|
 |
|
| |
|
|
|
| Validated accuracy of a novel urea
breath test for rapid Helicobacter pylori detection and in-office
analysis. |
| Olga Hegedusa, Johan Rydenb, Ann-Sofie
Rehnberga, Stefan Nilssonb, and Per M. Hellströma. |
| |
| Background |
| A novel 14C urea breath
test (UBT) was developed to detect the presence of Helicobacter
pylori by bench analysis in office, enabling the practitioner
to readily reveal H. pylori infection. |
| |
|
| |
| Aim |
| To validate the novel UBT (Helicobacter™)
versus conventional UBT. |
| |
|
| |
| Methods |
| Pretreatment (n=203) and post-treatment
(n=147) detection of H. pylori. Additional tests with encapsulated
14C-urea (n=37) were validated. After Intake of liquid or encapsulated
14C-urea, exhaled 14CO2 in breath was trapped in benzethoniumhydroxide/ethanol,
or absorbed to LiOH-soaked pads on a dry cover surface (Heliprobe
BreathCard™). The amount of absorbed 14C was detected
using a ß-scintillator or two Geiger-M?ller counters operating
in parallel (Heliprobe™ Analyzer). |
| |
|
| |
| Results |
| For pretreatment detection, we found
full concordance between the UBTs, with 100% sensitivity and
specificity (CI 95-100% and 97-100%, respectively) and strong
agreement (r=0.80, CI 0.75-0.85;?=1, CI 0.86-1.14; P<0.0001).
Similarly, for post-treatment follow-up detection, sensitivity
and specificity were 100% (CI 85-100% and 97-100%, respectively)
with significant agreement (r=0.48, CI 0.34-0.59; ?=1, CI 0.84-1.16;
P<0.0001). The use of encapsulated 14C-urea did not change
agreement between the tests. Sensitivity and specificity were
100% (CI 72-100% and 87-100%, respectively) with strong agreement
between the tests (r=0.71, CI 0.50-0.84; ?=1, CI 0.68-1.32;
P<0.0001). |
| |
|
| |
| Conclusion |
The novel Heliprobe UBT, with either
liquid or encapsulated 14C-urea, seems equi-efficacious
to conventional UBT in fulfilling its role as the non-invasive
gold standard for detection of H. pylori. Eur J Gastroenterol
Hepatol 14:1-8 2002 Lipponcott Williams & Wilkins.
European Journal of Gastroenterology & Hepatology 2002,
14:1-8 Keywords: breath test, duodenal ulcer, dyspepsia, gastritis,
Gram-negative bacteria, Helicobacter pylori, peptic ulcer disease,
stomach ulcer, urea, urease.
aDepartment of Medicine,
Unit of Gastroenterology and Hepatology, Karolinska Hospital,
Karolinsk Institute and bNoster System AB, Stockholm, Sweden.
Correspondence to Per M. Hellström, MD, PhD, Department
of Gastroenterology and Hepatology, Karolinska Hospital, SE-171
76 Stockholm, Sweden
Tel:+46 851773877; fax:+46 51772058;
Email:Per. Hellstrom@medks.ki.se
This investigation was supported by the Karolinska Institute
and Noster System AB.
Received 24 July 2001 Accepted 16 January 2002 |
| |
|
top |
| Introduction |
The urea breath test (UBT) was first
described by Graham et al. [1] in 1987. The idea of developing
a 14C-UBT to detect the presence or absence of Helicobacter
pylori was a result of two publications appearing in letter
form in the Lancet. The first, from Tytgat and co-workers [2],
discussed the finding that H. pylori possessed a urease enzyme
that was extremely powerful. The second, from Marshall and Langton
[3], described how patients infected with H. pylori tended to
have lower urea and higher ammonia concentrations in gastric
juice than did non-infected people. The later development of
the UBT in our other’s laboratories has proven the UBT
to be a highly standardized, sensitive and specific test [4].
