Colitis caused by Escherichia coli 0157 : H 7 : A study of six cases

The clinical and pathologic features of six pa tients with proven vcrotoxin-producing Escherichia coli colitis are described . C lin ical data, 25 biopsy ~recunens and two autopsies from these patien ts are reviewed. A ll presented with crnmpy abdomina l pain and bloody d iarrhea. Colonoscopic find ings included eJema, erythema, pseudomembranes and hemorrhage. O n hiopsy, two patients had ischemic colitis, one had pseudomembranous colitis and th ree had a com1'ination of concurrent ischemic and pseudomem branous colitis. Four cases showed fibrin-platelet thrombi in mucosa! capillaries and submucosal a rte rio les. The classical patte rn of infect ious colit is was no t seen in these cases. Other niimpecific changes inc lud ed patchy mucosa! e<lema, congest ion, focally rrominent interstitial hemorrhage and mild, patch y increase of the lymphoplasmocytic component of the lamina. Ischemic necrosis was present in 10 of 25 biopsies (40% ), pseudomembranes in seven of 25 biopsies (28%), and four of25 riopsies (1 6%) showed both. Colon from one aut0psy revealed edema, pseudo membranes and intramura l infarc tio n . Concurrent thro mbotic thro mbocytopenic purpura was clinically documented in three of six patients. It is concluded that, in the context of hemorrhagic colitis, the following observations are indicative of E coli Ol 57:H7: the combination of pseudomcmbranous and 1schemic colitis; ischemic colitis in a yo ung pat ient; or pseudo membranous colit is with Closcridium difficile-negative culture and toxin. Multiple biopsies are reyuired to demonstrate the full-blown features. E coli 0 157:H 7 colit is should he aJJed to the differen tial diagnosis of submucosal edema. Can J Gastroenterol 1990;4(4): 141-146 (pour resume, voir page 142)

v: EROTOX!N-PRODl.X::TNG ESC/ !ERIC/ /IA coli O 157:117 i~ a n important cau~e of hemorrhagic colitis which is he ing di agnosc<l with in creasing frequency ( l ).The w mpl1cat io ns a rc po tent iall y life t h rea te nin g: he mo lyri v ure mt c syndrom e is we ll doc ume nted , and , more rece ntl y, th rombotic thrombocywpenic purpura has been observed (2)(3)(4).
The i<le nrifi cation of diagno~t ic pa t h n lo g 1c a l les io n s t h at s ugge st verotoxin -produc ing E coli in fec tion is important hecause clin ical markers of th e disease may be negari ve or inapparent.For instance, in the adu lt age group, hacteria arc shed in t he ~tool only fo r a few Jays to a week after onset o f sy mp toms (4,5) .T h e a ssay fo r veroto xm is positive for o nl y a few addit io nal days, a nd i~ not available in ma n y te rtiary ca re hos pitals.Thus, de lay by the pat ient in seeking me<l ical attention, or by cl inicians in ohta ini ng stool for c ulture, reduces the likelihood of isolating the organism.Thrombo tic t hrombocyto pen ic pur pura may dominate the clinical p ict ure to t he ext e nt tha t co lit is is overlooke d a ltogeth er (4).
The autho rs pre viously reported two cases of E coli 0 157: H7 colitis complicated by thrombotic thrombncytopeni c rurpura anJ briefly described the unusual finding of simultnneous ischcmic and pscu<lomcmhranous colitis (3).In the present rcpon, several pathological alteratiom arc descri bed, which, in comhination, strongly suggeM the diagnosis of verotoxin-producing E coli.

