Original research article
Fourteen-day safety and acceptability study of 6% cellulose sulfate gel:
a randomized double-blind Phase I safety study
Jill L. Schwartza,4, Christine Maucka, Jaim-Jou Laib, Mitchell D. Creininc,
Vivian Brached, Susan A. Ballaghe, Debra H. Weinerb, Sharon L. Hillierc,
Raina N. Fichorovaf, Marianne Callahana
aDepartment of Obstetrics and Gynecology, CONRAD, Eastern Virginia Medical School, Arlington, VA 22209, USA
bFamily Health International, Research Triangle Park, NC 27709, USA
cDepartment of Obstetrics, Gynecology and Reproductive Sciences, University of Pittsburgh and Magee-Womens Research Institute,
Pittsburgh, PA 15213, USA
dProfamilia, Santo Domingo 10401, Dominican Republic
eDepartment of Obstetrics and Gynecology, CONRAD, Eastern Virginia Medical School, Norfolk, VA 23507, USA
fDepartment of Obstetrics, Gynecology and Reproductive, Biology, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA 02120, USA
Received 28 December 2005; revised 31 January 2006; accepted 23 February 2006
Background: Topical microbicides against the human immunodeficiency virus (HIV) 1 that are nonirritating to the female genital epithelium
are urgently needed to slow the heterosexual spread of HIV infection. Products that are also effective contraceptives provide additional
benefits. Cellulose sulfate (CS) is a noncytotoxic antifertility agent that exhibits in vitro antimicrobial activity against sexually transmitted
pathogens, including HIV.
Methods: We performed a multicenter, Phase I, placebo-controlled, randomized study to evaluate the genital toxicity of CS. Two cohorts of
healthy women used 3.5 ml of 6% CS gel or 3.5 ml of K-Y Jelly, vaginally, bid, for 14 days. The first cohort was sexually abstinent, and the
second cohort was sexually active.
Results: CS was associated with only a slightly higher odds ratio (OR) of symptoms of minor urogenital irritation compared to the inactive
lubricant K-Y Jelly (OR=2.02, 95% confidence interval=0.90–4.53). In addition, there were minor shifts in some genital flora, but there
was no evidence of greater inflammation as evidenced by few colposcopic findings, decreased influx of polymorphonuclear cells and
minimal changes in proinflammatory cytokines. Moreover, both products appeared acceptable to most women. Product leakage was
identified as more of a problem in sexually abstinent women, but less so in women using the product for sexual intercourse, as would be the
case in actual practice.
Conclusion: CS was safe for twice-daily use for 14 days. CS is appropriate for future studies in effectiveness trials.
D 2006 Elsevier Inc. All rights reserved.
Keywords: Cellulose sulfate; Contraception; Microbicide; Sexually transmitted infection; Clinical trial
About half of all human immunodeficiency virus (HIV)
infections worldwide are found in women, yet the only
approved female-initiated barrier method with claims of
reduction of sexually transmitted pathogen transmission,
including HIV, remains the female condom. Microbicides
are vaginally applied chemical barriers that offer women
who at risk for HIVand other sexually transmitted infections
(STIs) an HIV-reducing method not requiring the cooper-
ation of their partners. The spermicide nonoxynol-9 (N-9), a
surfactant that destroys cellular and microbial membranes, is
an effective contraceptive . However, N-9 is not effective
for the prevention of HIV-1 and other STIs, and, with
frequent use, may increase a woman’s risk due to genital
epithelial disruption . Alternative methods — such as
vaginal microbicides, whose antimicrobial effects are not
mediated by nonspecific cytotoxic mechanisms and are,
therefore, nonirritating to the genital epithelium and
0010-7824/$ – see front matter D 2006 Elsevier Inc. All rights reserved.
4 Corresponding author. Tel.: +1 703 524 4744; fax: +1 703 524 4770.
E-mail address: firstname.lastname@example.org (J.L. Schwartz).
Contraception 74 (2006) 133–140
compatible with the mucosal immune barrier — are urgently
needed to slow the spread of heterosexual HIV infection.
Cellulose sulfate (CS) is such a noncytotoxic microbicide
that functions as an entry inhibitor. It exhibits antimicrobial
activity against sexually transmitted pathogens, inhibits
sperm function in vitro  and acts as a contraceptive when
applied vaginally in rabbits . CS blocks cell surface re-
layers and dendritic cells, blocks the gp120–CD4 coreceptor
primary isolates and laboratory-adapted strains .
