Am. J. Trop. Med. Hyg., 80(1), 2009, pp. 20–23
Copyright © 2009 by The American Society of Tropical Medicine and Hygiene
Improvement in Detection of Enterotoxigenic Escherichia coli in Patients with Travelers’
Diarrhea by Increasing the Number of E. coli Colonies Tested
Thushara Galbadage, Zhi-Dong Jiang,* and Herbert L. DuPont
The University of Texas School of Public Health at Houston, Houston, Texas; The University of Texas Medical School at Houston, Houston,
Texas; St. Luke’s Episcopal Hospital, Houston, Houston, Texas; Baylor College of Medicine, Houston, Texas
Abstract. No cause of one-third of travelers’ diarrhea (TD) cases can be detected despite microbiologic assessment.
We propose that these pathogen-negative TD cases include undetected enterotoxigenic Escherichia coli (ETEC). As a
standard in diagnostic microbiology, samples of five E. coli colonies were tested to detect ETEC from stool cultures. We
compared the sensitivities and the number of ETEC detected by 20 colonies with those of 5, 10, and 15 colonies. The
percent detection of ETEC with five E. coli colonies was significantly different from that using 20 E. coli colonies. Of
the 116 subjects studied, the number positive for ETEC for 5, 10, 15 and 20 colonies tested were 22 (19.0%), 37 (31.9%),
45 (38.8%) and 46 (39.7%), respectively. Based on our results, we recommend the testing of at least 10 E. coli colonies for
optimum detection of ETEC in patients with TD.
INTRODUCTION
have ETEC that simply go undetected. We proposed that an
increase in number of E. coli colonies selected and tested
will more accurately detect cases of TD caused by ETEC.
Our proposition was further supported by studies showing
a positive antibacterial drug effective for in the treatment of
pathogen-negative TD.15,16
Enterotoxigenic Escherichia coli (ETEC) is the leading
cause of travelers’ diarrhea (TD) and is identified in close
to half the case-patients with diarrhea.1–4 Unlike most other
bacterial pathogens causing diarrhea, which can be detected
by culturing patient stool specimens on selective media,
E. coli pathogens cannot be distinguished in this manner. Nonvirulent E. coli is part of the biota living symbiotically and lining the gastrointestinal tract of healthy persons. Therefore,
pathogenic E. coli strains causing diarrhea needs to be differentially identified with molecular techniques being widely
used for this purpose.
ETEC causes diarrhea by colonizing the small intestine and
produces one or two enterotoxins, a heat-labile enterotoxin
(LT) and a heat-stable enterotoxin (ST).4 Previous methods
for the detection of ETEC and other E. coli pathogens causing
diarrhea included several types of DNA hybridization assays,
also known as colony hybridizations.5–7 Currently, polymerase
chain reaction (PCR) techniques are commonly used in
enteric diagnostic laboratories for the detection of ETEC.8–11
Iijima and others recently suggested that quantitative realtime PCR (qPCR) will better identify diarrhea-producing
E. coli (DEC).12
Conventionally, five E. coli-like colonies selected from
stool cultures of patients with TD are tested for ETEC
and other DEC.1,2 This procedure has been standardized to
use five E. coli colonies for ETEC toxin assay without scientific validation.13,14 The underlining assumption has been
that most E. coli present in stool specimens of patients with
acute diarrhea are of the pathogenic form. This assumption
may or may not be correct because the extent of DEC present may be related to duration of illness at time of study.
A study showed that ratios of DEC to non-pathogenic E. coli
in stools of patients with diarrhea were low, with the exception of ETEC and Shiga toxin–producing E. coli (STEC),
which were present in proportions as low as 40% of total
E. coli.12 Moreover, more than one-third of patients with TD
studied remain negative for all known enteropathogens.1 We
hypothesized that some of these pathogen-negative patients
MATERIALS AND METHODS
Patients and E. coli colonies tested. De-identified stool
specimens collected from 116 patients with TD acquired
in Guadalajara, Mexico, during the summer of 2006 were
used for our study. A case-patient was defined as a foreign
traveler who passed unformed stools three or more times
during a 24-hour period with at least one abdominal symptom
of enteric infection.