PATIENTS AND METHODS
Sixteen patients wnh pos iti ve Moo! cultures for E coli 0 157: 11 7 were identified from the records of the microbio logy labora tories of two teaching hospitals over a r eriod of one-and-ahalf yea rs.Six of these patients had a total of 25 colonic biopsies.T wn of the six patients died and the autopsy material was reviewed.The biopsies were seria lly sectio ned and stamed with hematoxylin and cosin.C linica l features, abdominal x-rnys, and endoscopic findings of these six patients were reviewed.RESULTS Clinical features: All six patients presented with a o ne to five day history of crampy abdominal pain, blnody diarrhea, nau~ea ,md vomiting.There was no history of trave l outside of Canada.Two patients received a few doses of c rythromycin.Four of the ca~cs were sporadi c.In the o ther two cases, a milder, self-limited form of diarrhea developed in other family members and soc ial contacts.Three patients developed thrombotic thmmbocycopcnic purpura (as defined by the presence of thmmbocyropenia, microangiopathic hemo lytic anemia, fever, normal proth rombin and partial rhromboplastin times, normc1I fibrinogen levels, and declining neurolog ical and rena l functions).Dilated, edemato us small and large howel ( with occasional thumbprinting) were seen on standard a bdominal x-rays (Figure I) .C olo noscopy showed erythema , edema and hemorrhage in all six patients and pseudomembranes in two (Figure 2).There was pancolitis in o n e patient with predominance on the right side.O nl y the left colon was visua li zed in the remaining five paticms.However, bleeding above rhc level of th e colo noscope was repo rted int woof these fi ve patients.di agnosed whe n the type 2 lesion desc ribed by Price a nd Davies (6) wa, identified (Figure 3h) .This as well ,h the other histological types I and~ were nmed in fourpmients.Mucosa !biopsic, in five patients revealed isc hem1c colitis defined ns va riable depth of coagulative necrosis, with or without type 3 pscudnmembranes (Figure k).
Ncut roph iliL infi lt ration of gland and lamina was not a feature of the biopsies, and o nly the rare microfocu, could be identified.The lymphopla,mocytic component of the lamina was only foca lly and mildly increased .Submucosa was included in two bior sies fro m the sa me patient and was markedly edemato us.
Two patient~ di ed; both were elderly and both cases were complicated hy thrombotic rhromhocytopenic purpura.At a utopsy, the colon of one patient s howed o nly ede ma and petechial hemorrhages.The second patient died fo llowing central ne rvous system compl icatiom of thrombo ti c thrombocytopenic purpura.In this case, the colon s h owed extensive isc hcmic nec rosis with focal pseu<lomcmhrane formation (Figure 4 ).

