Feldman M, Cryer B, Lee E, Peterson WL. Role of Seroconversion in Confirming Cure of Helicobacter pylori Infection. JAMA. 1998;280(4):363–365. doi:10.1001/jama.280.4.363
From the Departments of Internal Medicine (Drs Feldman, Cryer, and Peterson) and Pathology (Dr Lee), University of Texas Health Science Center at Dallas, and the Medical (Drs Feldman, Cryer, and Peterson) and Laboratory (Dr Lee) Services of the Department of Veterans Affairs Medical Center, Dallas, Tex.
Context.— The role of serologic testing to confirm cure of Helicobacter pylori infection after antimicrobial therapy is not completely
Objective.— To determine the utility of serologic testing in confirming cure of H pylori infection more than 1 year after therapy.
Design.— A prospective, before-after interventional trial.
Setting.— An outpatient clinical research laboratory in an academic, urban Veterans
Affairs medical center.
Participants.— Twenty-three otherwise healthy men and women with active H pylori infection demonstrated by gastric biopsy and with positive H pylori serologic findings.
Intervention.— A 14-day course of bismuth, tetracycline, and metronidazole.
Main Outcome Measures.— Determination of IgG serum antibodies to H pylori at baseline, 1 month, 3 months, and approximately 18 months after
completion of therapy compared with serial gastric mucosal biopsy specimens
with stains for H pylori and for histologic examination
as the criterion standard.
Results.— Fifteen (65%) of 23 subjects were cured of their H pylori infection as assessed by gastric biopsy, with elimination
of gastritis; median antibody levels declined from 92.5 U/mL at baseline to
undetectable levels at 18 months. The other 8 subjects (35%) were not cured
and had persistent gastritis at 18 months; median antibody levels declined
from 130.6 U/mL at baseline to 89.7 U/mL at 18 months. Sensitivity and specificity
of seroconversion (from a positive to negative test result) in detecting cure
of H pylori infection were 60% and 100%, respectively.
Conclusion.— Undetectable antibody levels beyond the first year of therapy accurately
confirm cure of H pylori infection in initially seropositive
healthy subjects, with reasonable sensitivity.
IN 1994, a National Institutes of Health consensus conference recommended
that patients with peptic ulcers who are infected with Helicobacter pylori should be treated with antimicrobial drugs.1,2 Moreover, in the past few years more
and more individuals infected with H pylori without
ulcers are being treated.3- 5
If there is a desire to confirm cure of H pylori
infection early after therapy, urea breath testing or a gastric biopsy, obtained
either endoscopically or nonendoscopically,6
are required, since H pylori serum IgG antibody tests
remain positive for many months after cure.7
However, urea breath tests and gastric biopsy are time-consuming and expensive,
and biopsy is invasive. If one waits 6 months after therapy and performs a
quantitative serologic test on paired serum samples, a 20% or more reduction
in serum IgG may be informative of cure.7 While
the sensitivity and specificity of this paired-quantitative serologic approach
for confirming cure have been reported to be between 80% and 90%,7 serum storage for prolonged periods is not practical.
Furthermore, many laboratories do not report quantitative serum antibody levels,
but instead report results qualitatively as "positive" or "negative."
Many physicians are interested in confirming cure (or persistence) of H pylori infection only on recurrence of symptoms, which
may occur a year or more after antimicrobial therapy. Perhaps by then, enough
time has elapsed for antibody titers to become undetectable in cured individuals,
rendering seroconversion (defined in this context
as a change from a positive to a negative serologic test result) clinically
helpful. In one study, however, less than 50% of subjects who had been cured
(as assessed by urea breath tests) seroconverted during the second year of
therapy.8 Nevertheless, only 15 patients were
followed beyond 15 months in that study.8 To
further evaluate the role of late serologic testing in confirming cure, we
measured H pylori serum antibodies and obtained gastric
mucosal biopsy specimens in 23 men and women before and for approximately
18 months after completion of therapy with bismuth, tetracycline, and metronidazole.
Using tissue stains of H pylori as the criterion
standard, we calculated the sensitivity and specificity of H pylori seroconversion in confirming cure of H pylori infection.
Twenty-three adults (10 men and 13 women) with no history of peptic
ulcer and no chronic upper gastrointestinal tract symptoms were enrolled in
this prospective study during 1993 and 1994. They ranged in age from 25 to
85 years (median, 48 years; mean, 49 years). Thirteen were African American,
8 were white, 1 was Hispanic, and 1 was American Indian. Each had a positive
baseline H pylori serum antibody test result (>20
optical density units per mL [U/mL]), using a fluorescence immunoassay to
detect IgG antibodies to H pylori in serum (FIAX
test kit, Bio Whittaker, Walkerville, Md; sensitivity and specificity, 99%
and 97%, respectively6). Each subject also
had a baseline gastric mucosal biopsy demonstrating H pylori organisms.6,9 Studies
were approved by an institutional review board. Informed written consent was
obtained from each subject.