The collected stool samples were processed in our field laboratory for enteric bacterial pathogens (including Salmonella
spp., Shigella spp., Vibrio spp., Campylobacter jejuni, Yersinia
enterocolitica, Aeromonas spp., and Plesiomonas shigelloides)
and parasites (Giardia lamblia, Cryptosporidium species, and Entamoeba histolytica) by previously described
methods.17
After we cultured these stools separately on selective
MacConky agar, 20 colonies of E. coli for each patient were
selected and saved in peptone stubs. The study sample consisted of 116 sets of 20 E. coli colonies. The sets of 20 E. coli
colonies for each stool specimen were randomly numbered
from 1 to 20 and divided into four sub-groups of five colonies:
1–5, 6–10, 11–15, and 16–20.
Testing of 20 colonies per subject for ETEC. Each patient
with TD was tested for ETEC in four steps, corresponding to
the four sub-groups of five E. coli colonies. First, five colonies
(colonies 1–5) of all 116 stool cultures were tested for ETEC,
by identifying ST or LT toxin–encoding genes using PCR (see
below for methods). Patients who were positive for ST, LT, or
ST/LT for these five E. coli colonies were considered positive
for ETEC. The remaining 15 E. coli colonies for each of these
patients positive for ETEC were not tested further. Patients
who were negative for ST and LT for the first five E. coli colonies were taken to the second step, where the next five colonies (colonies 6–10) of each patient were tested for ETEC
using PCR. Similarly, those patients positive for ST, LT, or ST/
LT in the second five E. coli colonies were considered positive
* Address correspondence to Zhi-Dong Jiang, 1200 Herman Pressler,
RAS E-739, Houston, TX 77030. E-mail: Zhi-Dong.Jiang@uth
.tmc.edu
20
21
IMPROVED DETECTION OF ENTEROTOXIGENIC ESCHERICHIA COLI
for ETEC and was not tested beyond the first 10 E. coli colonies. The patients that remained negative for ETEC were
taken to the third step. The same process was repeated for
the third set of five E. coli colonies (colonies 11–15) and then
for the fourth set (colonies 16–20). After testing for ETEC
in up to 20 E. coli colonies, patients negative for ST and LT
toxin–encoding genes were considered to be truly negative
for ETEC.
DNA extraction and pooled DNA preparation. In groups
of five colonies per patient, E. coli from peptone stubs were
grown separately on MacConky agar plates at 37°C overnight
and collected individually into 200 µL of deionized water in
1.5-mL microfuge tubes.18 Each sample in microfuge tubes
were heated at 100°C for 10 minutes, subjected to vertex
for 10 seconds, and centrifuged at 12,000 rpm for 2 minutes.
Supernatants containing the DNA extracts were transferred
into separate 0.5-mL microfuge tubes and labeled. Pooled
DNA samples were prepared by mixing 10 µL of each of
the five DNA extractions corresponding to a patient into
another 0.5-mL microfuge tube. Six DNA extracts for each
patient, five for the individual E. coli colonies, and one
for the pooled DNA sample were stored in their 0.5-mL
microfuge tube at −20°C.
PCR amplification and ETEC identification. Conventional
PCR was used to amplify the ST and LT toxin–encoding genes
in the pooled DNA extracts of E. coli colonies prepared for
each patient. The PCR mixture per DNA extract contained
2.0 µL of 10× buffer, 4.0 µL of 5Q buffer, 0.8 µL of 2.5 mM
MgCl2, 1.5 µL of 2.5 µM dNTPs, 1.0 µL of forward primer, 1.0
µL of reverse primer, and 6.7 µL of double-distilled deionized
water. A total of 0.5 µL of Taq DNA polymerase was used per
100 µL of PCR mixture. In each PCR tube, 3 µL of pooled
DNA extract was combined with 17 µL of PCR reaction mixture. Amplification of ST and LT were done separately using
only one set of primers at a time. The primer sequences used
were ST-F 5′-TGGATGCCATGTCCGGAGGT-3′, ST-R
5′-CAACAGGTACATACGTTACAGAC-3′, LT-F 5′-GAGTACTTCGATAGAGGAACTCA-3′, and LT-R 5′-GATCCGGTGGGAAACCTGCT-3′. Two thermocycler were used to run
the PCR starting at 95°C for 2 minutes, then 10 cycles at 94°C
for 30 seconds, 62°C for 30 seconds, and 72°C for 1 minute,
then 30 cycles at 94°C for 30 seconds, 56°C for 30 seconds,
and 72°C for 1 minute, a final step at 72°C for 1 minute, and
holding the PCR products at 4°C until removed. A positive
control, E. coli H10407 (no. 35401; American Type Culture
Collection, Manassas, VA) and a negative control (without a DNA extract) were used with each PCR cycle. Gel
electrophoresis was used to separate amplified ST and LT
toxin–encoding gene sequences present. Running gels were
prepared with 1.5% agarose containing trace amounts of
ethidium bromide (3 µL per 100-mL gel). Gel lanes contained a mixture of 3 µL of PCR product, 2 µL of 6× loading buffer (0.5 mM EDTA, 20% sodium dodecyl sulfate, 5%
bromophenol blue, 1% glycerol), and 5 µL of 1 M Tris, pH
8.0, 0.5 M EDTA. Gels were run for 30 minutes, at 100V,
and digital images of the bands were taken under ultraviolet light.
