REVIEWS OF INFECTIOUS DISEASES • VOL. 11, SUPPLEMENT 1 • JANUARY-FEBRUARY 1989
© 1989 by The University of Chicago. All rights reserved. 0l62-0886/89/11OH)()51$02.00
Vaginal Tampon Model for Toxic Shock Syndrome
Marian E. Melish, Susanna Murata,
Cynthia Fukunaga, Kanitha Frogner,
and Carol McKissick
From the Department of Pediatrics, John A. Burns School
of Medicine, University of Hawaii, Kapiolani Medical Center
for Women and Children, Honolulu, Hawaii
Recognition of the association of toxic shock syndrome (TSS), menses, and tampon use came early
in the experience with this syndrome [1, 2] and was
quickly confirmed [2, 3]. Case-control studies demonstrated a significantly increased risk of TSS associated with use of tampons as compared with no
use of tampons [3, 4]. The association of one tampon brand of a unique chemical composition with
TSS was significantly higher than the association of
all other tampon brands [4-6]. Epidemiologic studies
have demonstrated a direct association of risk for
TSS with the fluid absorbency of the tampon, as
measured by an in vitro technique [4]. Many studies
of the effect of tampons on the production of toxic
shock syndrome toxin 1 (TSST-l) in culture have been
performed [7-13]. Animal models that reproduce the
major physiologic changes of TSS have been developed [14-19]. In these models, either staphylococcal extracellular products or purified TSST-l are alThis work was supported by a grant from Playtex Family Products, Inc.
Please address requests for reprints to Dr. Marian E. Melish,
Department of Pediatrics, John A. Burns School of Medicine,
University of Hawaii, Kapiolani Medical Center, 1319 Punahou
Street, Honolulu, Hawaii 96826.
S238
lowed to diffuse from an internal focus or depot such
as an infection chamber, osmotic pump, or subcutaneous tampon. We have developed a rabbit model
that allows us to evaluate the multiplication of
staphylococci, the production of TSST-l, and the development of clinical TSS-like illness following sequential insertion of vaginal tampons in a pattern
that simulates human tampon use during menstruation.
Materials and Methods
Staphylococci. Staphylococcus aureus strain
FRI1169, which is tyrosine-requiring and TSST-lsecreting, and strain S411, which is a TSST-l-positive
tryptophan auxotroph, were the TSS-associated
strains employed. Two TSST-l-negative, coagulasepositive S. aureus strains, S256 and S253 (which had
been isolated from children with life-threatening bacteremia) were used as control, non-TSS strains.
Radioimmunoassay (RIA) for TSST-J. TSST-l
was quantitated by a double-antibody RIA. Purified TSST-l was labeled with 125 1with use of chloramine-T [20]. 125I-Labeled TSST-l was diluted to a final concentration of 1 ng/mL. A competitive-binding,
double-antibody, liquid-phase RIA was performed
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
The effects of tampon composition, inoculum size, and simulated menses on production
of toxic shock syndrome toxin 1 (TSST-I) and toxic shock syndrome (TSS) were evaluated
in a rabbit model that simulates tampon use in humans. Three small generic compressedfiber tampons were successively inserted vaginally (remained in place 4.5 hours X 2; overnight x I). Tampon no. 1 was inoculated with live TSST-l-positive staphylococci plus
5 mL of saline or simulated menses (defibrinated rabbit blood plus 2.5 g of bovine serum
albumin/dL) immediately after insertion; saline or simulated menses alone were used with
tampons no. 2 and 3. The vagina was washed after removal of tampon no. 3. TSS-like
illness was produced consistently in animals with carboxymethyl cellulose/polyester foam
tampons, which supported higher organism counts and greater TSST-I production in association with subsequent tampons. Cotton and rayon tampons were not associated with
as much clinical illness, organism growth, or TSST-I production. Simulated menses supported toxin production and clinical illness when the inoculum was one-tenth that required for controls. Sham tampon insertion was associated with TSS-like illness in two
of 10 rabbits; thus, other factors may promote TSS in the absence of vaginal tampons.
This model reliably reproduces menstrual TSS, since one-time vaginal inoculation with
TSST-l-positive staphylococci in the presence of blood and certain tampons leads to TSS,
and may be useful in evaluating catamenial products and in understanding other factors
important in TSST-I production in vivo and the development of TSS.
Vaginal Tampon Model for TSS
White rabbits >7 months of age weighing 4-5.5 kg
were purchased from a local supplier, housed separately in the research animal facility, and fed water
and rabbit chow ad libitum.
Experimental procedure. Rabbits were anesthetized with ketamine (35 mg/kg) and xylazine (6 mg/
kg) and were occasionally given valium (0.5 mg)
when necessary to increase relaxation. Blood was obtained for baseline studies via the medial ear artery.
An individually preweighed tampon was inserted
through the urogenital sinus into the vagina above
the urethra by means of a modified plastic applicator (figure 1); 5 mL of fluid with or without 1 mL
of staphylococcal suspension was inoculated in the
vagina just below the tampon by a pipette inserted
through the applicator. A string for tampon removal
remained at the vulvar opening. After 4.5 hours, the
animal was again anesthetized, the tampon was re-
Figure 1. Dissection of rabbit urogenital sinus, bladder,
vagina, and bicornuate uterus, together with tampon applicator. The applicator is inserted above the bladder, and
a 1.5-g tampon is extruded into the vagina. Fluid with or
without staphylococci is then inoculated through the applicator.
