| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
a Author for correspondence. Fax 313-493-6213; e-mail Acutler{at}pol.net
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsDiscussionReferences |
|---|
Key Words: indexing terms: peptic ulcer • Helicobacter infections/diagnosis • immunoassay
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsDiscussionReferences |
|---|
Noninvasive techniques to detect bacterial infection include urea breath tests (UBT) and anti-H. pylori antibody detection by serologic methods.1 The presence of antibodies to H. pylori strongly correlates with histologic evidence of infection in untreated patients (5)(6). Serologic detection of H. pylori antibodies with current methodologies is expensive and time consuming, and requires transport of the specimen to a reference laboratory. An office-based test to screen patients with upper gastrointestinal symptoms for H. pylori would expedite diagnosis and treatment of the infection. Such a test should be inexpensive, easy to perform, accurate, and produce results rapidly, i.e., during the encounter with the patient.
The immunochromatographic test (ICT) method of anti-H. pylori antibody detection is based on the principle of reverse-flow immunochromatography. The Flexsure® HP kit (SmithKline Diagnostics, San Jose, CA) includes the high-molecular-mass cell-associated protein (HM-CAP), an antigen highly specific to H. pylori [7], as the target antigen to detect H. pylori-specific IgG antibodies in human serum. The aim of our study was to determine the sensitivity, specificity, and positive and negative predictive values of this ICT for the detection of IgG antibodies to H. pylori in human serum.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsDiscussionReferences |
|---|
phase i: evaluation in select cohort
Patient population
. Patients presenting to a
gastroenterologist at a large clinic in Detroit for
esophagogastroduodenoscopy from 1990 to 1993 underwent multiple tests
to define their H. pylori infection status. Serum samples
collected as part of the evaluation had been numbered and stored in a
-70 °C freezer. Patients had given prior consent for future
investigations involving stored sera.
Establishment of
H. pylori status. Each
patient's infection status had previously been determined by using a
panel of six tests: (a) rapid urease test (CLO®
test only), (b) antral histology with hematoxylin and eosin
stain, (c) antral histology with silver stain,
(d) ELISA for IgG antibody, (e) ELISA for IgA
antibody, and (f) [13C]UBT. From a
large bank of sera, serum samples from donor patients that were either
positive or negative by all six tests were selected. These banked serum
samples were then used to evaluate the ICT by a single investigator
(E.L.Y.) blinded to the results of all tests used to establish H.
pylori status.
phase ii: evaluation in prospective cohort
Patient population
. Adult (
18 years) patients who met
entry criteria and were scheduled for esophagogastroduodenoscopy
between August 1994 and March 1996 for the evaluation of upper
gastrointestinal symptoms were offered participation. Patients with
exposure to antibiotics, proton pump inhibitors, or bismuth compounds
in the previous 4 weeks were excluded. Patients with previous treatment
for H. pylori were excluded. Informed consent was obtained
before endoscopy.
Establishment of
H. pylori status. Infection
status was determined in all patients by using three tests:
(a) rapid urease tests (CLO test or PyloriTek®
test); (b) antral histology with hematoxylin and eosin
stain; silver stain was used at the pathologist's discretion; and
(c) [13C]UBT.
H. pylori infection was defined by either the presence of H. pylori organisms associated with chronic active inflammation on antral histology or by the combination of both a positive UBT and a positive rapid urease test. Absence of H. pylori infection required both negative antral histology and a negative UBT. Rapid urease testing, if available in these patients, also had to be negative. The ICT was performed by a single investigator (S.M.) blinded to other H. pylori test results with separated sera from these patients as described below.
In phase I, only sera with agreement among all six H. pylori tests were used. Hence, the ICT was evaluated against a strict standard that a priori required H. pylori IgG and IgA antibodies to be in accordance with H. pylori infection status. The test was then evaluated in a more clinical situation against a three-test "gold standard." No tests to determine antibodies against H. pylori were performed in phase II.
tests for detection of h. pylori
Invasive tests
. Antral biopsies were obtained 1–3 cm
from the pylorus and used for rapid urease testing and histology.
1) Rapid urease test: This test involves the use of a single gastric tissue sample to detect antral urease activity. The CLO test (Delta West, Bentley, W. Australia) was exclusively used in patients who participated in phase I. In phase II, either the CLO test or the PyloriTek test (Serim Research, Elkhart, IN) was used to detect antral urease activity. Sensitivity and specificity of these two tests is similar (8). Both tests were performed to the manufacturer's specifications.
2) Histology: Two antral tissue samples were submitted for histopathologic analysis. H. pylori infection was established if typical-appearing organisms were detected in the presence of chronic active gastritis (1). A silver stain (Warthin–Starry stain) was used in all tissue specimens to identify H. pylori organisms in phase I. In phase II, a silver stain (Steiner's stain) was done, upon the pathologist's discretion, when typical organisms were not seen upon hematoxylin and eosin stain when chronic active inflammation was present.
