Parsonnet J, Shmuely H, Haggerty T. Fecal and Oral Shedding of Helicobacter pylori From Healthy Infected Adults. JAMA. 1999;282(23):2240–2245. doi:10.1001/jama.282.23.2240
Author Affiliations: Division of Infectious Diseases and Geographic Medicine, Department of Medicine, School of Medicine (Drs Parsonnet and Shmuely and Mr Haggerty), and the Division of Epidemiology, Department of Health Research and Policy (Dr Parsonnet), Stanford University, Stanford, Calif.
Context Helicobacter pylori commonly infects humans;
however, its mode of transmission remains unknown.
Objective To determine how humans—the primary host for H pylori—shed the organism into the environment.
Design Controlled clinical experimental study conducted from February through
Setting Clinical research unit of a hospital in northern California.
Patients Sixteen asymptomatic H pylori–infected
and 10 uninfected adults.
Intervention A cathartic (sodium phosphate) and an emetic (ipecac) were given to
all infected subjects and an emetic was given to 1 uninfected subject.
Main Outcome Measure Confirmed H pylori isolates cultured from stool,
air, or saliva before and after catharsis and emesis and from vomitus during
emesis. Isolates were fingerprinted using repetitive extragenic palindromic
(REP) polymerase chain reaction and species identity was confirmed by sequencing
the 16s ribosomal RNA gene.
Results All vomitus samples from infected subjects grew H
pylori, often in high quantities. Air sampled during vomiting grew H pylori from 6 (37.5%) of the 16 subjects. Saliva before
and after emesis grew low quantities of H pylori
in 3 (18.8%) and 9 (56.3%) subjects, respectively. No normal stools and only
22 (21.8%) of 101 induced stools grew the organism, although 7 (50.0%) of
14 subjects had at least 1 positive culture (2 stool culture samples were
contaminated by fungus and were not included). Fingerprints of isolates within
subjects were identical to one another but differed among subjects. No samples
from uninfected subjects yielded H pylori.
Conclusions Helicobacter pylori can be cultivated uniformly
from vomitus and, occasionally, from saliva and cathartic stools. The organism
is potentially transmissible during episodes of gastrointestinal tract illness,
particularly with vomiting.
Helicobacter pylori causes peptic ulcers and
has been implicated in the etiology of distal gastric cancer.1
However, it is not known how H pylori is transmitted.
This uncertainty stems from 4 unresolved questions: (1) how does the organism
leave its host and enter the environment? (2) where in the environment does
the organism reside? (3) when do people acquire infection? and (4) are all
people susceptible to infection? While most epidemiological evidence supports
direct person-to-person transmission, the manner in which this occurs is unknown.2- 4
Helicobacter pylori is thought to reside normally
only in the stomach; thus, the organism is presumed to enter the environment
in feces, saliva, or vomitus. Helicobacter pylori
is a relatively fastidious organism, however, making its identification in
clinical specimens difficult. Experienced laboratories may recover H pylori from only 50% to 70% of infected gastric biopsies.5,6 From stool, saliva, and vomitus—which
can be heavily colonized by more robust organisms—recovery of H pylori is even more difficult.7- 10
Thus, many clinical studies have relied on polymerase chain reaction (PCR)
for H pylori identification.11,12
Unfortunately, PCR cannot distinguish between DNA from viable cells and nonviable
organisms. A new PCR-based method, immunomagnetic separation (IMS) with PCR,
may remedy this problem by preferentially amplifying DNA within intact cells.13- 15
In this study, using both culture and IMS PCR, we evaluated whether H pylori could be recovered from feces, vomitus, and saliva
of asymptomatic, infected adult volunteers. Since some studies suggest that H pylori is excreted only in diarrheal stools, we cultured
stools both before and after administration of a cathartic. In addition, we
sampled the air during episodes of vomiting. In this manner, we hoped to elucidate
how H pylori enters the environment to invade new
We recruited healthy volunteers by advertising on radio stations and
in business establishments, clinics, and churches. We preferentially recruited
in minority (black and Hispanic) populations known to have high prevalence
of H pylori in northern California.16
During 3 weeks of announcements, we received 379 inquiries and interviewed
132 potential participants; 103 were eligible to participate. Exclusion criteria
included age older than 55 years; pregnancy; history of ulcer disease, gastrointestinal
tract bleeding, or severe dyspepsia; routine use of cathartics; recent use
of antibiotics, histamine antagonists, or proton pump inhibitors; and prior
treatment for H pylori infection.