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
as follows: 0.05 ng of 125I-Iabeled TSST-l was incubated with a suitable dilution of human TSST-l
antiserum (1:1,000) and serial twofold dilutions of
unknown sample at room temperature for 48 hours.
The mixture was then incubated with normal human
serum (1:180) and the second antibody- goat antihuman IgG - at room temperature for 24 hours.
Tubes were centrifuged at 7,000 rpm for 15 minutes,
the supernatant was decanted, and the tubes were
wiped above the precipitate with a cotton swab. The
radioactivity of the precipitate was counted in a
Beckman autogamma counter (Fullerton, Calif.).
For each assay a standard curve was prepared with
use of duplicate serial twofold dilutions (5 ng-0.04
ng/mL) of unlabeled, highly purified TSST-l. The
amount of TSST-l present in the sample was calculated from the standard curve. The minimum quantity of TSST-l detectable by this assay is 0.6 ng/mL.
Interassay variation on the same samples tested
repeatedly over a wide range of TSST-l concentrations is <7CtJo.
Tampons. Compressed-fiber tampons weighing
1.5 g composed of either combed cotton or nonenhanced viscose rayon were produced for this study
by Playtex Family Products, Inc. (Paramus, N.J.).
These generic tampons are of a different design than
commercially available tampons manufactured by
Playtex Family Products. The mean in vitro absorbency of the tampons was measured by standard
procedures in a syngyna device. The mean absorbency of cotton tampons was 5.67 g of fluid (range,
5.38-6.21 g) and that of rayon tampons was 5.53 g
(range, 5.13-5.96 g). Carboxymethyl cellulose/polyester foam (CMC/foam) tampons were constructed
in our laboratory by sorting CMC/foam chips from
a commercially purchased CMC/foam tampon and
reassembling a smaller, 1.4 g, tampon in the original wrap using the same proportions of CMC and
foam chips as in the original. Syngyna absorbency
was 6.04 g (range, 5.47-6.56 g). All tampons were
placed within "regular"-size plastic inserters used
with commercially available tampons.
Simulated menses. Simulated menses consists of
defibrinated rabbit blood mixed with 2.5 g of bovine serum albumin /dL (Sigma, St. Louis, Mo.) to
yield a final protein concentration of 8.5 g/dL (albumin concentration, 3.5 g/dL; hemoglobin concentration, 9-11 g/dL) [21]. Blood was taken from
animals with no detectable antibody to TSST-l as
determined by a radioimmunobinding assay.
Animals. Sexually mature female New Zealand
S239
Melish et al.
S240
Results
Effects ofrepeated anesthesia and tampon insertion. Six animals were subjected to repeated insertion of a cotton tampon with 5 mL of simulated menses but without inoculation of staphylococci. These
animals all remained well, were afebrile, and did not
develop the clinical (conjunctivitis, diarrhea) or behavioral (hunched posture, inactivity) signs associated with model TSS [14-19]. Minor elevations in levels of BUN and a depression in levels of calcium were
noted (figure 2). No animal had S. aureus recovered
from the vagina or had TSST-l detected in plasma,
removed tampons, or vaginal washes. In subsequent
results, individual animals were considered to have
clinical abnormalities consistent with TSS only if the
BUN and creatinine values were >2 SD above the
mean values for control animals and if calcium
values were >2 SD below the mean values.
17
16
15
::~,~':"':":"'._"'_''_"'_'_ J
11
10
9
60
..J
o
........
Cl
:2
z
:::»
CD
50
: ....,,~I·~t---.
",
:
20 ;'
' .. '. ". '"
"
,
f'-
'L~""",:
1.
72----9....6---1~20
10 OL---2....4 ---48'O"'-----....
HOUR
Figure 2. Concentrations of calcium and blood urea
nitrogen (BUN) among two control groups of rabbits. The
pairs of dotted lines outline concentrations (± 95070 confidence intervals) among six animals that underwent
repeated tampon insertion and received simulated menses but no organisms. The large closed circles identify the
mean concentrations (± 95% confidence intervals) among
10 animals that underwent sham tampon insertion and received simulated menses and 109 cfu of Staphylococcus
aureus strain FRIll69 (toxic shock syndrome toxin
I-positive).
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
moved by gentle traction upon the string, and a second tampon was inserted. Another 5 mL of sterile
fluid, without staphylococci, was inoculated. Tampon no. 2 was allowed to remain in place for 4.5 hours
and was then removed under anesthesia as before.
Tampon no. 3 was inserted and then another 5 mL
of sterile fluid. Tampon no. 3 was allowed to remain
in the vagina for 14 hours to simulate overnight usage. After the final tampon was removed, the vagina
was irrigated four times with 5 mL of normal saline. The lavage fluid was collected and pooled for
subsequent quantitation of TSST-l ("vaginal wash").
The animal was then observed frequently but underwent no further intervention. Blood was obtained
for quantitation ofTSST-l and hematologic and clin-·
ical chemistry testing at times 0, 4, 8, 24, 48, and
72 hours, and 5 days from the onset of the experiment.
