J7 Clin Pathol 1996;49:107-111

Detection of Helicobacter pylori in gastric biopsyand resection specimens

M Ashton-Key, T C Diss, P G Isaacson

AbstractAims-To compare the sensitivity of de-tecting Helicobacter pyloni in gastricbiopsy and resection specimens usingtinctorial and silver impregnation stains,immunohistochemistry and the poly-merase chain reaction (PCR).Methods-Formalin fixed, paraffin waxembedded tissue from 33 gastric biopsyspecimens (26 showing chronic gastritisand seven showing low grade mucosa as-sociated lymphoid tissue (MALT) lymph-oma) together with blocks of uninvolvedmucosa from gastrectomy specimens forMALT lymphoma (five cases) were stud-ied. Consecutive sections were stainedusing haematoxylin and eosin, Giemsa,the Warthin-Starry silver stain, and apolyclonal antibody directed against Hpyl-orn using an immunoperoxidase techniquefollowing heat induced antigen retrieval.PCR analysis of DNA extracted from afurther section was carried out using prim-ers which amplified a 411 base pair frag-ment of the urease A gene.Results-Hpylorn was detected in 14 (37%)sections stained with haematoxylin andeosin, 21 (55%) with Giemsa, 23 (61%) withWarthin-Starry, and 25 (66%) stained withthe antibody. Seventeen (45%) cases werepositive on PCR. Immunohistochemistrywas positive in all cases in which H pyloriwas detected by other methods.Conclusion-Immunohistochemistry us-ing an immunoperoxidase technique fol-lowing heat induced antigen retrieval fordetecting H pyloni in gastric biopsy andresection specimens is highly sensitive andeasy to use.(J Clin Pathol 1996;49:107-11 1)

Keywords: Helicobacter pylorz, detection, immuno-histochemistry, PCR.

cial stains such as Giemsa or the Warthin-Starry silver stain. None of these stains isspecific for the organism and all may be difficultto interpret because of non-specific staining ofmucus, debris and water bath contaminants.Moreover, the Warthin-Starry stain, which isconsidered to be the most sensitive, is tech-nically demanding and frequently not re-producible. Optimal interpretation of thesestains requires careful examination of the sec-tions at high magnification.More recently, immunohistochemistry, in

situ hybridisation and the polymerase chainreaction (PCR) have been proposed as al-ternative specific detection methods,912-18 butthese techniques have not been compared withroutine tinctorial stains. In this study we havecompared traditional detection methods,using haematoxylin and eosin, Giemsa andWarthin-Starry stained sections with immuno-histochemistry using a commercially availableanti-H pylori antibody and PCR on DNA ex-tracted from paraffin wax sections.

MethodsFormalin fixed, paraffin wax embedded tissueblocks from gastric biopsy specimens showingchronic gastritis (26 cases) and low grade B cellmucosa associated lymphoid tissue (MALT)lymphoma (seven cases) together with blocksof uninvolved mucosa from gastrectomy speci-mens for MALT lymphoma (five cases, threehigh grade and two low grade), were retrievedfrom the files of the Histopathology De-partment of UCL Medical School, London,and the histology reviewed. Chronic gastritiswas defined as the presence of an infiltrate ofplasma cells and lymphocytes, with or withoutactive inflammation and lymphoid aggregates.All sections were stained with haematoxylin andeosin, Giemsa and the Warthin-Starry silverimpregnation method.

Department ofHistopathology,University CollegeLondon MedicalSchool, LondonM Ashton-KeyT C DissP G Isaacson

Correspondence to:Dr M Ashton-Key,Department ofHistopathology, UniversityCollege London MedicalSchool, Rockefeller Building,University Street, LondonWC1E 6JJ.Accepted for publication18 October 1995

Helicobacter pylori infection is associated withgastritis and has also been implicated in thepathogenesis of gastric carcinoma and lymph-oma.'-' There are a number of methods fordetecting H pylon', including the breath test,the urease test and culture, but histologicaldetection in a gastric biopsy is the commonestand among the most sensitive."' Accurate de-tection of the organism is essential for correctpatient management and it is particularly im-portant to confirm eradication of bacteria fol-lowing treatment. The spiral shaped bacteriacan be seen in routine haematoxylin and eosinsections but are more easily detected with spe-