A 6-day study conducted on women in the United States
suggested that 6% CS was as safe and acceptable as the
control, K-Y Jelly . Similarly, CS was found to be safe
when applied to the penis for seven consecutive days . A
magnetic resonance imaging study of the distribution of CS
gel showed that a volume of 3.5 ml, compared to a volume
of 2.5 ml, improved spread throughout the vagina without
significant leakage .
The objective of this study was to assess whether CS is
safe enough in preserving genital epithelial integrity and
function, in comparison to a marketed sexual lubricant, to
warrant further development.
2. Materials and methods
This was a randomized double-blind Phase I study
performed at three centers: the CONRAD Clinical Research
Center at the Eastern Virginia Medical School (Norfolk, VA,
USA), the Magee-Women’s Hospital (Pittsburgh, PA, USA)
and Profamilia (Santo Domingo, Dominican Republic).
Procedures were reviewed with the study personnel for all
the study. Sample size was based on convenience, as
appropriate for a Phase I vaginal product safety study. Across
all sites, 30 women were enrolled in each of two treatment
groups. The study was first conducted within a cohort of 30
sexually abstinent women before being conducted in a cohort
of 30 sexually active couples. An independent data monitor-
sexually abstinent cohort and was ready to progress to the
sexually active cohort. All study sites progressed to the
sexually active portion of the study. Appropriate human
subjects approval was obtained at each center.
The formulations tested were 3.5 ml of 6% CS gel
(150 mg of CS)1and 3.5 ml of K-Y Jelly.2All gels were
packaged in identical, single-use applicators manufactured
by HTI Plastics (Lincoln, NE, USA).
Healthy female volunteers who were 18–50 years old,
were in good health, had regular menstrual cycles, had a
normal Pap smear, were not at risk for pregnancy and
provided signed informed consent were enrolled. Women in
the sexually active cohort were required to have had a
mutually monogamous relationship for at least 3 months and
to have been protected from pregnancy by either male or
female sterilization or a nonintravaginal hormonal contra-
ceptive. Nonpregnant female volunteers without vaginal
candidiasis, bacterial vaginosis (BV) and/or urinary tract
infection, and with normal (F20%) complete blood count
(CBC), chemistry panel and activated partial thromboplastin
time were enrolled. Female volunteers were ineligible if
they have had a hysterectomy, were breastfeeding or were
within 2 months from their last pregnancy outcome, or were
allergic to the study product N-9 or latex (for the sexually
active cohort). Female volunteers who tested positive for
Trichomonas vaginalis, Neisseria gonorrhoeae or Chla-
mydia trachomatis; or who were recently diagnosed with, or
treated for, any STI or pelvic inflammatory disease; or who
had genital symptoms were ineligible. A history of drug or
alcohol abuse, chronic or current use of antibiotics, or prior
exposure to CS precluded participation.
Male volunteers were eligible for enrollment if they were
at least 18 years old, provided signed informed consent and
had been in a mutually monogamous relationship for at least
3 months with an eligible female partner. Male participants
were ineligible if they were allergic to the study product N-9
or latex; if they had been recently diagnosed with, or treated
for, any STI; or if they had recent symptoms consistent with
Random allocation sequences were created by a statistician
not otherwise involved in the study, using a program based on
the SAS3(SAS Institute, Inc., Cary, NC, USA) RANUNI
function. Randomization employed the random permuted
blocks method and was stratified by center and by cohort.
Allocation sequences were provided to the product supplier,
which packaged the products into identical white wrappers
identified by participant number. A data check confirmed
that participants were randomized in chronological order.
Each female volunteer was seen in four visits. On the
screening visit, volunteers were instructed to abstain from
all vaginal activities beginning 72 h prior to the enrollment
visit, which was scheduled to fall within the follicular phase
(Cycle Days 5–10) of the menstrual cycle. All colposcopies
were carried out according to the World Health Organiza-
tion/CONRAD Manual for the Standardization of Colpo-
scopy for the Evaluation of Vaginally Administered
Products Update 2000 .
