TABLE OF CONTENTS
Cases
Clinical
Questions
Guidelines
Diagnosis:
Choosing a Test
Comment
on Office Kits
Comment
of Salivary IgG Testing
Diagnostic Performance of H. Pylori Tests: A Comparison
Results
Comparison
of Tests
Likelihood Ratios, Pretest and Posttest Probability
User's Guides to the Medical Literature
Cost
Comment on Cost
Revisiting
the Cases
Summary
helicobacter pylori
a diagnostic approach
Raquel S. Watkins, M. D.
Wake Forest University
Internal Medicine Grand Rounds
September, 1998
Considerable confusion exists over the appropriate diagnostic testing for
Helicobacter Pylori. Although it was discovered and cultured by Warren and Marshall in
1982, the identification of this gram-negative spiral shaped organism dates back to 1889.(12)
More than a century after its initial identification and more than sixteen years
after it was finally cultured, H. pylori continues to revolutionize our approach to
dyspepsia.
27 year old Hispanic man with no
significant past medical history.
S: stomach pain, gnawing sensation
R.O.S: pain usually relieved by food, severe enough to
wake him from sleep, present for about one month prior to presentation, no melena, no n/v,
no family history of malignancy, no tobacco, no medications
Diet: spicy foods, moderate caffeine intake, infrequent ETOH, no NSAIDS
O: vital signs stable
Abm: mild epigastric tenderness, normal bowel sounds,
(-) masses, (-) HSM,
Heme(-)
I/P:Differential: dyspepsia: gastritis, gerd, pud, less
likely pancreatitis
Check amylase, lipase
Stop caffeine, alcohol, spices
Prescription given for Zantac 150 Q hs
Return W/I one month or prn.
He returned within four weeks with the same complaints
Reported compliance with Zantac
H. pylori serology checked and was positive
MOC therapy given
No abdominal discomfort on return in two weeks
50 year old African-american
woman with a past medical history of depression
S: new onset abdominal pain, burning sensation, relieved
with food
R.O.S: early satiety, no melena, no nausea, no weight
loss, no tobacco, no ETOH, no NSAIDS, family history of GI "tumor"
Diet: no change recently
Meds.: St John’s Wort for two years
O: Vital signs stable
Abm.: mid-epigastric tenderness with palpation, normal
bowel sounds, no rebound, no guarding, Heme (-)
I/P: gastritis:
H. pylori serology sent and was positive
MOC therapy samples given without resolution of symptoms
Referred for endoscopy
Endoscopy pending
In an office based setting, when dealing with a
dyspeptic patient:
- What is the diagnostic performance of H. pylori tests?
- How should the primary care physician approach making the diagnosis?
Maastricht Consensus Report(5)
A panel of representatives of European physicians issued the Maastricht
Consensus Report in 1997. They declared that H. pylori diagnosis begin at the general
practitioner level. This recommendation charges the primary care physician to become
familiar with the diagnostic performance of H. pylori tests, and the variables involved in
making the diagnosis. In addition, they issued a recommended approach to the management of
the dyspeptic patient in the community.
Dyspeptic
patient
First primary care visit |
| <45 years old |
OR |
>45 years old |
| no *alarm symptoms |
|
*alarm symptoms |
review history
Test for H. pylori
° 13 C UBT
or
° lab serology
|
|
refer |
| if H. pylori (+) then treat |
|
*alarm symptoms: weight loss,
anemia, dysphagia, palpable mass, malabsorption, bleeding
Fig. 1 (5)
American Gastrointestinal Association: Medical
Consensus Statement:
Management of Dyspepsia Fig 2(2)
*(Figure is missing)
In choosing a test, the primary care physician is faced with many important
issues. The patient’s history and clinical presentation are key. For example prior
treatment for H. pylori infection, recent treatment with proton pump inhibitors, or the
presence of warning sign (anemia, gastrointestinal bleeding, and weight loss) should
impact on the choice of testing. Cost and the test’s accuracy and precision must also
be considered.