IMMUNOHISTOCHEMISTRYParaffin wax sections were heated in sodiumcitrate buffer (pH 6 0, 0 01 M) in a pressurecooker at 130°C for two minutes prior to im-munostaining with anti-H pylori antibody(Dako, High Wycombe, UK) at a 1 in 1000dilution using a standard Streptavidin-biotincomplex method.'920

MICROSCOPIC EVALUATIONSections were evaluated for the features ofchronic gastritis, which included the presenceof an infiltrate of plasma cells and lymphocytes

107

on Decem

ber 12, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.49.2.107 on 1 F

ebruary 1996. Dow

nloaded from

Ashton-Key, Diss, Isaacson

Detection ofH pylori summary of the results obtained with the different detection methods used

Detection method [number ofpositive cases (00)]

HaematoxylinSpecinien type and eosin Giemsa Warthin-Stany Immunoperoxidase PCR

Gastritis (n =26) 10 (38) 13 (50) 14 (54) 15 (58) 12 (46)MALT lymphoma biopsy (n =7) 4 (57) 7 (100) 6 (86) 7 (100) 5 (71)Gastrectomy (n =5) 0 (0) 1 (20) 3 (60) 3 (60) 0 (0)Total (n =38) 14 (37) 21 (55) 23 (61) 25 (66) 17 (45)

in the lamina propria with or without acuteinflammation and lymphoid aggregates, and/or the presence of low grade B cell MALTlymphoma. The haematoxylin and eosin,Giemsa, Warthin-Starry, and immunoperoxi-dase stained sections were carefully examinedfor the presence ofHpylori-like organisms usinga x 60 objective with x 10 eyepieces. Eachstain was assessed on a different day withoutreferring to the results of the other stains.

POLYMERASE CHAIN REACTIONDNA was extracted from serial sections of theparaffin wax embedded material as describedpreviously.2' PCR amplification of a 411 basepair fragment from the urease A gene of Hpylori was carried out as described by Claytonet al.22PCR was carried out in a reaction mix con-

taining 10mM Tris (pH 9), 50mM KC1,200 mM of each dNTP, 250 ng of each primer(HP 1: 5'-GCCAATGGTAAATTAGTT-3';HP2: 5'-CTCCTTAATTGTTTTTAC-3'),0 1% Triton X-100, 5 tld DNA extract, and 0 5units Taq polymerase (Promega, South-ampton, UK; added at 50°C following initialdenaturation) in a total volume of 50,ul. ThePCR conditions were as follows: 40 cycles of93°C for 45 seconds, 50°C for 40 seconds and72°C for one minute 50 seconds, followed by10 minutes at 72°C.In each experiment a positive control Hpylori

extract and a negative control (no DNA) wererun in parallel with the test cases. Productswere run on 2% agarose gels, stained with

a;Figure 1sections.

ethidium bromide and viewed under ultravioletlight. Cases yielding a 411 base pair productwere scored as positive.

ResultsThe table summarises the results. The presenceof chronic gastritis was confirmed in the 26gastric biopsy specimens and in the uninvolvedmucosa from the five MALT lymphoma gast-rectomy specimens. A diagnosis of low gradeB cell MALT lymphoma was confirmed inseven gastric biopsy specimens.Hpylori was identified in 14 sections stained

with haematoxylin and eosin (10/26 biopsyspecimens with chronic gastritis, four/seven bi-opsy specimens with MALT lymphoma, zero/five sections from gastrectomy specimens). Hpylori could be identified with greater frequencyin sections stained with Giemsa; 21 were pos-itive (13/26 biopsy specimens with chronic gast-ritis, seven/seven biopsy specimens withMALTlymphoma and one/five gastrectomy sections).H pylori could be detected at a still greaterfrequency in Warthin-Starry stained sections;23 were positive (14/26 biopsy specimens withchronic gastritis, six/seven with MALT lymph-oma and three/five gastrectomy sections). Onstaining with immunoperoxidase, 25 cases werepositive (15/26 biopsy specimens with chronicgastritis, seven/seven with MALT lymphomaand three/five gastrectomy sections). In all casesthe bacteria were more prominent and easierto detect in the immunostained sections than insections stained tinctorially. With the tinctorialand silver stains a high power or even an oil

b(a) Gastric biopsy specimen showing chronic gastritis. (b) H pylori can be seen in the immunostained

108

on Decem

ber 12, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.49.2.107 on 1 F

ebruary 1996. Dow

nloaded from

Detection ofH pylori in gastric biopsy and resection specimens

a bFigure 2 Gastrectomy for MALT lymphoma. H pylon hidden within mucus in Giemsa stained sections (a) areobvious in the immunoperoxidase stained sections (b).

