The intact cervicovaginal epithelium appears to be a significant barrier to HIV transmission, as indicated by the relatively low efficiency of HIV acquisition after vaginal intercourse,1 and by the observation that the reliable transmission of simian immunodeficiency virus (SIV) via intravaginal inoculation requires approximately 10 000-fold more virus than that required by the intravenous route.2 Furthermore, it is biologically plausible to expect that disrupting the physical integrity of the epithelium will compromise this barrier to viral transmission. Epithelial barrier disruption could be caused by cytotoxic intravaginal products (therapeutic, cosmetic, or prophylactic), by secondary inflammatory reactions to intravaginal products, or by physical disruption (trauma).
Intravaginal agents, including microbicide candidates, can damage genital tract epithelium,3–5 and a recent clinical trial of a nonoxynol-9 microbicide candidate indicates that such damage may result in increased susceptibility of women to the acquisition of HIV.5 Animal model evidence supports this possibility: mice treated with nonoxynol-9 and other detergents have been shown to be more susceptible to vaginal challenge with HSV-2,6 and nonoxynol-9 also increases susceptibility to rectal challenge with this virus.7 Similar results were observed with a second sexually transmitted disease (STD) pathogen, Chlamydia trachomatis, wherein exposure to a single dose of the microbicide chlorhexidine digluconate subsequently caused a 100-fold increase in susceptibility to vaginal challenge.8 Moreover, the vaginal epithelium of mice exposed to a single application of chlorhexidine later became markedly friable; vaginally inserted swabs became covered with blood. Together, these results indicate that microbicide candidates can cause epithelial compromise, which manifests both as an increased propensity to bleed, and increased susceptibility to pathogens.
Trauma may also compromise the epithelial barrier, and increase susceptibility. Evidence of a possible increase in the susceptibility of women to HIV in the context of coital trauma is provided by several observational studies that detected an association between postcoital vaginal bleeding and HIV infection.1,9
Disruption of sufficient severity to allow the leakage of blood into the lumen of the genital tract implies a depth of disruption beyond the basal lamina to the capillary level. Breaches of sufficient size to allow red blood cells (RBCs) out would be of sufficient size to allow the entry of HIV (∼70 times smaller diameter) and would also probably be large enough to facilitate the entry of HIV-infected cells. Such breaches may be sufficient to release blood and increase susceptibility, but might be below the detection limits of colposcopy. Furthermore, the upper tract cannot be assessed by colposcopy, and tools to detect epithelial disruption may be useful, because microbicides are now known to travel to the upper tract.10 Moreover, the loss of barrier function in the upper tract may affect susceptibility to HIV and other STDs because it has been shown that potentially infectious semen can ascend to the endometrial cavity.11
Epithelial disruptions can result in clinically evident vaginal hemorrhage, but more subtle disruptions might result in subclinical microhemorrhage not evident to the woman, or on clinical examination. We have therefore developed sensitive methods to detect human hemoglobin and intact RBCs in cervicovaginal secretions, to explore their utility in the detection and quantification of cervicovaginal micro-hemorrhage. We applied these and a previously available assay (Hemastix) to cervicovaginal lavage specimens obtained from healthy premenopausal female volunteers, recruited for studies intended to observe the frequency of coital microtrauma. We report here the results characterizing and comparing the three assays, and propose the utility of these methods for assessing epithelial disruption induced by vaginal microbicides and other intravaginal products. We expect that this approach may also allow the assessment of epithelial disruption of the upper tract, which cannot be observed by colposcopy.
Patients and clinical procedures
Menstrual cycle study
To learn which portion of the menstrual cycle would provide a relatively low background for microhemorrhage assays, six participants were recruited to obtain lavages at many times during a single menstrual cycle. Students were recruited at the Johns Hopkins University, and were premenopausal, and in general good health. They were instructed to perform lavage procedures at or near the end of visibly discernable bleeding of their next menses, and repeat the procedure every 4 days thereafter for a total of four occasions. The study was approved by the Review Board on the Use of Human Subjects of Johns Hopkins University.
Coital microtrauma studies
Participants were recruited from among patients and staff at the Johns Hopkins Bayview Medical Center, were between the ages of 18 and 45 years, were premenopausal, in general good health without ongoing significant cervicovaginal abnormality by history, were sexually active in a monogamous heterosexual relationship with a current partner for 3 or more months with a coital frequency of four or more times per month not using barrier contraceptives, and with typical menstrual period length of 26 or more days without intermenstrual spotting in the past three cycles. The studies were approved by the Johns Hopkins Medicine Institutional Review Board.
