Am. J. Trop. Med. Hyg., 57(6). 1997. pp.

656659
Copyright 0 1997 by The American Society of Tropical Medicine and Hygiene

LONGEVITY OF ANTIBODY RESPONSES TO A SALMONELLA TYPHI-SPECIFIC

OUTER MEMBRANE
PROTEIN: INTERPRETATION
OF A DOT ENZYME
IMMUNOSORBENT
ASSAY IN AN AREA OF HIGH TYPHOID FEVER ENDEMICITY
K. E. CHOO, T M. E. DAVIS, A. ISMAIL, ANDK. H. ONG
Department of Paediatrics and Department of Medical Microbiology
and Parasitology,
School of Medical Sciences,
Universiti Sains Malaysia, Kubang Kerian, Kelantan, Malaysia: Department of Medicine, Fremantle Hospital,
University of Western Australia, Fremantle, Western Australia

Abstract.
The objective of this study was to investigate the longevity of positive dot enzyme iminunosorbent
assay (dot EIA) results for 1gM and IgG to a Salmonella typhi outer membrane protein in Malaysian children with
enteric fever. The patients were children one month to 12 years of age with clinical evidence of typhoid fever, positive
blood or stool cultures for S. typhi, and/or a positive Widal test result who were admitted over a two-year period to
General Hospital (Kota Bharu, Malaysia). These patients received standard inpatient treatment for enteric fever in
cluding chloramphenicol
therapy for 14 days. Dot EIA tests were performed as part of clinical and laboratory as
sessments on admission, at two weeks, and then at 3, 6, 9, 12, 15, 18, and 21 months postdischarge.
Assessment of
the longevity of positive dot EIA 1gM and IgG titers was done by Kaplan-Meier analysis. In 94 evaluable patients,
28% were dot ELA 1gM positive but IgG negative on admission, 50% were both 1gM and IgG positive, and 22%
were 1gM negative and IgG positive. Mean persistence of 1gM dot EIA positivity was 2.6 months (95% confidence
interval = 2.03.1 months) and that of IgG was 5.4 months (4.56.3 months). There were no significant differences
between the three subgroups. Thus, positive 1gM and IgG results determined by dot EIA within four and seven
months, respectively, following documented or suspected enteric fever in a child from an endemic area should be
interpreted with caution. In other clinical situations, the dot EIA remains a rapid and reliable aid to diagnosis.

Although a positive bacterial culture remains the gold

standard for the diagnosis of typhoid fever, indirect diag
nostic tests are also in common use. The most established
of these is the Widal test, which continues to be evaluated
in field studies as an adjunct to clinical assessment and bac
terial isolation.' 2 Recently, a dot enzyme immunosorbent
as
say (dot EIA) using a 50-kD outer membrane protein (OMP)
from Salmonella typhi has been developed as a cost-effective
serologic test alternative that has the additional advantage of
providing a result within hours of blood sampling.3
The dot
EIA allows separate visual assessment of the presence of
specific IgG and 1gM antibodies to the OMP in a standard
1: 100 dilution of serum by a characteristic color change and
has been reported to be at least as specific and sensitive as
the Widal test in children with typhoid fever.5
In other infections, the kinetics of the antibody response can
vary with the infecting microorganism, the clinical state of the
patient at the time of diagnosis, and the response to treatment.
In Shigella dysentery affecting adults, for example, serum 1gM
titers peak within weeks of presentation and decrease to initial
levels after 23months, while IgG levels still are double the
initial titers at this time.6 The duration of significant 1gM titers
in Borrelia

burgdorferi

infections

can

sometimes

be

positive in Malaysian
highly endemic area.

is a function

of the clinical

features

Thus,

for different

infective

agents

typhi

or a positive

Widal

test

result

(significant

titer

1:

80)2 were eligible for recruitment. Informed consent was ob

mined from a parent or guardian of each child for partici
pation of the child in the study, including regular attendance
at outpatient assessments after discharge from the hospital.
The study was approved by the Ethical Committees of both
the Ministry of Health (Malaysia) and Universiti Sans Ma
laysia.
Methods. Clinical management of each child included mm
tial clinical assessment, routine hematologic and biochemical
tests and cultures, commencement
of rehydration either oral
ly or intravenously if required, and chioramphenicol
therapy
(administered orally, or intravenously if unable to take oral
ly, at a dose of 75 mg/kg/day) for 14 days. Blood and fecal
culture for S. typhi and the Widal test were performed as
described previously.5 Briefly, blood samples were taken us
ing aseptic technique, incubated in broth with regular sub
culture on blood agar over a period of seven days. Stool
cultures were plated on MacConkey
and deoxycholate-ci
trate agar both directly and after initial incubation in selenite
broth. Species identification in positive cultures was made
using a panel of biochemical and agglutination tests (Well
come Reagents, Kent, United Kingdom). The Widal test was
carried out using antigens obtained from Wellcome Diag
nostics (Dartford, United Kingdom) and standard serial di
lutions of serum in normal saline starting at 1:40.
Rapid serodiagnosis
was also performed at presentation

many

of the ill

and clinical

fever from a

Patients. All children between one month and 12 years

of age admitted to the Department of Paediatrics, General
Hospital (Kota Bharu, Kelantan State, Malaysia) in 1991 and
1992 with fever and positive blood or stool cultures for S.

ness.7 Persistence of Pseudomonas aeruginosa anti-lipopoly

saccharide antibodies appears to be independent of the severity
of the infection, antigemc stimulus, and initial serologic re
sponse.8

with entenc

PATIENTS AND METHODS

years, while the relationship between IgG and 1gM in the serum
at presentation

children

situa

tions, the duration of detectable specific 1gM and IgG may

vary. This may have implications for the interpretation of se
rologic test results, including the Widal test itself, in areas
where the infection is prevalent.9
To improve the interpretation of the dot EIA in the field,
we have evaluated how long 1gM and IgG test results remain

656

LONGEVITY

OF ANTIBODY

RESPONSE

657

TO S. TYPHI OMP

Percent positive

T@me 1
Details of the groups of patients classified by initial dot-enzyme
immunoassay results

Group 1
*1gM

2
Group 1
(IgM+IIgG)Group (1gM and lgG+)Group

(lgMIlgG+)Number2647Age

@l9G]

(years),21mean
3.0Sex(MIF)12/1421/2612/9Duration
3.17.5
SD8.0

2.37.1

fever(days),
of
7Blood mean SD12
culture posi

71

712

1
.5

(%)54%57%52%Widal
tive
0 antigen Ii

.5
5,'..

ter,
median1:3201:1601:80*(range)(01
1,280)(01:640)Widal
:1,280)(01 :
H antigen ti
ter,
median1:6401:3201:320(range)(1
,280)(01:1,280)a
:401
:640)(01 : 1
@

p < 0.02 versus GroupsI

Group 2

and 2.

ff@gM

lgG

St

using the dot EIA.4 Briefly, 1: 100 dilutions of patient serum

were added to standard aliquots of 0.3 @i.gof purified OMP
dotted on nitrocellulose
strips. After 1 hr. horseradish per
oxidaseconjugated antiserum to human IgG or 1gM was
added. The presence of antigen-antibody
complexes was as
sessed visually from the resultant color change in compari
son with that of positive control sera. A positive dot EIA
IgG or 1gM result was a titer of at least 1: 100 because only
one dilution was used.
Patients were hospitalized at least 14 days or until fever
clearance. Follow-up assessments,
including the dot EIA,
were scheduled at two weeks, and then at 3, 6, 9, 12, 15,
18, and 21 months postdischarge or until dot EIA results for
both 1gM and IgG were negative. Stool or rectal swab cul
tures were also taken at each assessment to identify carriers
of S. typhi.
Statistical
analysis.
Data analysis was performed
by
means of parametric statistics using the computer software
package SPSS for Windows (SPSS, Inc., Chicago, IL) and
results are given the mean and standard deviation (SD) un
less otherwise stated. Assessment of the duration of a posi
tive dot ELA result was done by Kaplan-Meier analysis and
the log-rank test was used to determine differences in mean
persistance

A two-tailed

of 1gM

and

IgG

levels

level of significance

between

patient

St

V
St
St

icx
Group 3
*1gM

lgG

80

60 e@'

St
St
St

40

5,
.5.