Noninvasive tests
.
1) [13C]UBT: After an overnight fast, every patient
underwent a 30- or 60-min breath test. The increase in mole fraction of
isotope tracer 13CO2 compared with baseline was
expressed as atom percentage excess units
{[(13CO2/12CO2) +
13CO2] x 100}. A UBT result of 6 atom
percentage excess was considered positive. The sensitivity and
specificity ranges of this cutoff value were 95–99% and 94–99%,
respectively (9)(10)(11).
In phase II, [13C]UBT was performed with the
following technical differences: (a) Posturea ingestion
breath sample was collected only at 30 min; (b) an atom
percentage excess value of 5 was considered positive and <5 was
negative. With these cutoff values, the [13C]UBT is
highly accurate, reliable, and reproducible. Sensitivity and
specificity ranges were 95–98% (12)(13). In
both phases the UBT was performed after endoscopy.
2) ELISA: ELISAs (PyloriStat; BioWhittaker, Walkersville, MD) to detect
the IgG and IgA antibodies to H. pylori were performed in
phase I of the study by using sonicates of pooled H. pylori
strains, as previously described (5). Absorbance of 1.0
was considered seropositive and a value of <1.0 was considered
seronegative. Sensitivity and specificity of these assays (range) were
92–96%
(6)(9)(12)(14). ELISAs
were not used in phase II of the study.
3) ICT: A commercially available ICT kit (FlexSure HP) was tested against characterized sera in both phases of the study. The kit contained a test card with a "pink pad" containing colored particles bound to anti-human IgG, a "blue pad" containing a colored dye, a "test line" containing purified H. pylori antigen (HM-CAP), and a "control line" containing conjugate-specific antibodies. In phase I, banked serum (stored at -70 °C in sealed containers) was thawed to room temperature and tested in the ICT kit. In phase II, patients undergoing endoscopy had ~5 mL of venous blood collected aseptically. Serum was separated and utilized in the ICT kit. No serum samples were frozen before testing in phase II. The FlexSure HP test was performed as follows.
Two drops of a buffer solution (PBS with 0.25 g/L sodium azide) were added to the pink pad to reconstitute the conjugate. One drop of undiluted serum from either the banked sample in phase I or the fresh sample in phase II was then added to the pad with the blue dye. The blue dye-colored serum migrated up the chromatographic test strip through a band of immobilized H. pylori antigen (HM-CAP) in the test line. If antibodies specific to H. pylori were present in the serum, they bound the antigen on the test line. The card was then folded, transferring the conjugate from the pink pad and initiating reverse-flow chromatography. The conjugate then migrated down the test strip, passing through the test line and control line. If antibodies to H. pylori had been captured on the test line, the conjugate bound to both the test and control lines, producing two visible lines (a positive test). If antibodies to H. pylori were absent from the serum being tested, the conjugate bound only to the control line (a negative test). Results were read visually at 4 min. The control line provided assurance that the test was functioning properly.
The FlexSure HP product insert reports 99% reproducibility of results in a blinded single-site study (5 serum samples, tested 10 times each on two lots of product by three technicians working independently) and 99% reproducibility of results in a multisite study (5 serum samples, tested 10 times on two lots of products by individual technicians working in three geographically separated locations). In phase I of our study, banked serum samples with inaccurate FlexSure results were retested with the kit to check reproducibility. No serum samples from phase II of the study were retested.
statistics
Sensitivity, specificity, and positive and negative predictive
values were calculated in the traditional method (15). All
ages are reported as mean ± SD.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsDiscussionReferences |
|---|
Of 107 sera, 68 were H. pylori positive (prevalence 63.5%) as established by the six tests. The remaining 39 (36.5%) sera were H. pylori negative on all six tests.
Complete agreement between the ICT and H. pylori status occurred in 102 of 107 serum samples (overall accuracy 95.3%). There were 65 true-positive, 37 true-negative, 2 false-positive, and 3 false-negative results. The ICT had sensitivity, specificity, and positive and negative predictive values of 96% [95% confidence interval (CI) 88–99%], 95% (83–99%), 97% (90–100%), and 93% (80–98%), respectively, in well-characterized banked sera with a H. pylori prevalence rate of 63.5%.
The five banked serum samples with inaccurate H. pylori results were retested with the FlexSure test kit. Results were fully reproducible, and no change in determined antibody status was found.
phase ii
Participating were 96 patients (45 men, 51 women) with mean age
51.1 ± 14.2 years. There were 50 Caucasians, 42
African-Americans, and 4 patients of Middle Eastern descent.