On the enrollment visit, eligible volunteers were
randomized to receive one of the two study products and
were given 27 applicators for twice-daily use, at approxi-
mately 12-h intervals, for 14 days, excluding the morning of
the third visit. The first application took place in the clinic,
1The CS gel contained the following inactive ingredients: glycerin,
methylparaben, propylparaben and water.
2The K-Y Jelly (Personal Products Co., Skillman, NJ, USA) contained
chlorhexidine gluconate, methylparaben gluconodelta lactone, glycerin,
hydroxyethyl cellulose, purified water and sodium hydroxide.
3SAS is a registered trademark of SAS Institute, Inc., in the United
States and in other countries.
J.L. Schwartz et al. / Contraception 74 (2006) 133–140
followed 15 min later by colposcopy. The sexually active
participants were instructed to have vaginal intercourse,
using the male condoms (nonspermicidal) provided, at least
twice weekly and to time the intercourse in conjunction
with the last gel insertion at least 6 h prior to the third and
fourth visits. The third and fourth visits were scheduled to
take place within 6–18 h after the 14th and final gel
A vaginal smear was obtained from the vaginal wall for
Gram stain, air-dried on a glass slide and transported to the
Magee-Women’s Research Institute (MWRI) according to
standard procedures for Nugent scoring . Smears were
also examined for the number of polymorphonuclear
leukocytes (PMN) to evaluate for increases in inflammatory
cells with product use. The score used a point scale from 0
to 4+ in the slide examination of five fields under oil
immersion (?1000) [(0)=no PMN/field; (1+)= b1 PMN/
field; (2+)=1–4 PMN/field; (3+)=5–30 PMN/field;
(4+)= N30 PMN/field].
A second swab was obtained for semiquantitative vaginal
cultures for 11 organisms and was placed in anaerobic
media (Port-a-Cul; Becton Dickinson Corp., Cockeysville,
MD, USA) and transported overnight to the MWRI, where it
was evaluated according to a previously published proce-
dure . Semiquantitative growth has been correlated with
quantitative log growth according to the scale: (1+)=
102–104; (2+)=105; (3+)=106; (4+)=107–108.
The vaginal walls and cervix were lavaged with 10 ml of
normal saline. The collected fluid was centrifuged for 10 min
Cytokine data were not collected from participants in the
Brigham and Women’s Hospital (Boston, MA, USA).
Standardized procedures were developed for the measure-
ment of cytokines TNF-a, IL-1a, IL-1h, IL-1 receptor
antagonist (IL-1RA) and IL-8 in saline cervical vaginal
lavage (CVL) using commercial photometric or lumines-
IL, USA), a multilabel microplate counter Victor 2 (Perkin
Elmer Life Sciences, Boston, MA, USA) and WorkOut
Version 1.5 Wallace Software (DAZDAQ Ltd., Brighton,
East Sussex, England, UK). Cytokine standards were
recovered from CS gel and K-Y Jelly solutions to determine
product interference with cytokine detection systems. The
intra-assay and interassay coefficients of variation of
cytokine recovery from the CVL matrix were determined
by standard spiking in pooled CVLs. The criteria for
elevated cytokine values were defined as moderate (mean
at baseline+1 SD) and high (mean at baseline+2 SD).
Study objectives were evaluated in the treated population
(women providing at least some follow-up data). The
primary safety outcome was the proportion of women with
any evidence (signs and symptoms) of urogenital irritation
during 14 days of twice-daily use. Signs (as seen on pelvic
examination and colposcopy) and symptoms (as reported
by subjects; i.e., genital pruritus, genital and pelvic pain,
abnormal bleeding and other/vaginal lacerations) were
evaluated as separate sub-endpoints. Determination of
which symptoms were considered evidence of urogenital
irritation was carried out by a clinician not performing study
procedures and who reviewed blinded event descriptions.
All noniatrogenic, new or worsening colposcopy findings
were counted as signs of urogenital irritation.
