Journal of Microbiological Methods

Transcription

1 Ž. Journal of Microbiological Methods elsevier.comrlocaterjmicmeth Ne immunoassay for the detection of Helicobacter pylori infection compared ith urease test, 13 C breath test and histology: validation in the primary care setting C.F. Weijnen a,), H.A. Hendriks a,b, A.W. Hoes a, W.M. Vereij b, T.J.M. Verheij a, N.J. de Wit a a Julius Center for General Practice and Patient Oriented Research, UniÕersity Medical Center Utrecht, Location Stratenum, 6th Floor, UniÕersiteitseg 100, 3584 CG Utrecht, Netherlands b Stichting Artsenlaboratorium, Utrecht, Netherlands Received 17 April 2001; accepted 23 April 2001 Abstract Helicobacter pylori plays a major role in peptic ulcer disease and, as a result, testing for H. pylori infection among patients ith dyspepsia has often been advocated. The aim of the study as to determine the diagnostic accuracy, the analytical performance, and optimal cut-off point of a ne serological assay, the Pyloriset EIA-G III for the detection of H. pylori infection in the primary care setting. For 113 primary care patients ith dyspepsia urea breath test, CLOe test, histology and serology tests ere performed. Diagnostic accuracy of the Pyloriset EIA-G III as evaluated against a reference standard of a carbon urea breath test Ž CUBT., CLOe test and histology Ž from gastric biopsies.. Precision, linearity and correlation of the serological assay ith the CUBT and former Pyloriset ere also determined. At the optimal cut-off level of 40 Urml, the positive predictive value as 92.1%, negative predictive value 96.3%, sensitivity 87.5%, and specificity 93.9%. The ithin-run precision as high. The recovery data ere good. The correlation of both CUBT and the former Pyloriset EIA-G and the Pyloriset EIA-G III as high. At the cut-off level of 40 Urml, the ne Pyloriset EIA-G III is a reliable method to detect H. pylori infection in the primary care setting. q 2001 Elsevier Science B.V. All rights reserved. Keyords: Helicobacter pylori; Serology; Diagnostic tests; Family practice 1. Introduction ) Corresponding author. Tel.: q ; fax: q address: c.f.eijnen@med.uu.nl Ž C.F. Weijnen.. The bacterium Helicobacter pylori is knon to play a major role in the development of peptic ulcer disease. Infection ith this spiral, urease producing bacterium causes histological gastritis and is an important risk factor for the development of gastric r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. Ž. PII: S