Upon removal tampons were weighed to determine
the quantity of fluid absorbed and were placed in
a sterile 35-mL syringe that had been modified by
drilling additional holes near the tip. Fluid was expressed by compression of the tampon, and the concentration of TSST-l, the pH of the fluid, and the
number (cfu) of staphylococci were determined by
RIA, pH meter, and plating of serial dilutions on
mannitol salt agar (BBL Microbiology Systems,
Cockeysville, Md.), respectively. In initial experiments, cotton tampons were repeatedly immersed in
10 mL of saline after initial expression of fluid and
were compressed to remove fluid for 5 cycles. The
fluid extracted from these tampon washes was analyzed for TSST-l concentrations. Small additional
quantities of TSST-l were recovered in the initial
wash, decreasing to undetectable amounts «0.6 ng/
mL) by wash no. 5. Because the total amount of
TSST-l recovered from each tampon by repetitive
washing was negligible, we assumed that we could
calculate the total amount of TSST-l per tampon
with reasonable accuracy by multiplying the TSST-l
concentration (in ng/mL) in the fluid extracted in
the original compression step by the increase in
weight (or gram of fluid) of the tampon during insertion.
Statistical methods. For the comparison of clinical chemistry responses, the mean values for blood
urea nitrogen (BUN), creatinine, and calcium (in
mg/dL) per group were calculated together with the
95070 confidence intervals. The comparison of the
total number of organisms (cfu) and total amount
of TSST-l per tampon was made by one-way analysis of variance (ANOVA).
Vaginal Tampon Model for TSS
Table 1. Effect of saline or simulated menses on production of toxic shock syndrome toxin 1 (TSST-l) and
the development of toxic shock syndrome-like illness in
rabbits with cotton tampons infected with Staphylococcus
aureus strain FRI 1169.
No. (cfu) of
strain FRlll69
(no. of tested
animals)
No.of animals
affected (010)
Mean TSST-l
(ng/tampon) in
indicated tampon
TSS illness Mortality
2
3
48
16
631
196
826 59
2,293 187
770
18
382
741
682
Saline
1011 (6)
10 10 (12)
5 (83)
5 (42)
3 (50)
0
Simulated menses
10 10 (4)
109 (9)
108 (5)
3 (75)
5 (55)
3 (60)
1 (25)
4 (44)
1 (20)
469
190
sion of calcium level (9.6 mg/dL), and moderate elevation of triglyceride level (378 mg/dL) (figure 2).
TSST-l was present in vaginal washes at 24 hours in
four of 10 animals in amounts ranging from 6 ng
to 269 ng.
Effect ofsimulated menses. The effect of simulated menses on clinical signs of TSS and on TSST-l
production was studied with use of cotton tampons
and various concentrations of FRI1169 staphylococci
(table I). The presence of simulated menses was consistently associated with a higher frequency of TSSlike illness and with the production of greater amounts
of TSST-l with each concentration of staphylococci.
Effect oftampon composition. Studies comparing tampon composition utilized an inoculum of 109
cfu of FRI1169 and simulated menses. All animals
that received CMC/foam tampons developed clinical and chemical signs of TSS;one of five died. Five
of nine animals that received cotton tampons developed chemical dysfunction; four died. One of six
animals that received rayon tampons developed a
TSS-like illness and died. The mean BUN values for
the FRI1169, simulated menses, CMC/foam tampon
group differed significantly from those for the control group (no organisms, simulated menses, cotton
tampons) at 8, 24, 48, and 72 hours, with significantly lower calcium values at 24 and 48 hours (figure 3). The mean BUN and creatinine values for the
FRI1169, simulated menses, cotton tampon group
significantly exceded those for the no organism,
simulated menses, cotton tampon group at 24 hours
only; mean calcium values were significantly depressed at 24 hours only. Mean BUN values for the
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
Clinical effects of TSST-l-negative staphylococci.
The effects of two TSST-l-positive, TSS-associated
strains, FRI1169 and S411, and two TSST-l-negative,
non-TSS strains were compared with the use of cotton tampons. Staphylococci were inoculated along
with 5mL of simulated menses immediately after insertion of tampon no. 1; only simulated menses was
inoculated after insertion of tampons no. 2 and no.
3. Four of six animals inoculated with 109 cfu of
strain S411 and five of nine inoculated with 109 cfu
of FRI1169 developed a TSS-like illness, as determined by clinical and clinical chemistry criteria. (Fever, conjunctival injection, diarrhea, and progressive lethargy were present; BUN and creatinine values
were >2 SD above the mean and calcium values >2
SD below the mean for animals subjected to anesthesia and repeated placement of tampons and simulated menses without organisms.) Two of six animals
given S411 and four of nine animals given FRI1169
died. One of six animals inoculated with S411 had
TSST-I detected in plasma (3.8 ng/mL at 24 hours)
and two of nine inoculated with FRI1169 had TSST-l
detected in plasma (1.3 ng/mL at 4 and 24 hours in
one rabbit and 2.0 ng/mL at 24 hours [time of death]
in another). All had TSST-I detected in tampon fluid
and vaginal washes. None of nine animals inoculated
with 109 cfu of either strain S256 (six) or S253 (three)
developed fever or any clinical signs of illness, and
all had BUN, creatinine, and calcium values within
2 SD of control values. No TSST-I was detected in
the plasma, tampon fluids, or vaginal washes from
these animals, although their vaginal cultures were
persistently positive for staphylococci. Mean BUN,
creatinine, and calcium values at 24 hours for animals
receiving TSS-associated staphylococci exceeded the
95 % confidence intervals around the means found
in animals inoculated with non-TSS staphylococci.
