Evaluation of IgA and IgG serology for the detection of Helicobacter pylori infection

HELICOBACTER PYLORI HAS, IN A short period of time, become one of the most important gastrointestinal infections. Since its identification in the 1980s this organism has dramatically affected gastroenterology, and it is now felt to play major roles in the recurrence or pathogenesis of duodenal ulcer disease, chronic antral gastritis, gastric ulcer disease and, possibly, nonulcer dyspepsia (1-3). Patients with gastric adenocarcinoma have also been shown to exhibit a higher prevalence of H pylori infection (4-6). The importance of H pylori makes it imperative to develop a safe, noninvasive and simple method of detection. Histological staining and microbiological culture of antral tissue remain the gold standards for H pylori detection (1,7). Unfortunately, both these methods require gastroscopy, and hence are expensive, time-consuming and not without complications. The biopsy urease or campylobacter-like organism (CLO) test requires less time because the results are available at gastroscopy (8,9). However, endoscopy is still necessary, with its associated disadvantages. Noninvasive methods of H pylori detection include serology and breath urea testing. Serological testing requires a few millilitres of serum compared with breath urea testing where ingestion of 13Cor 14C-labelled urea, followed by breath sampling 20 to 30 mins later to trap expired carbon dioxide, is needed. The breath tests also involve use of an expensive mass spectrometer with the 13C method, and raCLINICAL GASTROENTEROLOGY

diation exposure (180 mGy for helicobacter infected patients, which is approximately equivalent to the exposure from a chest x-ray, and 70 mGy for uninfected patients) with the 14 C method (1,7,10).
The aims of this study were to evalu-ate objectively two commercial kits for the detection of H pylori antibodies, to compare immunoglobulin (Ig) A versus IgG antibodies in terms of sensitivity and specificity for ongoing H pylori infection, to determine the optimal serum threshold values for each and to assess the usefulness of serology testing one month after H pylori eradication in patients who have not had pretreatment serology.

