Use of electron microscopy in examination of

J Clin Pathol 1991;44:313-316

Use of electron microscopy in examination offaeces and rectal and jejunal biopsy specimens

GMM Connolly, D S Ellis, J E Williams, G Tovey, B G Gazzard

Department ofMedicine,Westminster Hospital,Dean Ryle Street,Horseferry Road,London SW1GMM ConnollyB G GazzardElectron Microscopyand HistopathologyUnit, London SchoolofHygiene andTropical MedicineD S EllisJ E WilliamsG ToveyCorrespondence to:Dr B G Gazzard

Accepted for publication21 November 1990

AbstractThe stools and rectal biopsy specimensof 44 patients with AIDS and diarrhoeawere examined by culture, light micro-scopy, and electron microscopy. In 13patients examination of rectal biopsymaterial and faecal samples showed no

pathogen, but in two of these, micro-sporidiosis was found by electronmicroscopical examination of jejunalbiopsy specimens. This organism was

also identified electron microscopicallyin one of the further five jejunal biopsysamples taken from patients with a

known cause of diarrhoea. Blastocystishominis infection was identified electronmicroscopically in six patients, all ofwhom had cryptosporidiosis additionallyseen by light microscopy. Four of thesesix patients remained well for longperiods, with only moderate diarrhoea,and follow up showed no evidence ofblastocystis infection. In only four of 11patients found to have cryptosporidiumin their stools at light microscopy were

organisms found at electron microscopy.Viral inclusions were only identified atelectron microscopy in one of 10 patientswith an opportunistic viral infection seenat light microscopy (cytomegalovirusn = 7, herpes simplex virus n = 3). Noadditional viral pathogens were detectedin either stools or rectal biopsy materialby electron microscopy.

It is concluded that routine electronmicroscopic examination of stoolsamples or rectal biopsy material takenfrom patients with AIDS and diarrhoeais unnecessary and does not increase theyield of potential pathogens comparedwith standard microbiological tech-niques and histology.

In some patients with AIDS and diarrhoea,no pathogenic organism can be detected instools or in rectal biopsy specimens.' Thepathogenesis of diarrhoea in such patients isof considerable interest: possibilities includeinfection of gut mucosal cells with the humanimmunodeficiency virus (HIV),2 neuronaldamage in the gut wall,3 or thirdly, a pathogensuch as Microsporidium or a viral infectionundetected by standard microbiological tech-niques (Abstract presented at the Fifth Inter-national Conference on AIDS, Montreal,1989).The purpose of this paper was to determine

the value of electron microscopy in detecting

unsuspected pathogens in HIV antibodypositive patients with diarrhoea, particularlythose in whom no microbiological cause hadbeen found by routine methods.The 44 patients in this study were all HIV

antibody positive by two different methods4and had had diarrhoea (at least three liquidstools every 24 hours) for longer than a

month. All belonged to group IV subclass Aor C according to the Centers for DiseaseControl Classification System for AIDS.5One hundred consecutive patients present-

ing between March and July 1989 were

screened for diarrhoea by microbiologicalexamination of stools and light microscopicexamination of rectal biopsy material. The first35 of these patients (group I) were included inthis study whatever the results of the screen-

ing tests. Of the subsequent 65 patients, onlythe nine (group II) in whom no potentialpathogen was uncovered by routine screeningwere included.

All patients in this study had stool samplesand rectal biopsy specimens examined byelectron microscopy. It was not thoughtethical to perform jejunal, biopsies on allpatients, but they were examined by electronmicroscopy in 18. These included the 13patients (four from group I) who underwentthe procedure for diagnostic purposes as no

other pathogen had been uncovered, and a

further five patients (all with cytomegalovirusinfection) who had had a biopsy as part ofroutine investigations for coincident upperabdominal dyspepsia.

All faecal samples were cultured routinelyand examined for ova, cysts, and parasites.They were stained by a modified Ziehl-Neelsen method with concentration forCryptosporidium.6 The faeces were furtherprocessed for mycobacterial species byinoculation on to a Lowenstein-Jensen slopeand into Kirschner medium, and these were

examined weekly for 10 weeks.Jejunal biopsy material was obtained by a

Watson capsule attached to the end of a flexi-ble gastroscope and positioned beyond thefourth part of the duodenum, under directvision.

Rectal biopsy samples were obtained fromat least two sites from all patients during rigidsigmoidoscopy. The material was divided forelectron microscopic and light microscopicexamination.