Effects ofsham insertion oftampons, TSS staphy10COCCl~ and menses. Ten animals were inoculated
with 109 cfu of TSS-associated staphylococci
(FRI1169) vaginally along with 5 mL of simulated
menses after insertion of an empty tampon applicator, which was removed after inoculation (sham insertion). Sham insertion and placement of 5 mL of
simulated menses was performed after 4 hours and
8 hours, but no tampons were left in place. Two of
these 10 animals developed clinical and chemical abnormalities compatible with TSS; one died at 24
hours with a rapidly progressive TSS-like illness associated with marked elevation in BUN (112 mg/dL)
and creatinine (4.0 mg/dL) levels, moderate depres-
S241
S242
Melish et af.
17
16
~
0
........
()
2
2
:::>
~
...................
15
14
13
12
11
10
9
60
~
0
~
2
50
I
I
I
40
CD
,
I
Z
:::>
/1'- ---1'"'" ....
I
30
•
'J.
Discussion
'.
20
10
..
a
24
,
.
....
72
48
96
120
HOUR
Figure 3. Concentrations of calcium and blood urea
nitrogen (BUN) among two groups of animals that underwent repeated insertions of carboxymethyl cellulose/polyester foam (.A, n = 5) and rayon tampons (_,
n = 6), respectively, plus simulated menses and 109 cfu
of Staphylococcus aureus strain FRIll69 (toxic shock syndrome toxin I-positive) compared with control (cotton
tampon, simulated menses, no organisms). The pairs of
dotted lines outline concentrations (± 95070 confidence
intervals) for control an.imals; other values are the mean
concentrations (± 95% confidence intervals).
FRIl169, simulated menses, rayon tampon group and
those for the no organism, simulated menses, cotton tampon controls did not differ significantly at
any time point (figure 3). The total amount of TSST-1
in each successive tampon, the concentration of organisms in each tampon, and the amount of TSST-1
recovered by vaginal washing after removal of tampon no. 3, with an indication of which animals developed TSS and which died, are presented in table
2. Mean concentrations of TSST-1 and organisms in
tampon no. 1 did not differ significantly in any of
the three tampon groups. In tampons no. 2 and no.
3, the mean TSST-1 and bacterial concentrations were
significantly higher in CMC/foam tampons than in
either cotton or rayon tampons. In tampon no. 3
We have attempted to create an animal model for
menstrual TSS. In this model TSS-associated staphylococcal strains produce adequate amounts ofTSST-1
to cause a TSS-like illness when they are placed in
the rabbit vagina in the presence of simulated menstrual fluid and repetitively inserted tampons. Features of this model similar to those of menstrual TSS
are that (1) illness is more likely to occur in the presence of tampons, (2) illness occurs after tampons are
repetitively inserted in a pattern that simulates daytime and nighttime tampon use, and (3) the magnitude of the inoculum capable of providing TSS-like
illness is similar to some of the quantitative measurements of staphylococci found in the vagina of
normal women [22, 23]. It is likely that many local
conditions found in menstrual TSS, including pH,
O 2 and CO2 , and availability of nutrients, are closely
approximated in this model.
Features of this model that differ from those in
the human situation are (1) the hormonal milieu of
the rabbit, which is unlike that in a woman at menses; (2) the tampons used in these experiments,
which, although easily accommodated, are relatively
large in relation to the diameter of the rabbit vagina
and the size of the rabbit; and (3) the anatomy of
the urogenital tract of the rabbit. The TSS-like illness induced in this model is essentially similar to
that in other rabbit models of TSS that involve focal infection or toxin depot placement [14-19]. Rabbits that developed TSS-like illness had a stereotypical pattern of illness, including fever, conjunctival
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
70
mean TSST-1 and bacterial concentrations were significantly higher in cotton than in rayon tampons.
Clinical TSS-like illness was correlated with both the
total amount of TSST-1 in tampon no. 3 and the bacterial concentration in tampon no. 3. Ten of the 11
animals with ~500 ng of TSST-1 in tampon no. 3
developed TSS-like illness, while only one of nine
with <500 ng of TSST-1 in tampon no. 3 became ill.
All 12 animals with ~10S cfu of S. aureus in tampon
no. 3 became ill; none of eight with <105 cfu developed TSS-like illness. Blood cultures were repeatedly
negative for all animals that had vaginal placement
of staphylococci and tampons. Free TSST-1 at concentrations of 0.6-3.4 ng/mL was detected intermittently in the blood of two of five rabbits with CMC/
foam tampons and in two of nine rabbits with cotton tampons.
Vaginal Tampon Model for TSS
S243
Table 2. Quantities of toxic shock syndrome toxin 1 (TSST-1) and concentrations of Staphylococcus aureus in
sequentially inserted tampons of different composition in rabbits also inoculated vaginally with simulated menses
and 109 cfu of S. aureus strain FRI1169 (TSST-l-positive).
No. (cfu/mL)
of S. aureus in
indicated tampon
Total TSST-l (ng)
Tampon no.
Tampon composition,
animal no.