Confirmation and detection of ETEC-positive colonies. The
PCR was conducted with pooled DNA extracts of groups of
five E. coli colonies per patient. Once an ST- or LT-positive
sample was detected in one of the pooled DNA extracts, this
result was confirmed and determined as to which individual
colonies were positive. This confirmation was conducted by
repeating the PCR for either ST or LT on individual DNA
extracts of the pooled DNA extracts that were positive for
ST or LT. This method indicated how many ST- or LT-positive
colonies were present within a sample of five E. coli colonies
tested. In cases with an ST- and an LT-positive pooled DNA
extract, the PCR for both ST and LT was repeated separately
on individual DNA extracts. The same colony or DNA extract
had to be positive for ST and LT to be identified as positive
for ST/LT.
Statistical analysis and assay comparison. The results were
grouped into four groups: colonies 1–5, colonies 1–10, colonies 1–15, and colonies 1–20. The PCR assay using 20 E. coli
colonies had the best probability of detecting all cases with
diarrhea caused by the ETEC. Therefore, the proportion of
ETEC detected by the 20 E. coli colony assay was compared
with that of the 5-, 10-, and 15-colony assays. Patients positive for ST, LT, ST/LT, or ETEC (total) in the first five colonies were assumed to be positive in the 10-, 15-, and 20-colony
assays. Likewise, this assumption was also made for those
positive for the second five colonies and the third five colonies.
The sample of 116 case-patients with TD was large enough
for a normal approximation, and the z-test was used to compare population proportions. After we applied the Bonferroni
correction for multiple comparisons of data, the significance
level (P value) for the comparison was set at 0.0167. There
were 12 patients with one missing E. coli colony, 2 with 2 missing E. coli colonies, and 1 with 3 missing E. coli colonies of
their sets of 20 E. coli colonies saved. This finding was caused
either by drying-up of the peptone media or the absence of an
E. coli colony in the peptone media. These missing colonies
had no significant affect on the data and were not included in
the statistical analysis.
RESULTS
Detection of ETEC and assay sensitivities. The 5-colony
assay detected 22 (19%) of 116 patients positive for ETEC:
15 positive for ST, 6 positive for LT, and 1 positive for ST
and LT (Table 1). The 10-colony assay detected 37 (31.9%)
patients positive for ETEC: 26 positive for ST, 10 positive for
LT, and 1 positive for ST and LT. The 15-colony assay detected
45 (38.8%) patients positive for ETEC: 26 positive for ST, 17
positive for LT and 2 positive for ST and LT. The 20-colony
assay detected 46 (39.7%) patients positive for ETEC: 26 positive for ST, 17 positive for LT, and 3 positive for ST and LT.