For the first Coital Microtrauma Study (CMT-1), after a screening and enrollment visit, participants were scheduled to return for a precoital colposcopy visit between day 18 and day 22 of their menstrual cycle. They were asked to be abstinent for 3 or more days before this visit. During this precoital colposcopy visit, participants were taught and performed a lavage of the introitus, and then of the vagina, with saline, using a syringe and catheter (see below). The labia, introitus, vagina and cervix were then examined by colposcopy with 15x magnification. A speculum pelvic examination was performed, and in some cases, a second lavage procedure was performed after the examination to assess possible trauma from the examination. Participants were instructed to repeat the lavage procedure in the evening 3 days after the colposcopy examination, and the following morning to have intercourse, repeat the lavage procedure approximately 20 min after intercourse, and return for post-coital colposcopy within 2–3 h of intercourse.
For the second Coital Microtrauma Study (CMT-2), participants were instructed in and performed the lavage procedures described below at a screening and enrollment visit, but had no colposcopy or speculum examination. They were instructed to obtain pre and postcoital lavages approximately 1 h before and 20 min after intercourse on three occasions. Each of these study intercourses was to occur between day 18 and day 22 of the menstrual cycle and be preceded by 3 or more days of abstinence. One was to take place without condoms, one with an unlubricated latex condom, and one with a latex condom lubricated with a spermicide-free aqueous lubricant.
Participants in both CMT-1 and CMT-2 were asked to call the clinic at the onset of their next menses. This was to enable the recognition of early menses as a possible cause of increased blood detected during the final study sampling.
Baseline sample assays of CMT-1 and CMT-2 are reported here. Pre- and postcoital comparisons will be reported elsewhere.
Participants lavaged the introitus with 10 ml saline dispensed directly from a syringe, and collected in a sample collecting basin positioned under a toilet seat. The vagina was next lavaged using a second aliquot of saline dispensed by a syringe via a 18F urinary catheter inserted 3–4 in into the vagina and collected in a separate basin. Two milliliters of each sample was transferred from each basin to a tube containing an equal volume of fixative (10% neutral buffered formalin) and stored at room temperature, and the remainder of the sample was transferred to another tube containing 0.5 ml sterile 5% bovine serum albumin (BSA) in phosphate buffered saline (PBS), frozen within 1 h, and stored frozen until analysis. (BSA was added to the collected lavages as described above because pilot experiments showed the need to add a carrier protein to the lavage fluid to achieve full sensitivity and reproducibility in the hemoglobin enzyme-linked immunosorbent assay; ELISA.)
Blood detection assays
Visual detection of blood
Freshly-drawn, EDTA-anticoagulated, venous blood was serially diluted in saline, and 50 JLLI drops were placed on a flesh-colored surface and visually compared with an adjacent 50 JLLI drop of saline without blood.
Commercially available, heme-based assay
Several established heme-based clinical assays are available, including tests for fecal occult blood, and dipsticks to detect blood in urine. Both are fairly sensitive, but have somewhat subjective endpoints, and are either non-quantitative, or only give rough quantitation. We elected to test Hemastix on cervicovaginal secretions because this product is in widespread clinical use. It is inexpensive, quick, and simple to use, and is more quantitative than the Hemoccult assay.
Hemastix test strips (Bayer Diagnostics, Tarrytown, NY, USA) were exposed to thawed, sonicated lavage samples (see sample procedures for ELISA below), and were scored according to the manufacturer's instructions.
Because of the poor quantitation of the available heme-based assays described above, we developed two alternative assays, an RBC enumeration assay, and an ELISA for human hemoglobin, with the goal of creating fully quantitative assays with greater sensitivity and more objective endpoints.