2
3
Vaginal wash
2
3
1,439
4,795
2,546
1,454
2,078
2,462
(1,249-3,675)
2,924
569
1,974
211
50
1,049
(47-2,051)
16,255
6,139
5,470
3,394
2,010
6,654
(1,733-11,575)
76
7,423
1,869
7,839
373
3,516
(167-6,864)
108
108
109
109
108
109
108
108
108
1()6
108
107
108
108
107
Cottont
1
2
3
4
5
6
7
8
9
Mean (95% confidence
interval)
700
883
1,010
4,479
2,906
3,448
3,985
3,071
156
2,293
(1,242-3,344)
1,246
100
20
21
96
63
27
24
92
188
(0-448)
2,148
1,098
1,024
841
832
533
146
33
14
741
(301-1,181)
770
620
359
77
512
92
75
49
86
293
(111-475)
108
108
108
107
108
106
107
107
107
lOS
106
106
106
106
10"
10"
10"
108
107
107
105
105
105
10"
1()2
10"
Rayont
1
2
3
4
5
6
Mean (95 % confidence
interval)
925
2,785
1,043
589
651
310
1,050
(340-1,760)
580
61
<1
<1
16
<1
110
(0-295)
49
<1
<1
<1
<1
<1
8
(0-24)
23
11
9
<1
<1
<1
7
(0-14)
107
107
107
108
107
107
107
105
106
103
104
103
105
103
103
102
10"
10 1
10"
* All animals developed toxic shock syndrome (TSS)-like illness. Rabbit no. 1 died on day 3; all other animals survived. For
tampons 1,2, and 3, the mean pH of fluid was 7.3, 7.3, and 6.9, respectively, and the mean amount of fluid (mL/g of tampon)
was 5.3, 4.2, and 4.2, respectively. CMC/foam = carboxymethylcellulose/polyester foam.
t Animals no. 1, 2, 3, and 5 developed TSS-like illness and died <21 hours, <47 hours, <45 hours, and on day 4 after inoculation,
respectively; all other animals survived. For tampons 1, 2, and 3, the mean pH of fluid was 7.8, 7.8, and 7.4, respectively, and
the mean amount of fluid (mL/g of tampon) was 6.7, 5.7, and 5.2, respectively.
t Animal no. 1 developed TSS-like illness and died on day 4; all other animals survived. For tampons I, 2, and 3, the mean
pH was 7.6, 7.6, and 7.5, respectively, and the mean amount of fluid (mL/g of tampon) was 5.8, 5.0, and 5.1, respectively.
injection, hunched posture, progressive inactivity,
substantial elevation in BUN and creatinine values,
lipemia with elevated triglyceride levels, marked
hypocalcemia occasionally associated with tetany,
and marked hypoalbuminemia.
Blood cultures were negative, even in animals with
fatal illness. In some animals with TSS-like illness,
free TSST-l was present in the low concentrations
(1-10 ng/mL) typically found in human TSS and in
other TSS animals models [19]. Although some
animals developed visible erythema on the ears and
in the skin underlying the sparse hairs at the nape
of the neck, we considered this observation too subjective for scoring. Most animals developed leukocytosis with lymphopenia; we have also found this feature in animals injected with non-TSS-associated,
TSST-l-negative staphylococci and therefore do not
consider it a specific finding. Because most animals
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
CMC/foam*
1
2
3
4
5
Mean (95070 confidence
interval)
5244
of the absorbent contraceptive sponge [27]. Further
experiments are planned to assess the role of vaginal oxygen tension, mucosal trauma, and presence
of nutrients in the production of TSST-l in this
model. We may be able to define a threshold level
of conditions necessary for the production of TSSlike illness. Certain tampons may be found to augment the likelihood of TSS development, while
others may be found to have no additional effect or
even to retard the development of TSS when the basic threshold conditions are present.
These preliminary experiments suggest that the
chemical composition or factors related to chemical composition of tampons have an effect on the
production of clinical TSS in this in vivo model and
on intravaginal staphylococcal growth and TSST-l
production. Generic CMC/foam tampons, similar
to previously available commercial tampons in composition and construction, were associated with a
TSS-like illness in all animals tested and produced
mean concentrations of TSST-l and bacteria in tampons no. 2 and no. 3 and in the final vaginal washings that were significantly higher than those found
with either cotton or rayon tampons. This finding
is in agreement with data from multiple case-control
epidemiologic studies, which demonstrated that the
risk of developing TSS was higher when tampons
of similar composition were used than when any
other tampons were used [4, 5, 28, 29].
Although tampon absorbency has been correlated
positively with risk of TSS in epidemiologic studies
[4], it is not clear whether absorbency itself is the
critical factor or whether absorbency is related to
other important factors not yet fully evaluated. For
example, absorbency may be directly related to O 2
content of tampons. The role of O 2 on TSST-l production in this model is obvious interest for three
major reasons. (1) Repeated in vitro observations
have demonstrated the necessity of oxygen for TSST-l
production during staphylococcal growth. (2) Wagner has demonstrated that a dramatic effect of the
placement of tampons in menstruating women is the
transformation of the anaerobic atmosphere of the
vagina to one that has an aerobic P0 2 for prolonged
periods (up to 8 hours) [25]. (3) Todd has demonstrated aerobic conditions in focal abscesses in patients with TSS [30]. In our studies, in vitro absorbency of two tampons and in vivo absorbency in all
three tampon types tested were similar despite significant differences in TSST-l production. In addition to fiber type, various types of treatment that
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
in this series did not become ill or recovered from
TSS illness after the tampon was removed and the
concentrations of bacteria and toxin were reduced
by vaginal lavage, we included in our definition of
TSS-like illness the requirement that BUN and creatinine levels be greater and calcium levels be lower
by 2 SD than the values found in similarly anesthetized and manipulated control animals. No animals
that received one of the two non-TSS-associated
TSST-l-negative staphylococcal strains in the vaginal model developed azotemia or hypocalcemia, the
two most reliable objective indicators of experimental TSS in rabbits in our experience. These experiments demonstrate that TSST-1 can be absorbed
from the vagina, a finding confirming those of Kush- '
naryov et al. [24], who found kinetic and electronmicroscopic evidence of receptor-mediated transport
a of TSST-1 in vaginal epithelial cells. De Azavedo
et al. [16] also demonstrated the production of a TSSlike illness in response to TSST-l absorbed from a
genital tract focus.