The primary outcome was assessed separately for three
intervals: immediately (15 min) after initial product applica-
tion, after 7days of product use and after an additional 7days
of product use, although the analysis evaluated the outcome
in vaginal health, as assessed by wet mounts, Gram stains,
vaginal cultures and cytokine analyses. For safety endpoints
measured at multiple time points (other than cytokine
measurements), product groups were statistically compared
using generalized estimating equation (GEE) models imple-
mented by the GENMOD procedure in SAS, which controls
for center and cohort, and treats time of measurement as a
repeated measure. The analysis used the link function of logit
for binary endpoints, cumulative logit for ordinal endpoints
evaluated using exact measures (data not shown). Cytokine
data were compared between groups using exact Wilcoxon–
Mann–Whitney tests stratifying on center and cohort,
although a post hoc analysis used GEE. Adverse events
(AEs) were coded using MedDRAversion 5.1 (International
Federation of Pharmaceutical Manufacturers Associates,
Geneva, Switzerland). For symptoms reported by more than
four women, proportions of women experiencing these AEs
were compared between groups using an exact Mantel–
Haenszel test pooling center and cohort.
The acceptability of the product was assessed using a
questionnaire. The key acceptability item was prespecified
as whether or not the participant said she would buy the
product for either contraceptive use or for STI prevention.
Proportions were compared between groups using exact
logistic regression controlling for center and cohort.
Recruitment occurred from October 2002 to September
2003; follow-up ended on October 2003. Sixty women, 30 in
randomized to K-Y Jelly; the imbalance occurred because
one set of supplies was tampered with by customs officials
and removed. In the sexually active cohort, 14 volunteers
were randomized to CS gel and 16 were randomized to K-Y
Jelly; the imbalance occurred because the sequence had been
generated balanced across 14 possible participants per cohort
per site, but not across the 10 participants actually enrolled.
Since all 60 women provided data and since there were no
allocation errors, the treated population is identical to the
J.L. Schwartz et al. / Contraception 74 (2006) 133–140
Nineteen of 30 (63%) participants in the CS group
compared to 15 of 30 (50%) in the K-Y group had at least
one protocol violation, which was a minor deviation that did
not affect the results or overall safety of the study. All
participants completed the study using their assigned product
at least 25 times out of a total of 27 expected applications and
were seen at all scheduled visits, except for two participants
who used only 14 applicators because they discontinued after
1 week: one (CS sexually abstinent group) who requested
discontinuation due to moderate discomfort and one (K-Y
sexually abstinent group) who discontinued after 1 week of
product use because customs officials tampered with
supplies. All participants in the sexually active cohort
reported four to six coital acts during follow-up, and all but
one of the sexually active participants (K-Y group) reported
using condoms for each act of intercourse. Participant
characteristics are summarized in Table 1.
3.1. Signs and symptoms of urogenital irritation
Overall, about equal percentages of women, 10 of 30
(33%) in the CS group and 8 of 30 (27%) in the K-Y group,
reported 20 and 13 symptoms of urogenital irritation,
respectively (Table 2). The most common symptom of
urogenital irritation was pain, which accounted for more
than half of the complaints. Three participants assigned to
the CS group experienced very slight or slight intermen-
strual bleeding: (a) spotting related to oral contraceptives;
(b) postexamination spotting unrelated to product; and (c)
blood-tinged discharge possibly related to product.
One participant (CS group) complained of moderate
pelvic pain from study examination and moderate vaginal
discomfort during the insertion of study applicators. All
other urogenital complaints were considered of very slight
or slight severity. Symptoms of urogenital irritation were
more commonly reported by participants in the sexually
active group compared to the sexually abstinent cohort
(Table 2). There were no serious, unexpected or product-
use-related adverse experiences.
Nineteen (63%) participants in the CS group had 37
colposcopic findings, and 14 participants (47%) in the K-Y
group had 28 colposcopic findings (Table 3). No colpo-
scopic finding with deep epithelial disruption was observed
at any point during this study. Each product group had five
participants with at least one finding of superficial epithelial
disruption. Of the 65 noniatrogenic findings observed
during follow-up, 57 were considered by site investigators
to be at least possibly product-related. The most common
colposcopic finding was erythema, which predominated in
sexually abstinent women regardless of product group and
accounted for the greater number of findings in the abstinent
women compared to that in the active women.
Across all product use periods, participants in the CS
group tended to have slightly higher odds of experiencing
any evidence of urogenital irritation compared to those in
the K-Y group [odds ratio (OR)=1.36, 95% confidence
interval (95% CI)=0.70–2.61]. This increase was predom-
inantly based on an increase in the symptoms (OR=2.02,
95% CI=0.90–4.53) rather than in the signs (OR=1.18,
95% CI=0.62–2.22) of urogenital irritation (Table 4).
