Detection of antibody against Helicobacter pylori in the saliva of patients with dyspepsia

There is a need to develop noninvasive assays to detect Helicobacter pylori infection in the gastric mucosa, Current dogma predicts that the presence of antibody within saliva should accurately reflect contemporary colonization of the gut mucosa. This study examined the clinical value of a saliva enzyme-linked immunoadsorbent assay (ELISA) for anti-H pylori antibody, compared with the serum ELISA assay, and found the sensitivity of the saliva assay was 89%, specificity 94%, accuracy 93%, positive predictive value 89% and negative predictive value 94%. Assessment following eradication therapy demonstrated that salivary antibody was a more sensitive indicator of colonization than was serum antibody. The immunoglobulin G antibody in saliva correlated best with colonization, and regression analysis was most consistent with a local production of antibody. These results indicate that detection of antibody in saliva contributes to diagnosis and management of H pylori infection.

Depistage de l'anticorps dirige contre Helicobacter pylori dans la salive de patients atteints de dyspepsie RESUME : ll devicnc necessa ire de mettre au point des methodes no n effra cti vcs pour depi tc r !'infec ti on a Helicobacter pylori dans la muqucusc gastriquc.clon l'hyporh ese qui a cours actuellement, la presence de l'anticorps dans la sa live pourrait rcflctcr unc co lonisati on ac ti ve de la muquc u e digc rive.Ccttc ctud c s'e t pcnchcc sur l'urilite clinique du re •r ELISA pour la recherche de l'anricorps anti-H pylori dans la sa livc p lut6t quc dan le sang ct a note pour ccs tests sur de cchantillo n de sa livc, un degre de scnsibilite de 89 %, de specificite de 94 % ct de precisio n de 93 %, ainsi qu 'une va lcur de previsibilitc positive de 89 % cc de prcvisibilitc negative de 94 %.
Les test~ effec tues sui te a un trnitemenr d'eradication o nt confirme que l'anticorp conrenu dan la sa live c t un marqueur plus prccis de la colonisation que ne l'est l'anticorps cnntenu clans le sa ng, L'a nricorps IgG present dans la sa li ve ctait e n mcilleure correlati on avec la colonisati n et l'analy c de regress io n concord ait le plu avec une producti on loca le d'anticorps.Ces resul tats indiq uent que le depistagc de l'anti co rps clans la sa livc contribu e au diag nostic er a u trairc mcnt de  H EU C013ACTER PYLO RI IS A MAJOR cause of gastriti s ( 1,2 ) a nd erad icatio n of in fec tio n is now recomme nded , at least in patie n ts wit h assoc iated pe pt ic ulcer di sease (3) .Inc reas ingly, at te ntio n has bee n paid to t he di ag nos is of infect io n using no nin vasive test to e nab le both se lect ive endoscopy (4 ,5) a nd t he in itiatio n of logica l manageme n t-S uch tests sho uld ass ist the furth er e pide mio log ica l assessme n t of the effec ts of age, race and e nv ironme n tal fac tors on infect io n with H /Jylori inc lud ing the link be tween c h ro nic in fect ion and gast ri c ca nce r ( 6, 7).