To date, the study of the effect of tampons on
TSST-l production has been performed under in
vitro conditions, with microbiologic nutrient media,
and using incubation periods that are considerably
longer than the average daytime or nighttime continuous use of a tampon. Variable and sometimes
contradictory results have been reported [7, 8, 11-13].
Depending upon their design, in vitro experiments
can give insight into one or another facet of the factors that affect bacterial growth, toxin production
or dissemination, and disease production. None of
these systems can give a complete picture of the complex interactions that occur in menstruating women.
In vitro growth conditions differ strikingly, particularly with regard to the availability of nutrients, local fluid dynamics and tissue pressure; and P0 2 ,
Peo 2 , and pH.
Our observation that sham insertion may also be
associated with a TSS-like illness is most interesting
and suggests that factors such as admittance of O 2
into the usually anaerobic vaginal environment during tampon insertion [25], possible minor mucosal
alterations associated with tampon insertion [26],
and the presence of menstrual fluid as a nutrient
source are important factors that can lead to the
production of TSST-1 in the absence of absorbent
material. This also is of interest because of the recent demonstration that a nonabsorbent vaginal foreign body - a contraceptive diaphragm - has an
epidemiologic association with TSS as strong as that
Melish et al.
Vaginal Tampon Model for TSS
References
1. Todd J, Fishaut M, Kapral F, Welch T. Toxic shock syndrome
associated with phage group I staphylococci. Lancet
1978;2:1116-8
2. Davis JP, Chesney PJ, Wand PJ, LaVenture M, the Investigation and Laboratory Team. Toxic shock syndrome:
epidemiologic features, recurrence risk factors and prevention. N Engl J Med 1980;303:1429-35
3. Shands KN, Schmid GP, Dan BB, Blum D, Guidotti RJ, Hargrett NT, Anderson RL, Hill DL, Broome CV, Band JD,
Fraser DW. Toxic shock syndrome in menstruating women:
association with tampon use and Staphylococcus aureus
and the clinical features in 52 cases. N Engl J Med 1980;
303:1436-42
4. Osterholm MT, Davis JP, Gibson RW, Mandel JS, Wintermeyer LA, Helms CM, Forfang JC, Rondeau J, Vergeront JM, the Investigation Team. Tri-State Toxic-Shock Syndrome Study. I. Epidemiologic findings. J Infect Dis
1982;145:431-40
5. Schlech WF, Shands KN, Reingold AL, Dan BB, Schmid GP,
Hargrett NT, Hightower A, Herwaldt LA, Neill MA, Band
JD, Bennett JV. Risk factors for development of toxic shock
syndrome: association with a tampon brand. JAMA
1982;248:835-9
6. Kehrberg MW, Latham RH, Haslam BT, Hightower A, Tanner M, Jacobson A, Barbour AG, Noble V, Smith CB. Risk
factors for staphylococcal toxic shock syndrome. Am J
Epidemiol 1981;114:873-9
7. Schlievert PM, Blomster DA, Kelly JA. Toxic shock syndrome
Staphylococcus aureus: effect of tampons on toxic shock
syndrome toxin 1 production. Obstet Gynecol 1984;64:
666-71
8. Tierno PM, Hanna BA. In vitro amplification of toxic shock
syndrome toxin-l by intravaginal devices. Contraception
1985;31:185-94
9. Mills JT, Parsonnet J, Tsai YC, Kendrick M, Hickman RK,
Kass EH. Control of production of toxic-shock-syndrome
toxin-l (TSST-l) by magnesium ion. J Infect Dis 1985;
151:1158-61
10. Schlievert PM. Effect of magnesium on production of toxic
shock-syndrome toxin-l by Staphylococcus aureus. J Infect Dis 1985;152:618-20
11. Lee AC, Crass BA, Bergdoll MS. Investigation by syringe
method of effect of tampons on production in vitro of toxic
shock syndrome toxin 1 by Staphylococcus aureus. J Clin
Microbiol 1987;25:87-90
12. Reiser RF, Hinzman SJ, Bergdoll MS. Production of toxic
shock syndrome toxin 1 by Staphylococcus aureus restricted
by endogenous air in tampons.· J Clin Microbiol 1987;
25:1450-2
13. Robbins RN, Reiser RF, Hehl GL, Bergdoll MS. Production
of toxic shock syndrome toxin 1 by Staphylococcus aureus
as determined by tampon disk-membrane-agar method.