One clinical diagnosis of BV was made (CS group)
during the study and, as per the analysis plan, was not
counted toward the endpoint of urogenital irritation. Only
one participant (K-Y group) developed an abnormal Nugent
Participant characteristics (pooling cohorts)
CharacteristicsCS (n=30) K-Y (n=30)
Age [mean+SD (range)]
Education (school years completed)
Ethnicity [n (%)]
Race [n (%)]
More than one race/other
Partner status [n (%)]
Living with partner
Not living with partner
Prior use of spermicide [n (%)]
24 (80.0 )
27 (90.0 )
AESexually abstinent Sexually active
CS (n=16) K-Y (n=14)CS (n=16) K-Y (n=14)
Other (vulvar laceration)b
Total number of events/women
with at least one eventc
aIncludes genital burning, irritation, urinary symptoms, discomfort, dyspareunia, pelvic pain and increased sensitivity.
bCounted as a colposcopic finding and as a urogenital event in the analysis because it was reported by the participant and was seen on examination.
cThe total number of women may not equal the column sum because a given woman may have more than one type of urogenital event.
J.L. Schwartz et al. / Contraception 74 (2006) 133–140
score (8) during the study. Since there were very few
positive outcomes, comparison statistics could not be
computed on yeast, BV or T. vaginalis using GEE.
Treatment with both CS and K-Y Jelly appeared to
decrease the proportion of women with PMN on Gram
stain. In order to take into account baseline variations in the
microflora, planned analyses were repeated by replacing
raw values on Visits 3 and 4 with their respective change
from baseline (the bchange scoreQ). Overall, there were no
differences in alterations in H2O2-negative lactobacilli,
Gardnerella vaginalis, Enterococcus, anaerobic gram-
negative rods, Mycoplasma hominis or group B strepto-
cocci comparing CS and K-Y Jelly (Fig. 1). Women in both
treatment groups had decreased levels of H2O2-positive
lactobacilli. Participants in the CS group had statistically
significantly increased odds for Escherichia coli and
Staphyloccocus aureus and had decreased odds for
Ureaplasma compared to those in the K-Y group (Fig. 1).
3.3. Cytokine evaluation
Of the 40 participants in whom cytokines were measured,
5 of 21 (24%) exposed to CS gel and 9 of 19 (47%) exposed
to K-Y Jelly had moderate or high levels during the study
(baseline, 7 days and/or 14 days), but only two participants
with moderate or high levels in each group had no elevated
levels at baseline. Three participants with abnormal Nugent
scores at baseline and one with an abnormal Nugent score
after 1 week of K-Y Jelly use had moderate to high cytokine
levels at baseline, suggesting inflammatory conditions prior
to product use.
None of the cervicovaginal samples had TNF-a
values above the established precision limit of detection
(N22 pg/ml). Median IL-1a and IL-1h changed minimally
from baseline. Median IL-8 levels decreased after product
application in both cohorts. When baseline variation was not
taken into account, a noticeable difference was found
between the CS and K-Y groups for IL-1RA (pb.0001 on
by high median IL-1RA in the K-Y sexually active cohort at
baseline. However, when statistical comparison was repeated
on change scores, there was no such difference between
treatment groups in any of the cytokines (p=.29–.98 on Visit
levels decreased or remained the same during product use
(from baseline to 14 days) (Fig. 2).
Sexually abstinentSexually active
CS (n=16) K-Y (n=14) CS (n=16) K-Y (n=14)
Total number of findings/women
with at least one findingc
6 (38) 10
aFindings with disrupted blood vessels.
bFindings with disrupted epithelium, all of which were considered superficial.
cThe total number of women may not equal the column sum because a given woman may have more than one type of finding.
Comparison of product groups on primary endpoint: urogenital irritation
CS group versus K-Y group Any evidence of urogenital irritation Symptoms of urogenital irritationa
Signs of urogenital irritation
After first usec
After 7 daysc
After an additional 7 daysc
After all usesd
aThere were no symptoms of urogenital irritation reported at 15 min after the first gel application at the study sites.
bSince the sample size was not based on statistical considerations, no formal hypothesis testing was performed, and the results of statistical tests should not
be interpreted as formal justification for rejecting or failing to reject hypotheses, but only as supplementary information to guide clinical review. All p values
are exploratory and should be interpreted cautiously in light of the relatively large number of statistical tests and low statistical power. For this reason, actual p
values are reported (i.e., for the evaluation of objectives, there is no reference to an a level for Type I error) and are not adjusted for multiple comparisons. All p
values are two-tailed.
cExact logistical regression with center, cohort and product group as covariates.
dGEE analysis with center, cohort and product group as independent predictors, and with time as repeated measure.