Jejunal and rectal biopsy specimens were

fixed in 15% unbuffered formol saline andprocessed using a routine paraffin wax systemwith vacuum impregnation. Sections were cut

313

Connolly, Ellis, Williams, Tovey, Gazzard

Table I Final cause of diarrhoea in 44 patients

Final diagnosis (n =) Bacteriology Histology Electron microscopy

Cryptosporidium (7) 7 3} rpoprdu seen in 4

Cryptosporidium (4) and cytomegalovirus 4 4 J

Cytomegalovirus (2) - 3 No CMV seenHerpes simplex virus (2) - 2 1Mycobacterium avium intracellulare (1) 1 1 1Candida (1) 1 1 1Non-Hodgkin's lymphoma (2) - 2 2 Preneoplastic changesBacterial infection (5) 1 Campylobacter(4Salmonella species _ 3 Bacterial infectionGiardia lamblia (6) 4 4 1Microsporidia (2) - - 2 on jejunal biopsyMicrosporidia + CMV (1) - 1 CMV 1 on jejunal biopsyNo pathogen ( 11)

at 5 gIm, with at least three levels being taken.Haematoxylin and eosin and Ziehl-Neelsenstains were performed routinely.Cytomegalovirus (CMV) inclusion bodieswere sought in all sections and confirmedusing immunoperoxidase techniques suitablefor paraffin wax sections.7

Severe inflammation was defined as a pro-nounced inflammatory infiltration in the laminapropria.

Electron microscopy was carried out onspecimens fixed in glutaraldehyde (faecalmaterial was embedded in 2% agar),post-fixed in osmium tetroxide, and afterdehydration and embedding in Araldite,stained with lead citrate and uranyl acetate.

ResultsIn the whole series of 44 patients, electronmicroscopic screening yielded only nineadditional pathogens. In six patients Blasto-

cystis hominis infection was found in stoolsamples together with cryptosporidiosis, and inthree patients microsporidial infection ofjejunal biopsy material was uncovered, whichwas in addition to cytomegalovirus colitis inone case (table 1).Two potential pathogens (microsporidial

infection in both cases) were uncovered byelectron microscopy in the 13 patients in whomno pathogen was found by routinemicrobiological screening of stools or lightmicroscopic examination of rectal histologicalsections.

MICROBIOLOGYMicrobiological examination of the stools wasthe single most useful investigation thati iden-tified all the cases of cryptosporidial infection(n = 1 1) and four of the six cases of giardiasis.Two other opportunistic pathogens presumedto have caused the diarrhoea (Candida andMycobacterium avium intracellulare [MAI])

Figure 1 Blastocystishominis organism identifiedin stools ofpatients byelectron microscopy.

¶-

k

314

I

0.

Use of electron microscopy in examination offaeces, and rectal andjejunal biopsy specimens

Table 2 Clinicalfeatures of 11 patients with cryptosporidiosis; six with blastocystisinfection in addition

Electron microscopy Weight Volume of Total CD4Case loss at diarrhoea cells SurvivalNo B hominis Cryptosp* presentation (kg) ml/24 h x 109/l (months)

1 + - 5 2500 20 5t2 + - 15 1000 250 173 + - 3 800 200 184 + - 2 500 200 185 + - 2 500 150 186 + + 5 1000 30 9t7 - + 8 2500 20 3t8 - + 5 1500 45 129 - + 12 2000 20 9t10 - 10 1500 30 711 - - 3 500 30 18

*All these patients had cryptosporidiosis identified by light microscopical examination of stoolsamples.tDeath.

and five non-opportunistic pathogens (Salmon-ella and Campylobacter species) were identifiedby stool culture (table 1).

LIGHT MICROSCOPICAL EXAMINATION OF RECTALBIOPSY SPECIMENSLight microscopical examination of rectalbiopsy specimens was ofmost value in identify-ing opportunistic viral infection and confirmingthe diagnosis of tumour in two patients. Viralcolitis was diagnosed in nine patients;cytomegalovirus (CMV) inclusions were seenin seven (four of whom also had crypto-sporidium infection), and herpes simplexinclusions were identified in two.

Light microscopical examination failed todetect evidence of cryptosporidium in four ofthe 11 patients where this organism was iden-tified in the stools, and did not discover anyother patients with this infection.