Of 96 patients, 57 (59.4%) were infected with H. pylori as characterized by visualization of H. pylori on antral biopsies or by the combination of a positive UBT and positive rapid urease test. The remaining 39 (40.6%) patients were negative on all tests and were deemed free of infection. The ICT correctly identified the H. pylori status of 84 of 96 patients (overall accuracy 87.5%). There were 54 true-positive, 30 true-negative, 9 false-positive, and 3 false-negative results. Sensitivity, specificity, and positive and negative predictive values of the ICT in this population were 95% (95% CI 85–99%), 77% (61–89%), 86% (75–93%), and 91% (76–98%), respectively.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsDiscussionReferences |
|---|
Many factors affect the performance of a test used to screen a population. These include the prevalence rate of the disease, accuracy of the "gold standard" test(s) used to establish presence or absence of the disease, and the operational characteristics inherent to the methodology used to screen the population. There are several possible explanations for the reduction in specificity in phase II of the study.
Antibiotics are frequently prescribed for common infections (upper respiratory infections, urinary tract infections, etc.). Monotherapy with commonly used antibiotics, such as metronidazole or clarithromycin, can achieve H. pylori eradication rates in up to 17–20% of patients (19)(20). Serum antibody tests remain positive for a significant period, or perhaps indefinitely, after H. pylori eradication (21). Remote ingestion of certain antibiotics in potential study candidates could therefore result in eradication of H. pylori. Such patients would have negative tests for urease activity (UBT and rapid urease test) and (or) nonvisualization of bacteria on histology in the presence of detectable serum antibodies (a positive ICT). Such a situation would result in "false-positive" antibody test results and could contribute to a lower specificity of the ICT. Other conditions that could potentially lower the specificity of antibody-based testing include recent ingestion of antibiotics, bismuth compounds, and proton pump inhibitors. These agents alter gastric mucosal inflammation, bacterial distribution, and urease activity (22)(23)(24), making histologic detection of H. pylori more difficult and thereby increasing the chances for false-negative H. pylori infection status. The use of such products was an exclusion criterion for our study.
Various antigen preparations have been used to develop tests for detection of antibodies to H. pylori. The ICT in this study has the HM-CAP antigen to detect anti-H. pylori IgG antibody. The HM-CAP antigen is highly specific, yet cross-reactivity with other bacteria is always possible and could partly explain the observed decrease in specificity in phase II.
There is no single agreed-upon "gold standard" test for the establishment of H. pylori infection. As such, research results can be affected by the manner in which bacterial infection is established. In our study, we used a combination of tests to eliminate biases inherent in individual test methodologies. Given our test requirements to determine H. pylori infection, it is unlikely that the lowered ICT specificity in phase II was due to missed bacterial infection. Nevertheless, in phase I, only sera with agreement among all six tests, including H. pylori IgG and IgA antibodies, was used. In phase II, infection status was determined by a three-test panel without regard to antibody status. Some spectrum bias between the two phases of the study is possible, in that H. pylori infection status was more clearly established in phase I and less clearly delineated for phase II sera.
Identification of H. pylori infection has become critical in the management of patients with gastroduodenal ulcer disease. The pathogenesis of H. pylori-related gastroduodenal diseases (gastritis, ulcers, neoplasia) is not clearly understood, but eradication of this infection favorably alters the natural history of gastroduodenal ulcer disease. Currently, serology plays an important role in two situations, i.e., screening of various populations to understand the epidemiology of H. pylori infection, and to identify those patients with gastroduodenal ulcer disease who are infected. Infected patients with active or quiescent gastroduodenal disease require antibiotic therapy (25). Absence of H. pylori infection, as manifested by negative serological tests, in patients who are not exposed to nonsteroidal antiinflammatory drugs almost completely excludes the presence of gastroduodenal ulcer disease as well as the precursor lesions of gastric carcinoma.
New whole-blood test technology has recently been approved for commercial distribution. However, whether this new methodology will be as accurate as the serum-based tests is unclear at present (26). Further clinical application of the whole-blood tests is required.
The ICT method used in this study is commercially available (FlexSure HP), inexpensive ($10 to $12 per kit), and easy to perform. Positive and negative controls are built into each test card. The positive predictive value of 86% implies that the test could be used to identify H. pylori infection in patients with upper gastrointestinal symptoms, though a small number of false-positive tests would result in some overtreatment. The negative predictive value of 91% suggests that negative results could be used to rule out infection. Because results are available in 4 min, clinical decisions regarding patient care could be made during the visit. Antibody testing is expected to continue to improve and play an ever increasing role in the primary clinician's diagnostic evaluation.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Presented in part at the American Gastroenterological Association Scientific sessions in 1995 [Digestive Disease Week (DDW), San Diego] and 1996 (DDW, San Francisco).
1 Nonstandard abbreviations: UBT, urea breath test; ICT, immunochromatographic test; HM-CAP, high-molecular-mass cell-associated protein; and CI, confidence interval. 
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  R. J. F. Laheij, H. Straatman, J. B. M. J. Jansen, and A. L. M. Verbeek Evaluation of Commercially Available Helicobacter pylori Serology Kits: a Review J. Clin. Microbiol., October 1, 1998; 36(10): 2803 - 2809. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