DISCUSSION
The sa I ient pathologic changes seen in these six cases of E coli Ol 57:H7 infection were isc hcmic col it is and all three morpho logical types of pseudo• membra nou s co liti s.Half of the patie nts h ad th e unusual finding of simultaneous pseudomembranous and ischemi c colitis.None of the biopsies showed the feature~ of infec tious colitis.Since ischemic colitis was iJcnrificd in only 10 of 2 5 biopsies anJ pseudomembranous colitis in only seven of 25, 1r is evident that muluple samples arc necessary for demonstralion of lhe lesions and for aJequale evalumion of rhe colonic mucosa.
Concurrenl bowel infarction and pscudomembranc formatinn have been Jescnbed in a subset of ischemic bowel disease under Jiagnostic terms such as nonocclusive inteslinal ischemia and pscudomcmhranous entcmcolilis.They occur a~ complications of various conditions such a~ shock, postoperative states and heart failure (6-9).ln contrast with E coli 0157:1-17 colilis, small bowel involvement is common and the prognosb is poor.On ncca~ion, pseuJo-membrane~ can abo be seen assoc iated wnh ischemic colitis in mesenteric artery occlusion.Usually only type 3 pscudomembranes, and not the full spectrum of types l to 3, are present in CAN J GASTROENTl:RL)L Vm 4 No 4 MAY/JUNE 1990 E co/i0157:H7 colitis occlusive ischemic colitis (8).In con-w1st, mucosa!hemorrhagic necrosis of the type seen in ischcmic bowel disease, as well as intramural and transmural infarction, arc not a feature of antibiotic-associated pseudomem hranous colitis.Thus, in the context of hemorrhagic coli tis, the simultaneous occurrence of pscudnmembranous colitis and ischcmic colitis seems to be a d istinctive finding of E coli Ol 57:H7 colitb.
The clin ical prcsentntton m many previously reported cases of E coli 01 57:H7, as well as in the present stu<ly, suggested ischcmic rather than infectious colitis (1,3,4).Three of the present six patiems were under 50 years of age, anti mucosa !ischemia was a component of the hisropmholngy in five cases.lschcmic colitis is unusual in young patients.Thus, its presence in young patients would also suggest a diagnosis of£ coli 0157:l 17.Conversely, the possibility of an infectious etiology may be overlooked in the e lderly patient presenting clinically with ischemic colitis.
A hiopsy showing pseuJomembranous o r pscudomembranous and 1schemic colitis may he the only clue ro E coli Ol 57:117 colitis.The tliagnosis of E coli Ol57:Il7 is suggested in any patient with biopsies showing pseudomem branous colitis but nega1 ive Clostridium difficile cultures or toxin.
Other pathological findings seen raJiologicall y and/or cntloscopically may also be helpfu l in pointing to the diagnosis of E coli Ol 57:I 17 .The differential diagnosis ofsuhmucosal edema of the colon b essentially limited to inflammatory bowel disease (particularly Cmhn 's disease), ischemic colitis, and pseudomembranous colitis.Since the basic pathology of E coli Ol57:H7 coli tis was that of ischemic, pscudomembranous, or pseudomemhranous and ischemic colitis, it is not ~urprising t hat subrnucosal edema was also present.Other colonic lesions as-soci<1ted with 1~chem1c colitis were also found in the prescm study: focal interstitial hemmrhage (six cases) anJ small vessel fibrin-platelet thrombi (four cases).The rare section showed capillary thrombi in otherwise normal mucosa.3a 3e 3d Figure 3) Colonic bio/Jsies.3a Segment of colonic mucosa where pseudomembranous colitis ( long arrow) is continuous wi1h ischemic co/iris ( short arrow).3b On higher magnification, the pseuclomembranous component is formed of rhe typical 'fire cracker' lesions and streaming exudate .3c There is full thickness infarction of the mucosa in 1he ischemic />orrion of the biopsy with thrombi in submucosal vessels and mucosa!capillaries (arrows).3d Several biopsies show a lesser degree of 1schemia identified only by aitenuation of the superficial portion of a few glands (arrows).3e Microf oci of superficial hemorrhages are common and result from rupture of large, di/aced mucosa/ capillaries ( short arrows ouiline capillary, long arrow the dismpiion) T he colonic patho logy obse rved here has been reported to a limited extent in the literature.One case reported by Morrison et al (2) was associated with pseudomembranes identified his-144 rologically but not recognized on endoscopy.Richardson et al ( LO) described three cases of E coli 0157:H7 colitis complicated by h e mo lytic-urem ic syndrome in the pediatric age group.
Pse udome mbran e fo rm a tio n was present in one case, (again seen only on mic roscopic examina tio n) and ischemic necrosis (personal communic;i.tion) in two cases (10) .None of these  Minimal findings on biopsy with rodiolog,colly or endoscopicolly proven submucosol edema or hemorrhage Biopsy-proven ischemlc colitis in on elderly patient report e<l cases s h owed pseu<lomembranous and 1sche mic colitis.In all other reported cases, the patho logical alterations in mucosa!biopsies showed a spectrum, including: no altcratinns; nonspecific changes including interstitial hemorrhage, mi ld increase of the lymphoplasmocytic component of the lamina propria and edema; neurrophilic infiltrates of the lamina propria and g la nd~ in the fashion of infectious col itis; and apL)pto:,is o f su rfa ce epithelium ( 1,4,5,11 ).It is surpnsing that on ly four of a ll of the reported cases s howed ischemic colitis or pseudo, membranous colitis, and that pseudomembranou:, and ischem ic colitis was not seen in any report of E coli O I 57:H7 infection; othcrassociated features such as submucnsal edema a nd hemorrhage CAN] GASTROENTERl)L VL)L 4 NO 4 MA Y/JLJNE 1990 were prominent findings in most cases.Possible explanat ions for ch is di screpancy include mucosa I sa mpling limited to th e distal colon (wirh the mo re seve re prox imal d isease nor visuali zed endoscopically) and a patient population with a milder form of colitis than observed in the present stud y.Indeed, th ree of the six cases were complicated by thrombotic thrombocywpen ic purpura and two of the six patients died, a far higher percentage of complications than in ocher se ri es.
The relationship betwee n pse udom e m bra n o us col ir is and isc h e mic colitis has heen extensively d eba ted.So m e authors co n si d e r pseu<lome mbran o u~ col itis as pan of the spectrum of isch emic colitis, whereas o th ers regard pseudomembrano us colitis and ischemic colitis as distinct a nd sepa rate e ntiti es.T he findings

Eml sec al
JescribcJ in the present six cases suggest that pseuJomemhranous col it is and 1schemi c colitis arc inJeed o n a cnntinuum of the :mme disease.The mechan1,m hy whicL the pseudomembrane forms has been suggested by Norris in h is review of the spectrum of ischemic bowel disease ( 12).Whereas total cessation of h loodfl ow to bowel results in coagulative necrosis, a n acute inflammatory in filtrate Jcvclops in bowel mucosa when colo nic bloodtlow ceases and is then re-established.N orris suggests that thb process evnlves m pseuJomem bra ne format ion.
The nuthurs have d emonstrated in these cases of E coli Ol 57:H7 colitis, disruption a nd /or thrombosis of mucosal cap1llan es and, to a lesser extent, thrombosb of small submucosal arterioles.Lysb of these thrombi or collatera l blood supply from the rich mucosa!plexus to the ,,chemic mucosa may result in reflnw uf blood, acute in-""nuated wnh r\chenchw coli 0157:H7.Ann Intern Med l 984; LOI :738-42.

TABLE 1 Algorithm for diagnosis of Escherichia co/i0157:H7 infection
Clostridium difficile culture/toxin Pseudomembronous colitis in elderly with c linical picture of ischemlc colitis Diagnosis of E co/i0157:H7 should be included in the differential diagnosis if: