MAJOR ARTICLE
Hyperosmolar Sexual Lubricant Causes Epithelial
Damage in the Distal Colon: Potential Implication
for HIV Transmission
Edward J. Fuchs,1 Linda A. Lee,2 Michael S. Torbenson,3 Teresa L. Parsons,1 Rahul P. Bakshi,1 Anita M. Guidos,1
Richard L. Wahl,4 and Craig W. Hendrix1
Divisions of 1Clinical Pharmacology and 2Gastroenterology, Department of Medicine, and Departments of 3Pathology and 4Radiology,
Johns Hopkins University School of Medicine, Baltimore, Maryland
Background. Many sexual lubricants are hyperosmolar. Hyperosmolar enemas induce epithelial damage, and
enema use has been associated with an increased risk of HIV infection. To inform the development of rectal
microbicide formulation, we evaluated the effects of hyperosmolar gels on the rectal mucosa.
Methods. Two commercial lubricants were compounded into iso-osmolar and hyperosmolar mixtures (283
and 3429 mOsm/kg, respectively). Each gel was radiolabeled with 500 mCi of 99mTechnetium–diethylene triaminepentaacetic acid, and 10 mL was given rectally to 10 subjects in random sequence. Sigmoidoscopy by an
endoscopist blinded to treatment assignment was performed 90 min later to obtain luminal and mucosal samples.
Urine radiolabel detection was used to assess mucosal permeability.
Results. Epithelial denudation 10 cm from the anus occurred to a greater degree with the hyperosmolar gel
than with the iso-osmolar formulation (median toxicity grade, 2.50 vs. 1.17 out of 3, respectively; P p .009). The
hyperosmolar gel was also associated with lower isotope luminal concentration at 10 cm, compared with the isoosmolar gel (median, 8.9% vs. 54.6% of administered concentration, respectively). Mucosal permeability measured
through 12 h was reduced with the hyperosmolar gel (P p .037).
Conclusion. Rectally applied hyperosmolar gels induce greater epithelial denudation and luminal secretion
than iso-osmolar gels. Because denudation plausibly increases the risk of HIV transmission, hyperosmolar gels
make poor rectal microbicide formulations, and hyperosmolar sexual lubricants may increase susceptibility to HIV
infection.
Despite the fact that the majority of newly reported
cases of HIV/AIDS in the United States are among men
who have sex with men (MSM), little attention has
focused on developing topical HIV microbicides for
rectal use [1]. Similarly, it has been reported that up
Received 15 August 2006; accepted 13 October 2006; electronically published
23 January 2007.
Potential conflicts of interest: none reported.
Financial support: Johns Hopkins University Center for AIDS Research (grant
P30 AI42855); General Clinical Research Center at Johns Hopkins–National
Institutes of Health (NIH)/National Center for Research Resources (grant M01RR000052); NIH (Midcareer Investigator Award in Patient-Oriented Research K24
AI01825 to C.W.H.).
Presented in part: Microbicides 2006, Cape Town, South Africa, 23–26 April
2006 (abstract PB20).
Reprints or correspondence: Edward J. Fuchs, Div. of Clinical Pharmacology,
Johns Hopkins University School of Medicine, Harvey 502, 600 N. Wolfe St.,
Baltimore, MD 21287-5554 ([email protected]).
The Journal of Infectious Diseases 2007; 195:703–10
2007 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2007/19505-0016$15.00
DOI: 10.1086/511279
to 25% of women have engaged in receptive anal intercourse [2–5]. Because it is likely that any vaginal
HIV microbicide will also be used rectally, vaginal formulations should be evaluated for rectal use. Largescale trials using nonoxynol-9 (N-9) have demonstrated
that the failure to appreciate the potential deleterious
effects of an HIV microbicide formulation can result
in unforeseen toxicity and development delays [6]. Although the surfactant properties of N-9 are the likely
explanation, commercial N-9 products are frequently
hyperosmolar (table 1).
Recently, Phillips et al. [7, 8] demonstrated that rectal
application of N-9 resulted in sloughing of surface epithelia. Hyperosmolar fluids have been shown to induce
similar changes in the distal colon [9–13]. Because
many water-based personal lubricants are hyperosmolar
(table 1), such formulations, applied rectally, may induce similar damage.
Small volumes of rectally applied hyperosmolar luHyperosmolar Epithelial Damage
• JID 2007:195 (1 March) • 703
Table 1. Osmolality of some common over-thecounter lubricants and nonoxynol-9 (N-9)–containing contraceptives.
Product
Median, mOsm/kg
Tap water
FemGlide
3
42
Semen
Gynol II (2% N-9)
340
1182
Conceptrol (4% N-9)
K-Y Plus (2% N-9)
Fleet enema
Wet Original/sterile water (1:1)
1257
2037
2127
2215
K-Y Jelly
Astroglide/sterile water (1:1)
2424
3126
ID Glide
PrePair
3429
4026
NOTE. All samples were measured on a Wescor vaporpressure osmometer. Osmolality of tap water, semen, and
Fleet enema are provided for reference.
bricants undergo volume expansion, which implies that fluid
is secreted into the colonic lumen [14]. Early epidemiologic
studies suggested that rectal douching is a strong predictor of
HIV and hepatitis B virus infection, and MSM focus groups
indicated that hyperosmolar enemas are commonly used for
such douching [15–18, 19]. Furthermore, colonic inflammation
has been associated with increased luminal permeability of
small-molecule radioisotopes [20–23]. Although there is no
direct evidence that a tonicity gradient causing surface denudation and changes in colonic mucosal permeability increases
susceptibility to HIV infection, it is plausible. Demonstration
of a hyperosmolar effect toxic to the colonic epithelia would
have profound implications for rectal microbicide development
and, possibly, for safer lubricant selection. We performed the
present study to explore the potential epithelial toxicity due to
rectal application of commercially available hyperosmolar gels.
SUBJECTS, MATERIALS, AND METHODS
Study participants. After institutional review board (IRB) approval was obtained, 10 male subjects (all MSMs), including 8
HIV-seropositive and 2 HIV-seronegative men, were recruited,
gave informed consent, and were enrolled. Subjects reported
no history of rectal disease or surgery or of recent diarrhea,
defined as ⭓3 loose stools per day within 5 days of entry. All
subjects refrained from receptive anal intercourse for 5 days
before each admission.
In a 2-period crossover design, subjects received an unmodified 10-mL dose of ID Glide (Westridge Laboratories; 3429
mOsm/kg [pH 4.79]) and a 10-mL dose of a FemGlide (Cooper
Surgical)/ID Glide combination gel (283 mOsm/kg [pH 6.77])
per rectum in randomized sequence.
The combination iso-osmolar mixture was established using
704 • JID 2007:195 (1 March) • Fuchs et al.
the method described below. The lubricants chosen contain no
detergents with potentially confounding N-9 effects. Each gel
was compounded with 500 mCi of 99mTechnetium–diethylene
triaminepentaacetic acid (99mTc-DTPA) for assessment of intraluminal concentration and colonic permeability. The sequence of gel assignment was randomized to avoid confounding results from treatment effect. Crossover to the second arm
was conducted with at least 2 weeks between admissions.
Subjects were admitted to an inpatient General Clinical Research Center and placed on a liquid diet for 10 h before gel
dosing (12 h before endoscopy). The following day, subjects
received a 250-mL tap-water enema 8 h before dosing. This
enema type and volume is commonly used before anal intercourse [19]. The cleansing enema was administered 8 h before
endoscopy to minimize trauma to the colonic epithelium; studies have shown that epithelial repair occurs rapidly and is nearly
complete by 8 h [8, 24, 25]. Subjects expelled the enema immediately after administration.
The lubricants were compounded with radioisotope and
loaded into dosing syringes by a commercial radiopharmacy
(Cardinal Health) and delivered to the nuclear medicine facility
at Johns Hopkins Hospital. The radioactivity of dosing syringes
was measured in a dose calibrator (CRC 15-W; Capintec) before
and after dosing. Subjects were placed in a supine position,
and the compounded lubricant was injected intrarectally using
a lightly lubricated Luer applicator (product 35-1107; Professional Compounding Centers of America).
Subjects remained supine until the completion of flexible
sigmoidoscopy. Beginning with voiding to empty the bladder
just before gel dosing, timed urine samples were collected at
0–4-, 4–8-, 8–12-, and 12–24-h intervals. Bowel movements
within 24 h after dosing were also collected. To control for the
amount of isotope remaining in the colonic lumen that were
associated with each timed urine sample, interval urine collection was terminated when a bowel movement occurred, and a
new collection was initiated to complete the designated interval.
Within 1.5 h of dosing, subjects underwent conscious sedation.
Sigmoidoscopy was then performed by the study gastroenterologist (L.A.L.), who was blinded to treatment assignment. A
flexible endoscope (model CFQ160S; OlympusAmerica) was introduced to each rectum and advanced to a distance 10 cm
beyond the anal verge. A cytology brush (model 60315; Ballard
Medical Products) was introduced through the endoscope port,
extended from the catheter sheath, and swept in a back-andforth motion against one side of the colonic wall. The brush
was withdrawn into its sheath and removed from the sigmoidoscope. The brush was reextended from the sheath and
clipped off at the neck into a preweighed 4-mL sterile vial. This
procedure was repeated to sample the opposite side of the
colonic wall.
The sigmoidoscope was advanced ∼2.5 cm. Clinical biopsy
forceps (Microvasive 1599; Boston Scientific) were passed
through the endoscope port and opened, and a single pinch
biopsy sample was obtained. The forceps were withdrawn
through the endoscope, and the biopsy sample was carefully
removed and placed into a preweighed 4-mL sterile vial that
contained formalin preservative. With the sigmoidoscope remaining in position, 2 additional pinch biopsy samples were
obtained in the same manner from circumferentially equidistant sites of the colonic wall, 12.5 cm from the anus. The
sigmoidoscope was then advanced to 40 and 42.5 cm, where
brush and biopsy sampling were repeated, respectively. For simplicity, the 12.5- and 42.5-cm sites are referred to as 10 and 40
cm in the results.
Gamma counting. All brush and biopsy samples were immediately taken to the laboratory and weighed and counted on
a gamma counter (Cobra-II Auto-Gamma Counter; Packard)
within a 110–150-keV energy window. Fluid volume retained
on the cytology brushes was determined by weighing the sterile
vials containing brushes immediately after sampling and after
drying. Brush and biopsy sample gamma counts were weightand decay-corrected to reflect counts per gram or per milliliter,
relative to the time of dosing. Isotope concentrations for the
brush and biopsy samples were calculated for each endoscopic
distance (10 and 40 cm) and treatment arm. Values for each
location and treatment arm were averaged and expressed as a
fraction of the dose administered for each subject.
After isotope decay to background, the formalin-preserved
biopsy samples were embedded in paraffin, sectioned, and
stained with hematoxylin-eosin. The slides were read by a gastrointestinal pathologist (M.J.T.) who was blinded to treatment
assignment and sampling level. A median of 4 sections per slide
was prepared (range, 2–10), and all fields containing surface
mucosa were examined. In aggregate, sectioned biopsy epithelial surfaces were a median of 12 mm in length (range, 4–24
mm/slide). All sections on each slide were reviewed, and a
global score for the biopsy sample was assigned. A categorical
scale of epithelial surface denudation was defined as follows:
0, no surface denudation; 1, less than one-third denudation; 2,
one- to two-thirds denudation; 3, more than two-thirds denudation. Biopsy samples from the same treatment assignment
and anatomic level were averaged to determine the grade of
surface denudation observed at that site.
Stool samples were collected in preweighed containers, mixed
with a minimum of 50 mL of water, and stirred to homogeneity.
Then, 1-mL aliquots of urine and stool suspension were
counted in a gamma counter as described above. Assay results
were volume- and decay-corrected and converted to microcuries. Urine and stool results were expressed as a fraction of
the initially administered dose of isotope for the midpoint of
each urine interval or episodic bowel movement, respectively.
Colonic permeability was determined on the basis of urine
radioactivity and expressed as a percentage of total radioactivity
in the colon, both with and without adjustment for loss of stool
radioactivity.
Osmolality measurements. The osmolality of a sampling
of sexual lubricants and vaginal contraceptive gels was assessed
using a vapor pressure osmometer (Model 5520 Vapro vapor
pressure osmometer; Wescor). Semen samples were also tested
after being obtained from consenting volunteers under a separate IRB-approved protocol. Three separate measurements of
each sample were obtained, and median values were determined. The osmolality of samples 13200 mOsm/kg was determined by dilution after the linearity of multiple concentrations
was established. Similarly, the mixture of FemGlide and ID
Glide selected for the iso-osmolar gel was established by linear
regression of 6 concentrations of gel mixtures (r 2 p 0.995). A
1:12 ratio of ID Glide to FemGlide resulted in an iso-osmolar
compound that was then used in the randomized study.
Statistical analysis. Paired comparisons between gel assignment and anatomic location were made for permeability,
biopsy grade, and concentration of isotope outcome variables
using the Wilcoxon signed-rank test; correlations between biopsy and brush gamma activity were assessed using Pearson’s
correlation coefficient (SPSS version 9.00; SPSS).
RESULTS
Median values of a sample of commercial lubricants, products
containing N-9, tap water, Fleet enema and human semen are
shown in table 1. With the exception of one lubricant
(FemGlide), all products tested ranged between 4 and 14 times
the physiologic osmolality of the colon, ∼290 mOsm/kg. Physiologic osmolality is defined by intraluminal infusions of solutions measuring 288 mOsm/kg that induce no net secretion
or absorption of water from the colonic lumen, unlike infusions
of lower and higher osmolality [11]. By contrast, tap water is
hypo-osmolar, and human semen ranges from 17% to 43%
greater than physiologic osmolality.
All subjects tolerated the study procedures well, and no significant adverse events were noted. The median time from gel
dosing to endoscopy was 1.25 h for the iso-osmolar treatment
arm (range, 0.80–1.70 h); that for the hyperosmolar arm was
1.20 h (range, 0.67–1.35 h). The median isotope dose administered was 470 mCi for the iso-osmolar arm (range, 385–532
mCi), compared with 469 mCi for the hyperosmolar arm (range,
413–590 mCi). Neither the median time to sigmoidoscopy nor
the dose of isotope retained was statistically different between
treatment assignments.
In paired comparisons, a statistically significant difference in
gel concentration between the iso-osmolar and hyperosmolar
arms was noted 10 cm distant from the anus (figure 1). The
median iso-osmolar concentration was 54.6% (range, 22.3%–
130.2%) of the administered dose concentration, compared
Hyperosmolar Epithelial Damage • JID 2007:195 (1 March) • 705
Figure 1. Fraction of administered dose per milliliter as measured by radioisotope concentration for each treatment arm and location of sampling.
Results are decay-corrected. For each condition, side-by-side box plots represent the median (central bar), interquartile range (box), and maximum and
minimum values (whiskers), excluding outliers 13 box lengths outside the box (in which case, values are represented by an asterisk). A concentration
gradient is observed in the iso-osmolar (iso) arm, with the greatest concentration of dose remaining at 10 cm. By contrast, no significant gradient is
appreciated between the 10- and 40-cm levels in the hyperosmolar (hyper) arm.
with 8.9% (range, 0.3%–17.6%) in the hyperosmolar arm
(P p .005). The median isotope concentration in the iso-osmolar arm was 9.90 times greater at 10 cm than that at 40 cm
(figure 1) (P p .005). There was no significant difference between concentrations at the 10- and 40-cm levels in the hyperosmolar arm. The same relationship in gamma counts per
gram of tissue was found for biopsy samples, with lower concentrations noted at 40 cm after iso-osmolar dosing, compared
with those for other conditions and sites (data not shown). In
aggregate, the brush and biopsy results were highly correlated
(r p 0.633; P ! .01); site- and gel-specific correlations of brush
and biopsy were also significant, except for hyperosmolar gel
at 10 cm (iso-osmolar 10 cm, r p 0.930 and P ! .001; isoosmolar 40 cm, r p 0.650 and P p .042; hyperosmolar 10 cm,
r p 0.100 and P p .784; hyperosmolar 40 cm, r p 0.788 and
P p .007).
Except for a single subject who had scattered patchy infiltrates seen in both treatment arms, no other gross mucosal
abnormalities were observed during endoscopy. Surface denudation was most severe in the hyperosmolar arm at the 10cm level, where the median epithelial denudation grade was
2.5 (range, 1.5–3.0). This was significantly worse than the grade
of denudation observed in the iso-osmolar arm at the same
level (median, 1.17; range, 0.33–2.00; figure 2) (P p .009). The
706 • JID 2007:195 (1 March) • Fuchs et al.
most common finding in biopsies at 10 cm after the application
of hyperosmolar gel was grade 3 denudation showing nearly
complete loss of epithelial integrity (figure 3). By contrast, after
the administration of iso-osmolar gel, the most common histologic findings were either completely intact surface epithelium
or minimal surface changes, including minor lifting of small
sections of intact epithelial cells with some loss of epithelial
cell mucin content. At 40 cm, comparison of histologic grading
revealed no significant difference between treatment arms. Similarly, there were no differences in denudation grade between
the 10- and 40-cm sites in the iso-osmolar arm.
The greatest difference in cumulative urine isotope accumulation was observed during the 8–12-h interval or at the
10-h midpoint (iso-osmolar median, 3.42%; range, 1.45%–
23.22%; hyperosmolar median, 2.79%; range, 1.13%–4.53%;
P p .037) (figure 4). Ten hours after dosing, 9 of 10 subjects
had a lower percentage of isotope absorbed after administration
of the hyperosmolar gel. By 18 h, 1 additional subject’s isotope
accumulation in urine increased above the paired iso-osmolar
value. Therefore, in 8 of 10 subjects at 18 h, the median accumulation of isotope in urine remained higher when the isoosmolar gel was used (4.20% vs. 3.12% of dose absorbed;
P p .074). This difference was also seen when we controlled
Figure 2. Grade of surface denudation derived from the average of 3 biopsy samples per site, by treatment arm and level of sampling. Significantly
greater denudation was observed with hyperosmolar (hyper) gel at 10 cm, compared with that at 40 cm or that with iso-osmolar (iso) gel at any
location (P ! .007). The most common finding was grade 0 or 1 surface denudation in the iso-osmolar arm at both levels and in the hyperosmolar
arm at 40 cm. The median grade in both the iso-osmolar and hyperosmolar 40-cm levels closely approximated the lower quartile (lower quartile, 0.63;
median, 0.67). The most common finding in the hyperosmolar arm at the 10-cm level was grade 3 surface denudation.
for loss of radioisotope from bowel movements (data not
shown).
DISCUSSION
We found that hyperosmolar gels induced greater denudation
of epithelia at 10 cm in the rectum, compared with iso-osmolar
gels. Coincident with this, the concentration of radiolabel found
in the lumen 10 cm from the anus was less than that observed
with the iso-osmolar formulation, consistent with a greater
dilutional effect secondary to enteric secretion induced by the
hyperosmolar gel. These findings suggest that the administration of a hyperosmolar gel is associated with epithelial “injury”
most pronounced at the site of the initial, most-concentrated
exposure to an osmotic gradient. The dramatic epithelial
changes are not likely so severe in situ but indicate the relative
impact of the gels on epithelial integrity, which is made more
Figure 3. Representative hematoxylin-eosin stains of colonic mucosal biopsy samples showing varying degrees of epithelial disruption. A, Normal
colonic mucosa with an intact epithelial layer (Ep; grade 0, iso-osmolar gel, 10-cm level). B, Epithelial “lifting” with mucin-depleted cells separating
from lamina propria (Lp, arrows; grade 1, iso-osmolar gel, 10-cm level). C, Missing entire epithelial layer, with complete exposure of underlying Lp
and only deep colonic crypts (Ccs) remaining (grade 3, hyperosmolar gel, 10-cm level). Original magnification: A and B, ⫻64; C, ⫻40.
Hyperosmolar Epithelial Damage • JID 2007:195 (1 March) • 707
Figure 4. Difference in isotope accumulation by paired comparison between iso-osmolar and hyperosmolar conditions. Results shown as the midpoint
of each urine collection interval (median value; whiskers represent the upper and lower quartiles on the Y-axis, and the X-axis is the midpoint of
each urine collection interval). Paired comparisons demonstrate more isotope in urine with the iso-osmolar gel, which reached statistical significance
at the 10-h interval, when 9 of 10 subjects had more isotope accumulation in urine with the iso-osmolar condition (P p .037 ). The inset shows each
individual, with the percentage change on a log scale.
apparent by subsequent tissue processing. Although we did not
directly measure gel migration in the present study, our previous observations of the rate of retrograde distribution suggest
that little gel would have migrated to 40 cm by the sampling
time that we used; hence, lower concentrations are expected to
be seen at this distance [14]. The results found when the isoosmolar concentration was used support this expectation.
We hypothesize that the concentration findings are due to a
secretory response from the colonic mucosa that dilutes the
hyperosmolar gel at 10 cm and facilitates retrograde movement
to 40 cm, consistent with results for comparable gel concentrations at 10 and 40 cm in the hyperosmolar arm. The biopsy
concentration comparisons were also consistent with this finding and were highly correlated with the luminal brush results.
The exception to this was the nonsignificant correlation between brush and biopsy at 10 cm after hyperosmolar gel, which
could have been altered by the degree of fluid influx at this
location.