Red blood cell assay
In this assay, RBCs were fixed, permeabilized, and stained with anti-hemoglobin antibody, and were counted with a fluorescent microscope. Briefly, at the time of assay, fixative was removed from the cellular lavage samples by centrifugation washes with PBS, and the cells were permeabilized by a 1 h incubation with cold 1:1 acetone/methanol, washed with PBS, and stained with a 1:10 dilution of FITC-labeled affinity purified goat anti-human-hemoglobin (Bethyl Laboratories, Montgomery, TX, USA) and 0.4 μM ethidium homodimer-1 (Molecular Probes, Eugene, OR, USA) to stain nucleated cells. After washing by centrifugation, RBCs were enumerated by epifluorescent microscopy in a hemocytometer using filters appropriate for FITC/TRITC. Pilot studies showed that the intensity of RBC staining did not decrease during storage at room temperature for at least 6 months.
ENZYME-LINKED IMMUNOSORBENT ASSAY DETECTION OF HUMAN HEMOGLOBIN
At the time of the assay, samples were thawed, and 0.25 ml was transferred to a microcentrifuge tube and sonicated to release hemoglobin from RBCs [2 min in Bronson (Danbury, CT, USA) model 450, immersed in circulating water in a cup-horn transducer, power setting 7, duty cycle 90%, shown in pilot experiments to give 100% RBC lysis]. Lysed samples were diluted in PBS/3% BSA as required, guided by rough quantitation from Hemastix. Samples were assayed in duplicate on 96-well microtiter plates coated overnight with 2 |J.g/ml unlabeled sheep anti-human-hemoglobin antibody (Bethyl Laboratories). Bound hemoglobin was subsequently detected by incubation with a 1:1000 dilution of peroxidase-conjugated sheep anti-human hemoglobin antibody (Bethyl Laboratories) and then developed with colorometric substrate (o-phenylenediamine dihydrochloride). The hemoglobin concentration (ng/ml) was calculated by comparing the optical density of each sample well read at 450 nm, with a standard curve run on the same plate and prepared using a purified hemoglobin standard (Bethyl Laboratories) diluted in PBS/3% BSA.
Samples testing negative (below threshold) were assigned values of half the threshold value for the assay. Medians and interquartile boundaries were determined. Data for RBC and hemoglobin assays were log-transformed, and non-parametric correlations (Spearman r) between assays were calculated, because even the log-transformed data were not normally distributed. Log-transformed values for hemoglobin in pre- and post-speculum/colposcopy examinations were compared by two-tailed, paired t test.
Visual detection of blood
Dilutions of blood at and below 1:100 were visually recognizable as blood-tinged, whereas a dilution of 1:200 was indistinguishable from saline.
This method was simple and rapid, and provided readings of negative, trace, 1+, 2+, 3+, and greater than 3+ by comparison with the chart provided with the test strips. However, some resultant colors fell between adjacent illustrated color values, forcing an arbitrary choice. The threshold sensitivity listed on the package insert was 5000–20 000 RBCs/ml.
Red blood cell assay
The labeled RBCs were easily detectable as bright green round cells with distinct borders under fluorescent microscopy using FITC or FITC/TRITC filter sets, and were completely free of staining with ethidium homodimer-1, (which stains cells containing nuclear DNA bright red under the FITC/TRITC filter set, and was thus helpful in distinguishing RBCs from occasional green fluorescent artifacts observed in nucleated epithelial cells). The volume of fluid examined microscopically was equivalent to one-tenth of a milliliter of the original lavage fluid, and thus the threshold for detection was approximately 10 RBCs per milliliter of lavage fluid. Approximately 20 samples could be assayed in one day.
Hemoglobin enzyme-linked immunosorbent assay
Pilot experiments to determine the detection threshold of blood lavage fluid showed that the detection efficiency in lavage fluid compared with buffer was 80%, and the detection limit was 5 ng/ml (data not shown). Approximately 45 samples could be assayed in one day.
Hemoglobin in vaginal lavage as a function of menstrual cycle
This pilot study was undertaken to guide the timing of sampling in subsequent studies. Six women each returned between two and four vaginal lavage samples for hemoglobin determinations. The hemoglobin levels are shown in Figure 1, and show that hemoglobin levels fell to a nadir at or shortly after midcycle.
EFFECT OF SPECULUM EXAMINATION ON HEMOGLOBIN DETECTION
Speculum/colposcopic examinations did not result in increased hemoglobin detection in lavages. The geometric mean hemoglobin concentrations for the eight paired sets of vaginal lavages obtained immediately before and after speculum/colposcopy examinations were 19 and 10 ng/ml, respectively (P = 0.16). (The reduction in means of the post-speculum samples relative to the pre-speculum samples may be attributable to depletion caused by the first lavage.) Similarly, there was no increase in hemoglobin in introital lavages after speculum examinations (data not shown). Further evidence that speculum/colposcopy examinations did not raise blood concentrations is the observation that the geometric mean hemoglobin for 19 precoital lavages in CMT-1 that were preceded by speculum/colposcopy examinations 3 days before were virtually identical to the geometric mean of the 17 precoital vaginal lavages in CMT-2, wherein no speculum examinations were performed (78 versus 79 ng/ml, respectively).
The characteristics of the assays are summarized in Table 1, and the median values and quartile boundaries for baseline samples in the coital microtrauma studies are shown in Figure 2.
Visible detection is simple but insensitive and non-quantitative. The Hemastix method is much more sensitive than visual detection, but gives limited (discontinuous) quantitation. Moreover, the relative durability of heme would be expected to limit its utility as a marker for recent bleeding. The RBC assay has a very high sensitivity, detecting blood at many million-fold dilution of whole blood, and at more than a million-fold below the visual detection threshold. The hemoglobin assay was also highly sensitive, although slightly less than the RBC assay.
The medians for these baseline vaginal lavage samples are 32 ng/ml hemoglobin, 39 RBCs/ml, and a Hemastix reading of 1+ for heme. Although the full ranges are broad as a result of a few samples with very high values, most pre-intervention samples gave readings sufficiently low to provide a background against which subsequent micro-hemorrhage might be detected.
All three quantitative assays correlated with each other (Spearman r = 0.59 for hemoglobin versus RBCs, 0.60 for hemoglobin versus Hemastix, and 0.53 for RBCs versus Hemastix) as expected because all detect blood-related analytes. However, as also expected, these correlations were modest, because the analytes are diverse in character and would have different persistence times in the genital tract (see Discussion). The percentages of all samples that fell below the assay threshold were 18% for Hemastix, 23% for RBCs, and 27% for hemoglobin. These observations also suggest differences in longevity of the analytes, because otherwise sub-threshold values would be expected to be far more infrequent with the most sensitive (RBC) assay (see Discussion). Finally, the median values for hemoglobin and RBCs from Figure 2 suggest that only a small percentage (∼4%) of the hemoglobin detected would be accounted for by the number of RBCs detected (each containing 30 pg hemoglobin), suggesting that RBCs are far more labile than hemoglobin.
Menstrual cycle study
Figure 1 shows the hemoglobin concentration in vaginal lavages as a function of the menstrual cycle day. Hemoglobin concentrations reached a nadir by approximately midcycle (day 14), and indicate that the luteal phase is the portion of the cycle most suitable to obtain the lowest baseline values.
Disruption of the female genital tract epithelium is a plausible risk factor for HIV and perhaps other sexually transmitted infection acquisition, and concern over such disruption is one of the main motivations for including colposcopy as part of the assessment of microbicide safety evaluations.4 A disruption of sufficient severity to allow the leakage of blood into the lumen of the genital tract implies a depth of disruption beyond the basal lamina to the capillary level. For these reasons, an assessment of epithelial damage by the quantitation of hemorrhage or microhemorrhage seems particularly pertinent to the studies of epithelial change that might increase HIV or other STD susceptibility.
Although menstrual bleeding and other intercurrent conditions that result in hemorrhage in the female genital tract create background conditions that impose certain limitations when using measurements of microhemorrhage in lavages to assess epithelial integrity, these limitations appear manageable. Background levels are generally low in the luteal phase of the cycle, as observed in our menstrual cycle study, as well as in the larger sample of baseline observations in our coital microtrauma studies. The excessive proximity of a subsequent menses can be effectively ruled out as a likely source of blood in a ‘post-exposure’ sample by a telephone contact to ascertain the actual date of onset of the following menses. Finally, our data indicate that careful speculum examinations and colposcopy do not cause increases in microhemorrhage in cervicovaginal lavages.
The new assays described here show promise as tools to detect microhemorrhage. They are sufficiently sensitive to measure the hemoglobin or RBC concentrations in vaginal lavages from most women who are not using intravaginal products, even in the luteal phase of the menstrual cycle, when these values fall to their lowest levels. This indicates that the new assays have achieved sufficient sensitivity for this application, and that further increases in sensitivity would have limited utility for the present purpose. Moreover, in contrast to previously available heme-based assays, the new assays are fully quantitative. In addition to their application to the detection and quantitation of trauma, we believe they will also prove useful in the evaluation of epithelial disruption caused by microbicide toxicity.