The two major categories for the diagnosis of H. pylori infection include
invasive and non-invasive testing. Invasive tests require endoscopy and include culture,
biopsy, histology and stain, and the Clo test. Non-invasive tests include quantitative or
qualitative serology (anti H. pylori IgG or IgA) which can be done with either lab ELISA
or office kits, the urea breath test, and salivary testing. Ideally culture is usually the
gold standard when dealing with bacterial infections, but culturing H. pylori is a tedious
procedure and is subject to errors in storage and handling and therefore biopsy with
histology and stain is considered the gold standard. (8,14)
Graham et al conducted a prospective study on 551 patients (196 with symptoms
and 355 without) to compare the FlexSure office kit to QuickVue, and ELISA against a
positive 13C-UBT as diagnosis of H. pylori infection. FlexSure
office kit was comparable to the ELISA (Sens. 96%/Spec. 95%) and was easy and quick (5
minutes.) to use. (9)
Fallone et al conducted a prospective study on 106 patients referred to tertiary
center for endoscopy, to compare Helisal Kit, which uses an ELISA technique to determine
IgG antibodies in saliva. Saliva and serum were compared to gastric histology. They found
that salivary IgG titers correlated significantly with serology but had a very low
sensitivity (66%) and specificity (74%) when compared to histology. (7)
More studies are needed prior to recommending its use.
There is no comparison of all currently available tests. Several reviews have
addressed the comparative accuracy of common tests to detect H. pylori infection but some
either lacked noninvasive testing, used individuals with and without prior treatment, used
a single parameter to define infection status (not the gold standard), or looked at fewer
than one hundred patients. (4)
The largest study and the best available evidence was published by Cutler et al.
(4) This study attempted to go beyond the limitations of
previous studies.
- The study was designed as a blinded comparison of 6
independent tests (7 sets of readings) for detecting H. pylori. The objective was to
compare the sensitivity, specificity, and negative and positive predictive value of the
most widely available tests for the diagnosis of H. pylori. Culture and whole blood was
not included.
- 268 patients who presented for endoscopy to a large
gastrointestinal clinic in Detroit Michigan between 1989-1992 were identified. None of
the patients were previously treated for H. pylori.
- Each patient underwent multiple tests for H. pylori:
- Each had an endoscopy and 3 antral biopsy specimens were obtained:
- One biopsy was used for the Clo test and the other two for histology
for acute and chronic inflammation, and Warthin-Starry silver stain
- A single experienced pathologist who was blinded to the results of the
other tests used to assess H. pylori status evaluated all histology specimens.
- Each patient had 3 non-invasive tests:
- a 13 C-UBT and serological assays for IgG and IgA antibodies
to H. pylori were performed by ELISA
- Serological assays were performed blindly with respect to
patient infection status
- To overcome the lack of a true diagnostic standard, Cutler et al. used
concordance of > 4 of the 7 test results to represent infection with H. pylori
- 173 of 268 patients or 65% had > 4 positive parameters
data for
graphs compiled from (4)
| Comparison of Tests: Table
1 |
PARAMETERS |
Sensitivity,
Specificity, Positive and Negative Predictive Values, *Likelihood Ratios of Seven
Diagnostic Assays for H. pylori infection among 268 patients from Detroit, Michigan . |
Sens.(%) |
Spec
(%) |
PPV |
NPV |
LR(+) |
LR(-) |
| Invas ive |
Chronic Antral Inflammation |
100 |
66.3 |
84.4 |
100 |
3.0 |
0 |
| Acute Inflammation |
86.7 |
93.7 |
96.2 |
79.5 |
13.8 |
0.1 |
| Warthin-Starry
Stain |
93.1 |
99.0 |
99.4 |
88.7 |
93 |
0.7 |
| Clo Test |
89.6 |
100 |
100 |
84.1 |
|
0.01 |
| Noninvasive |
Urea Breath Test |
90.2 |
95.8 |
97.5 |
84.3 |
21.5 |
0.01 |
| Serum IgG |
91.3 |
91.6 |
95.2 |
85.3 |
10.9 |
0.09 |
| Serum IgA |
71.1 |
85.3 |
89.8 |
61.8 |
4.8 |
0.03 |
(1,4,18)
The patients in this study had at least a 65% prevalence
or pre-test probability of H. pylori infection. When comparing tests, sensitivity and
specificity are not enough. As clinicians, when we order tests, we have to ask: Of
everyone with a positive test, what percent will have disease? (PPV) or Of everyone with a