2 236 ( ) C.F. Weijnen et al.rjournal of Microbiological Methods adenocarcinoma and lymphoma ŽNIH, 1994; Marshall and Warren, 1984; Parsonnet, 1993; Eslick et al., Screening the hole general population for infection ith H. pylori does not appear to be cost-effective, but case finding in certain risk groups, i.e. patients ith active ulcers, a history of ulcers, or gastric mucosa-associated lymphoid tissue lymphoma is indicated ŽPeterson et al., 2000; EHPSG, Many invasive and non-invasive methods are available for the detection of H. pylori infection. Invasive methods require an endoscopy to obtain biopsies of gastric tissue, in hich H. pylori can be diagnosed by urease activity, histology or culture of the bacterium. Endoscopy is an inconvenient and expensive method of H. pylori testing Žapproxi- mately EURO 250 per endoscopy including invasive H. pylori testing. and reliable non-invasive methods could be of great help, notably in the primary care setting. Noninvasive techniques to detect bacterial infection include carbon urea breath tests Ž CUBT., antigen stool tests and anti-h. pylori antibody detection by serological methods ŽVaira et al., 1999; Savarino et al., The antigen stool test is promising but needs further validation in different patient settings. The 14 C-UBT is a simple and reliable test, but the radioactive component restricts its practical use. The 13 C-UBT does not have the disadvantage of radioactivity, but requires the availability of an expensive mass spectrometer, resulting in a 13 total cost of EURO 45 per C-UBT Žpersonnel, equipment and materials.. The test characteristics of most office tests, such as hole blood tests, are disappointing ŽQuartero et al., 2000; Jones et al., A fast and reliable test method to diagnose H. pylori infection is to test for antibodies to the antigen of H. pylori. The enzyme immunoassay is the most common used serological test, because it is a reliable, fast and lo cost technique Žper test approximately EURO 11 for personnel, equipment and materials.. Many serological kits for the detection of H. pylori-specific IgG antibodies are no commercially available Ž Laheij et al., 1998; Wilcox et al., The aim of this study as to investigate the analytical performance and reliability of a ne serological assay, the Pyloriset EIA-G III, for the detection of H. pylori infection in the primary care setting. We investigated its test characteristics and analytical performance against a reference standard of CUBT and invasive tests, and against the former Pyloriset, and determined the optimal cut-off level for the ne test. 2. Materials and methods 2.1. Patients Patients ere recruited from general practices in the city of Utrecht, the Netherlands. Eligible for the study ere patients that presented themselves to their GP ith dyspepsia lasting for at least 2 eeks and ere G 18 years old. Pregnant patients and patients ith pulmonal or cardiac comorbidity ere excluded. The patients ere referred to the local primary care laboratory for serological screening for H. Ž pylori infection Pyloriset EIA-G and Pyloriset EIA-G III, Orion Diagnostica, Espoo, Finland.. At the same visit, the patients also underent a 13 C urea breath test ŽPylobactelle, BSIArTorbet, Chatham, U.K..; the breath-test samples ere analysed and results ere expressed as 13 CO r 12 2 CO2 concentra tions Žan increase in CO2r CO 2, concentrations from baseline of more than 3.5 as required for an established H. pylori infection.. Subsequently, patients ere referred to the local hospital for endoscopy, during hich biopsy samples ere taken for histology Žusing Giemsa or haematoxylinreosin stain., and a rapid urease test Ž CLOe, Australia Diagnosis ith Pyloriset EIA-G III The nely developed EIA kit, Pyloriset EIA-G III assay uses microtiter ells coated ith inactive H. pylori antigens. In the present study, the assay procedure steps ere automated using the Biolab 300 Ž Meridian Diagnostics, Cincinnati, USA.. Serum samples ere diluted Ž 1:201. ith serum dilution buffer. Four undiluted calibration sera and diluted samples ere added to the ells, mixed, and incubated at 258C for 30 min. The plate as ashed three times ith ashing buffer, after hich conjugate Žperoxidase conjugated anti-human IgG Ž rabbit. as added to each ell. After mixing and a second incubation at 258C for 30 min, the plate as ashed

3 ( ) C.F. Weijnen et al.rjournal of Microbiological Methods X X again. Substrate Ž3,3,5,5 -tetramethylbenzidine. as then added to each ell and the plate mixed and incubated at 258C for 10 min. The reaction as stopped by adding the stopping solution Ž 1 M H SO. 2 4 and the absorbance of the assay as read at 450 nm. The optical densities of four reference standards ere used to plot a standard curve Ž straight line. by hich the H. pylori IgG antibody levels in patient samples ere quantified. The results ere expressed in arbitrary units per milliliter. Reference standards 1 4 represent 10, 20, 120, 640 Urml, respectively. The absorbance readings are proportional to the logarithm of the antibody concentration. Folloing the manufacturer s interpretation of the assay, the result should be considered positive for H. pylori antibodies if the Urml of the serum is equal or higher than that of the calibrator serum 2 Ž G 20 Urml.. To compare the ne ith the former H. pylori assay, the arbitrary units ere multiplied ith a factor 10 Ž recommended by the manufacturer. to reset these values in titer values as used in the former Pyloriset EIA-G ne kit. The total procedure time for the Pyloriset EIA-G III as 80 min Definition of the reference standard Patients ere considered to be H. pylori infected if at least to of the folloing three tests ere positive: Ž. 1 rapid urease test Ž CLOe.; Ž C urea breath test Ž CUBT.; and Ž 3. histology Analytical performance of the Pyloriset EIA-G III Precision Within-run precision as determined by using sera on three levels. Replicate measurements Ž n s 20. ere performed in one run for each level. This procedure as processed using a single reagent lot. The ithin-run precision data are expressed as coefficients of variation Ž CV, %.. Beteen-run precision as determined using sera on three levels. Replicate measurements Ž ns5. ere performed in different runs. A single reagent lot as used during the measurements. The beteen-run data ere expressed as CVs Linearity The linearity of the Pyloriset EIA-G III as assessed by calculation of the recovery of a repeatedly diluted high concentrate sample Ž ; 800 Urml Effect of rethaing sera on detection ith Pyloriset EIA-G III The possible effect of rethaing sera on the IgG antibody levels as tested ith sera at three levels. The sera have been thaed and frozen again for five times and analysed. The possible effect of rethaing is expressed as CV Correlation of tests Correlations beteen the qualitative test results Ž. i.e., positive or negative of the Pyloriset EIA-G III ith the CUBT, and Pyloriset EIA-G ne ere determined using Cohen s kappa Žmeasurement of agreement Statistical analysis The test characteristics ere reported in terms of sensitivity, specificity, positive and negative predictive values. Measurement of agreement as reported in terms of Cohen s kappa. Statistical analysis as performed using SPSS version 9.0 for Windos. 3. Results 3.1. Patients Beteen April 1999 and January 2000, 133 primary care patients ith dyspepsia ere included, and referred for H. pylori testing and endoscopical diagnosis. For 113 patients, complete data on both the EIA and the reference standard ere available. The H. pylori infection rate, according to the reference test Ž consisting of CUBT, CLO and histology., as 31.7% Validation against reference standard The observed negative and positive predictive values, sensitivities and specificities of the Pylori-