The detectio n of scrum a ntibod y is a sensitive a nd spec ific assay fo r infection with ad van rngcs over urea breath tests, t he c urre ntl y ava il able a lte rnative to sero logica l di agnos is (8,9 ).Indeed, it has been recomme nded t hat sc rum a nt ibody assay sho uld replace c ndo •copy as t he ini tial investigat io n of d yspeps ia in subjects yo unger th an 4 5 yea rs (4 ), Sc rum a nti bod y lcve b , however, respond slow ly to a cha nge in H pylori sta tus (1 0) and have ce rta in technical diffic ul t ies in pat ien t manage me n t, leav ing roo m fo r a n a lte rnative non invas ive assay to de tect infect io n .W e, th e refore, compa red the prese nce and level of sa li vary a n tibod y to H /)ylori with a serum test in subjects wi th dyspepsia to dete rmine the va lue of a sali va a nti bod y assay in mon itoring infec ti on, PATIENTS AND METHODS O ne h undred and thirty-fo ur patients with dyspepsia ( 58 ma les, 76 fe males; mean age 58,9 yea rs [ra nge 22 to 31) were referred for endoscopy.infection with H pylori was established via biopsy of the gastric antrum by demonstration of bacteria by histology or by detection of urea e production.Subjects with a positive urease test were offered eradication therapy consisting of colloida l bismuth subcirrate (2xl08 mg), metronidazole (400 mg) and amoxycillin trihydrate (500 mg).Bismuth was administered every 12 h for four weeks while metronidazole and amoxyci llin trihydrate were administered every 8 h for two weeks.The treated group was re-endoscoped eight weeks after beginning antimicrobial therapy and reasessed for the presence of H /)ylori.In addition, unstimulated saliva and •crum were collected at study start and at the post-treatment visit.lmmunoglobulin G (IgG) antibody to H pylori was measured using an enzyme-linked immunoadsorbent assay (ELISA).Separate assay systems have been developed for serum and saliva (11 ).These assays u e a purified high molecular weight antigen, and the saliva assay has been commercially developed by Cortecs Diagnostics (Deeside, United Kingdom).Comparison of saliva ELISA for anti-H pylori antibody with the serum ELISA assay: The database obtained from the 134 patients was used to determine the clinical value characteristics of salivary antibody detection against the sero logical results.Sensitivity of the saliva assay was 89%, specificity 94%, accuracy 93%, po itive predictive value 89% and negative predictive value 94%.Regression analysis of lgG anti-H pylori antibody from paired serum and saliva samples (Figure 2) (r2=0.55)was most consistent with a contribution to antibody in saliva from local synthesis.The pretreatment level of salivary antibody was a major determinant of the time taken to fall in response to antim icrobial treatment (Tables 3,4) .Thus for four-week post-treatment negative sa liva tests, the mean pretreatment antibody level was 1.5 ELISA units (EU)/mL compared with 3.3 EU/ml for those who remained positive but at le s than 50% of pretreatment values.For those not fa lling below the 50% level, the pretreatment level was 5.0 EU/ml.Three patients in this la t group tested at six months showed leve ls less than 50% (n=2) or negative (n= 1) for saliva antibody .