Of the 103 eligible subjects, 62 subjects elected to participate. After
providing written informed consent, each subject contributed a serum sample
for H pylori IgG. The first 15 of the 27 seropositive
subjects were invited to undergo 13C breath testing for H pylori (Meretek Corp, Nashville, Tenn), physical examination, blood
cell counts, blood chemistries, and stool occult blood testing. Two asymptomatic
subjects identified as H pylori IgG positive in a
previous study asked to participate and were also invited for breath testing
and physical examination. All 17 seropositive subjects were confirmed to be
infected with H pylori by breath test but 1 was excluded
from further participation because of anemia. The remaining 16 subjects (the H pylori–infected group) were invited to undergo
the clinical experiment described herein. Ten H pylori
serology– and breath test–negative volunteers were identified
(the H pylori–uninfected group). Subjects were
paid for their participation in the study.
Helicobacter pylori–infected subjects
were admitted to the general clinical research unit, where they were administered
45 mL of sodium phosphate solution in 90 mL of water, followed by 720 mL of
water. We chose sodium phosphate as the cathartic because it has a rapid time
of onset, acts on both the small and large bowels, and has been used previously
to facilitate diagnosis of gastrointestinal tract pathogens.17
We collected all stools during the 8 hours following cathartic administration
and immediately transported them to the laboratory.
After an overnight fast, infected subjects were administered 5 mL of
ipecac followed by at least 480 mL of water. Prior to emesis, we obtained
a saliva sample (the subject spit into a cup) and placed a Mattson-Gavin air
sampler 0.3 m away from the subject. For the duration of the emesis period,
we sampled air onto sheep blood trypticase soy agar plates with a fluoropore
filter centrally covering half the plate's diameter. We replaced the plate
every 30 minutes to prevent desiccation. For 10 subjects, we placed a second
air sampler 1.2 m away to determine the radius of bacterial aerosolization.
Samples were transported to the laboratory for processing within 10 minutes
of emesis. After vomiting had subsided, we collected a second sample of saliva.
The 10 uninfected control subjects provided normal stool and saliva
samples for analysis. One uninfected control also underwent the emesis portion
of the experiment.
The protocol was approved by the Stanford University Administrative
Panel on Human Subjects in Medical Research.
We diluted stool samples to a 20% suspension in phosphate-buffered saline
(PBS) and sieved the suspension through a 250-µm strainer. We plated
a 200-µL portion of the suspension on sheep blood trypticase soy agar
supplemented with polymyxin B (3.3 µg/mL), amphotericin B (50 µg/mL),
bacitracin (200 µg/mL), nalidixic acid (10.7 µg/mL), and vancomycin
(100 µg/mL). We cultured a second portion (1 mL) of the suspension using
the method described by Kelly and colleagues.8
Using these methods, we successfully recovered H pylori from inoculated stools at concentrations as low as 102
organisms/mL (American Type Culture Collection strain 43579).14
Sensitivity of H pylori detection varied, however,
depending on the strain of H pylori inoculated (range,
We neutralized vomitus specimens to a pH of 7.0 and diluted the specimen
with PBS at a 1:1 ratio; 200 µL was cultured on antibiotic plates as
described herein. Using these methods, we were able to detect H pylori inoculated into vomitus at concentrations as low as 10 organisms/mL,
depending on the strain of H pylori used.14 Ipecac (0.125 mL per milliliter of sample) decreased
sensitivity of H pylori detection 10-fold. Saliva
samples were diluted 1:1 in PBS and plated as described.
