Campylobacter pylori detection in gastric mucosa : Association with gastritis

ln order to evaluate the association between Camphlobacter pylori 
and gastritis, two biopsies were taken from the duodenal bulb, antrum, body and 
fundus (and from lesions if there were any) in 100 consecutive patients referred to this 
gastroscopic clinic For each site, one biopsy was for histology and C pylon detection by 
Warthin-Starry staining, and the second biopsy was for culture. In addition, for each 
patient a gastric brushing was Gram stained. Twenty-one patients were excluded 
from the study. Among the remaining, 13 patients had positive biopsy culture for C 
pylori and, of these, 30 (91%) had gastritis (including 23 with active chronic gastritis). 
This culture sensitivity increased with the number of biopsies. Forty-two patients 
had a positive brushing specimen, of which 30 (71%) had gastritis. Gram stain on a 
brushing specimen had a sensitivity of 84.8%, in comparison with the biopsy culture. 
Of the 23 patients with positive Warthin-Starry stain , 19(83%) had an histology of 
gastritis. There was a strong correlation between the presence of C pylori in the stomach 
mucosa and the gastritis. The incidence of C pylori associated gastritis is similar in 
Quebec to other parts of the world. The biopsy culture is a simple and specific test, 
and at least two biopsies are necessary for a good sensitivity. Gram stain on a brushing 
specimen is an adequate test for rapid detection of C pylori in the stomach.

R Fl'()RT<; 01 SPIRAi llACTl'RIA IN THE human s tomach have occurrcJ ~poradically over the l,1s1 cen t ury ( 1.21However, it was only in 1982 that a microacroph ilic campylobacter-like organism was isolated by Marshall and Warren ( 'll from gastric antra l b iopsy specimens This mitial publica tion sparked off worlJ wide enth usiasuc research into this bac terium now named Camfrylobacrer fry/on (4.5).Since th.is time, a numherofscuJ • ies have shown an association between the p resence of C /;ylori in gastric mucosa a nd h istologicall y confi r med gastri tis However, there is sri ll some d iscussion whether the organ ism has a ca usative role or is simply a secondary invader.
With chis sw dy, it wa~confi rmed that C pylori was correlated with gastritis anJ then differen t methods for C pylori decec• tion in the huma n srnmc1ch were eval uated T he literatu re was also review<.:d.emphasiz ing principal argurn<.:nts in favour o( th<? causative relationship of the b;icteri u m wi th gnsrritis.Fina lly, some basic nmions for a C trylori de tection pro• tocol in patients undergoing endoscopy are suggested.

PATIENTS AND METHODS
Endoscopy: One hu ndred consecutive patients were included wh o were referred for gastroscopy on cl inical grounJs from January co March [987.lnformed consent was obtained from all patients for endoscopy an d biopsies.Biopsies were obtained from the duodenal b ulb, antru m, body, fund us and from lesions if there were any.Two biopsies were taken in each a rea.one fo r culture and the other for h istology, giving eight to JO tissue specimens from each patient.A lso. one brushing speci men from the anrrum was spread on a glass slide and air d ri ed.An endoscopist recorded the cl in ical hismry, med ication and drug use.alcohol co nsum p t ion, clinical d iag n osis and endoscop ic fin d ings.Microbiology: Samples for culture were immediately immersed in 5 ml of thioglycolate b roth , transported to the laboratory within 2 h of collection and macerated in 2 ml of the same broth .One drop was the n p lated on b lood agar (trypticase soy agar with sheep blood 5'\,) and choce,,are agar (GC agar, bio-X 1% and hemoglobin 1%) a nd incubated microaerophilically in anaerobic jars fo r seven days at 37°C.T hegaspak (Campy-Pa k, BB L) was replaced every 24 h.C pylori, when isolated.was identified using d irect examination with Gram sta in , oxyJase, cata lase, oxydario nfcrmenrarion test with pu rple bromocresol.nitrate reduction , h ip p urace hydrolysis, urease (C h ristensen's u rea med ium), indole, and susceptibility to cephalothin and nalidixic acid.The b rushing specimens were fixed and stained by Gram method.Histology: T he b iopsies for h istological examination were fixed in fo rmalin and routinely p rocessed.Sections were stained with he m atoxylin and eosin a nd assessed for gastritis usi ng the sa me criteria as Marsha ll (3 ). that is normal stomach, chronic gastritis or active ch ronic gastritis.Chronic gastritis indicates inflammation with no increase in polymorphon uclea r le u kocy tes b u t with increased or normal number o f lym phoid cells and edema, congestion or cell damage.Active chron ic gastri tis is indicated if polymorp ho n uclear leu kocytes arc increased in number, if a few infiltrate one gland neck or pit, or if they a rc scattered throughout the superficia l epithelium.ln addition, other sections from all b iopsies were stained by the Warth in-Srnrr y method (6) and exa m ined for prese n ce or absence of small curved bacilli on the surface of the epitheli u m.D ata ana lysis: The resu lts were recorded in each departmen t (gastroentcrology.microbiology and pathology) independen tly.For statistical analysis of the fin dings.th e Fisher's exact tesr or rhe xz test was used depending o n sam ple size.For the study of the correlation between the presence of C /rylori and gascri u:,, o n ly b iopsies from the antrum, bod y and fu nd us were considered.A patien t was considered positive when there was at least one positive site for C pylori by culture a nd/or h istology.