4 238 ( ) C.F. Weijnen et al.rjournal of Microbiological Methods Table 1 Characteristics of the Pyloriset EIA-G III at different cut-off levels in relation to the reference standard Ž 95% confidence interval. Cut-off value 20 Cut-off value 30 Cut-off value 35 Cut-off value 40 PPV 69.8 Ž Ž Ž Ž NPV 98.5 Ž Ž Ž Ž Sensitivity 97.4 Ž Ž Ž Ž Specificity 80.5 Ž Ž Ž Ž PPV: positive predictive value. NPV: negative predictive value. set EIA-G III assay are given in Table 1. Test characteristics ere determined at different cut-off levels Žabove hich an individual as considered to be H. pylori infected.. At cut-off levels for infection varying from Urml, positive predictive values ranged from 69.8% to 92.1%, negative predictive values from 96.3% to 98.6%, sensitivities from 87.5% to 97.4%, and specificities from 80.7% to 93.3% Analytical performance of the Pyloriset EIA-G III Precision The ithin-run precision data, expressed as CVs of the mean levels of 10, 133 and 503 Urml, ere 1.6%, 8.4% and 11.2%, respectively. At the mean levels of 11, 93 and 614 Urml, the beteen-run precision CVs ere 4.8%, 7.5% and 28.7%, respectively Linearity The calculated recovery data are shon in Table 2. Good linearity data ere observed, ith recovery data above 80% for all the to-, four- and eight-time dilutions. Table 2 Table demonstrating the results of dilution in terms of recovery using the Pyloriset EIA-G III Dilution Value Ž Urml. Recovery Ž %. A Ar Ar Ar Effect of rethaing sera on detection ith Pyloriset EIA-G III The CVs measured for the samples at mean levels of 10.5, 162 and 892 Urml, that had been thaed and frozen again, ere respectively 3.4%, 13.5% and 28.4% Correlation of tests The correlation Žcomparing qualitative test re-. sults beteen the breath test and Pyloriset EIA-G III as high: Cohen s kappa of This result as similar for cut-off levels of 35 or 40 Urml for the Pyloriset EIA-G III as level of proven infection. The correlation beteen the former Pyloriset EIA- G and the Pyloriset EIA-G III as high; Cohen s kappa of 0.97 or 0.98 Žusing cut-off values of 40 or 35 Urml, respectively.. 4. Discussion The ne Pyloriset is a reliable method to detect H. pylori infection in primary care. Test characteristics validated against a high quality reference standard are excellent, and correlation of results ith those of CUBT and the former EIA is good. The recommended cut-off value needs to be carefully reconsidered. The cut-off value of the former Pyloriset assay as given by the manufacturer as a titer of 300. Surprisingly for the Pyloriset EIA-G III assay, a cut-off value of 20 Urml is recommended by the manufacturer, here a value of 30 Urml as expected according to the factor 10 difference beteen the to assays. For use in clinical practice, supporting clinical decisions for individual