All plates were microaerophilically incubated at 37°C. Suspicious
colonies were confirmed as H pylori. When possible,
we estimated the number of colony-forming units (CFUs) of H pylori per milliliter of sample.
We bound purified polyclonal rabbit anti–H pylori IgG (Dako, Glostrup, Denmark) to magnetized polystyrene beads precoated
with sheep anti–rabbit IgG as described by the manufacturer (Dynal,
Oslo, Norway). We then mixed the stool, vomitus, and saliva suspensions with
30 µL (1.8 × 106 beads) of the coated beads for 1 hour
at 4°C. We separated the beads from the solution using a magnetic particle
concentrator, discarded the solution, and resuspended the beads in 1 mL of
PBS with 0.1% bovine serum albumin. After 3 such separations and washings,
1 portion of the separated bead-bacteria complex was resuspended in 30 µL
of sterile distilled water, boiled for 10 minutes to lyse the bacteria, briefly
chilled on ice, and frozen until analyzed by PCR. A second portion of the
bead bacteria complex was resuspended in 100 µL of PBS and cultured
as described herein. Because we found that IMS did not improve culture sensitivity,
we did not culture the bead-bacteria complex after the first 5 subjects.
Polymerase chain reaction of the IMS-separated bead-bacteria complex
was performed using primers specific to the H pylori
16s ribosomal RNA (rRNA) gene as previously described.18
A 139– base pair (bp) band on agarose gel electrophoresis indicated
the presence of H pylori in the sample. Negative
controls included sterile distilled water and immunomagnetic beads without
added samples. In inoculation experiments, IMS PCR detected 33 H pylori organisms/mL in stools and 3 organisms/mL in vomitus.14
To establish air sampling methods, we aerosolized H pylori (109 organisms/mL) in a biosafety hood while sampling
air at varying intake speeds onto plates and filters. Culture could detect
between 106 and 107 aerosolized organisms, with heaviest
growth occurring at a 1.7 m3/h intake speed. Filter strips were
placed in 100 µL of sterile water; half were then sonicated. Filters
then underwent 6 cycles of freeze-thaw lysis. We tested the filter solutions
for the 16s rRNA gene using PCR as described herein. Fluoropore filters (Millipore,
Bedford, Mass) consistently yielded H pylori without
sonication (sensitivity = 103 organisms) and were chosen for this
Isolates were confirmed as H pylori biochemically
(oxidase, urease, and catalase positive) and by morphology under light microscopy.
From each positive culture, we subcultured 1 colony and amplified the H pylori 16s rRNA gene as described herein.18
If the 16s rRNA gene amplified, we then fingerprinted the isolate using repetitive
extragenic palindromic (REP) PCR as previously described.19,20
Repetitive extragenic palindromic PCR fingerprints of isolates from vomitus,
stools, air, and saliva within and among subjects were compared.
To confirm species identity, isolates with unique REP PCR fingerprints
were sent in a blinded fashion to Midi Labs (Newark, Del) for sequencing of
the first 500 bp of the 16s rRNA gene.21 For
PCR-positive, culture-negative samples, the 139-bp 16s rRNA amplicon was sequenced
in our laboratory.
The mean age of the 16 infected subjects was 38.7 years (range, 22-53
years) and 9 (56.3%) were women. The mean age of the 10 uninfected subjects
was 39.0 years (range, 29-49 years) and 6 (60%) were women.