Today, the presence of H. pylori in the gastrointestinal tract
is most conveniently detected non-invasively using either the
13C- or 14C-UBT [1, 5, 6]. The UBT detects
the metabolically active bacteria, which makes the test suitable
for pretreatment detection as well as for monitoring results
after eradication treatment against H. pylori. The validity
of the UBT is generally high, with a reported sensitivity of
90-98% and specificity of 92-100% [5, 7-10]. With the aid of
an acidified urea cocktail, the diagnostic precision of the
method is enhanced [4]. Even if the UBT has become the diagnostic
procedure of choice for detection of H. pylori infection, the
detection procedure using a liquid CO2-tapping medium
and ß–scintillation has hampered the further development
of UBT to a handy diagnostic tool. Naturally, the 13C-UBT is
always an alternative method with high diagnostic precision
and accuracy [1, 4, 6-8], but it seems to complicate the analysis
by requiring a centralized, expensive mass spectrometer with
a continuous need for maintenance and calibration.
In the search for a UBT for detection of H. pylori fulfilling
common demands of easiness, handiness and quick in-office analysis,
we have developed a simplified 14C-UBT system (Heliprobe™,
Noster System AB, Stockholm, Sweden). The objective of the present
was to evaluate whether this novel UBT, based on a new collection-and-analysis
method, has a comparable diagnostic performance and accuracy
with the conventional 14C-UBT method as the 14C-UBT
previously developed and validated in our laboratory [4]. Furthermore,
we aimed to determine the optimal diagnostic cut-off level for
the novel UBT in two different situations: (1) for pretreatment
of H. pylori in antibiotic-naïve patients, and (2) for
follow-up detection after eradication treatment. We designed
and performed a clinical study in which each patient swallowed
an acidified 14C-urea solution and breath samples
were collected via both the conventional UBT and the Heliprobe
system. In addition, a preliminary evaluation was made replacing
the urea solution with an encapsulated form of 14C-urea
(PYCap, Tri-Med Specialties Inc., Charlottesville, Virginia,
USA), which should further simplify the novel UBT method. |
| |
|
|
top |
| Materials and methods |
| The novel 14C-urea
breath test system |
The new Heliprobe UBT is a completely
‘dry’ system consisting of two components, the Heliprobe
Breath-Card™ and the Heliprobe Analyzer™ (Fig.1).
The Heliprobe BreathCard is a flat, credit-card-sized collection
vehicle that absorbs exhaled CO2 via chemical bounding
to pads soaked in LiOH. The collection process is simple; the
patient breathes into a mouth-piece on the card until a pH-sensitive
indicator changes colour from orange to yellow as an indication
of CO2 saturation of the pads. The breathing time
varies depending on the number of breaths into the card, the
average time being approximately 1-2 min. Since the exhaled
CO2 is bound chemically to the pads, the card can
be stored for several years without loss or deterioration of
its CO2 content.
With the Heliprobe Analyzer, the traditionally used liquid ß–scintillator
has been replaced with an instrument containing two built-in
Geiger-M?ller counters operating in parallel. This technology
swap has made it possible to design a cheap, small (laptop-sized),
and fully automatic analyzer that can be operated by the nurse
or physician in the clinic (Fig.1).
The Heliprobe BreathCard is simply put into the slot of the
Heliprobe Analyzer. By pressing the start button, a fully automatic
test sequence is initiated and run for 250 s. The result of
the measurement is presented on a liquid-crystal display (LCD)
and on a printer. The analysis is based on the number of emitted
ß–particles that hit the two Geiger-M?ller counters
during the 250-s measurement cycle and is presented as counts
per min (cpm) together with the test result ‘negative’,
‘equivocal’ or ‘positive’.
The cut-off levels between the different test results are based
on the obtained cpm values. The diagnostic cut-off is programmable
to different levels by setting lower and upper limits. A cpm
value below the lower limit is presented as a negative result,
values between the lower and upper limits are presented as equivocal,
and values above the upper limit are positive. By setting the
lower and upper limits to the same value, equivocal results
can be avoided. The Heliprobe Analyzer is continuously compensating
for background radioactive variations, thereby eliminating this
source of error. |
| |
|
|
top |
| The conventional 14C-urea breath
test system |
The conventional UBT is based on trapping
exhaled CO2 in a Hyamine solution (1 ml1.0-mol/l
benzetho-niumhydroxide in methanol (Hyamine®, Sigma Chem.