After baseline studies confirmed H pylori seropositivity
and active infection on gastric biopsy, each subject was treated with a 2-week
course of bismuth subsalicylate (524 mg 4 times daily), tetracycline (500
mg 4 times daily), and metronidazole (250 mg 4 times daily). One subject intolerant
to tetracycline received amoxicillin (500 mg 4 times daily) instead. No antisecretory
therapy was prescribed. Each subject was interviewed by telephone halfway
through this 2-week course to encourage compliance and to record adverse effects.
They were also interviewed at the end of the 2-week course of therapy.
One month, 3 months, and approximately 18 months (15-23 months) after
completion of therapy, subjects returned to the laboratory for remeasurement
of H pylori serum antibodies, for mucosal biopsies
of the gastric body, and for mucosal biopsies of the gastric antrum (month
3 and month 18 only). Gastric mucosal biopsy specimens were obtained through
a modified nasogastric tube under fluoroscopic guidance, a technique that
does not require intravenous sedation or endoscopy.6,9
Biopsy specimens were interpreted by one of us (E.L.) who was provided no
clinical information. The presence or absence of stainable H pylori organisms using a hematoxylin and eosin stain was reported.9 If there was a question as to the presence of H pylori, a Giemsa stain was also performed.10
The presence or absence of chronic gastritis, using previously described criteria,9 was also assessed and gastritis was graded in intensity
as absent (score=0), mild (score=1), moderate (score=2), or severe (score=3).
Serum samples were processed in "real time" (ie, they were run in the laboratory
immediately after collection, rather than stored). Antibody levels were reported
as positive (>20 U/mL), borderline (15-20 U/mL), or negative (undetectable,
ie, <15 U/mL). The actual antibody concentration (U/mL) was also recorded
for positive and borderline test results. Subjects whose serologic results
changed from positive to negative were considered to have seroconverted.
Cure of H pylori infection was diagnosed when
mucosal biopsy specimens from both the gastric body and the antrum no longer
showed H pylori organisms 18 months after therapy.
Persistent infection was diagnosed when H pylori
organisms were still visible on gastric biopsy specimens 18 months after therapy.
Data were analyzed with Systat 6.0.1 (SPSS Inc, Chicago, Ill, 1996).
We used the Wilcoxon signed rank test to compare mean gastritis scores before
vs after therapy, and paired t tests to compare mean
antibody concentrations; because H pylori antibody
levels were not normally distributed, their square root was used. Two-sided P values less than .05 were considered significant.
Helicobacter pylori infection was cured in
15 subjects (65%). The other 8 subjects (35%) still had H pylori infection approximately 18 months after antimicrobial therapy.
None of these 8 subjects had had elimination of H pylori infection on earlier biopsy specimens and then became reinfected.
Thus, H pylori infection in these 8 subjects was
persistent, not recurrent.
Table 1 presents mean (±SEM)
gastritis severity scores at baseline (body and, in some cases, antrum), month
1 (body), month 3 (body and antrum), and month 18 (body and antrum) in the
2 groups of subjects (cured vs not cured). Cure of H pylori infection was associated with significant declines in gastritis severity
scores, with no gastritis in the gastric body or antrum in any of the 15 subjects
at 18 months (P=.002 and .003 vs baseline, respectively).
In the 8 subjects whose H pylori infection was not
cured, there was nevertheless a 50% decrease in the mean gastritis severity
score in the gastric body at month 1 (P=.06 vs baseline).
However, by month 3, and at month 18, severity of gastritis in the gastric
body in these 8 subjects with persistent infection increased to baseline scores
(P=.06, month 3 vs month 1; P=.02,
month 18 vs month 1). Antral gastritis severity was similar at baseline and
at month 18 in these 8 subjects who were not cured of their infection.
Table 2 displays quantitative H pylori serum antibody levels in each of the 23 subjects,
separated into those cured of their infection and those not cured, as well
as the median antibody levels for the 2 groups. In the cured group, there
was no change in serum antibody level at month 1, after which levels declined
approximately 50% at month 3 (P=.03, month 3 vs month
1). Serum antibody levels at month 18 were significantly lower than at baseline,
month 1, and month 3 (P<.005). Nine of 15 subjects
whose H pylori infection had been cured had undetectable H pylori serum antibody levels at 18 months (ie, 60% had
seroconverted). In the 6 remaining cured subjects, there was a significant
decline in H pylori serum antibody levels at 18 months
(P<.01, month 18 vs baseline).
There were no statistically significant changes in H pylori serum antibody levels in the 8 subjects whose H pylori gastritis had not been cured 18 months after therapy (subjects
16-23, Table 2). None of these
8 subjects seroconverted 18 months after therapy, although one (subject 20)
had a borderline antibody titer.