J.L. Schwartz et al. / Contraception 74 (2006) 133–140
About 72% of women in the CS group and 80% in the
K-Y group said that they would buy the products to prevent
pregnancy, and about 62% in the CS group and 70% in the
K-Y group would buy the products to prevent STIs. There
was no noticeable difference in the odds of the preplanned
acceptability endpoint, bwould not buy the product for
either STI or pregnancy prevention,Q between the CS and
K-Y groups (pb.84).
All of the women found the product either beasyQ or
bvery easyQ to use, and most had no problems inserting it.
Of the 29 participants who used CS gel, 11 did not feel the
gel, 13 found the gel sensation pleasant and 5 found the gel
unpleasant. Of the 30 participants who used K-Y Jelly, 18
did not feel the gel, 6 found the gel sensation pleasant or
very pleasant and 6 found the gel sensation unpleasant.
More women in the sexually abstinent group than in the
sexually active group found the sensation unpleasant.
Product leakage was the most common problem encoun-
tered with both products, and most women reported at least
very slight product leakage. Of note, 7 of 16 (44%) women
in the CS sexually abstinent cohort, 4 of 12 (33%) in the
K-Y sexually abstinent cohort, 2 of 11 (18%) in the CS
sexually active cohort and 4 of 15 (27%) in the K-Y
sexually active group reported that leakage would prevent
them from using the gel in the future. However, there was
no noticeable difference in the proportion of overall vaginal
leakage of moderate or severe intensity between the CS and
K-Y groups (p=1.00).
3.5. Total AEs and systemic safety
More than four women experienced AEs in each of two
body system categories: gastrointestinal disorders and
reproductive system and breast disorders. There were no
obvious differences between the CS and K-Y groups in
proportions experiencing these types of AEs (pz.22).
Therewas a totalof 37 AEs in15 (50%) participants inthe
CS group compared to 23 AEs in 13 (43%) participants in the
K-Y group, including the urogenital AEs already described.
The most common AE that was not considered urogenital
irritation was nausea, which was more common in the CS
group. None of these complaints was thought to be related to
the study product by the site study investigator, and none was
severe. Two moderate AEs, Gardner duct cyst (CS abstinent
group) and herpes simplex virus (K-Yabstinent group), were
not counted toward the genital endpoint because they are not
relevant to the assessment of product safety.There were three
additional individuals in the CS group who reported four
moderate nonurogenital AEs, including neck and back pain,
nausea and headache. One male participant in each product
(which was probably product-related) and lower abdominal
pain (which was considered unrelated).
There were no apparent differences between groups in
mean values at follow-up in regard to laboratory parameters.
There were 15 abnormal systemic safety laboratory values
on the final visit among five participants in the CS group
and six participants in the K-Y group. Except for two
participants who had a mild inflammatory process (based on
their CBC) that was associated with a viral syndrome and
shifts in the vaginal flora, these abnormal values were not
This study was conducted to assess the safety of CS
when used vaginally twice daily for 14 days in a sexually
abstinent cohort and in a sexually active cohort. There were
slightly increased odds of experiencing any evidence of
urogenital irritation in the CS group compared to the K-Y
group, primarily because of increased odds of symptoms.
About one third of the women in the CS group and one
quarter of women in the K-Y group reported slight
symptoms of urogenital irritation, which were not clinically
significant but did increase the OR to an almost significant
level (pb.09), even in this small study. It is unlikely that
intermenstrual spotting is a product effect of CS as two of
the three participants who experienced spotting had other
contributing factors. Although nausea occurred more
commonly in the CS group, a causal relationship is
unlikely; still, future studies should continue to explore
Slightly more women in the CS group had at least one
colposcopic sign of genital irritation during follow-up. In an
earlier study, the only cervicovaginal findings that
were associated with HIV infection were grossly observed
epithelial breaches . In this study, no findings of deep
disruption of the epithelium were observed in either group.