Ft ,.

i-.- N..At..;;a

4F-

bFigure 3 The coccoidform ofH pylori in haematoxylin and eosin stained sections (a) can be identified with certaintyusing immunohistochemistry (b).

immersion objective was always necessary toidentify the organisms; specific staining of Hpylori in immunostained preparations couldeasily be seen using the x 10 or x 20 objective,even when very few organisms were present(fig 1).

Several patterns of distribution of H pyloriwere identified. Organisms attached to thebrush border of the gastric foveolar epithelialcells or within the superficial mucus were mosteasily seen. In some cases, however, the bacteriawere masked in haematoxylin and eosin andGiemsa stained sections by inspissated mucusor by being positioned flat, closely apposed tothe epithelial surface. These organisms weremore easily seen with the Warthin-Starry stainbut were strikingly obvious in immunostainedpreparations. Some difficulty was encounteredin differentiating Hpylori from debris in Giemsaand Warthin-Starry stained sections wherewater bath contaminants also posed a problemin occasional cases. Similarly, coccoid or-

ganisms, which were particularly seen in sec-tions from resection specimens, caused someuncertainty. The specificity of immunostainingobviated all of these difficulties (figs 2 and 3).When DNA extracted from the sections was

amplified using PCR, Hpylori were identified in17 cases (12/26 biopsy specimens with chronicgastritis, five/seven MALT lymphomas andzero/five gastrectomy sections).

DiscussionPrevious studies have emphasised that stainingwith haematoxylin and eosin is not adequatefor detecting H pylori and that staining withGiemsa or the use of the Warthin-Starry orrelated silver impregnation techniques are themethods of choice.'01' However, the latter tech-niques are not specific for H pylori and, asconfirmed in our study, both generate falsepositive results because of bacterial con-taminants or debris. Moreover, many laborat-

109

:-4..;L.

.:.t

on Decem

ber 12, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.49.2.107 on 1 F

ebruary 1996. Dow

nloaded from

Ashton-Key, Diss, Isaacson

ories find that the Warthin-Starry stainingtechnique is difficult to reproduce reliably. Asit has been recognised that H pylori causeschronic gastritis and has been implicated in thepathogenesis of gastric carcinoma and lymph-oma, the importance of identifying H pylori inbiopsy specimens has increased. As an example,Hpylori eradication is used as a first line therapyfor low grade gastric MALT lymphoma.3 Theconfirmation of H pylori eradication and thedetection of reinfection are important for cor-rect patient management. For these reasons,pathologists have sought more reliable methodsfor detecting H pylori in biopsy specimens,including immunohistochemistry, PCR and,more recently, in situ hybridisation.

Antibodies reactive against H pyloni havebeen available for several years,12"14 althoughimmunohistochemistry as a detection methodfor H pylori has not become popular. Negriniet al'2 raised several monoclonal antibodiesagainst Hpylori and were able to show specificstaining of the organism in gastric brushingsand frozen sections ofgastric biopsy specimens.Their antibodies were unreactive in formalinfixed, paraffin wax embedded material, makingthem unsuitable for routine use. Loffeld etall3 compared H pylori detection using culture,Giemsa and immunoperoxidase staining in-corporating polyclonal rabbit anti-H pyloriantiserum. Of these three methods, immuno-peroxidase staining was the most sensitiveand culture the least sensitive. However, despitefinding that immunoperoxidase staining wasmore sensitive than staining with Giemsa,they considered Giemsa staining very sensitiveand recommended it as the routine detectionmethod.Using the Dako polyclonal rabbit anti-Hpyl-

oni antibody with prior trypsinisation, we wereinitially unable to stain H pyloni in tissue sec-tions successfully due to excessive backgroundstaining of epithelium and mucus. We over-came this problem by using heating, ratherthan trypsin, as the antigen retrieval method.This enabled us to use the antibody at a dilutionof 1 in 1000 with a proportional reduction inbackground staining. Our laboratory has foundthat this method is easy to use, less demandingthan Warthin-Starry staining, and that it pro-duces reliable results which are easy to in-terpret. Low numbers or even single organisms,often difficult to detect using traditional stains,are easily identified in immunostained sections.A further advantage of immunoperoxidasestaining is the ability to identify the coccoidform of H pyloni, which cannot be identifiedconfidently using traditional stains and whichcannot be cultured.Molecular techniques including PCR and in

situ hybridisation have both been suggestedrecently as specific methods for the detectionof H pyloi.9 15-18 PCR has become particularlypopular and has been shown to be a veryspecific detection method, although sensitivitydoes vary depending on the primers used, thesource ofDNA and number ofbacteria present.In this study PCR was more sensitive thanstaining with haematoxylin and eosin but lesssensitive than the other methods investigated.