Severe non-specific rectal inflammation wasalways present with viral colitis and bac-

Figure 2 ThreeMicrosporidian meronts atan early stage ofdevelopment lying beneaththe duodenal microvilli ofa patient with AIDS.

teriological infection but was absent from thosewith cryptosporidiosis, microsporidiosis, or noidentified pathogen.

ELECTRON MICROSCOPICAL EXAMINATIONElectron microscopical examination of stools,rectal, or jejunal biopsy material showed Bhominis infection in the stools of six patients (fig1). In all of these cryptosporidium hadpreviously been identified by standard tech-niques and by electron microscopy in one. Fourof these six patients remained relatively welldespite lack of specific treatment and two diedof overwhelming cryptosporidial infection(table 2). The organism could no longer beidentified in the stools of three of the foursurviving patients between three and sixmonths later. The fourth patient was lostto follow up. There was no tendency forpatients with blastocystis in addition tocryptosporidium to have more severe clinicalmanifestations (table 2).

Microsporidial infection was diagnosed byelectron microscopical examination of jejunalbiopsy samples in three patients (fig 2) and intwo was the only potential pathogen. All threepatients had stool volumes in excess of 1 1/dayand one died within two weeks of diagnosis; theother has been lost to follow up. The thirdremains seriously ill with CMV colitis.

In the two cases of lymphoma diagnosed bylight microscopy bizarre nuclei in the laminapropria and excessive lipids in the mito-chondria suggested neoplastic change.

Electron microscopical examination of rectalbiopsy material did not diagnose any additionalpatients with cryptosporidial infection. In onlyfour of the 11 patients diagnosed by lightmicroscopy was there electron microscopical

'I

315

~ I

Connolly, Ellis, Williams, Tovey, Gazzard

evidence of infection on rectal biopsy in threeand in faecal material in one.

No viral pathogens were found in the faecalsamples of the 44 patients. Viral inclusionswere only seen in the rectal biopsy specimen inone patient who was known to have herpessimplex colitis.

DiscussionThe range of causes of diarrhoea in these 44HIV positive patients is similar to that of otherpublished studies.' Unfortunately, the additionof electron microscopy of stool samples andrectal biopsy material to standard investigationonly yielded one pathogen of dubious clinicalimportance-B hominis. Electron microscopicexamination of a small number of jejunalbiopsy samples, however, did show microsp-oridial infection in three patients.

Blastocystis hominis can be identified readilyby light microscopical examination of stoolsamples when Field's stain is used.8 This hasnot been done routinely in our laboratory, butthe high prevalence of blastocystis infectionuncovered indicates the need to reconsider thispolicy. Field's stain was not used in this studyas blastocystis infection was only discoveredwhen all the stool samples were examined byelectron microscopy in one batch. B hominiscarriage was confined to those patients who hadcryptosporidial infection. The presence of Bhominis in these patients was of dubiouspathogenic importance. The patients did nothave a more severe clinical course than thosewith cryptosporidiosis alone. None receivedspecific treatment, and three of the four surviv-ing patients had no evidence of blastocystisinfection at follow up, but had cryptosp-oridium as a cause for continuing diarrhoea.Blastocystis carriage in diarrhoea has beenassociated with a high mortality in somestudies, but not in others where the diarrhoearesolved promptly with appropriateantimicrobial treatment.9 l0The only pathogen detected in this study by

electron microscopy which could not have beendiscovered by routinely available techniqueswas microsporidium, although it has recentlybeen suggested that even this organism may bevisualised using thin plastic (resin) sections ofjejunal biopsy specimens examined at electronmicroscopy. There is no positive proof thatmicrosporidial infection produces diarrhoea,but it has only been found so far in symp-tomatic HIV positive patients in whom thereare associated abnormalities of villus architec-ture which may produce diarrhoea. Further-more, it seems to be particularly common inpatients in whom no other cause for diarrhoeacan be found. Thus although it was only seen inone of 100 jejunal biopsy specimens taken fromHIV positive patients with a variety of knowncauses for diarrhoea," it was found in 30% of agroup of patients from New York in whomother causes for diarrhoea had been excluded(Abstract presented at the Fifth InternationalConference on AIDS, Montreal, 1989).The lack of detection of additional viral

pathogens in faecal material from our patientsagrees with previous observations that virusesusually identified by electron microscopy, suchas rotavirus, have not been seen in HIV positive