20

@, p

24

810

12

1416

18

20

groups.
Time (months)

was used.
FIGURE 1.

RESULTS

Of 151 children admitted during the two-year study period

with probable enteric fever, 94 (45 males and 49 females
between the ages of two and 11 years) were included in the
analysis. These 94 evaluable patients were those from whom
complete data were available for inpatient management, who
had at least one assessment after discharge, and who were
neither readmitted with typhoid fever nor had a positive stool
culture during follow-up.
On the basis of dot ELA results at the time of presentation,
patients were divided retrospectively
into one of three sub
groups. Group 1 consisted of those who were initially 1gM
positive and IgG negative, Group 2 comprised those who
were both 1gM and IgG positive on admission, and Group

Percentage

of patients

with

positive

dot enzyme

im

munosorbent assay 1gM and/or IgO results during follow-up for up

to 21 months postdischarge. Group 1 patients are those who were
only 1gM positive initially, Group 2 were both 1gM and IgO positive
at presentation,

and Group

3 had only

positive

IgO titers

on admis

sion.

3 patients were those who were 1gM negative but IgG pos
itive. Details of these groups are summarized in Table 1. The
groups had similar demographic characteristics
and features
related to typhoid fever except that those who were initially
1gM negative (Group 3) had significantly
lower Widal 0
antibody titers than those in the other two groups.
Figure 1 shows the serial changes in positivity for both
1gM and IgG antibodies by dot EIA over the period of fol
low-up. In Group 1, there was a decrease in the percentage

658

CHOO AND OTHERS

100

TABLE 2
[@Group
- Group

80

Mean survival [95% confidence intervals] in months for 1gM or IgG

antibody detected by dot-enzyme immunoassay in those patients

who were positive

A Censored

at any time during follow@up*

a)
>

positiveAll

U)
0
0.

60

1gM positive

patients2.6

lgG

5.4 [4.56.3]
(n = 82; 3 censored)

[2.03.1]

censored)Group (n = 75; 2

at

12.3

4.7 [2.96.5]

[1.63.0]

censored)Group (n = 26; 2
22.7
[1.93.4]
censored)Group (n
47; 0
3

a) 40
C.)

20

(n = 14; 0 censored)
5.8 [4.66.9]

(n = 47; 3 censored)
4.9 [3.06.81

(n = 2)
a Censored

cases

are those

who

(n = 21; 0 censored)
were

positive

initially

or during

follow-up

but defaulted

from further assessmentbefore the test results became negative.

[@Group 1

a)
>

- - Group

S..

Group

IgG antibodies (P = 0.075) with lower mean

Groups 1 and 3 compared with to Group 2.

a Censored

values

in

DISCUSSION

In
0
0.

CD
at
C
a)
C.)
a)
0.

10

12

14

16

18

20

Time (months)
FIGURE 2.