There are several possible reasons for this.Although serial sections were used for the PCRonly a small amount ofDNA is sampled. Thiswill detect the organisms if they are numerousbut if only scanty bacteria are present the DNAsample analysed may not include any bacterialDNA. This potential source of false negativeresults could be reduced by increasing theamount ofDNA sampled either by repeat PCRsor larger aliquots of extract in each reaction.There are disadvantages to both of these pro-cedures; increasing the number of PCR re-actions is time consuming and increasing theamount of extract in each reaction may inhibitthe PCR. The 411 base pair DNA fragmentamplified with these primers is relatively largefor amplification from paraffin wax sections andthis may have resulted in some false negativeresults. This could be avoided by using differentprimers which amplify a smaller fragment.Using fresh tissue and primers which amplifieda 210 base pair fragment, Fabre et all7 wereable to detect H pyloni infection in the sameproportion of biopsy specimens as usingGiemsa stained sections. In this study we haveshown that immunohistochemistry is superiorto Giemsa staining and therefore is also likelyto be superior to PCR using primers to a smallerDNA fragment.

In situ hybridisation, using both RNA andDNA probes, is a sensitive and specific methodfor detecting Hpylori in tissue sections.'5 16 Likeimmunohistochemistry, in situ hybridisationshows strong specific staining of the bacteriaand permits direct comparison with histology.Compared with immunohistochemistry, how-ever, in situ hybridisation has several dis-advantages; it is a more complex techniquewhich requires specially trained technical staffand few routine laboratories have access tomolecular facilities whereas most already per-form immunohistochemistry.

In conclusion, we have shown that usinga heat mediated antigen retrieval technique,immunohistochemistry is a highly sensitivemethod for identifyingHpyloni in gastric biopsyspecimens. It is more sensitive than traditionalstains and considerably easier to interpret. Werecommend immunohistochemistry as themethod of choice for the identification of Hpylori in gastric biopsy specimens. However, inmost histopathology departments the numberof gastric biopsy specimens is large and theresources for immunohistochemistry are lim-ited. Although it would be ideal, it is clearlynot practical to perform H pyloni immuno-histochemistry on every gastric biopsy speci-men. For routine use immunohistochemistrycould be reserved for the following threegroups: (1) biopsy specimens which showchronic gastritis and are negative for H pylon?using haematoxylin and eosin and Giemsastains; (2) post-treatment biopsy specimens forMALT lymphoma, to ensure eradication ther-apy has been successful; and (3) biopsy speci-mens in which coccoid forms or otherorganisms not definitely identifiable as Hpyloniusing routine stains are present.

110

on Decem

ber 12, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.49.2.107 on 1 F

ebruary 1996. Dow

nloaded from

Detection ofH pylori in gastric biopsy and resection specimens

1 Marshall BJ, Warren JR. Unidentified curved bacilli in thestomach of patients with gastritis and peptic ulceration.Lancet 1984;i:1311-15.

2 Goodwin CS, Armstrong JA, Marshall BJ. Campylobacterpyloridis, gastritis and peptic ulceration. Clin Pathol1986;39:353-65.

3 Wotherspoon AC, Doglioni C, Diss TC, Pan L, MaschiniA, de Boni M, et al. Regression of primary low-grade B-cell gastric lymphoma of mucosa-associated lymphoidtissue after eradication ofHelicobacter pylori. Lancet 1993;342:575-7.

4 Wotherspoon A, Ortiz-Hidalgo C, Falzon MR, IsaacsonPG. Helicobacter pylori-associated gastritis and primaryB-cell gastric lymphoma. Lancet 1991 ;338: 1175-6.

5 Parsonnet J, Friedman GD, Vandersteen DP, Chang Y,Vogelman JH, Orentreich N, et al. Helicobacter pyloriinfection and the risk of gastric carcinoma. N Engl Med1991;325:1127-31.