The 5-colony assay had the lowest sensitivity, and sensitivities
increased with use of more E. coli colonies. The 10-, 15-, and
20-colony assays were 1.68, 2.05, and 2.09 times, respectively,
Table 1
Number of ETEC toxin-encoding genes detected by PCR assays, using
5, 10, 15, and 20 colonies of Escherichia coli, from 116 stool cultures
of patients with travelers’ diarrhea*
No. (%) of stools with positive results (n = 116)
Toxin detected
5 colonies
10 colonies
15 colonies
20 colonies
Heat stable
Heat labile
Both toxins
Total ETEC
15 (12.90)
6 (5.1)
1 (0.9)
22 (19.0)
26 (22.4)
10 (8.6)
1 (0.9)
37 (31.9)
26 (22.4)
17 (14.7)
2 (1.7)
45 (38.8)
26 (22.4)
17 (14.7)
3 (2.6)
46 (39.7)
* ETEC = enterotoxigenic Escherichia coli; PCR = polymerase chain reaction.
22
GALBADAGE AND OTHERS
more sensitive than the 5-colony assay. If the 20-colony assay
was assumed to have a sensitivity of 100%, the sensitivities of
the 5-, 10-, and 15-colony assays would be 47.8%, 80.4% and
97.8%, respectively.
Comparison of the 5-, 10-, and 15-colony assays with the
20-colony assay. Results of the 5-colony assay were significantly different from those obtained with the 20-colony assay
in the number of ST (15 versus 26; P = 0.0072), LT (6 versus
17; P = 0.0019), and ETEC (22 versus 46; P < 0.0001) forms
detected among the 116 case-patients. However, there was
no significant difference between the results of the 5-colony
assay and the 20-colony assay in detecting ST/LT (1 versus 3;
P = 0.1210). Results of the 10-colony assay were not significantly different from those obtained with the 20-colony assay
in the number of ST (26 versus 26; P = 0.5000), LT (10 versus 17;
P = 0.0295), ST/LT (1 versus 3; P = 0.1210), and ETEC (37
versus 46; P = 0.0438) forms detected. Similar to the 10-colony
assay, the results of the 15-colony assay were not significantly
different from those obtained with the 20-colony assay in the
number of ST (26 versus 26; P = 0.5000), LT (17 versus 17;
P = 0.5000), ST/LT (2 versus 3; P = 0.2793), and ETEC
(45 versus 46; P = 0.4247) forms detected.
Proportion of ETEC to E. coli stool cultures. The proportions (95% confidence intervals [CIs]) of ST, LT, and ST/LT
toxin–encoding genes detected in patients positive for ETEC
to total E. coli colonies present in these patients were 0.39
(0.29–0.49), 0.22 (0.18–0.32), and 0.11 (0.00–0.31), respectively.
The proportion (95% CI) of ETEC to total E. coli was 0.31
(0.24–0.39) for the sample of patients in this study. It should be
noted that in calculating the proportion estimates, the probability samples were not the same size, having 5, 10, 15, and at
times, 20 colonies, depending upon the step at which the first
positive sample was identified.
Mixed infection. Eight patients (7%) with ETEC infections were co-infected with another pathogen in this study.
Co-infection with EAEC and ETEC accounted for six of the
eight mixed infections. One patient with ETEC infection was
co-infected with Cryptosporidium parvum and another was
co-infected with Salmonella spp.
DISCUSSION
Our study showed a significant increase in the number of
ETEC strains detected in patients with TD when 20 E. coli
colonies were tested for ST and LT toxin-encoding genes than
when five E. coli colonies were tested. The sensitivity of the
PCR assay using 20 colonies was 2.09 times higher than that
of the assay using five colonies. The 19.0% of ETEC-positive
patients detected by the 5-colony assay was significantly different from the 39.7% of ETEC-positive patients detected
by the 20-colony assay (P < 0.0001). More than half the possible ETEC-positive patients (52.2%) went undetected with
the 5-colony assay. Use of a probability sample of five E. coli
colonies showed low sensitivity (< 50%) for detecting ETEC
in patients with TD.
The sensitivity of the 20-colony assay was 1.24 times higher
than that of the 10-colony assay. However, the 31.9% of
ETEC-positive patients detected by the 10-colony assay was
not significantly different from the 39.7% of ETEC-positive
patients detected by the 20-colony assay (P = 0.0438). There
was also no statistically significant difference between the
10- and 20-colony assays for detection of ST, LT, and ST/LT
individually. Increasing the number of colonies from a 15 to 20
detected only one more ST/LT-positive patient, but this difference was not statistically significant.