RESULTS
From the [00 patients, 21 were excluded (two for having taken antibiotics in the week preceding the endoscopy, th ree fo r upper d igestive neoplasia, o ne for incomple te clinical data, one fo r u ninterpre table cu lt ure and 14 for peptic ulcers).T he nu mber of peptic ulcers was too small to study adequately any association with C pylori.
C /rylori was isolated from 7 5 biopsie:.o f 3 3 patien ts.T he bacteria were curved or S-shaped Gram-negative rods.ln electronic microscopy, they had three to five sheathed flagella arising from one end of the cell.Wi th the Gram stain on the brushing specimen , a good correl ation between gastritis and presence of C pylori was obtained n:.1ble 3 l In comparison with th e culture, the Gram sensitivity was 84.8"1, since 28 of the 33 positive patients by cultu re were also Gram-positive.
Sections stained with the Warthi n -Starry method were not all easy to interp ret.When there was a d oubt.o ften cau~ed by artefacts.the specimen was considered negative.However, a correlat ion between gastritis and C /rylori was establish ed by ch is method (Table > ).
Only 23 ca~es were posi tive but 8n, had gastritis and on ly l7"o had a norma l histology.
Fina ll y, the endo~copic findings were compared with the histological diagnosis.There were 20 cases with endoscopical diagnosis of gastritis.One-half of th ese patients h ad gawiti~ by h istology  C pylori associated gastritis is d iagnoscd by hi sro logi cal and/or micro bio logical examin ation of gastric biopsy specimen.T he cult ure of biopsy specim en is the most specific test.The typical idcntific;ition profile o f the bacteriu m provides very li ttle risk of false-posi ti ve culture.The tech nique is simple and may be performed in a rou tine laboratory.Numerous causes of false-negative result arc poss ible, such as recent antibiotic therapy, bur can be avo id ed witho ut difficulty (7 ).
As demonstrated in Table 2, the culture sensitivity inc reases with th e number of biopsies.H owever, although the number of positive patie n ts is similar whateve r the observed site, it ca nnot be concluded that rhc choice of b iopsy site docs not influence the sensitivity of the cultu re.As the biopsy forceps were not sterilized between each area and as the order of biopsies was always the sa me, a crosscontamination of cu ltu re from preceding samples was theoreticall y possible.particu larly from th e antrum.w h ich is positive in most positive patients.to one of these areas.U n fortunately, their method was not perfect since the endoscope lumen was not ste rili zed between each biopsy.The pri ncipal d isadva n tage with the culture is the delay of several d;iys before obtaining results.A Gram-stained brushing sreci m cn is useful because results may be available with in I h , ie, whe n the patien t is still in the gastroscopy clinic.Presen t results demonstrated a good sensitivity of84.8% in comparison with culture but a decreased specificity since 10 patients with negati ve cultu re and normal hi stology were positive o n Gramsta in.Unfortunately, it is impossible to establish if these positive Gram stai ns are true posi tive since there wa s no culture from the brushing specimens.Also .present resu lts ca nnot be co mpared wi th others since most s tu dies used Gramstaining on b iopsy specimens instead of brushing specimens and none evaluated the test specifica ll y.