Distribution analysis according to parameters of infection with H pylori:
Eradication was not achieved by four weeks in three patients (Table 1, patient 10; Table 3, patients 5 and 6).Salivary antibody fe ll in all three at four weeks.This fa ll in antibody level probably reflects a response to partial eradication of bacteria as judged by im provement in both clinical and laboratory (seroconversion and reversal of a po itive urease test) parameters.

DISCUSSION
Anti-H pylori IgG antibody has been detected in the saliva of patients infected with H pylori.The concordance with serum antibody confirmed its value as a diagnostic a ay fo r H pyloi1 infection in patients with dyspepsia.Reasse sment four weeks after completing a four-week erad ication program demon traced the saliva assay to have advantage over the erum test as a practical monitor of successful therapy.A level of independence of lgG anti• body levels in saliva was consistent with a component of local secretion from B lymphocytes migrating from the gut mucosa.
Both antibody assays used the same H pylori antigen, thus allowing com-   disappearance of serum antibody was about half that of salivary antibody.The rate of fall in serum antibody concentration, however, was faster than rates previously reported using tests that contained crude bacterial extract for antigen (10).Patient with high saliva (and serum) antibody levels took longer to revert to negative following successful eradication therapy.With standard criteria, three patients failed to eradicate H pylori, all of whom had a fall in salivary antibody.Two of these patients had histological, serological and clinical evidence of a partial response to antimicrobial therapy.Two of the three had high pre-therapy levels of antibody, suggesting that high levels of antibody may predict a risk group for failure LO eraJicate bacteria.An early 'dip' in serum antiboJy following eradication therapy has also been noteJ in patients with a partial response to therapy ( lO).A lo nge r period of observation is needed to define the kinetics of rhe salivary ant ibody response to erad ication treatm nt ancl to determine any clinical va lue in identifying subject at risk of fa iling eradication therapy.An important group of patients with dyspepsia was urease-and histology negative (5.2% of total stud y group) but had H pylori antibody in both serum and saliva.ix of the seven patients in this group had chronic gastritis, two of whom also had changes in acute inflammation .lt is likely thar gastritis in the e patients is associated will, a low level of infection not detected by clasical rests.This concept is consistent with the obse rvation of reversal of damage with antimicrobial therapy in the ahsence of detectable infection ( 12), and of the ,ign ifi cant numbers of patients with peptic ulcer without identifiable cause (13,14).Eradication therapy monitored with salivary anribody levels and polymerase chain reaction ana lysis of biopsy material (15) may estab lish infection as a basis of gastritis in a proportion of these patients.The source of the [gG antibody in aliva is not clear but is consistent with previous demonstrations of lgG antibody following mucosa!stimu lation with microbial antigens ( 16, l 7).ln animal models, secretory [gG antihody correlates with the kin tics of local lgA secretion anJ does not clo ely fo llow serum levels (18), suggesting that local synthesi • from gut-derived lymphocytes may contribute to the lgG antibody pool in •ecretiom.Further support for local secretion of [gG was gained from the observation of lgG-containing plasma cell within the respiratory muco a following intestinal immunization (19).This concept is consistent with observations in the current study of a poor correlation between [gG antibody levels in saliva and scrum, and ;:i difference in elimination kinetics between serum an<l saliva.Analysis of the suhclass distribution of lgG antibody to H /Jylori in the saliva of infected patients showed clear differences from the pattern ob erved in homologow, serum samples (20).We have shown that lgA anti-H pylon antibody is irregularly found in the aliva of infected patients (unpublished data).The variable nppearance of antibody isotypes in saliva in different mucosa!infections may reflect patterns of antigen presentation to Peyer's patches or specific characteristics of antigen epitopes.
It is considered that the salivary antibody ELISA should prove u •eful in the clinical assessment of patients with dyspepsia and in monitoring the result, of eradication treatment.ali vary anti-H pylori antibody is a more reliable index of current infection than serological a es, ment.
ubjects were defined as having 'evidence of H pylori infection' if any of the four indicators -urease test, or histological evidence of bacteria, serum antibody or salivary antibody -were po itive.The characteristics of 'positive' subjects are shown in Figure l.

TABLE 2 Conversion
from positive to negative results for anti-Helicobacter pylori antibody at one month post-treatment va-negative (Table1), more than twice the percentage that became seronegative.If a reduction in antibody titre of more than 50% was assessed (Table2), less than half the serum test resu lts were reduced by at least 50% while the result for saliva (83%) was of the ame order

Detection of H pylori antibody in saliva
8 No 7 DECEMBER J 994

TABLE 3 Patients with pretreatment positive saliva anti -Helicobacter pylori antibody who at one month post-treatment have positive saliva antibody and saliva antibody less than 50% pretreatment level Antral mucosa biopsy Saliva Histopathology antibody• Serum antibody t
'Positive saliva'20.10enzyme-linkedimmunoadsorbent assay units/mL (EU): 1 Positive serum '22.50 EU: tLess than 50% of pretreatment level; +Positive; -Negative: DU Duodenal ulcer: GU Gastric ulcer; PRE Pretreatment: POST Post-treatment

TABLE 4 Patients with pretreatment positive saliva anti-Helicobacter pylori antibody who at one month post-treatment have positive saliva antibody and saliva antibody not less than 50% pretreatment level Antra l mucosa biopsy Saliva Histopathology antibody• Serum antibody t H pylori
6m Six month; •Positive saliva '20.70 enzyme-linked immunoadsorbent assay units/mL (EU); 1 Positive serum '22.50 EU: 1 Less than 50% of pretreatment level: +Positive: -Negative: DU Duodenal ulcer: GU Gastric ulcer: PRE Pretreatment: POST Post-treatment CAN J GASTROENTEROL VOL 8 No 7 DECEMBER 1994