PATIENTS AND METHODS
Adult patients presenting for upper gastrointestinal endoscopy at the Montreal General Hospital between November 1992 and May 1993 who were not taking acid suppression medication or antibiotics were asked to par-ticipate in this prospective study.Informed consent was obtained in all patients in accordance with the guidelines set forth by the Institution's Ethics Review Board.Initially subjects underwent four antral biopsies for histological detection of H pylori. Part way through the study it became clear that two biopsies were sufficient for an accurate diagnosis (32), and all subsequent subjects underwent only two antral biopsies.A serum sample was also drawn for serological testing at the time of endoscopy.There were 27 sets of samples from such patients.Nine additional samples were drawn from patients who had completed treatment for H pylori infection one month before serological sampling.These patients had not undergone serological testing before treatment.Eradication was successful in six of these nine patients, all of whom had been positive for H pylori on histology in the past, had had duodenal ulcers and had undergone different eradication treatment regimens for H pylori. Successful eradication was defined as the absence of H pylori organisms on histology three months after termination of treatment.
IgG and IgA antibody levels were measured from the serum samples by following the directions provided in two commercial kits: Pyloriset EIA-G and Pyloriset EIA-A (Orion Diagnostica, Espoo, Finland).These kits use enzyme immunoassay techniques to identify antibodies directed at the acid glycine extract, which comprises a number of protein antigens, including the species-specific 120 kDa protein, the presumptive flagellar 61 to 62 kDa protein and the outer membrane 29 to 31 kDa protein, but only reduced amounts of the major cross-reactive 54 kDa protein (33).Briefly the ELISA involves dilution of the patient's serum (1 to 201 for IgG and 1 to 101 for IgA) with the provided serum dilution buffer, placement of this diluted solution into the provided H pylori antigen-coated wells and incubation for 60 mins.The wells are then washed, leaving only the serum antibodies that had bound to the antigen-coated wells.The provided enzyme conjugate, which binds to the antibody, is placed into the Mesure sérologique de l'IgA et de l'IgG dans le dépistage de l'infection à Helicobacter pylori DONNÉES DE DÉPART : La sérologie est un moyen sûr, simple et non effractif de déterminer la présence de Helicobacter pylori.Les rapports sur le degré de sensibilité de spécificité ont toutefois varié considérablement et l'on ignore si une sérologie positive est le reflet d'une infection antérieure (traitée au cours des deux dernières années) ou d'une infection évolutive.OBJECTIFS : Évaluer l'utilité de trousses commerciales de dépistage de l'infection évolutive à H. pylori et de son éradication.Les trousses, Pyloriset EIA-A et EIA-G, utilisent des méthodes immuno-enzymologiques pour mesurer les titres d'immunoglobuline (IgA et IgG respectivement).PATIENTS ET MÉTHODES : Des patients adultes et consentants adressées pour une endoscopie des voies digestives supérieures ont subi des biopsies antrales à des fins histologiques ainsi que des prélèvements pour analyse sérologique.Les échantillons sériques ont été divisés en trois groupes : groupe 1 -17 échantillons de patients atteints de H. pylori décelé à l'histologie; groupe 2 -13 échantillons provenant de patients présentant une histologie négative et aucun traitement récent d'éradication (moins de trois mois); et groupe 3 -6 échantillons de patients chez qui H. pylori avait été éradiqué avec succès un mois avant l'épreuve sérologique.La capacité des épreuves sérologiques sur l'IgA et l'IgG à différencier entre la présence ou l'absence de H. pylori a été évaluée à l'aide d'analyses par tables de contingence.Les points de clivage des titres idéaux ont été mesurés à l'aide de l'analyse des courbes de caractéristiques opérationnelles.RÉSULTATS : Les titres de clivage idéaux pour l'IgA et l'IgG ont été estimés à 300 et 900 U respectivement.La sensibilité à l'égard de l'IgA a été de 82 % (IC 95 % : 57-96), la spécificité à 85 % (55-98) et la précision diagnostique à 83 % (65-94).Pour l'IgG, la sensibilité, la spécificité et la précision diagnostique ont été de 76 % (IC 95 % : 50-93), 85 % (55-98), et 80 % (61-92) respectivement.Les taux d'IgA un mois après l'éradication (groupe 3) étaient nettement inférieurs (P<0,05) aux titres mesurés chez les patients positifs à l'égard de H. pylori (groupe 1) contrairement au taux d'IgG.CONCLUSIONS : La sérologie est une méthode simple non effractive de dépistage de H. pylori, qui s'accompagne d'une bonne précision diagnostique.Les épreuves Pyloriset EIA-A présentent des caractéristiques de rendement similaires à celles du Pyloriset EIA-G.Un mois après le traitement pour éradication, le dépistage de l'IgA, mais non pas de l'IgG, peut être une bonne méthode pour vérifier si H. pylori a été éliminé.wells, followed by another incubation period of 60 mins.The wells are washed again and incubated with the provided substrate solution, which reacts with the conjugate until the addition of the provided stopping solution 30 mins later.This reaction causes a colour change in the solution within the well, depending on the amount of conjugate present, and reflects the concentration of antibody within the serum.The amount of colour change is then determined objectively by measuring the absorbance of the well with an ELISA reader (EMAX Precision Microplate Reader, Molecular Devices, California) set at 405 nm.Control serum solutions of known H pylori titre, which are provided with the kit, are also run in an identical manner alongside the patient serum samples each time the ELISA reader is used.The serum sample titre is obtained by comparing the measured absorbance from a curve constructed using the control samples.It is suggested by the manufacturer that any titre 500 U or more be considered positive for the presence of H pylori antibodies, and less than 500 U be considered negative.The investigator performing the serology testing was blinded to the histological results.
Antral biopsies were interpreted by an experienced pathologist blinded to the serology results using hematoxylin-eosin stains with or without silver stains for H pylori detection.In one patient, the tissue biopsy was insufficient for diagnosis but culture happened to have been done and was positive; hence, this patient was included in the group of positive biopsy results.Statistical analysis: All results are expressed as actual values or as mean ± SEM.Independent t tests, with Bonferroni adjustments, were used to assess statistical significance between groups.Test performances were analyzed with 2x2 tables.95% CI were calculated for proportions reflecting test performance using the standard normal approximation of the binomial distribution.The serology threshold value able to differentiate between the presence or absence of H pylori infection was assessed by using receiver operating characteris-tic (ROC) curve analysis (34)(35)(36).A ROC curve displays the false positive rate on the x axis (1 -specificity), and the true positive rate on the y axis (sensitivity) for the varying test thresholds, thus plotting the performance of a diagnostic test (34,36).For both IgG and IgA serology, five possible cut-offs were chosen: 300, 500, 700 and 900 U for both IgA and IgG as well as 200 U for IgA and 1100 U for IgG.The ideal cutoff values for IgA and IgG titre were chosen by determining the point lying geometrically closest to an ideal test with 100% specificity and sensitivity (the upper left corner of the graph) (35,36).A modified maximum likelihood program was used to compare the area under the fitted curve for the IgA serological test, which represents test performance, with the area beneath the curve for the IgG test (37,38).These comparisons were performed (where appropriate) with matched analysis.