Persistence

curves

for both 1gM- and IgG-positive

dot

enzyme immunosorbent assay results grouped by initial serologic

result. Censored
quate follow-up

data are from patients who did not complete

and were excluded from persistance analysis.

ade

of patients who were 1gM positive such that all were nega
tive by six months. Less than half of this group became IgG
positive during this time and the decrease in the number of
IgG-positive cases was slower than for 1gM. This latter trend
was also observed in Group 2. In Group 3, only two patients
or just under 10% became 1gM positive during follow-up.
Persistance curves for 1gM and IgG are shown in Figure
2 and the results of Kaplan-Meier
analysis are summarized
in Table 2. Only two (2.7%) of 75 patients who tested 1gM
positive during the study defaulted from follow-up before
they became antibody negative (Figure 2). In the case of
IgG, this was the case in only three (3.7%) of 82 cases. The
mean persistance of detectable 1gM antibody in patients who
tested positive at any time during the study period was 2.6
months, with the majority of patients becoming negative
within three months of the positive result. There were no
significant differences in mean persistance value between
subgroups (P = 0.88), but because of insufficient numbers,
Group 3 was not included in analysis. For IgG, the mean
persistance of antibody positivity was approximately
twice
as long as that for 1gM (5.4 months), with most patients
becoming negative six months after initial assessment. There
was a nonsignificant
trend to shorter persistance times for

Our results indicate that in Malaysian children with en

tenc fever, the mean duration of an initially or subsequently
positive 1gM dot EIA result by persistance analysis is just
under three months postdischarge.
For IgG positivity, the
mean duration is approximately double this figure. The 95%
confidence intervals for these means approach one month on
either side. Thus, positive 1gM and IgG results within four
and seven months, respectively, of documented or suspected
enteric fever would appear to obfuscate the initial serologic
assessment of a febrile child.
Patients groups classified according to initial dot EIA 1gM
and IgG profiles did not differ in their basic demographic
and clinical features. However, it is interesting that a signif
icantly lower Widal anti-O titer was found in Group 3 chil
dren. This suggests that despite the comparable duration of
fever before presentation in the three groups, Group 3 pa
tients presented at an immunologically
later stage in their
infection
creasing

(with
Widal

a negative
1gM result by dot EIA and a de
anti-O titer) than children
in the other two

groups. Nevertheless, the duration of either 1gM or IgG test

result positivity did not appear to be influenced by the im
munologic stage at which the children presented.
In Group 1 patients who were all 1gM positive and IgG
negative

by dot

EIA

at presentation,

the development

of a

positive IgG test result occurred in less than half during fol
low-up. This may have been due partly to the sampling sched
ule, with some patients possibly exhibiting a transiently pos
itive IgG test result between assessments. However, given the
mean and 95% confidence intervals for persistance time of an
IgG response in the three groups compared with the 2.5-3month gap between samples, such a contribution would ap
pear limited. Thus, there appears to be a proportion of patients
(approximately
10% of the total series) who do not develop
IgG antibodies to the OMP as detected by the dot EIA.
Two patients in Group 3 who were initially 1gM negative
and IgG positive subsequently developed positive 1gM test
results. This could indicate that these patients had a residual
positive IgG titer from a previous episode of enteric fever
and redeveloped significant 1gM titers from the new infec
tion. Other unexpected patterns also emerged during follow

LONGEVITY

OF ANTIBODY

up in individual patients, including a transiently positive IgG

test result at 18 months in a child with a negative result three
months before and three months after this timepoint. How
ever, factors such as exposure to other Salmonella species
without the patient necessarily developing overt symptoms
of infection may influence the results of the dot EIA.
There are few studies that have prospectively
examined
aspects of humoral immunity to bacterial infections in hu
mans. Nevertheless, the longevity of detectable 1gM and IgG
antibodies by dot EIA in the present study was comparable
with that reported in bacterial infections such as Shigella
flexneri6 and Yersinia

Specific IgG antibodies

do not always develop after IgMU or may be delayed in their
appearance,12

observations

that

are

consistent

with

the

se

rologic findings in individual patients in our series.