Co., St Paul, Minnesota, USA) and 1 ml 99.8% ethanol) kept in
a 20 ml scintillation vial. The patient exhales into the Scintillation
vial through a straw, which is connected to a water-lock to
eliminate the possibility of swallowing the solution. Phenolphthalein
(Sigma) was used as a colour indicator for saturation of the
benzethoniumhydroxide solution with 1 mmol CO2.
After saturation indicated by colour change from pink to colourless,
10ml, scintillation liquid Optiphase ‘Hi Safe’ (Wallac,
Fison Chem., Loghborough, Leicestershire, UK) was added, and
the sample was analysed in a liquid ß-scintillation counter
(Beta Rack 1215, Wallac). As a blank, we used 1 ml benzethoniumhydroxide,
1 ml ethanol and 10 ml scintillation liquid. As standard, 0.5
ml 14C-urea cocktail was added to be prepared scintillation
vial with benzethoniumhydroxide solution, ethanol and scintillation
liquid. Quench correction was applied by the external standard
ratio method to yield sample activities in disintegrations per
minute (dpm). The results were presented as CO2 recovery (%
dose recovered/mmol CO2, trapped multiplied by the
weight of patient) |
| |
|
|
top |
| Study design |
The study was carried out on consecutive
patients scheduled for diagnostic of H. pylori infection or
post-treatment follow-up at the Department of Gastroenterology
and Hepatology, Karolinska University Hospital, Stockholm, Sweden.
No attempts were made to select patients based on their diagnosis.
Patients could not take proton-pump inhibitors in the week before
the test was undertaken. Antacids, H2-receptor antagonists
and sucralfate were stopped 24 h before the test day, and antibiotics
and bismuth medications were stopped during the month before
the study. Patients were prompted to observe a 6-h fasting period.
To minimize exposure to oral microflora, patients were instructed
to brush their teeth well in the morning before the UBT, and
to swallow the acidified 14C-urea solution quickly.
The inclusion criteria for subjects in the study were age 18-85
years, upper-gastrointestinal discomfort or symptoms, and suspicion
of H. pylori infection. Exclusion criteria were pregnancy or
breastfeeding, previous gastric surgery, drug or alcohol addiction,
senile dementia, and a previous diagnostic UBT within 30 days.
The study design was approved by the local Ethics and Radiation
Safety Committees of the Karolinska Hospital. Each patient received
information about the investigative nature of the study, and
informed consent was obtained from each subject.
Ten minutes after ingestion of the solution, breath samples
were collected via both the conventional UBT and the Heliprobe
system. At 20 min, a second breath sample was obtained for the
conventional UBT. Since our aim was to develop a simplified
test procedure, we did not repeat this second test with the
Heliprobe system. |
| |
|
| |
| Preparation of the acidified
urea solution |
Liquid 14C-urea was obtained
commercially from the Swedish National Pharmacy (Apoteksbolaget
AB, Stockholm, Sweden) in 25- µCi vials at a concentration
of 5 µCi/ml and stored in refrigerator. Immediately before
the test, aliquots of 1 µCi(0.2 ml) were pipetted into
a plastic cup and 50 mL 0.05-mol/l citric acid water solution
was added.
In a separate study group, a 1-µCi encapsulated form of
14C-urea (PYCap, Tri-Med Specialties Inc.) was used instead
of the regular urea solution. |
| |
|
| |
| Determination of Helicobacter
pylori-positive and –negative cases |
| The conventional UBT was used to determine
H. pylori-infected patients. Optimal cut-off criteria were determined
in a previous validation study of the conventional method [4].
A 10-min test value above 0.80% mmol CO2-1 kg, or
a 20-min test value above 0.50% mmol CO2-1 kg, was
classified as positive. Patients with values equal to or below
the above-mentioned values were classified as negative. |
| |
|
|
top |
| Analysis and Statistics |
Examination for pretreatment detection
and post-treatment detection after eradication treatment were
analyzed separately. All evaluations were ‘done’
per protocol, but the effects of protocol violations were also
explored.