Helicobacter pylori seroconversion from positive
to undetectable levels (<15 U/mL) at 18 months had a sensitivity of 60%
for diagnosing cure of H pylori infection and a specificity
of 100% (Table 3). If the serum
antibody cutoff was raised, the sensitivity increased but the specificity
decreased, and at a cutoff of 100 U/mL the sensitivity was 100% but the specificity
was only 37.5%. Serology at earlier times (ie, at 1 month and 3 months) was
insensitive at low cutoff levels and nonspecific at high cutoff levels.
If a patient returns with recurrent symptoms after an attempt at eradication
of H pylori infection, it is reasonable to ask whether
or not the therapy has been successful. If therapy has been successful, other
causes for symptoms will need to be considered, while if therapy has not been
successful, retreatment may be appropriate. Largely from the work of Cutler
et al,7 it has been concluded that serology
can be a sensitive method for detecting cure of H pylori infection, but only beyond the first 6 months after therapy and only
if a baseline (before-therapy) serum sample can be stored and then sent to
the laboratory along with an after-therapy sample. Using this "paired serum
sample" approach, a 20% or greater decline in IgG antibody level has more
than 80% sensitivity and specificity for detecting cure. However, this approach
is not practical. Furthermore, many laboratories do not routinely report quantitative
antibody levels, but instead report serologic results qualitatively as "positive"
or "negative." In a follow-up study in the same subjects, Cutler and Prasad8 reported that only 29% to 45% of cured patients (as
assessed by urea breath tests) seroconverted from positive to negative test
results during the second year after therapy. However, only 15 patients were
followed beyond 15 months. Moreover, they did not report any patients with
persistent gastritis beyond the first year, so that the specificity of seroconversion
could not be determined.
Our study design differed from the studies of Cutler et al.7,8 First, we treated healthy subjects,
while they treated a more heterogeneous group of patients with duodenal ulcer,
gastric ulcer, and nonulcer dyspepsia. Second, we used gastric histology with
tissue stains for H pylori organisms to assess cure,
while Cutler et al used the C-urea breath test. While both techniques have
high sensitivity for detecting cure, our use of biopsies allowed us to also
confirm directly that cure was associated with elimination of gastritis (Table 1). Third, we studied not only individuals
who were cured, but also individuals with persistent gastritis after therapy,
allowing us to calculate both sensitivity and specificity
of seroconversion. Finally, we performed serologic tests in real time (as
serum samples were collected) to simulate clinical practice, while Cutler
et al stored serum samples and ran them together in a single assay. Both studies
had a 2-week regimen using bismuth, tetracycline, and metronidazole and both
used an IgG serologic test from the same manufacturer. Likewise, H pylori eradication rates in the 2 studies were similar (65% in ours
and 57% in the study by Cutler et al7,8).
The major finding of our study was that cure of H
pylori infection led not only to elimination of gastritis, but also
to seroconversion in 60% of subjects approximately 18 months after completion
of therapy. None of the 8 subjects whose H pylori
infection persisted after antimicrobial therapy seroconverted or had resolution
of gastritis. These data suggest that seroconversion beyond the first year
of therapy may reliably reflect cure, although 40% of the subjects who were
cured still had positive test results (ie, they had not seroconverted). In
these subjects, who were cured but remained seropositive 18 months after therapy,
IgG serum antibody levels nevertheless decreased by 41% to 94% from baseline
(median decrease, 72%), raising the possibility that a greater than 40% late
fall in serum antibody from baseline may be useful in assessing eradication,
even if serum samples are run in real time. However, this possibility is unlikely
to be the case, since 4 of our 8 subjects who were not cured nevertheless
had greater than 40% decreases in serum IgG antibody levels from baseline
after antimicrobial therapy (Table 2).
It is not clear why we found that seroconversion occurred in 60% of
cured subjects, while Cutler and Prasad8 found
that seroconversion occurred in less than 50%. This difference could have
been due to the markedly different patient populations or to chance. Our studies
suggest that serology, the simplest of all tests to diagnose H pylori infection, could also be used to assess adequacy of therapy
beyond the first year after treatment. During the first 3 months after treatment,
however, serology was insensitive for confirming cure in this study, as in
the earlier study by Cutler et al.7
For an individual more than 1 year beyond antimicrobial therapy, seroconversion
is a reliable indicator of cure of H pylori infection
and should probably be the test of first choice. A negative test result would
preclude the need for a urea breath test or gastric biopsy in the majority
of patients who are actually cured. Failure to seroconvert is nonspecific,
necessitating a urea breath test or gastric biopsy to confirm cure or persistent
infection. Because current regimens for eradicating H pylori infection are so effective (80%-90% success),11,12
an even higher percentage of subjects will probably seroconvert beyond the
first year than we observed in this study. Additional long-term studies of
seroconversion rates with these newer regimens are needed.