Our data allow a better characterization
of the clinical ap
plications and limitations of the dot EIA. Where enteric fever
is uncommon, the patient has had no significant febrile ill
nesses

over

the

previous

seven

months,

and

has

suggestive

clinical features, a dot EIA can prove to be a rapid and reliable

diagnostic aid at the time of presentation. In an endemic area
and when the child has had known or possible typhoid fever
in the relatively recent past, the dot EIA result must be inter
preted with caution. Of note, comparable kinetic information
to that presented in this report is not available for the Widal
test, indicating a further potential advantage of the dot EtA
over conventional serologic diagnostic techniques.
Financial support: The study was supported by an Intensification of
Research in Priority Area grant from the Ministry of Science, Tech
nology and Environment of Malaysia.
Authors' addresses: K. E. Choo, Department of Paediatrics, School
of Medical Sciences, Hospital Universiti Sains Malaysia, Kubang
Kerian, 16150 Kelantan, Malaysia. T M. E. Davis, Department of

Medicine, Fremantle Hospital, University of Western Australia, P0

Box 480, Fremantle, Western Australia 6160. A. Ismail and K. H.
Ong, Department of Medical Microbiology and Parasitology, School
of Medical Sciences, Universiti Sains Malaysia, Kubang Kerian,
16150 Kelantan, Malaysia.

RESPONSE

659

TO S. TYPHI OMP

Reprint requests: K. E. Choo, Depanment of Paediatrics, Hospital

Universiti Sains Malaysia, Kubang Kerian, 16150 Kelantan, Malaysia.
REFERENCES

1. Chow CB, Wang PS, Cheung MW, Yan WW, Leung NK, 1987.
Diagnostic value of the Widal test in childhood typhoid fever.
Pediatr infect Dis J 6: 914917.
2. Choo KE, Razif AR, Oppenheimer

SJ, Ariffin WA, Lau J, Abra

ham T, 1993. Usefulness of the Widal test in diagnosing

childhood

typhoid

fever in endemic

areas.

J Paediatr

Child

Health 29: 3639.

3. Ismail A, Ong KH, Kader ZA, 1991. Demonstration

of an an

tigenic protein specific for Salmonella typhi. Biochem Bio

phys Res Commun /81: 301305.
4. Ismail A, Kader ZS, Ong KH, 1991. Dot enzyme

immunosor

bent assay for the serodiagnosis of typhoid fever. Southeast

Asian J Trop Med Public Health 22: 563566.
5. Choo KB, Oppenheimer SJ, Ismail AB, Ong KH, 1994. Rapid
serodiagnosis
in an endemic

of typhoid
fever
area. Clin inftct

by dot enzyme
immunoassay
Dis 19: 172176.

6. Cohen D, Block C, Green MS. Lowell 0, Ofek I, 1989. Im

munoglobulin M, A, and 0 antibody response to lipopolysac
charide 0 antigen in symptomatic and asymptomatic Shigella
infections. J Clin Microbiol 27: 162167.
7. Hammers-Berggren

5, Lebech

AM, Karisson

M, Svenungsson

B, Hansen K, Stiernstedt 0, 1994. Serological follow-up after

treatment of patients with erythema migrans and neurobor
reliosis. J Clin Microbiol 32: 15191525.
8. Petras 0, Adam MM, 1987. Phases of the serological response
in Pseudomonas aeruginosa infections. Acta Microbiol Hung
34: 147157.

9. Schroeder SA, 1968. Interpretation of serologic tests for typhoid

fever. JAMA 206: 839840.
10. Granfors K, Viljanen M, Tillikainen A, Toivanen A, 1980. Per
sistence of 1gM, IgG, and IgA antibodies to Yersinia in Yer
sinia arthritis.

J infect Dis 141: 424429.

11. Elgh F, Wadell 0, Juto P. 1995. Comparison of the kinetics of

Puumala virus specific 1gM and 1gM antibody responses in
nephropathia

epidemica

as measured

by a recombinant

anti

gen-based enzyme-linked immunosorbent assay and an im

munofluorescence test. J Med Virol 45: 148150.
12. Kalish RA, Leong JM, Steere AC, 1995. Early and late antibody
responses to full-length and truncated constructs of outer sur
face protein A of Borrelia
immun 63: 22282235.

burgdoiferi

in Lyme disease.

infect