J Clin Microbiol 1987;25:1446-9
14. Scott DF, Kling JM, Kirkland 11, Best GK. Characterization
of Staphylococcus aureus isolates from patients with toxic
shock syndrome using polyethylene infection chambers in
rabbits. Infect Immun 1983;39:383-7
15. Arko RJ, Rasheed JK, Broome CV, Chandler FW, Paris AL.
A rabbit model of toxic shock syndrome: clinocopathologic features. J Infect 1984;8:205-11
16. de Azavedo JCS, Foster TJ, Hartigan PJ, Arbuthnott JP,
O'Reilly M, Kreiswirth BN, Novick RP. Expression of the
cloned toxic shock syndrome toxin 1 gene (tst) in vivo with
a rabbit uterine model. Infect Immun 1985;50:304-9
17. Parsonnet J, Gillis ZA, Richter AG, Pier GB. A rabbit model
of toxic shock syndrome that uses a constant, subcutane-
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
tampons undergo also might have an effect on TSST-l
production. In the studies of Lee et aI., the presence
of deodorant in tampons that were otherwise identical resulted in a decrease in the amount of TSST-l
produced [11].
The addition of magnesium and ascorbic acid to
tampons has also been advocated [31] as a possible
way of decreasing toxin production or of denaturing TSST-l. Further studies correlating TSS and
production of TSST-l with tampons of different absorbency, composition, additives, and O 2 content in
this in vivo model are planned.
In comparing experiments with simulated menses and inoculation of 109 cfu of TSS-associated
strains of S. aureus and cotton, rayon, or CMC/foam
tampons, respectively, it became apparent that the
number of cfu in tampons no. 2 and no. 3 and the
magnitude of total TSST-l in tampon no. 3 were major determinants predicting the development of TSS.
Ten of 12 animals with ~106 cfu of S. aureuslmL
in tampon no. 2 developed a TSS-like illness. Eleven
of 12 with ~105 cfu/mL in tampon no. 3 became ill.
Ten of 11 with levels of TSST-l ~500 ng in tampon
no. 3 developed TSS, while only one of nine with
<500 ng of TSST-l developed TSS. The ability of
a tampon to create an environment that permits the
persistence of elevated quantities of staphylococci
in the absence of repeated vaginal inoculation appears to be a key factor. This "carry-over" of higher
concentrations of organisms and persistent TSST-l
from tampon to tampon was universal for CMCI
foam tampons and appears to be directly related to
TSS-like illness in this model.
In studies to date, this rabbit system has proven
to be a reliable model for vaginally associated TSS.
We believe this model system will allow further
studies in the pathogenesis of tampon-associated
TSS and may be of value in the development and
testing of catamenial products. It may provide insights into factors important in staphylococcal multiplication, TSST-l production, and the development
of TSS to a greater degree than the in vitro systems
employed to date.
S245
S246
Discussion
DR. SHELDON LANDESMAN. I must applaud the
amount of effort that went into this. Did you prove
in this model that the tampon caused a statistically
significant increased likelihood of these animals developing TSS, compared with experiments without
the tampon?
DR. MARIAN MELISH. I do not think we are ready
to answer that question. The main thing that we have
with the tampon is the ability to measure the amount
of toxin present. I do think that the presence of
staphylococci, simulated menses, and repeated trauma to the vagina may be a sufficient explanation,
and it is entirely possible that odds ratios have come
down recently and that the types of tampons that
I am using do not promote TSS.
DR. LANDESMAN. Is there any evidence from this
study that tampons playa role in the development
of TSS in this animal model?
DR. MELISH. I think that they probably do, and
I think that it takes larger numbers to show that. The
finding that sham insertion or repeated instrumentation is also effective does not surprise me.
DR. CLAIRE BROOME. I thought your model work
was very interesting, and I look forward to further
studies. Could you tell us what kind of rayon fibers
you have used in the experiments done to date?
DR. MELISH Dr. Klesius, can you tell us about
the rayon fibers?
DR. ALAN KLESIUS. They were standard viscose
rayon fibers purchased from suppliers.
DR. MELISH. They were not polyacrylate rayon.
DR. EDWARD KASS. It seems to me that what is
crucial here is that you are not getting much absorption from the vagina. If you look at Dr. Parsonnet's
model, where the osmotic pump is subcutaneous,
only a relatively small amount of toxin produces
death. Here you have got a hugely greater amount.
DR. MELISH. No, we do not have much toxin at
all. We find microgram amounts similar to those used
in Dr. Parsonnet's model.
DR. JEFFREY PARSONNET. It is a little difficult
to compare, because Dr. Melish's experiment measures at one point in time, whereas in our model we
are infusing.
DR. MELISH. Of course the vagina does not provide a total barrier for the amount of toxin that can
be produced on tampons and in the vagina. The
amounts of toxin that are produced in this model
are somewhat disappointingly small if compared
with an I8-hour broth culture, where toxin is measured in tens or hundreds of micrograms. Here we
have, with one exception, not even tens of micrograms.
DR. KASS. There are now a number of ways that
have been published for getting vaginal epithelium
to dissociate. Most of these involve various chelation steps that break the little intercellular bridges.
Perhaps you will get a little more disease if you use
a chelating agent to increase the absorption of toxin.