Although 99mTc-DTPA has not been studied as a marker for
intestinal permeability, its use has been proposed for this purpose, because of its similarity in size yet lower radiation exposure than commonly used radioisotopes [26]. Paired comparisons in our study demonstrated that more isotope is
absorbed after iso-osmolar than hyperosmolar gel administra708 • JID 2007:195 (1 March) • Fuchs et al.
tion. This unexpected reduction in isotope absorption by hyperosmolar gels suggests a complicated effect of hyperosmolar
gel on the colonic mucosa that warrants further analysis. It
may be due, in part, to previously described hyperosmolarityinduced increases in mucin production that inhibit small-molecule penetration of the mucosa [10, 13, 27]. A net secretion
of fluid into the colonic lumen may also limit the radioisotope
from moving against a large fluid influx in the hyperosmolar
state.
Given that epithelial injury is seen within 1–1.5 h after lubricant application, damage is present not long after, if not
during, anal intercourse and potential exposure to infectious
seminal fluid. Whether damage to the physical epithelial barrier
increases susceptibility to HIV transmission is unknown, but
it is certainly plausible. In assessing the potential for increased
risk from hyperosmolar sexual lubricants, one must consider
whether sufficient lubricant used with anal intercourse comes
in contact with the epithelial surface to cause denudation and
whether a volume !10 mL induces similar toxic effects. One
published study indicated that up to 65% of MSM report using
⭓2 teaspoons of lubricant per sexual occasion; however, it is
not known how much gel was placed intrarectally and whether
this volume represents repeated application of smaller amounts
of gel [28]. A more recent study by the same authors dem-
onstrated that up to 35 mL of lubricant can be used comfortably
during receptive anal intercourse [29].
Our findings raise cautionary flags about the development
of hyperosmolar microbicide gel vehicles. More immediately,
given that the majority of lubricants we measured are hyperosmolar, the potential for hyperosmolarity-induced epithelial
damage and possibly increased HIV susceptibility raises significant concern. If this finding can be generalized to hyperosmolar enema preparations, in which the issues of volume
and mucosal access are more likely sufficient to induce mucosal
damage, the concern is even greater. Careful study to identify
safe alternatives for lubrication and bowel preparation is
needed, lest substitutes are selected that themselves pose an
unknown risk.
Although the lubricants we used in the present study were
not controlled for viscosity or pH, our prior observations
showed that viscous hyperosmolar gels were rapidly diluted to
a volume several times greater than that administered. This
volume expansion would reduce gel viscosity shortly after rectal
administration. The hyperosmolar gel used in the present experiment, ID Glide, has a pH of ∼4.8. The effect of pH on
rectal epithelium is unknown; however, animal studies evaluating rectal susceptibility to HIV infection used placebo gels
with a similar pH and found no demonstrable effect on the
epithelial layer [30]. Furthermore, lowering of the pH in the
colonic lumen has been associated with hastening epithelial
repair [31, 32]. We have subsequently tested the buffering capacity of ID Glide in vitro and found that it is fully neutralized
by 0.014 mEq of 0.1 N NaOH/g of gel, indicating that the
lubricant has low buffering capacity. Taken together, these facts
argue against gel acidity causing the observed cytotoxicity. The
lubricants that we studied do not contain N-9 or other detergents that would explain the results of the study. We do not
know whether any of the other ingredients found in ID Glide
could induce epithelial disruption; however, these ingredients
are frequently found in other personal lubricants.
Given that we sampled a single time point, no conclusion
can be drawn about the timing of repair of epithelial surface
changes. Other researchers have shown that repair likely begins
rapidly after injury, with the development of an intact barrier
by 2 h and a border histologically indistinct from baseline 8 h
after tissue insult [8, 24]. In future studies, it will be important
to understand the time course of potentially enhanced HIV
susceptibility after toxic insults, because any rectal microbicide
or lubricant gel may be reapplied in shorter intervals, possibly
resulting in cumulative epithelial damage.
It is interesting to note that a correlation has been observed
between the degree of leakage of lubricant from the vaginal
vault and increased gel osmolality—a finding consistent with
our observations of increased fluid flux with hyperosmolar gels
in the distal colon [33]. Additionally, because endocervical and
colonic tissue share a similar epithelial structure, the effect of
gel osmolality on the endocervix may also be of concern, particularly in the setting of cervical ectopy. Ectopy has, in some
instances, been associated with increased relative risk for HIV
infection, suggesting the importance of understanding whether
hyperosmolar lubricants and potential HIV microbicides induce similar changes with vaginal dosing [34, 35]. Thus, understanding the effect of osmolality-induced epithelial changes
on HIV susceptibility is critical to the development of safe
formulations of topical microbicides with the potential to affect
the outcome of future microbicide efficacy studies.