The results of the Heliprobe tests were divided into two classes,
H. pylori-positive and H. pylori-negative, based on the outcome
of the conventional UBT used for reference.
Normality was assessed by the Wilk-Shapiro W test. Descriptive
data are presented as mean ±Sd for normally distributed
data, and as the median and range for non-normally distributed
data. The Mann-Whitney U test was used for comparisons between
groups. Sensitivity and specificity together with 95% exact
(Clopper-Pearson) confidence limits for the proportion were
determined for all possible cut-off points with AccuROC ver.
2.3 (Accumetric Corporation, Montreal, Canada). Association
between the Heliprobe system and the conventional UBT was assessed
by correlation analysis (Spearman rank method) and explored
further by inter-rater agreement and Cohen’s unweighted
? statistics. A P-value of less than 0.05 was considered significant.
All statistics, except the sensitivity and specificity analyses,
were carried out using StatsDirect (CamCode, Ashwell, Hertfordshire,
UK). |
| |
|
| |
| Results |
| Pretreatment detection of Helicobactor
pylori |
The Characteristics of the 192 evaluated
patients in the pretreatment population are summarized in Table
1. Eleven additional patients did not meet the enrolment criteria
(8 taking proton-pump inhibitors in the week before UBT; 1 on
proton-pump inhibitor and antibiotics; 2 did not fast for 6
h) and were excluded from the main analysis.
Figure 2 shows the spread of the Heliprobe cpm values in the
H. pylori-positive and –negative groups. The H. pylori-negative
group consisted of 119 patients, and the H. pylori-positive
group consisted of 73 patients, revealing an H. pylori infection
prevalence of 38%. The minimum, median and maximum Heliprobe
values were 41, 400 and 1291 cpm, respectively, in the H. pylori-positive
group, and 0 1 and 25 cpm, respectively, in the H. pylori-negative
group. The Shapiro-Wilk W test indicated non-normality, and
the Mann-Whitney U statistical test revealed a highly significant
difference (median difference 391 cpm, CI 323-429 cpm; P<0.0001). |
| |
| Table1 Summary
of the study population for pretreatment detection of Helicobacter
pylori. |
| |
| |
Patients for
analysis |
| Patient characteristic |
H. pylori-positive |
H. pylori-negative |
All |
| Number of subjects (N) |
73 |
119 |
192 |
| Male/Female (N/N) |
43/27 |
46/73 |
93/99 |
| Age (years; mean ±
SD) |
52 ± 17 |
46 ± 16 |
48 ±16 |
| Weight (kg; mean ±
SD) |
72 ± 14 |
70 ± 14 |
71 ± 14 |
|
| |
| Sensitivity and specificity of the
Heliprobe system |
| Choosing a Heliprobe cut-off level in
the gap between the highest H. pylori-negative (25cpm) and the
lowest H. pylori-positive (41 cpm) gave 73 true positive, 0
false positive, 119 true negative, and 0 false negative measurements.
This created 100% sensitivity and specificity (CI 95-100% and
97-100%, respectively) of the Heliprobe system versus the conventional
UBT. |
Association between the Heliprobe system and the conventional
urea breath test |
| As shown in Figure 3, there was a significant
correlation between the Heliprobe cpm values and the conventional
UBT dpm values at 10 min, with a Spearman rank correlation coefficient
of 0.80 (CI 0.75-0.85; P<0.0001). An alternative association
analysis, the ? statistics, also revealed a significant concordance
between the two methods (? = 1, CI 0.86-1.14; P<0.0001). |
| |
 |
| Spread of Heliprobe test results
for the Helicobacter pylori-positive group (Hp-pos) and the
H. pylori-negative group (Hp-neg), as categorized by the conventional
14C urea breath test (UBT) method. The individual
test results in counts per minute (cpm) are shown together with
each group's media value. (a) Spread for the pretreatment population;
(b) spread for the post-treatment population. Not all points
are visualized adequately in the graphs due to superposition |
| |
|
|
|
top |
|