Stools collected prior to administration of cathartic from all 16 infected
and 10 uninfected subjects were negative by culture for H pylori. In 5 infected subjects but no uninfected subjects, IMS PCR
detected the H pylori 16s rRNA gene
From infected subjects, we collected 121 cathartic stools (mean per
subject, 7.6; range, 4-13), of which 115 were cultured; at least 4 samples
were cultured from each subject (Table 2). Cultures from 2 subjects were unevaluable due to fungal overgrowth;
cultures from 7 (50%) of the remaining 14 subjects yielded H pylori. Stools passed late in catharsis were more likely than early
stools to grow the organism. The amount of H pylori
shed in stool was quantifiable in 16 of 37 culture-positive stools from 5
subjects; the number of CFU/mL ranged from 5 to 2125. In 1 subject in whom
sequential stool cultures were quantified, the amount of H pylori appeared to increase in the later samples collected (500 CFU/mL
in sample 5, 725 CFU/mL in sample 6, and 2124 CFU/mL in sample 7).
By IMS PCR, 11 of 16 subjects had at least 1 cathartic stool positive
for H pylori. Stools excreted both early and late
during catharsis were equally likely to have the H pylori 16s rRNA gene fragment detected. Among the 5 subjects without H pylori DNA detected during catharsis, 2 had had H pylori DNA detected in stool prior to catharsis; an additional
subject with negative IMS PCR results had a positive stool culture. Thus,
stools from 14 (88%) of 16 subjects showed evidence of potentially viable H pylori. The 2 subjects without H pylori detected in their stools were those with contaminated culture plates.
We collected 85 vomitus samples from infected subjects (mean per subject,
5.3; range, 3-8). Five samples were contaminated with fungus and could not
be evaluated; the remaining 80 samples all grew H pylori. Cultures could be quantified from 38 samples representing 14 subjects.
The number of CFUs per specimen was high, with greater than 1000 CFU/mL of
vomitus in 31 samples and greater than 10,000 CFU/mL in 11 samples (range,
10-30,000 CFU/mL). Immunomagnetic separation PCR detected the H pylori 16s rRNA gene in all samples. The uninfected subject who was
administered ipecac vomited 3 times; all cultures and PCR assays from this
subject were negative for H pylori.
Saliva prior to emesis was positive for H pylori
by culture in 3 infected subjects (18.8%) and by IMS PCR in 7 infected subjects
(43.8%), including the 3 subjects with positive cultures. A half hour after
termination of emesis, saliva cultures were positive for H pylori from 9 infected subjects (56.3%) and IMS PCR results were
positive in 8 infected subjects (including 7 of the subjects with positive
cultures). Quantities of H pylori in postemesis saliva
tended to be low (4 quantified cultures had counts ranging from 50-500 CFU/mL).
Saliva samples from the uninfected controls were all negative for H pylori both by culture and by IMS PCR. Saliva from 1 uninfected subject
who underwent emesis was again negative for the organism after vomiting.
Air sampled prior to onset of vomiting did not yield H pylori by culture or PCR. After onset of vomiting, air sampled 0.3
m away from 6 subjects grew H pylori. In 5 of these
6 instances, the positive culture coincided with the first episode of vomiting;
in the sixth instance, the positive culture coincided with the fifth bout
of vomiting. In 2 of the air culture–positive cases, the filter was
also positive by PCR. Filters were additionally positive from 2 cases with
negative air cultures. No sample obtained 1.2 m from the subject yielded H pylori.
At least 1 H pylori isolate was available for
fingerprinting from 14 of the 16 subjects; 2 subjects had isolates available
from all 4 types of samples (stool, vomit, air, and saliva), 4 from 3 types,
4 from 2 types, and 4 from only 1 type. Fingerprints differed among subjects,
including between a wife-husband pair (TR26 and TR31) but were identical within
subjects (Figure 1).