Seventeen percent of women in each treatment group had at
least one finding of superficial disruption of the epithelium.
Thus, CS does not appear to cause more epithelial
disruption than K-Y.
Fig. 1. Microflora and PMN: CS versus K-Y Jelly. GEE analysis comparing
product groups, controlling for center and cohort, and treating time (7 and
14 days) as a repeated measure. The variable analyzed is change from
baseline to follow-up. *PMN values were obtained by Gram stain.
J.L. Schwartz et al. / Contraception 74 (2006) 133–140
At least one colposcopic finding was observed in 60% of
sexually abstinent women and in 47% of sexually active
women, which was somewhat reversed from an earlier
observational study in which colposcopic findings were seen
in 25% of women examined within 24 h of vaginal
intercourse compared with 14% of women who abstained
. It is possible that the gel provided some type of
protection in sexually active women, although other possi-
bilities are plausible.
Ensuring that a product does not significantly disrupt the
vaginal flora and does not increase urogenital infections has
become standard for the evaluation of microbicide candi-
dates. Semiquantitative vaginal cultures for the measure-
ment of such changes have been used in previous studies
with microbicide candidates, but the clinical relevance is not
entirely known. The increase in E. coli in the CS group and
the decrease in H2O2-positive Lactobacillus in both groups
were the most striking shifts in the vaginal flora. Both
changes have been previously reported with the use of N-9
and BufferGel [14,15], and with cervical caps and dia-
phragms with spermicides . However, whether these
differences will result in urogenital infections in a larger
clinical study is unknown. The increased OR observed with
S. aureus is predominantly due to a decrease of these
bacteria in the K-Y group rather than due to an increase of
these bacteria in the CS group. Furthermore, the reduction in
G. vaginalis and Ureaplasma in the CS group might be
Fig. 2. Comparison of cytokine levels. Top two rows: cytokine levels of sexually abstinent cohort. Comparison of baseline values versus values obtained
14 days after product (K-Y Jelly or 6% CS gel) use. Each line represents an individual subject. Bottom two rows: cytokine levels of sexually active cohort.
Comparison of baseline values versus values obtained 14 days after product (K-Y Jelly or 6% CS gel) use. Each line represents an individual subject.
J.L. Schwartz et al. / Contraception 74 (2006) 133–140
Our study is the first to report on the measurement of Download full-text
cervicovaginal cytokine levels to assess urogenital irritation
in response to vaginal products in a clinical safety study.
Cytokines regulate mucosal immune responses and, in the
vaginal environment, may play a critical role in the
pathogenesis of HIV-1 and other STIs . N-9-induced
vaginal inflammation has been linked to increased HIV
activity via IL-1-mediated mechanisms . A group of
cytokines, some of which have been shown to be released
from immortalized human vaginal epithelial cells in vitro
and from rabbit vaginal mucosa after exposure to surface-
active microbicide candidates , was evaluated to help
determine their potential as an in vivo system for monitoring
the undesirable proinflammatory effects of microbicides and
other vaginal products. In this study, neither product
appeared to have proinflammatory effects. The complicating
role of high baseline cytokine levels, possibly attributable to
subclinical inflammatory conditions, underscores the im-
portance of obtaining baseline values.
Participants were more likely to buy the product for
pregnancy prevention than for STI prevention, which fits
the careful selection of low-risk women who perceived
that they were not at risk for STIs. More women in the
sexually abstinent cohort found leakage to be a problem.
The purpose of testing the product in an abstinent cohort
was to establish safety before progression to gel use
during sexual activity. In actual practice, the gel would be
used during intercourse, during which some of the gel
would be distributed and leakage would be more expected
and better tolerated, as noted with the sexually active
women in this study.
CS was safe for twice-daily use for 14 days in sexually
abstinent and sexually active women. It was associated with
only slightly higher odds of reported symptoms of minor
urogenital irritation. In addition, there were minor shifts in
genital flora, but there was no evidence of greater
inflammation as evidenced by few colposcopic findings,
decreased influx of PMN and minimal changes in proin-
flammatory cytokines. Moreover, both products appeared
acceptable to most women. CS is appropriate for future
studies in effectiveness trials.