The low proportion of DEC found by Iijima and others12
when five E. coli colonies were tested is supportive of our
findings. Their study also showed that ETEC and STEC were
present in a proportion of approximately 40% among nonpathogenic E. coli, which was much higher than the observed
proportions for non-ETEC DEC. Their study and ours provided evidence that ETEC could be detected with less than
20 colonies. Their suggestion that real-time PCR is needed for
effective detection of DEC may not be true for ETEC because
we could identify most ETEC-positive sample with less than
20 E. coli colonies.
Use of the conventional assay with five E. coli colonies
would have categorized the ETEC diarrhea that we detected
in colonies 6–20 as pathogen-negative after screening for other
known enteropathogens. Our findings may at least partially
explain a portion of the pathogen-negative TD cases found in
studies of etiology of TD.1 Our results also help explain in part
the effectiveness of antibacterial drug therapy used to treat
pathogen-negative TD.15
Although increasing the number of colonies tested increased
the sensitivity and the percentage of ETEC detected, the specificity of the assays may decrease. The more colonies used,
ETEC will be detected even when present in low quantities.
Thus, this assay may detect ETEC that are not etiologically
important in the acute diarrhea episode, which represent falsepositives samples. Therefore choosing an optimum number of
colonies for detection of ETEC is important. However, in our
study, most ETEC-positive samples wee detected with the first
15 colonies.
We recently reported that fecal DNA remained stable on
hemoccult cards stored at room temperature for up to 14
months and could be used for PCR amplification for detection of diarrheagenic E. coli.19 Significantly more ETEC cases
were detected from hemoccult card DNA (38%) than from
fecal DNA (30%) or by culture that was followed by hybridization (10%) (P < 0.001). These results support those of the
current study, which indicate that ETEC may go undetected
when standard assay methods are used.
We recommend the use of at least 10 E. coli colonies for detection of ETEC in enteric diagnostic laboratories instead of the
five E. coli colonies often used. This modification will increase
the sensitivity of the current PCR assay and the number of
ETEC detected in patients with diarrhea patients and lead to a
lower percentage of unexplained pathogen-negative TD.
Received July 23, 2008. Accepted for publication September 28, 2008.
Acknowledgments: We thank Dr. Asha S. Kapadia (The University of
Texas School of Public Health [UTSPH]) for assisting with the statistical analyses, Dr. Ismail M. Meraz (UTSPH) for his contributions to
the initial laboratory procedures, and the students of The University
of Texas Medical School for collecting stool specimens used in this
study. This project fulfilled the requirement for MPH for Thushara
Galbadage at the University of Texas School of Public Health.
Authors’ addresses: Thushara Galbadage sand Zhi-Dong Jiang, The
University of Texas School of Public Health at Houston, Houston, TX
77225. Herbert L. DuPont, The University of Texas School of Public
Health at Houston, Houston, TX 77225; The University of Texas
Medical School at Houston, Houston, TX 77030; St. Luke’s Episcopal
Hospital, Houston, TX 77030; Baylor College of Medicine, Houston,
TX 77030.
IMPROVED DETECTION OF ENTEROTOXIGENIC ESCHERICHIA COLI
REFERENCES
1. Jiang ZD, Lowe B, Verenkar MP, Ashley D, Steffen R, Tornieporth
N, von Sonnenburg F, Waiyaki P, DuPont HL, 2002. Prevalence
of enteric pathogens among international travelers with diarrhea acquired in Kenya (Mombasa), India (Goa), or Jamaica
(Montego Bay). J Infect Dis 185: 497–502.
2. Ko G, Garcia C, Jiang ZD, Okhuysen PC, Belkind-Gerson J,
Glass RI, DuPont HL, 2005. Noroviruses as a cause of traveler’s
diarrhea among students from the United States visiting
Mexico. J Clin Microbiol 43: 6126–6129.
3. Paredes P, Campbell-Forrester S, Mathewson JJ, Ashley D,
Thompson S, Steffen R, Jiang ZD, Svennerholm AM, DuPont
HL, 2000. Etiology of travelers’ diarrhea on a Caribbean island.
J Travel Med 7: 15–18.
4. Sanchez J, Holmgren J, 2005. Virulence factors, pathogenesis and
vaccine protection in cholera and ETEC diarrhea. Curr Opin
Immunol 17: 388–398.