negative test what percent will be disease free? (NPV). Unfortunately, these numbers will
vary with disease prevalence. For example all we can say with regards to the PPV of 95.2
for the Serum IgG test is that in a population with a prevalence of 65%, 95.2% of people
with a + Serum IgG test will have H. pylori infection. However this doesn’t carry in
a population with a lower prevalence. One would expect more false + and a lower PPV. (18)
Fortunately, we have likelihood ratios. When comparing tests to decide which is best to
use likelihood ratios are helpful. A positive likelihood ratio is simply a ratio of the
probability of a positive test if disease is present/ probability of a positive test if
disease is absent. (18) " The best test to use for ruling in disease is
the one with the largest likelihood ratio of a positive test and the better test to rule
out disease is the one with the smallest likelihood ratio of a negative test. " (18)
Examination of the likelihood ratios above reveal that Warthin-Starry silver stain, Clo
test, Urea breath test and Serum IgG all have large likelihood ratios. A positive test on
any rules in disease and eliminates the need for further testing. The silver stain and Clo
test are as accurate as the Urea Breath test and Serum IgG in detecting Helicobacter
Pylori in previously untreated patients.
Of additional importance is the fact that the likelihood ratios above give on
information on likelihood of presence or absence of bacteria. They do not give information
on whether H. pylori is really responsible for the patient’s symptoms.
In order to use the likelihood ratio, a primary care
physician would need to establish a pretest probability of infection with H. pylori for
each patient who presents with dyspepsia. Prevalence rates can help physicians set pretest
probabilities. History and physical in addition to risk factors are sometimes helpful and
it appears as if race and age are important risk factors for infection with H. pylori. (13)
Malaty et al performed a case-controlled seroepidemiologic study of H. pylori
seroprevalence in 108 healthy Hispanic volunteer. 89 Hispanics were matched (1:1:1) with
African-americans and whites for age and socioeconomic status.
Does this article represent a believable estimate of the true value of each
test?
- Are the results valid?
- Was there an independent blinded comparison with a
reference standard?
- Concerns as to whether the single pathologist used was blinded, because although he was
blinded to the status of other results on each patient, it would be hard for him to be
blinded to the various histological findings on each slide.
- Did the patient sample include an appropriate spectrum of patients to whom the test will
be applied?
- Yes and No. Yes in that these patients had not been treated for H. pylori previously .
In addition there was a large African-American population, but only one Hispanic patient.
Trends in race, age, and GI conditions consistent with other literature.
- No in that there was a referral bias. Concerns as to whether the prevalence disease
higher than in my patient population
- Did the results of the test being evaluated influence the decision to perform reference
standard?
- Were the test methods described clear enough to permit replication?
- What are the results?
- Is data necessary for calculation of likelihood ratios
provided?
- Will the results help me in my patient care:
- Will the tests be reproducible and well interpreted in my
practice setting?
- Yes, but experienced pathologist needed. Also, local validation of serology kits and
office kits needed.
- Are the results applicable to my patients?
- Yes if practice is a large urban population and patients meet exclusion criteria of no
previous treatment for H. pylori.
- Will the results change my management?
- Yes, given large likelihood ratios (>10), results will move probability of disease
across treatment threshold.
- Will my patients be better off because of the test?
- Yes, tests do offer information that will be beneficial to patient care. Why? Target
disorder if left undiagnosed may be harmful (missing PUD, cancer, chronic gastritis). The
tests: endoscopy (Clo, stain), serology and UBT have acceptable risks and although impact
on patient outcome in certain situations (ie.non-ulcer dyspepsia is not certain, effective
treatment for eradication exists.