5 ( ) C.F. Weijnen et al.rjournal of Microbiological Methods Table 3 Percentage of patients incorrectly diagnosed in a population ith H. pylori infection rate of 40% at different cut-off levels Cut-off Cut-off Cut-off Cut-off level 20 level 30 level 35 level 40 False positive False negative patients, the optimal cut-off point should be guided mainly by the positive and negative predicted value of the test Ž PPV and NPV.. The fact that H. pylori diagnosis ill be mainly used in dyspepsia management to support ulcer detecting strategies Žin particular, endoscopy plus antibiotic treatment., puts even more emphasis on to need for a correct H. pylori test result, both positive and negative. Calculating the percentage of incorrectly diagnosed or missed H. pylori infections at different cut-off levels demonstrates the most efficient cut-off point in clinical dyspepsia management in primary care Ž Table 3.. At 40 Urml, only 5% of the patients are incorrectly diagnosed. Ž Validation of the former assay Pyloriset EIA-G Ne. in a similar primary care population resulted in a sensitivity of 91% and specificity 78% ŽLein-van den Broek et al., We used a more solid reference standard Žat least to out of three reference tests positive vs. only one reference test used by Lein-van den Broek et al., 1999., hich resulted in a sensitivity of 94.7% and specificity of 92.7%. The ne assay Ž EIA-G III. performs in a similar ay in terms of sensitivity, specificity, positive and negative predictive value as the former EIA-G Ne. A fe additional advantages of the ne EIA should be addressed. The total procedure time of the EIA-G III is 80 min vs. 160 min in the EIA-G Ne. In contrast to the former EIA, the EIA-G III has a straight calibration line, hich leads to accurate results in the hole calibration range. No disadvantages in comparison ith the former Pyloriset could be found. In conclusion, this ne serology assay is an accurate, reliable and inexpensive screening test for H. pylori infections in dyspeptic patients in a primary care population, but the cut-off level for use in primary care should be increased from 20 to 40 Urml. Acknoledgements This study as supported by a research grant from the Mediphos Medical Supplies Group ŽRen- kum, the Netherlands.. The authors ish to thank Ms. Sjannie Klerx, laboratory technician of the Stichting Artsenlaboratorium Utrecht, for the analysis of the sera. References Eslick, G.D., Lim, L.L., Byles, J.E., Xia, H.H., Talley, N.J., Association of Helicobacter pylori infection ith gastric carcinoma: a meta-analysis. Am. J. Gastroenterol. 94 Ž. 9, European Helicobacter Pylori Study Group, Current European concepts in the management of Helicobacter pylori infection. The Maastricht consensus report. Gut 41 Ž. 1, Jones, R., Phillips, I., Felix, G., Tait, C., An evaluation of near-patient testing for Helicobacter pylori in general practice. Aliment. Pharmacol. Ther. 11 Ž. 1, Laheij, R.J., Straatman, H., Jansen, J.B., Verbeek, A.L., Evaluation of commercially available Helicobacter pylori serology kits: a revie. J. Clin. Microbiol. 36 Ž 10., Lein-van den Broek, N.T., Numans, M.E., Buskens, E., de Wit, N.J., Smout, A.J., Verheij, T.J., Validation and value of an enzyme-linked immunosorbent assay for Helicobacter pylori in primary care. Scand. J. Gastroenterol. 34 Ž. 4, Marshall, B.J., Warren, J.R., Unidentified curved bacilli in the stomach of patients ith gastritis and peptic ulceration. Lancet 16 ŽŽ , National Institutes of Health, Helicobacter pylori in peptic ulcer disease. NIH Consensus Statement 12, Parsonnet, J., Helicobacter pylori and gastric cancer. Gastroenterol. Clin. North Am. 22 Ž. 1, Peterson, W.L., Fendrick, A.M., Cave, D.R., Peura, D.A., Garabedian-Ruffalo, S.M., Laine, L., Helicobacter pylori-related disease: guidelines for testing and treatment. Arch. Intern. Med. 8 Ž160Ž 9.., Quartero, A.O., Numans, M.E., de Melker, R.A., de Wit, N.J., In-practice evaluation of hole-blood Helicobacter pylori test: its usefulness in detecting peptic ulcer disease. Br. J. Gen. Pract. 50 Ž 450., Savarino, V., Vigneri, S., Celle, G., The 13 C urea breath test in the diagnosis of Helicobacter pylori infection. Gut 45 Ž Suppl. 1., I18 I22 Revie.

6 240 ( ) C.F. Weijnen et al.rjournal of Microbiological Methods Vaira, D., Malfertheiner, P., Megraud, F., Axon, A.T., Deltenre, M., Hirschl, A.M., Gasbarrini, G., O Morain, C., Garcia, J.M., Quina, M., Tytgat, G.N., Diagnosis of Helicobacter pylori infection ith a ne noninvasive antigen-based assay group. Lancet 3 Ž354Ž , Wilcox, M.H., Dent, T.H., Hunter, J.O., Gray, J.J., Bron, D.F., Wight, D.G., Wraight, E.P., Accuracy of serology for the diagnosis of Helicobacter pylori infection a comparison of eight kits. J. Clin. Pathol. 49 Ž. 5,