The authors wish to thank Christine Donahue, Lynn
Reid, Theresa Abercrombie and Leila Cochon for coordi-
nating the study at three different sites, and Christine
Colven and Susan Schmitz for monitoring the study.
Support for this subproject was provided by the Global
Microbicide Project (GMP), a program of CONRAD,
Eastern Virginia Medical School [GMP-02-44 and GMP-
02-45 subproject contracts with the Brigham and Woman’s
Hospital (MSA-02-304), and National Institutes of Health
grant M01RR000056 at the University of Pittsburgh]. Data
management and statistical analysis were carried out by
Family Health International.
 Raymond E, Chen PL, Luoto J. Contraceptive effectiveness and safety
of five nonoxynol-9 spermicides: a randomized trial. Obstet Gynecol
 Van Damme L, Ramjee G, Alary M, et al. Effectiveness of COL-1492,
a nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex
workers: a randomized controlled trial. Lancet 2002;360:971–7.
 Anderson RA, Feathergill KA, Diao XH, et al. Preclinical evaluation
of sodium cellulose sulfate (Ushercell) as a contraceptive antimicro-
bial agent. J Androl 2002;23:426–38.
 Anderson RA, Feathergill KA, Diao XH, et al. Contraception by
Ushercell (cellulose sulfate) in formulation: duration of effect and
dose effectiveness. Contraception 2004;70:415–22.
 Scordi-Bello IA, Mosoian A, He C, et al. Candidate sulfonated and
sulfated topical microbicides: comparison of anti-human immunode-
ficiency virus activities and mechanisms of action. Antimicrob Agents
 Mauck C, Weiner DH, Ballagh S, et al. Single and multiple exposure
tolerance study of cellulose sulfate gel: a Phase I safety and
colposcopy study. Contraception 2001;64:383–91.
 Mauck C, Frezieres R, Walsh T, Robergeau K, Callahan M. Cellulose
sulfate: tolerance and acceptability of penile application. Contracep-
C. Vaginal distribution of two volumes of the novel microbicide gel
cellulose sulfate (2.5 mL and 3.5 mL). Contraception 2005;72:65–70.
 World Health Organization and CONRAD. Manual for the standard-
ization of colposcopy for the evaluation of vaginally administered
products update 2000. Geneva, Switzerland/Arlington, VA7 WHO/
 Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial
vaginosis is improved by a standardized method of Gram stain
interpretation. J Clin Microbiol 1991;29:297–301.
 Ness RB, Hillier SL, Richter HE, et al. Douching in relation to
bacterial vaginosis, lactobacilli, and facultative bacteria in the vagina.
Obstet Gynecol 2002;100:765–72.
 Gentry YV, Hillier SL, Eschenbach DA. Evaluation of a rapid enzyme
immunoassay test for detection of Group B Streptococcus. Obstet
 Fraser IS, Lahteenmaki P, Elomaa K, et al. Variations in vaginal
epithelial surface appearance determined by colposcopic inspection in
healthy, sexually-active women. Hum Reprod 1999;14:1974–8.
 Watts DH, Rabe L, Krohn MA, Aura J, Hillier SL. The effects of three
nonoxynol-9 preparations on vaginal flora and epithelium. J Infect Dis
 Clarke JG, Peipert JF, Hillier SL, et al. Microflora changes with the
use of a vaginal microbicide. Sex Transm Dis 2002;29:288–93.
 Gupta K, Hillier SL, Hooton TM, Roberts PL, Stamm WE. Effects of
contraceptive method on the vaginal microbial flora: a prospective
evaluation. J Infect Dis 2000;181:595–601.
 Al-Harthi L, Kovacs A, Coombs RW, et al. A menstrual cycle pattern
for cytokine levels exists in HIV-positive women: implication for HIV
vaginal and plasma shedding. AIDS 2001;15:1535–43.
 Fichorova RN, Tucker L, Anderson DJ. The molecular basis of
nonoxynol-9 induced vaginal inflammation and its possible relevance
to human immunodeficiency virus Type 1 transmission. J Infect Dis
 Fichorova RN, Bajpai M, Chandra N, et al. Interleukin (IL)-1, IL-6,
and IL-8 predict mucosal toxicity of vaginal microbicidal contra-
ceptives. Biol Reprod 2004;71:761–9.
J.L. Schwartz et al. / Contraception 74 (2006) 133–140