patients as a cause of diarrhoea (Abstractpresented at the Fifth International AIDSConference, Montreal, 1989).12 Other studiesusing immune electron microscopy techniqueshave shown entero viruses in up to 9% ofHIVpositive patients with AIDS, but this was notmore common in patients who had diarrhoea.'3The optimal sequence of investigations for

diarrhoea in HIV positive patients is graduallybecoming clear. We have previously shownthat the sensitivity of microbiological examina-tion of stools and histological examination ofrectal biopsy specimens combined is in excessof90%, with a high positive predictive value.'4This confirms the findings of French studies'5which showed that the single most useful testwas stool examination followed by micro-biological examination of fluid obtained duringduodenal aspiration and that the sensitivity fordetection of cytomegalovirus infection wasimproved by examination of rectal biopsyspecimens. We believe that duodenal intuba-tion is a relatively uncomfortable techniqueand should be reserved for patients in whom nopathogen is uncovered by routinemicrobiological examination of stools or histo-logical examination of rectal biopsy specimens.If other studies confirm the high prevalence ofmicrosporidiosis in patients in whom no otherpathogen is found, examination of jejunalbiopsy material would also become an impor-tant investigation. It would be helpful to ascer-tain whether duodenal pinch biopsyspecimens, which are much easier to obtain,also provide evidence of this infection, andwhether examination by light microscopy canbe a sensitive technique for detection.

1 Connolly GM, Shanson D, Hawkins DA, Harcourt-Webster JN, Gazzard BG. Non-cryptosporidial diarrhoeainhuman immunodeficiency virus (HIV) infected patients.Gut 1989;30:195-200.

2 Nelson JA, Wiley CA, Reynolds-Kohler C, Reese CE,Margaretten W, Levy JA. Human immunodeficiency virusdetected in bowel epithelium from patients with gastro-intestinal symptoms. Lancet 1988;i:259-62.

3 Griffin GE, Miller A, Bateman P. Damage to jejunal intrinsicautonomicnerves in HIV infection. AIDS 1988;2:379-82.

4 Mortimer PP, Parry JV, Mortimer JY. Which anti-HTLVIII/LAV assays for screening and confirmation testing?Lancet 1985;ii:873-7.

5 Centers for Disease Control: Revision of the CDCsurveillancedefinitionforacquiredimmunodeficiencysyn-drome. Morbid Mortal Weekly Rep 1987;(suppl 36):15:35-155.

6 Garcia LS, Bruckner DA, Brewer TC, Shimizu RL. Tech-niques for the recovery and identification of crypto-sporidium oocytes from stool specimens. J Clin MicrobiolInfect Dis 1987;8:235-44.

7 Francis ND, Boylston AW, Roberts AHG, Parkin JM,Pinching AJ. Cytomegalovirus in gastrointestinal tracts ofpatients infected with HIV I or AIDS. J Clin Pathol1989;42:1055-64.

8 Moody AH, Fleck SL. Versatile Field's stain. J Clin Pathol1985;38:842-3.

9 Kain KC, Noble MA, Freeman HJ, Barteluk RL.Epidemiology and clinical features associated withBlastocystis hominis infection. Diagnost Microbiol InfectDis 1987;8:235-44.

10 Russo AR, Stone SL, Taplin ME, Snapper HJ, Doern GV.Presumptive evidence for Blastocystis hominis as a causeof colitis. Arch Int Med 1984;148:1064.

11 Canning EU, Hollister WS. Enterocytozoan bieneusi(microsporidia) prevalence and pathogenicity in AIDSpatients. Trans Roy Soc Trop Med Hyg 1990;84:181-6.

12 Smith PD, Lane C, Gelvie J, et al. Intestinal infections inpatients with the acquired immunodeficiency syndrome(AIDS). Ann Intern Med 1988;108:328-33.

13 Kaljot KT, Ling JP, Gold JW, et al. Prevalence of acuteenteric viral pathogens in acquired immunodeficiencysyndrome patients with diarrhoea. Gastroenterology 1989;97:1031-2.

14 Connolly GM, Forbes A, Gleeson J, Gazzard BG. The valueofbarium enema and colonoscopy in patients infected withthe human immunodeficiency virus. AIDS 1990;4:687-9.

15 Rene E, Marchie C, Requier B, et al. Intestinal infections inpatients with acquired immunodeficiency syndrome: Aprospective study of 132 patients. Dig Dis Sci 1989;34:773-80.

316