role of the caecum inthe hamster from that in the human that an anato-mical variation in the pathology is not too surprising.But what is difficult to explain are the markedly

different histological changes. We stress that amembrane or plaque is rarely found in hamsterdisease (Fig. 7) (Bartlett et al., 1977a; Lusk et al.,1978). The characteristic histology was either

473

474 A. B. Price, H. E. Larson, and Julie Crow

mucosal epithelial proliferation accompanied bysurface degenerative changes (Figs 4, 5) or mucosalhaemorrhage (Fig. 3), and not the focal superficialcrypt necrosis with exudation, typical of humanpseudomembranous colitis (Price and Davies, 1977).When ulceration was present in the hamster caecumthe covering inflammatory debris was consideredto be non-specific.

It is not clear how C. difficile infection producesthese changes. Gram stains and electron microscopyfailed to show organisms invading the mucosaand this is also true for human pseudomembranouscolitis (Steer, 1975). Oral administration of thetoxin does not produce lesions (unpublished data),while intraperitoneal injection produces fatalsystemic toxicity with widespread haemorrhagiclesions (Rifkin et al., 1978). Intracaecal injections oftoxin have been reported to give caecal lesions(Bartlett et al., 1977b) but in our experience withthis method the animals die within 24 hours fromthe systemic effects. Administration of largeamounts of C. sordellii antitoxin protects hamstersfrom death after clindamycin challenge butthe histological findings have not been reported(Rifkin et al., 1978). Certain clostridial toxins areknown to be vasoconstrictors and might induceischaemic damage (Marston, 1977). This wouldaccount for the haemorrhagic element seen in theenterocolitis of the hamster and for the similarityof human pseudomembranous to ischaemic colitis.Alternatively, the toxin may have a direct mucosaleffect (McDonel and Duncan, 1975) that wouldexplain the initial small necrotic foci seen in thehuman disease (Price and Davies, 1977). However,neither mechanism satisfactorily explains the pro-liferative lesions described in the hamster model.Mucosal proliferation as a response to bacterialinfection is unusual in humans but recognised incertain animal species. It is seen in wet-tail (proli-ferative ileitis) a common hamster intestinal in-fection with many similarities to the enteritis seenhere (Jacoby, 1978), and it also occurs in forms ofswine dysentery (Glock et al., 1974). Surprisingly theadministration of vancomycin alone caused mildcaecal changes (Fig. 8). In the most florid examplesthe pattern was difficult to distinguish from that ofmild toxin positive enteritis, but the caecal contentsremained toxin negative. The histological patternsdiverged in parallel with the appearances of toxin asseen in the experiment where the animals were killedat daily intervals (Table 2). It is tempting to postulatethat the morphological changes produced by clinda-mycin, vancomycin followed by C. difficile, andvancomycin alone (Figs 5, 8) are part of a singlespectrum. This graded response to antibiotics mighthave a close human parallel. It is well established

that certain patients who receive antibiotics developdiarrhoea (Tedesco, 1975); it is not known if thereare any mucosal abnormalities at this stage. Aminority of patients in this 'susceptible' state go onto develop a definite colitis which is either non-specific or pseudomembranous in type (Gibson et al.,1975). It is in the latter group that C. difficile isfound in the gut.

It is also important to know whether both thehuman and animal disease are endogenous orexogenous infections. The spectrum of proliferativehistological changes suggest a gradual developmentperhaps associated with an endogenous infection.However, theanimals givenvancomycin aloneremain-ed well if not exposed to the organism (Fig. 1), a fea-ture of particular importance to clinical managementsince vancomycin is used to treat human pseudo-membranous colitis (Tedesco etal., 1978). So far, workinhumans suggests that C.difficile isnotanendogenousmember of the normal adult flora (Larson et al.,1978) but more extensive epidemiological studies areneeded. Exogenous infection after clindamycinwould also best explain the results shown in Fig. 1.Our findings suggest that the hamster, despite

certain morphological differences, is still a goodmodel in which to study the wider spectrum ofantibiotic-associated intestinal enteropathy, in par-ticular the role of various antibiotics in the spectrumof clinical disease encompassing diarrhoea, non-specific colitis, and pseudomembranous colitis. It isalso a model for the conditions that allow C.difficile to become established as an intestinalpathogen.

We would like to thank the following for technicalhelp: Chris Sowter, David Gunner, Sheila Holt, theAnimal Histopathology Department, and the Div-ision of Comparative Medicine. We would also liketo thank John Clark for the photography and JillianJones for the preparation of the manuscript.

References

Bartlett, J. G., Chang, T. W., Gurwith, M., Gorbach, S. L.,and Onderdonk, A. B. (1978). Antibiotic-associatedpseudomembranous colitis due to toxin-producingClostridia. New EnglandJournal ofMedicine, 298, 531-534.