Under our experimental conditions, epithelial “injury” was
greater after exposure to hyperosmolar gels. The pattern of
dilution and our previous observations of gel migration are
consistent with the finding that injury is greatest at the site of
the initial, most concentrated exposure, and an enteric response
to the hyperosmolar gel results in mucosal fluid secretion and
dilution of intraluminal gel concentration. We do not know
whether epithelial disruption results in greater susceptibility to
HIV infection, but it is biologically plausible. Our findings may
have immediate implications for transmission risk if the mucosal changes observed are associated with increased HIV susceptibility and if the rectal mucosa is exposed to volumes of
sexual lubricants similar to those in our study. This concern
should also extend to hyperosmolar enemas, which may have
the same effect on the colonic epithelium. The epithelial
changes caused by hyperosmolar gels and their potential for
increasing susceptibility to HIV infection indicate that such
vehicles may be poor candidates in need of further study before
they may be safely used for microbicide development.
Acknowledgments
We thank Dr. Thomas Moench, ReProtect, Inc.; Dr. Richard A. Cone;
the Biophysics Department, Johns Hopkins University; and James Johnson
and Elizabeth Martinez, Division of Clinical Pharmacology, Johns Hopkins
University School of Medicine, for their technical support of the project.
References
1. Centers for Disease Control and Prevention. HIV/AIDS surveillance
report, 2004. Vol. 16. Available at: http://www.cdc.gov/hiv/stats/hasrlink
.htm. Accessed 31 March 2006.
2. Baldwin JI, Baldwin JD. Heterosexual anal intercourse: an understudied, high-risk sexual behavior. Arch Sex Behav 2000; 29:357–73.
3. Erickson PI, Bastani R, Maxwell AE, Marcus AC, Capell FJ, Yan KX.
Prevalence of anal sex among heterosexuals in California and its relationship to other AIDS risk behaviors. AIDS Educ Prev 1995; 7:
477–93.
4. Halperin DT. Heterosexual anal intercourse: prevalence, cultural factors, and HIV infection and other health risks, part I. AIDS Patient
Care STDS 1999; 13:717–30.
5. Misegades L, Page-Shafer K, Halperin D, McFarland W. Anal intercourse among young low-income women in California: an overlooked
risk factor for HIV? AIDS 2001; 15:534–5.
6. Van Damme L, Ramjee G, Alary M, et al. Effectiveness of COL-1492,
Hyperosmolar Epithelial Damage • JID 2007:195 (1 March) • 709
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
a nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex workers: a randomised controlled trial. Lancet 2002; 360:971–7.
Phillips DM, Taylor CL, Zacharopoulos VR, Maguire RA. Nonoxynol9 causes rapid exfoliation of sheets of rectal epithelium. Contraception
2000; 62:149–54.
Phillips DM, Sudol KM, Taylor CL, Guichard L, Elsen R, Maguire RA.
Lubricants containing N-9 may enhance rectal transmission of HIV
and other STIs. Contraception 2004; 70:107–10.
Meisel JL, Bergman D, Graney D, Saunders DR, Rubin CE. Human
rectal mucosa: proctoscopic and morphological changes caused by laxatives. Gastroenterology 1977; 72:1274–9.
Leriche M, Devroede G, Sanchez G, Rossano J. Changes in the rectal
mucosa induced by hypertonic enemas. Dis Colon Rectum 1978; 21:
227–36.
Billich CO, Levitan R. Effects of sodium concentration and osmolality
on water and electrolyte absorption from the intact human colon. J
Clin Invest 1969; 48:1336–47.
Kameda H, Abei T, Nasrallah S, Iber FL. Functional and histological
injury to intestinal mucosa produced by hypertonicity. Am J Physiol
1968; 214:1090–5.
Sakata T, von Engelhardt W. Influence of short-chain fatty acids and
osmolality on mucin release in the rat colon. Cell Tissue Res 1981;
219:371–7.
Fuchs EJ, Wahl R, Macura K, Leal J, Grohskopf L, Hendrix CW. Imaging the distribution of a rectal microbicide gel and semen surrogate
in the lower gastrointestinal tract [abstract PII-35]. Clin Pharmacol
Ther 2005; 77:P60.
Moss AR, Osmond D, Bacchetti P, Chermann JC, Barre-Sinoussi F,
Carlson J. Risk factors for AIDS and HIV seropositivity in homosexual
men. Am J Epidemiol 1987; 125:1035–47.