5. Begaud E, Germani Y, 1992. Detection of enterotoxigenic
Escherichia coli in faecal specimens by acetylaminofluorenelabelled DNA probes. Res Microbiol 143: 315–325.
6. Lund A, Wasteson Y, Olsvik O, 1991. Immunomagnetic separation
and DNA hybridization for detection of enterotoxigenic
Escherichia coli in a piglet model. J Clin Microbiol 29:
2259–2262.
7. Sommerfelt H, Bhan MK, Srivastava R, Bhatnagar S, 1993.
Evaluation of DNA-DNA hybridization for the direct detection of enterotoxigenic Escherichia coli in stool blots. Scand
J Infect Dis 25: 457–463.
8. Caeiro JP, Estrada-Garcia MT, Jiang ZD, Mathewson JJ, Adachi
JA, Steffen R, DuPont HL, 1999. Improved detection of enterotoxigenic Escherichia coli among patients with travelers’ diarrhea, by use of the polymerase chain reaction technique. J Infect
Dis 180: 2053–2055.
9. Ghosal A, Bhowmick R, Nandy RK, Ramamurthy T, Chatterjee
NS, 2007. PCR-based identification of common colonization
factor antigens of enterotoxigenic Escherichia coli. J Clin
Microbiol 45: 3068–3071.
10. Stacy-Phipps S, Mecca JJ, Weiss JB, 1995. Multiplex PCR assay
and simple preparation method for stool specimens detect
11.
12.
13.
14.
15.
16.
17.
18.
19.
23
enterotoxigenic Escherichia coli DNA during course of infection. J Clin Microbiol 33: 1054–1059.
Yavzori M, Porath N, Ochana O, Dagan R, Orni-Wasserlauf R,
Cohen D, 1998. Detection of enterotoxigenic Escherichia coli in
stool specimens by polymerase chain reaction. Diagn Microbiol
Infect Dis 31: 503–509.
Iijima Y, Tanaka S, Miki K, Kanamori S, Toyokawa M, Asari S,
2007. Evaluation of colony-based examinations of diarrheagenic Escherichia coli in stool specimens: low probability of
detection because of low concentrations, particularly during the
early stage of gastroenteritis. Diagn Microbiol Infect Dis 58:
303–308.
Black RE, Merson MH, Rowe B, Taylor PR, Abdul Alim AR,
Gross RJ, Sack DA, 1981. Enterotoxigenic Escherichia coli diarrhoea: acquired immunity and transmission in an endemic area.
Bull World Health Organ 59: 263–268.
DuPont HL, Olarte J, Evans DG, Pickering LK, Galindo E, Evans
DJ, 1976. Comparative susceptibility of Latin American and
United States students to enteric pathogens. N Engl J Med 295:
1520–1521.
DuPont HL, Haake R, Taylor DN, Ericsson CD, Jiang ZD,
Okhuysen PC, Steffen R, 2007. Rifaximin treatment of pathogen-negative travelers’ diarrhea. J Travel Med 14: 16–19.
Huang DB, DuPont HL, 2005. Rifaximin–a novel antimicrobial for
enteric infections. J Infect 50: 97–106.
Jiang ZD, Lowe B, Verenkar MP, Ashley D, Steffen R, Tornieporth
N, von Sonnenburg F, Waiyaki P, DuPont HL, 2002. Prevalence
of enteric pathogens among international travelers with diarrhea acquired in Kenya (Mombasa), India (Goa), or Jamaica
(Montego Bay). J Infect Dis 185: 497–502.
Persson S, Olsen KE, Scheutz F, Krogfelt KA, Gerner-Smidt
P, 2007. A method for fast and simple detection of major
diarrhoeagenic Escherichia coli. Clin Microbiol Infect 13:
516–524.
Grimes KA, Mohamed JA, Dupont HL, Padda RS, Jiang ZD,
Flores J, Belkind-Gerson J, Martinez-Sandoval FG, Okhuysen
PC, 2008. PCR-based assay using occult blood detection cards
for detection of diarrheagenic Escherichia coli in specimens
from U.S. travelers to Mexico with acute diarrhea. J Clin
Microbiol 46: 2227–2230.