Bartlett, J. G., Onderdonk, A. B., and Cisneros, R. L. (1977a).Clindamycin-associated colitis in hamsters: protectionwith vancomycin. Gastroenterology, 73, 772-776.

Bartlett, J. G., Onderdonk, A. B., Cisneros, R. L., andKasper, D. L. (1977b). Clindamycin-associated colitisdue to a toxin-producing species of clostridium in ham-sters. Journal of Infectious Diseases, 136, 701-705.

Cohen, L. E., McNeill, C. J., and Wells, R. F. (1973)Clindamycin-associated colitis. Journal of the AnmericanMedical Association, 223, 1379-1380.

George, W. L., Sutter, V. L., Goldstein, E. J. C., Ltudwig,S. L., and Finegold, S. M. (1978a). Aetiology of anti-

Morphology of experimental antibiotic-associated enterocolitis in the hamster 475

microbial-agent-associated colitis. Lancet, 1, 802-803.George, R. M., Symonds, J. M., Dimock, F., Brown, J. D.,

Arabi, Y., Shinagawa, N., Keighley, M. R. B., Alexander-Williams, J., and Burdon, D. W. (1978b). Identification ofClostridium difficile as a cause of pseudomembranouscolitis. British Medical Journal, 1, 695.

Gibson, G. E., Rowland, R., and Hecker, R. (1975). Diar-rhoea and colitis associated with antibiotic treatment.Australian and New Zealand Journal of Medicine, 5,340-347.

Glock, R. D., Harris, D L., and Kluge, J. P. (1974). Local-ization of spirochetes with the structural characteristics ofTreponema hyodysenteriae in the lesions of swinedysentery. Infection and Immunity, 9, 167-178.

Holdeman, L. V., and Moore, W. E. C. (1975). AnaerobeLaboratory Manual. Virginia Polytechnic Institute;Blacksburg, Virginia.

Jacoby, R. 0. (1978). Transmissible ileal hyperplasia ofhamsters. American Journal ofPathology, 91, 433-450.

Larson, H. E., Parry, J. V., Price, A. B., Davies, D. R.,Dolby, J., and Tyrell, D. A. J. (1977). Undescribed toxin inpseudomembranous colitis. British Medical Journal, 1,1246-1248.

Larson, H. E., and Price, A. B. (1977). Pseudomembranouscolitis: Presence of Clostridial toxin. Lancet, 2, 1312-1314.

Larson, H. E., Price, A. B., Honour, P., and Borriello, S. P.(1978). Clostridium diffilcile and the aetiology of pseudo-membranous colitis. Lancet, 1, 1063-1066.

Lusk, R. H., Fekety, R., Silva, J., Browne, R. A., Ringler,D. H., and Abrams, G. D. (1978). Clindamycin-inducedenterocolitis in hamsters. Journal of Infectious Diseases,

137, 464-475.McDonel, J. L., and Duncan, C. L. (1975). Histopatho-

logical effect of Clostridium perfringens enterotoxin in therabbit ileum. Infection and Immunity, 12, 1214-1218.

Marston, A. (1977). Intestinal Ischaemia. Edward Arnold:London.

Onderdonk, A. B., Hermos, J. A., and Bartlett, J. (1977).The role of intestinal microflora in experimental colitis.American Journal of Clinical Nutrition, 30, 1819-1825.

Price, A. B., and Davies, D. R. (1977). Pseudomembranouscolitis. Journal of Clinical Pathology, 30, 1-12.

Rifkin, G. D., Fekety, F. R., and Silva, J., Jr. (1977).Antibiotic-induced colitis implication of a toxin neutralisedby Clostridium sordellii antitoxin. Lancet, 2, 1103-1106.

Rifkin, G. D., Fekety, F. R., and Silva, J. (1978). Neutral-ization by Clostridium sordellii antitoxin of toxins im-plicated in clindamycin-induced cecitis in the hamster.Gastroenterology, 75, 422-424.

Scott, A. J., Nicholson, G. I., and Kerr, A. R. (1973).Lincomycin as a cause of pseudomembranous colitis.Lancet, 2, 1232-1234.

Small, J. D. (1968). Fatal enterocolitis in hamsters givenlincomycin hydrochloride. LaboratorY Animal Care, 18,411-420.

Steer, H. W. (1975). The pseudomembranous colitis associatedwith clindamycin therapy-a viral colitis. Gut, 16, 695-706.

Tedesco, F. J. (1975). Ampicillin-associated diarrhoea-Aprospective study. American Journal of Digestive Diseases,20, 295-297.

Tedesco, F., Markham, R., Gurwith, M., Christie, D., andBartlett, J. G. (1978). Oral vancomycin for antibiotic-associated pseudomembranous colitis. Lancet, 2, 226-228.