Other rapid detection tests have been previously investigated.Urcase tests performed directly o n b iopsy speci mens obmined diffe rent results depending on tech nique.lts sensitivity varied from 50% to 100°/o( 14,25,3 1-35 ) b u tthe rc is a commercially available urease test (CLO Test) with h igh s p ecifici ty and sensit iv ity ( 36, 3 7).Fu rch c rmorc, ;i rece n t study Jcmonstrntcd a correlatio n between the urease reacti o n time and the grades of inflammatio n a nd number of bacteria seen in biopsi es ( 32).T h e 11 C -ure a breath test also exploits the strong urease activity of C Jrylori and its initial resu lts arc encou raging (38).D e tection of the bacterium by p hase contrast m icroscopy ( 39) or immunofluorescc nce (40) is possibl e but, like the 11 C-u rea breath test, is no t available in m a ny hospitals.
As wi th most previous stud ies, the Warth in -S ta rry stain was used for h istolog ical detection o f C [lylori.Data wi th this stain we re less interesting than wi th cultu re a nd Gram stain.The test was time consu ming and difficu lt to interpret because th e re were ;i lot o f utefacts.Some p;ichologists suggest, as an alternative, th e Gie msa, hemacoxy li n an d eosi n or fluoresce nt acrid ine o ra nge tests ( 10,41.42 ).In a recent study, the bacterium was detected with th e hematoxylin and eosin stain in 95% of positive cu lture biopsies (29).CONCLUSION Despite several major reasons to consider C pylori as a pathogen, some continue ro refute this idea because the bacterium is found in apparently healthy subjects.However.as with othe r bacteria, such as group A streptococcus in the pharynx, a carrier state is possible.Although fu rther clinical trials o n C pylori infectio n treatment are required to confirm its parhogen icity, an investigation of different gastroduodcnal diseases, particularly gastritis, s hould include an attempt to detect C pylori with in the stomach in addition to the usual h istological colo u ration.At least two biopsies are necessary, one from the antrum.A rapid detection of the bacterium by Gram stained brushing specimen is very simple bur not specific enough.The rapid ureasc test deserves much consideration, however, it should be evaluated in the clinical setti ng before being used routinely.Any ki nd of rapid test should be confirmed by culture .Special stai ning of biopsies to find the bacterium is not indispensable if rapid rest and culture are done.
Haze ll et al ( 30), in a study of C pylori distribution with in the sto mach, (i n which they cleaned and ri nsed b iopsy fo rceps between each site) suggested taking biopsies fr o m both the a n trum and body because the bacterium may b e limited

TABLE 1
Correlation between Campylobacter pylori culture and histological examination of gastric mucosa in 79 patients

TABLE 2
•C pylon wos 1sofoted from cf feast one biopsy PARAl11S el al

TABLE 3
Correlatio n between Gram and Warthin-Starry results fo r C pylori and histological examination of gastric mucosa in 77 patients ( 7,15, is (24)(25)(26)pylobact.erspecieswithapositiveurease test(8 )and this test could su ffice in itself to presumptive id e ntification .A high correlation was fou nd between the presence of C [lylori and gastritis in the biopsies.Thirty of 33 (91 %) of the patie nts who tested positi ve for C pylroi T h ird , C pylori :-ippears particularly adapted to gastric epithelia l cells in wh ich ir is associated with well described speci fic lesions( 7,15, 16,(24)(25)(26).Further, few treatment studies h ;ive been done but the available data up to now a re very