6 Graham DY, Klein PD, Evans DJ jr, Evans DG, AlpertLC, Opekun AR, et al. Campylobacter pylori detectednoninvasively by the '3C-urea breath test. Lancet 1987;i:1174-7.

7 McNulty CAM, Dent JC, UffJS, Gear MW, Wilkinson SP.Detection of Campylobacter by the biopsy urease test: anassessment in 1445 patients. Gut 1989;30:1058-62.

8 Goodwin CS, Blincow ED, Warren JR, Waters TE,Sanderson CR, Easton L. Evaluation of culturaltechniques for isolating Campylobacter pyloridis fromendoscopic biopsies of gastric mucosa. Clin Pathol 1985;38:1127-31.

9 Weiss J, Mecca J, da Siwa E, Gassner D. A comparison ofPCR and other techniques for detection of Helicobacterpylori infection in dyspeptic patients. Clin Microbiol1994;32:1663-8.

10 Kolts BE, Joseph B, Achem SR, Bianchi T, Monteiro C.Helicobacter pylori detection: A quality and cost analysis.Am Gastroenterol 1993;88:650-5.

11 Miller NM, Sathar MA, Naran AD, Van den Ende, SimjeeAE, Manion G. Evaluation of various techniques to diag-nose Helicobacter pylori in patients with upper gastro-intestinal tract symptoms. S Afr Med 1991;80:575-8.

12 Negrini R, Lisato L, Cavazzini L, Manini P, Gullini S, Basso0. Monoclonal antibodies for specific immunoperoxidase

detection of Campylobacter pylori. Gastroenterology 1989;96:414-20.

13 Loffeld RJ, Stobberingh E, Flendrig JA, Arends JW. Hel-icobacter pylori in gastric biopsy specimens. Comparisonof culture, modified giemsa stain and immuno-histochemistry. A retrospective study. Jf Pathol 1991;165:69-73.

14 Cartun RW, Kryzmowski GA, Pedersen CA, Morin SG,Van Kruiningen HJ, Berman MM. Immunocytochemicalidentification ofHelicobacter pylori in formalin-fixed gast-ric biopsies. Mod Pathol 1991;4:498-502.

15 Van den Berg FM, Zijlmans H, Langenberg W, Rauws E,Schipper M. Detection of Campylobacter in stomachtissue by DNA in-situ hybridisation. Jf Clin Pathol 1989;42:995-1000.

16 Bashir MS, Lewis MS, Quirke P, Lee A, Dixon MF. In-situhybridisation for the identification of Helicobacter pyloriin paraffin wax embedded tissue. J Clin Pathol 1994;47:862-4.

17 Fabre R, Sobhani I, Laurent-Puig P, Hedef N, Yazigi N,Vissuzaine C, et al. Polymerase chain reaction assay forthe detection of Helicobacter pylori in gastric biopsy speci-mens: Comparison with culture, rapid urease test andhistopathological tests. Gut 1994;35:905-8.

18 Clayton C, Kleanthous K, Tabaqchali S. Detection andidentification of Helicobacter pylori by the polymerasechain reaction. J Clin Pathol 1991;44:515-16.

19 Guesdon JL, Ternyn CKT, Avrameas S. The use of avidin-biotin interaction in immunoenzymatic techniques.J Histochem Cytochem 1984;32:219-29.

20 Norton AJ, Jordan S, Yeomans P. Brief, high-temperatureheat denaturation (pressure cooking): A simple and effect-ive method of antigen retrieval for routinely processedtissues. 7 Pathol 1994;173:371-9.

21 Diss TC, Pan L, Peng H, Wotherspoon AC, Isaacson PG.Sources ofDNA for detecting B-cell monoclonality usingPCR. J Clin Pathol 1994;47:493-6.

22 Clayton CL, Kleanthous H, Coates P, Morgan D,Tabaqchali S. Sensitive detection of Helicobacter pyloriby using polymerase chain reaction. J Clin Microbiol 1992;30:192-200.

23 Bode G, Mauch F, Malfertheiner P. The coccoid forms ofHelicobacter pylori. Criteria for their viability. EpidemiolInfect 1993;111:483-90.

illl

on Decem

ber 12, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.49.2.107 on 1 F

ebruary 1996. Dow

nloaded from