The best available evidence may not be perfect, however a
critical appraisal allows us to be more informed about the data and how our patient
population meshes with that of the study.
US Northwest and National Reference Labs
| Tests |
Costs ($)* |
| Endoscopic |
|
| - histology |
50-100 |
| - Special
Stains |
200 |
| - Culture |
150 |
| - Urease
tests |
10 |
| Noninvasive |
|
| Antibody
tests |
|
| - Quantitative ELISA |
50-100 |
| -
Qualitative serum |
10-15 |
| - Qualitative whole blood |
10-15 |
| Urea breath
tests |
|
| - 13-C |
280 |
| - 14-C |
60 |
(8)
*costs for histology, special stains and cultures are
based on wholesale charges to clinical laboratories in the US
Northwest. For Noninvasive test and urease tests costs are based on charges from
national reference laboratories. (8)
Wake Forest University: Cost
| Test |
Cost ($)* |
| EGD w/
biopsy |
|
| - Professional fee |
737 |
| - Facility |
177 |
| Pathology |
|
| - H&E |
130 |
| - Silver |
192 |
| - Clo |
13 |
| Noninvasive |
|
| - UBT |
200 ^ |
| - qualitative serum |
42 |
*costs obtained from clinipac and preceptor
^no longer available
Very important factor in the equation. Ofman et al. in a
decision analysis compared empiric eradication with treatment depending on the results of
endoscopy in otherwise uninvestigated dyspepsia. Empiric therapy was less expensive with a
52% reduction in endoscopy but with a 252% increase in the use of antibiotics.
The 27-year-old Hispanic man who presented with
dyspepsia.
- His pretest probability based on age and ethnicity is about 38% (13).
- He is <45 years old and has no alarm symptoms, so therefore a noninvasive test is
indicated.
- He has never been treated for H. pylori before.
- Using the nonogram for interpreting diagnostic tests (11), if a qualitative
IgG serology ($10-40) is sent and is positive, our posttest probability is 85-90% he is
infected.
- If the test is negative however, with a LR(-) of 0.09, our posttest probability falls to
3%.
- If a similar patient presented to Reynolds today, I would send a lab serology and if
positive treat.
The 50 year old African-american woman with dyspepsia
- She has never been treated for H. pylori.
- Somewhat concerning is her history of early satiety and family history of GI
"tumor".
- Given these finding, one could argue to refer directly to endoscopy.
- However if a noninvasive strategy was applied, this woman would have a pretest
probability based on age and ethnicity of 90%.
- One could argue empiric therapy, because based on the LR(-) of 0.09, she would have a
posttest probability of 45% if her test was negative
Note that in both these patients these pre and posttest probabilities refer to
probability of H. pylori infection and not to whether the infection may indeed be causing
symptoms.
| Clinical Condition |
Recommended Test |
| New onset
dyspepsia |
|
| Age <45 |
Qualitative serology IgG (lab) |
| |
Office based
qualitative serology (avoid fingerstick samples) |
| Age >45 |
Endoscopy |
| |
Rapid urease
test +/- Qualitative serology IgG |
| Any age/Alarm symptoms** |
Endoscopy |
| History of PUD |
Qualitative
serology IgG (lab) |
| |
Office based qualitative
serology (avoid fingerstick samples) |
| On PPI |
Qualitative
serology IgG |
| |
Office based qualitative
serology |
| |
If endoscopy
indicated hold for 5 days, then Clo test |
| Post treatment |
|
| Dyspepsia |
Test if
recurrent symptoms |
| Recurrent symptoms |
UBT or endoscopy (as above) |
| Uncomplicated
duodenal ulcer |
Test if
recurrent symptoms |
| Complicated ulcer |
Confirm eradication |
| |
UBT or
*quantitative ELISA |
*quantitative Elisa requires that the sample used for
diagnosis of infection be stored and compared to the sample obtained after Rx
**Alarm symptoms include: GI bleed, weight loss, dysphagia, and anemia (3,4,7)
- What is the diagnostic performance of H pylori tests?
- Best available evidence is not perfect.