DR. MELIsH. I am comfortable with the notion
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
ous infusion of toxic shock syndrome toxin 1. Infect Immun 1987;55:1070-6
18. Rasheed JK, Arko RJ, Feeley JC, Chandler FW, Thornsberry
C, Gibson RJ, Cohen ML, Jeffries CD, Broome Cv. Acquired ability of Staphylococcus aureus to produce toxic
shock associated protein and resulting illness in a rabbit
model. Infect Immun 1985;47:598-604
19. Melish ME, Murata S, Fukunaga C, Frogner K, Hirata S,
Wong C. Endotoxin is not an essential mediator in toxic
shock syndrome. Rev Infect Dis 1989;1l(Suppll):S219-30
20. Greenwood FC, Hunter WM, Glover JS. 1311-labelled human
growth hormone of high specific radioactivity. J Biochem
1963;89:114-23
21. Bukr FK, Schweppe KW. Review on the biology of menstrual
blood. In: The biology of the fluids of the female genital
tract. Bukr FK, Schumacher GFB, eds. New York, Elsevier
North Holland, 1979:231-45
22. Martin RR, Buttram V, Besch P, Kirkland 11, Petty GP. Na- '
sal and vaginal Staphylococcus aureus in young women:
quantitative studies. Ann Intern Med 1982;96:951-3
23. Brown WJ. Variations in the vaginal bacterial flora. Ann Intern Med 1982;96:931-4
24. Kushnaryov VM, Bergdoll MS, MacDonald HS, Vellinga J,
Reiser R. Study of staphylococcal toxic shock syndrome
toxin in human epithelial cell culture. J Infect Dis 1984;
150:535-45
25. Wagner G, Bohr L, Wagner P, Petersen LN. Tampon-induced
changes in vaginal oxygen and carbon dioxide tensions.
Am J Obstet Gynecol 1984;148:147-50
26. Fox H. The pathology of tampon usage and of the toxic shock
syndrome. Postgrad Med J I985;61(Suppl 1):31-3
27. Schwartz B, Gaventa S, Broome CV, Reingold AL, Hightower AW, Perlman JA, Wolf PH, the Toxic Shock Syndrome Study Group. Nonmenstrual toxic shock syndrome
associated with barrier contraceptives. Rev Infect Dis
1989;11(Suppl 1):S43-9
28. Latham RH, Kehrberg MW, Jacobson JA, Smith CB. Toxic
shock syndrome in Utah: a case-control and surveillance
study. Ann Intern Med 1982;96:906-8
29. Helgerson SO, Foster LR. Toxic shock syndrome in Oregon.
Ann Intern Med 1982;96:909-11
30. Todd JK, Todd BH, Franco-Buff A, Smith CM, Lawellin OW.
Influence of focal growth conditions on the pathogenesis
of toxic shock syndrome. J Infect Dis 1987;155:673-81
31. Jacob J, Lau JR. Protective additive to vaginal products and
catamenials. United States Patent No. 4,585,792, April 29,
1986
Melish et af.
Vaginal Tampon Model for TSS
comparing strains more or less randomly without actually identifying precisely what the extracellular
products are. We now have two models in which
TSST-I is produced in the reproductive tract of the
rabbit - in the vagina and in the uterus. The TSST-I
gets across the epithelial surface in each in sufficient
quantity to have systemic effects. It is now time to
look at whether TSST-l in the reproductive tract is
damaging. I was wondering whether you have planned any histopathologic or electron-microscopic
studies in this fascinating model.
DR. MELISH. Most are in the planning stage. We
have only looked a little at histopathology, and I am
not prepared to discuss it at this time. Incidentally,
these animals have an almost entirely gram-positive
vaginal flora, consisting predominantly of peptostreptococci and lactobacilli. We did see some in
growth of gram-negative bacilli in some rabbits with
the tampon, but this was not common.
DR. ANTHONY CHOW. With respect to a-hemolysin, the isogenic strain that we used lacked ahemolysin and expressed TSST-l in high amounts.
DR. GARY BEST. We have had other reports at
this meeting that TSST-I production and a-hemolysin production tend not to go together.
DR. CHOW. However, TSST-I in this model does
produce the disease, an observation indicating that
TSST-I might be absorbed through the genital tract.
Downloaded from http://cid.oxfordjournals.org/ at Penn State University (Paterno Lib) on May 16, 2016
that TSST-I causes disease when it can infuse from
any place, induding a cotton ball placed in the vagina. This was a first attempt to look at what might
happen in the natural state, and I really do not want
to perturb the system any more than it is already perturbed.
DR. RICHARD MARPLES. I am slightly confused
over these rabbit studies because no one has mentioned a-hemolysin. The rabbit is particularly susceptible to intoxication with a-hemolysin, and antibody is protective. What was the status of the
a-hemolysin production of the strains used, and
would this to an extent explain some of the differences between the strains?
DR. MELISH. I used just two strains. I did not
think there was a significant difference, and I think
that I may be making a quantum leap from other
studies to say that it probably is TSST-I that is diffusing from these organisms. I did not investigate
a-hemolysin, and I do not know the rabbits' antibody status, but I think it is negative.
DR. MARPLES. There was a lot of work just before World War II on the effects of a-hemolysin on
rabbits.
DR. JOHN ARBUTHNOTT. The LD so for a-toxin
in rabbits is 64 hemolytic units, and there are 20
hemolytic units/ug. That gives you an idea of how
sensitive rabbits are to a-hemolysin. I think that is
one of the problems that you get into when you start
S247