RESULTS
Patient population: Thirty-six biopsies and serum samples were analyzed.The endoscopic diagnoses of these patients are shown in Table 1.The majority of the patients included in the study had duodenal ulcer disease.Ninety-three per cent of duodenal ulcer patients who had not received eradication therapy within the preceding three months were positive for H pylori on histology.Successful eradication of H pylori was noted in 67% of patients who had undergone eradication therapy with different regimens.Eleven subjects had a normal upper gastrointestinal endoscopy which had been performed for indications listed in Table 1.Two patients underwent endoscopic retrograde cholangiopancreatography (ERCP) for obstructive jaundice.The stomach and duodenum of these patients were carefully examined prospectively with the duodenoscope, and biopsies of the antrum were taken during the ERCP.
Overall, the presence of H pylori was histologically detected in 17 biopsies (group 1).Nineteen samples were negative for this organism including six samples obtained from patients who had been successfully treated out of the nine who took anti-H pylori medication.Because the serum samples for serology were obtained only one month after termination of treatment, and it is not known whether the antibody titre would have dropped by that time, these six samples were analyzed as a group separate from the remaining 13 negative samples.Hence, there are three groups of serum samples for serology: the first group consists of 17 samples from patients with positive biopsies, including the three who failed a recent CAN J GASTROENTEROL VOL 9 NO 2 MARCH/APRIL 1995 attempt at eradication therapy; the second comprises 13 samples from patients with negative histology and no recent (three months or less) eradication treatment; and the third group is six samples taken one month after successful eradication of H pylori.

IgA and IgG sensitivities and specificities:
The IgA and IgG titre for each serum sample in the three groups are shown in Figures 1 and 2, respectively.Sensitivities and specificities of both serological tests, using histology as the gold standard, were then calculated by using the aforementioned five different cut-off limits.These values are plotted on the ROC curves shown in Figure 3.The aim in using these curves is to maximize the number of true positives and simultaneously to minimize the number of false positives.The ideal cut-off values for IgA and IgG titre were 300 and 900 U, respectively.As shown in Table 2, when an IgA titre cut-off of 300 U was used the sensitivity was 82% (95% CI: 57 to 96), specificity was 85% (55 to 98), negative predictive value was 79% (49 to 95), positive predictive value was 88% (62 to 98) and the diagnostic accuracy was 83% (65 to 94).If the 500 U cut-off was used, as suggested by the manufacturer, the sensitivity, specificity, negative predictive value, positive predictive value and diagnostic accuracy were 71% (95% CI: 44 to 90), 92% (64 to 99), 71% (44 to 90), 92% (64 to 99) and 80% (61 to 92), respectively.For IgG serology (Table 2), using the 900 U cut-off yielded a sensitivity of 76% (50 to 93), specificity of 85% (55 to 98), negative predictive value of 73% (45 to 92), positive predictive value of 87% (60 to 98) and diagnostic accuracy of 80% (61 to 92).If the 500 U cut-off was used, as suggested by the manufacturer, the sensitivity, specificity, negative predictive value, positive predictive value and diagnostic accuracy were 82% (95% CI: 57 to 96), 62% (32 to 86), 73% (39 to 94), 74% (49 to 91) and 73% (54 to 88), respectively.The performance of the IgG test was not significantly different from that of the IgA test (area under ROC curve was 0.85 versus 0.82 respectively, P=0.70) (Figure 3).The mean IgA titre for the positive histology group was 850±129 U.The negative histology group had a lower IgA titre (379±220 U), but this trend did not achieve statistical significance (P=0.06, Figure 1).In this latter group there was a very high serological titre in one patient (labelled AD in Figures 1  and 2) found to be negative histologically.This was the only false positive IgA titre when the 500 U cut-off was used, and one of two false positives if the 300 U cut-off was used.The IgG titre had more false positives (five with the 500 U cut-off and two with the 900 U cut-off, Figure 2).The IgG titre from the negative histology group was significantly different from the positive histology group, with respective means of 612±210 and 2195±401 U (P<0.01).Use of serology in assessing eradication one month post-treatment: Group 3 had a mean IgA titre of 459±72 U.This was significantly (P<0.05)lower than the positive histology group (mean IgA titre of 850±129 U), suggesting that this antibody might be helpful for follow-up of H pylori status as soon as one month after cessation of therapy (Figure 1).IgG titre of this group, in contrast, still remained high, with a mean of 1732±525 U (Figure 2), not significantly different from the positive histology group (mean IgG titre 2195±401 U).