- The Clo test, Silver stain, Serology and UBT all have large positive likelihood ratios :
A positive test will move probability of disease across a threshold
- Office kits (qualitative IgG) are available and accuracy comparable to lab serology.
Never use fingerstick samples with the office kits.
- Insufficient data on salivary and urine tests.
- How should the Primary care physician approach making the diagnosis
?
- Utilize pretest probability and likelihood ratios (adds order to the chaos).
- Remember African Americans and Hispanics have a very high prevalence of H. pylori as
compared to Whites and it increases with age.
- Consider recommended guidelines, accuracy, precision and cost of test
- Realize that although H. pylori may be present, patient outcome even with diagnosis and
treatment in certain situations in not certain. Share this realization with your patient.
Thanks Dr. Joel Brugen, Dr. Ray Morrow,
Beth Sheppard, and Boo
…thinking of you mom
References
- ACP Journal Club: Non invasive tests were as accurate as
invasive tests for detecting Helicobacter pylori. 1996; Jan/Feb.
- American Gastroenterological Association. Medical Position Statement: Evaluation of
Dyspepsia. Gastroenterology. 1998; 114:579-581.
- Cutler, Alan. Testing for Helicobacter pylori in Clinical Practice. The American Journal
of Medicine. 1996; 100(5A): 31-41.
- Cutler et al. Accuracy of invasive and noninvasive tests to diagnose Helicobacter pylori
infection. Gastroenterology. 1995; 109:136-141.
- EHPSG. Current European concepts in the management of Helicobacter pylori infection. The
Maastricht Consensus Report. Gut. 1997; 41:8-13.
- Fallone et al. Determination of less costly methods of Helicobacter pylori detection.
Clin. Invest. Med. 1995;18(3): 177-185.
- Fallone et al. Detection of Helicobacter pylori Infection by Saliva IgG Testing.
American Journal of Gastroenterology. 1996; 91(6): 1145-1149.
- Fennerty M. Brian. A Review of Tests for the Diagnosis of Helicobacter Pylori Infection.
Laboratory Medicine. 1998; 29(9):561-566.
- Graham, David et al. Comparison of Rapid Serological Tests (Flexsure HP and QuickVue)
with Conventional Elisa for Detection of Helicobacter Infection. American Journal of
Gastroenterology. 1996; 91(5): 942-947.
- Jaeschke et al. For the Evidence-Based Medicine Working Group. Users’ Guides to the
Medical Literature. How to use an Article about a Diagnostic Test. A: Are the results of
the Study Valid? JAMA. 1994; 271(5): 389-391.
- Jaeschke et al. For the Evidence-Based Medicine Working Group. Users’ Guides to the
Medical Literature. How to use an article about a diagnostic test. B: What are the results
and will they help me in caring for my patients? JAMA. 1994; 271(9): 703-707.
- Kidd, M et al. A century of H. pylori: Paradigms lost-paradigms regained.
Digestion.1998; 59:1-15.
- Malaty, Hoda et al. Helicobacter pylori in Hispanics: Comparison with Blacks and Whites
of Similar Age and Socioeconomic Class. Gastroenterology. 1992;103(3): 813-816.
- Megraud, F. Advantages and disadvantages of current diagnostic tests for detection of H.
pylori. Scan. J Gastroenterology. 1996; 31 Suppl. 215: 57-62.
- Megraud, F. The Report of the Digestive Health Initiative International Update
Conference on Helicobacter pylori. Diagnosis and Candidates for treatment of Helicobacter
pylori infection. How should Helicobacter pylori infection be diagnosed? Gastroenterology.
1997; 113: S93-98.
- NIH Consensus Conference. Helicobacter pylori in Peptic Ulcer Disease. JAMA
1994;272:65-69.
- Ofman et al. Management Strategies for Helicobacter pylori Seropositive Patients with
Dyspepsia: Clinical and Economic Consequences. Ann Intern Med. 1997;126:280-291.
- Riegelman, Richard K., Hirsch, Robert P. Studying A Study and Testing a Test. How to
read the Health Science Literature. Little Brown. 1996