All 14 unique REP PCR isolates obtained from the 14 subjects were confirmed
as H pylori by sequencing of 500 bp of the 16s rRNA
gene with no more than 0.82% difference from the reference strain. For IMS
PCR–positive, culture-negative samples, the 16s rRNA amplicon sequence
was consistent with H pylori in 14 of 20 samples (Table 1). For the remaining 6 IMS PCR–positive,
culture-negative samples, we were unable to reamplify sufficient DNA to perform
In this study, we found that H pylori can be
cultured from both vomitus and stools of healthy H pylori–infected persons. Helicobacter pylori was
often present in high quantities in vomitus, with as many as 30,000 CFU/mL
of sample. Since the sensitivity of our vomitus culture was between 0.1% and
1%, we estimate that more than 106 organisms may be present in
each milliliter vomited. Thus, emesis could be a potent mechanism for discharging
millions of H pylori into the environment. Although
patterns of disease transmission by vomiting have not been systematically
studied, one would expect risk factors for transmission by vomitus to be similar
to those documented for H pylori, eg, close living
quarters, many siblings, and poor household sanitation and hygiene.23,24 The few documented cases of acute H pylori infection support gastric-oral transmission. Mitchell
and colleagues25 reported acute H pylori infection in 1-year-old twins 3 weeks following a sustained
vomiting illness in their H pylori–infected
mother. A second acute infection was reported in a researcher who routinely
processed gastric juice.26 Possible gastric-oral
transmission of H pylori was also reported following
mouth-to-mouth resuscitation of an infected person who had vomited.27
The process of vomiting also dispersed H pylori
into the air. Although other gastrointestinal tract pathogens, notably the
small, round, structured Norwalk-like viruses, can be transmitted by aerosol
during episodes of vomiting, we doubt this is a common mode for H pylori transmission.28,29
The short duration of contaminated aerosol (in the first minutes of the first
episode of vomiting) and the limited dispersion of organisms (less than 1.2
m) makes aerosol exposure unlikely.
Helicobacter pylori DNA has frequently been
amplified from both saliva and dental plaque,10,11,30
but only rarely has H pylori been cultured from the
recovered H pylori from saliva before emesis in 19%
of subjects and after emesis in 50% of subjects. To date, there is little
epidemiological data to support oral-oral transmission. Dental workers have
similar prevalence of H pylori as the average population.32 Most married couples demonstrate little concordance
of infection or strain type33,34
and treated patients are not reinfected by their untreated infected spouses.35 Thus, it remains to be seen whether organisms in
the mouth, which were typically present in low quantities compared with in
vomitus, represent a significant source of transmission.
Helicobacter pylori was less reliably cultured
from stools than from vomitus. This may, in part, be due to the lower sensitivity
of stool culture. In 50% of subjects, however, H pylori could be cultured from feces in the setting of rapid gastrointestinal
It can be argued that cathartic-induced diarrhea and emetic-induced
vomiting do not mimic gastroenteritis. It is likely that H pylori needs to be rapidly excreted from the proximal gastrointestinal
tract to be found viable in stools. Indeed, the lack of regulatory genes in H pylori implies that the organism cannot survive for long
periods outside its normal environment.36 Yet,
only pathogens involving the small bowel induce the watery diarrhea seen with
sodium phosphate administration. This being the case, colitic forms of gastroenteritis
and gastroenteritis with relatively slower intestinal transit may not transmit H pylori. Vomitus, on the other hand, was so uniformly
contaminated with high amounts of H pylori that it
is difficult to envision circumstances in which it would not be infectious.
This also raises the question of whether persons with chronic gastric regurgitation
or frequent vomiting from other medical conditions are high-risk H pylori transmitters.
In this study, we evaluated only healthy asymptomatic adults and found
that viable H pylori was excreted into the environment
by all infected subjects. Given the large number of infected hosts worldwide,
it is remarkable that so many remain uninfected. Barriers to acquiring infection—both
intrinsic to the host (eg, high gastric acidity, good nutrition) and extrinsic
to the host (eg, household and public sanitation and personal hygiene)—may
account for this phenomenon. We postulate that a declining incidence of gastroenteritis
that occurs as countries make the transition from developing to developed
may also contribute to the observed decline in H pylori infection in industrialized nations.37,38
Epidemiological investigations within households of persons with gastrointestinal
tract illness may provide important clues to understanding and controlling H pylori transmission.