Chmiel JS, Detels R, Kaslow RA, Van Raden M, Kingsley LA, Brookmeyer R. Factors associated with prevalent human immunodeficiency
virus (HIV) infection in the Multicenter AIDS Cohort Study. Am J
Epidemiol 1987; 126:568–77.
Coates RA, Calzavara LM, Read SE, et al. Risk factors for HIV infection
in male sexual contacts of men with AIDS or an AIDS-related condition. Am J Epidemiol 1988; 128:729–39.
Schreeder MT, Thompson SE, Hadler SC, et al. Hepatitis B in homosexual men: prevalence of infection and factors related to transmission. J Infect Dis 1982; 146:7–15.
Hylton J, Hendrix CW, Fuchs EJ. An assessment of sexual practices
affecting the feasibility of microbicide development among MSM [abstract 02667]. In: Microbicides, 2004 (London). Madison, WI: Omnipress, 2004:166.
O’Morain CA, Abelow AC, Chervu LR, Fleischner GM, Das KM. Chromium 51-ethylenediaminetetraacetate test: a useful test in the assessment of inflammatory bowel disease. J Lab Clin Med 1986; 108:430–5.
710 • JID 2007:195 (1 March) • Fuchs et al.
21. Rask-Madsen J, Schwartz M. Absorption of 51Cr-EDTA in ulcerative
colitis following rectal instillation. Scand J Gastroenterol 1970; 5:361–8.
22. Jenkins RT, Ramage JK, Jones DB, Collins SM, Goodacre RL, Hunt
RH. Small bowel and colonic permeability to 51Cr-EDTA in patients
with active inflammatory bowel disease. Clin Invest Med 1988; 11:
151–5.
23. Levitan R, Brudno S. Permeability of the rectosigmoid mucosa to tritiated water in normal subjects and in patients with mild idiopathic
ulcerative colitis. Gut 1967; 8:15–9.
24. Feil W, Lacy ER, Wong YM, et al. Rapid epithelial restitution of human
and rabbit colonic mucosa. Gastroenterology 1989; 97:685–701.
25. Tabet SR, Surawicz C, Horton S, et al. Safety and toxicity of nonoxynol9 gel as a rectal microbicide. Sex Transm Dis 1999; 26:564–71.
26. Bjarnason I, MacPherson A, Hollander D. Intestinal permeability: an
overview. Gastroenterology 1995; 108:1566–81.
27. Rubsamen K, Hornicke H. Influence of osmolality, short chain fatty
acids and deoxycholic acid on mucus secretion in the rat colon. Pflugers
Arch 1982; 395:306–11.
28. Carballo-Dieguez A, Stein Z, Saez H, Dolezal C, Nieves-Rosa L, Diaz
F. Frequent use of lubricants for anal sex among men who have sex
with men: the HIV prevention potential of a microbicidal gel. Am J
Public Health 2000;90:1117–21.
29. Carballo-Dieguez A, Exner T, Dolezal C, Pickard R, Lin P, Mayer KH.
Rectal microbicide acceptability: results of a volume escalation trial.
Sex Transm Dis 2006 [electronically published ahead of print].
30. Patton DL, Cosgrove Sweeney YT, Rabe LK, Hillier SL. Rectal applications of nonoxynol-9 cause tissue disruption in a monkey model.
Sex Transm Dis 2002; 29:581–7.
31. Rafter JJ, Eng VW, Furrer R, Medline A, Bruce WR. Effects of calcium
and pH on the mucosal damage produced by deoxycholic acid in the
rat colon. Gut 1986; 27:1320–9.
32. Venkatraman A, Ramakrishna BS, Pulimood AB. Butyrate hastens restoration of barrier function after thermal and detergent injury to rat
distal colon in vitro. Scand J Gastroenterol 1999; 34:1087–92.
33. Zeitlin L, Long N, Hosobuchi C, et al. Leakage of 3 commercial vaginal
gels in women [abstract A14]. Topical Microbicides Development and
Evaluation Workshop. In: Microbicides, 2003 (Washington, DC). Madison, WI: Omnipress, 2003.
34. Moss GB, Clemetson D, D’Costa L, et al. Association of cervical ectopy
with heterosexual transmission of human immunodeficiency virus: results of a study of couples in Nairobi, Kenya. J Infect Dis 1991; 164:
588–91.
35. Myer L, Wright TC Jr, Denny L, et al. Nested case-control study of
cervical mucosal lesions, ectopy, and incident HIV infection among
women in Cape Town, South Africa. Sex Transm Dis 2006; 33:683–7.