DISCUSSION
The clinical importance of H pylori infection with regard to peptic ulcer disease is no longer doubted (1-3).It is important that a safe, simple, noninvasive and effective method of identifying this organism be easily available to clinicians.Serology is certainly safe, simple and noninvasive, and we have shown that it can discriminate between H pylori infected and noninfected patients.IgG titre was significantly lower in patients who were negative for H pylori on histology, and IgA titre just barely failed to show significance because of a small sample size with one outlying patient, who had the highest IgA and IgG titres in group 2. Perhaps this was a false negative histological study rather than two false positive serological studies.This patient had cirrhosis and nonulcer dyspepsia.If this patient is excluded, the difference in the IgA titre means between the histologically negative and positive groups become highly significant (P=0.00006).Despite this patient's results, the test performances were good, with IgA and IgG serology displaying similar sensitivity and specificity (82% and 76%, 85% and 85%, respectively).Also, although the data analyzed using mean IgA titres did not achieve statistical significance, the strength of the association in the 2x2 table (P=0.001,c 2 =10.72, df=1, odds ratio = 25.7 [95% CI: 2.8-308]) is strong enough to justify consideration of the use of IgA serology in the detection of H pylori.
We found that the best cut-off values in our population, which consisted of patients referred for upper endoscopy, were 300 U for IgA and 900 U for IgG.This is in contrast to the manufacturer's recommended cutoff of 500 U for both these tests.The reason for this discrepancy may result from the fact that the population studied by the manufacturer was different in age, ethnicity and endoscopic diagnosis (16).
One of the most useful indications for a noninvasive test of H pylori detection is to assess eradication after attempted treatment.The use of serology has always been questioned in this setting because of the uncertainty of anti-body persistence after eradication.We have shown that the IgG titre remains elevated one month after successful eradication and is thus not useful to assess eradication one month after cessation of treatment (Figure 2).IgA titre, on the other hand, decreased significantly one month after cessation of successful treatment (Figure 1).Thus, the drop in antibody level may be an adequate indication of successful treatment.These findings are consistent with Gobert and colleagues' ( 17) study where IgM and IgA levels were found to have significantly decreased after six weeks of therapy.Vaira et al (39) and Veenendaal et al (40) also demonstrated significant declines in IgA titre after eradication.However, Kosunen et al (41) found that, although significant, the decrease in IgA titres was not as consistent as, and did not occur more rapidly than, that of IgG.The reason for this discrepancy is not entirely clear, but may be related to the differing diagnostic tests used in the studies.In all of these studies, the IgG titre showed a trend towards declining levels, but the decrease was not significant until 18 to 24 weeks after treatment.Thus, IgG levels only became a reliable indicator of successful treatment four to six months after treatment (18,41,42).IgA could therefore be a  more clinically useful test because seroconversion would be detected early on, and hence obviate the need for prolonged follow-up.Further study with a larger number of patients, followed preand posteradication, is required to determine the optimal IgA titre cut-off or percentage drop in titre necessary to identify successful eradication one month after treatment.This study was not designed to determine the exact clinical role of serological testing compared with other diagnostic techniques in the detection of H pylori.However, serology is an effective method of making the diagnosis of ongoing H pylori infection if one uses the optimal cut-off values of 300 U for IgA titre and 900 U for IgG titre.The Pyloriset EIA-A assay has test performance characteristics similar to the Pyloriset EIA-G, and the former may be useful in establishing eradication as soon as one month after cessation of treatment.Further studies will need to determine whether serol-ogy alone is sufficient in certain clinical settings to determine subsequent clinical strategy.

ACKNOWLEDGEMENTS:
The authors are indebted to C Wickham for her assistance in data collection.Dr Barkun is a Chercheur Clinicien Boursier of the Fonds de la Recherche en Santé du Québec.Research was supported, in part, by Kabi Pharmacia, Sweden.

Figure 1 )Figure 2 )
Figure 1) Individual serology titres (filled circles) for immunoglobulin (Ig) A antibodies in patients with positive histology for Helicobacter pylori, negative histology for H pylori and one month after successful H pylori eradication.The mean titre for IgA antibodies in each group± SEM ( --) is shown next to the group of individual titres.The dashed line represents the titre above which the manufacturer recommends the serology be considered positive for H pylori antibodies, ie, 500 U.The dotted line represents the 300 U cut-off.AD refers to the patient with the highest titre in the histologically negative group.The mean IgA titre was significantly lower (P<0.05) in the treated group compared with the positive group

Figure 3 )
Figure 3) Receiver operating characteristics (ROC) curve for immunoglobulin (Ig) A and IgG serology.The sensitivity and 1 -specificity of both IgA and IgG tests are plotted for five different cut-off values.The cut-off values are indicated with each point

TABLE 1 Endoscopic diagnosis of patients used in this study
*These 17 patients comprise group 1; † These six patients comprise group 3 (the 13 remaining negative patients comprise group 2)