On the basis of our studies to date, we make the following recommendations for the use of tests of microhemorrhage in the evaluation of microbicides. The earliest time appropriate for baseline measurements is approximately at midcycle (day 14). To avoid overlap of the product exposure period with the onset of the subsequent menses, exposure periods should thus be limited to 7–10 days, thus allowing completion by day 21 or day 24.
We found that the introital and vaginal lavage procedures gave similar results, and we recommend vaginal lavage alone for subsequent studies. Vaginal lavage is more easily incorporated into standard clinical examination procedures, particularly when performed during pelvic examination rather than with the self-sampling catheter method used in our studies. Moreover, it samples more extensively than the introital lavage procedure, and is often already routinely obtained for the measurement of other analytes.
The Hemastix method is the simplest and most straightforward assay, and although less sensitive and quantitative than the RBC and hemoglobin ELISA, it may be adequate to detect bleeding if the magnitude of the effect is high. Hemastix also proved useful in our studies in choosing appropriate dilutions to be tested in the hemoglobin ELISA. If the Hemastix method is used alone for measuring microhemorrhage, those samples that test more than 3+ should be diluted and retested as needed to allow improved quantitation in the upper ranges of heme concentration.
However, the hemoglobin ELISA or RBC enumeration assays described here seem more promising than heme assays because of their greater sensitivity, better quantitation, and the greater statistical power of continuous variables. We observed that despite the lesser sensitivity of the Hemastix assay, a lower percentage of Hemastix values were below the threshold than RBC or hemoglobin assays. This is compatible with a greater stability of heme compared with the RBCs or hemoglobin. It is known that RBCs are unstable at pH below 5,12 and hemoglobin is probably also less stable than heme. RBCs and hemoglobin are thus probably more reliable markers of recent hemorrhage as opposed to heme, an accumulated blood breakdown product of uncertain age and possible long persistence from an earlier event such as preceding menses.
The new assays are not commercially available, but are easily within the expertise of most research laboratories familiar with standard ELISA techniques or with access to a centrifuge and fluorescent microscope. Of the two, the ELISA is the easiest to learn and standardize, and is less labor intensive, especially for large numbers of samples.
We caution that the possibility of interference as a result of residual microbicide in lavages must be investigated when using any of these assays after microbicide exposure. For example, RBCs may be lysed by microbicidal detergents, lytic peptides, or acid buffers, perhaps making the hemoglobin assay more advantageous in these settings. Conversely, polyanionic microbicides are known to bind to a wide variety of proteins, and the RBC assay might be preferable when these agents are being evaluated. Pilot studies will be needed to determine the existence and extent of such potential interferences, and possible alterations of procedures to lessen or avoid them.
We conclude that the new microhemorrhage assays are highly sensitive when applied to cervicovaginal or introital lavage samples, and that background levels of hemoglobin are sufficiently low during the luteal phase to justify exploring the measurement of microhemorrhage as a means to detect and quantitate epithelial disruption. These assays may thus provide useful tools in the evaluation of microbicide toxicity, and may be complementary to colposcopy and the quantitation of inflammatory markers in cervicovaginal lavages discussed elsewhere in this symposium.
This work was supported by National Institutes of Health grant PO1AI45967. The authors would like to thank Evanson Mukara, Lesley Donoho, and Shannon Smith for technical assistance.
1. Padian NS, Shiboski SC, Jewell NP. The effect of number of exposures on the risk of heterosexual HIV transmission. J Infect Dis, 1990;161:883–887.
2. Sodora DL, Gettie A, Miller CJ, et al. Vaginal transmission of SIV: assessing infectivity and hormonal influences in macaques inoculated with cell-free and cell-associated viral stocks. AIDS Res Hum Retroviruses
, 1998;14 (Suppl. 1):S119–S123.
3. Roddy RE, Cordero M, Cordero C, et al. A dosing study of nonoxynol-9 and genital irritation. Int J STD AIDS
4. Goeman J, Ndoye I, Sakho LM, et al. Frequent use of menfegol spermicidal vaginal foaming tablets associated with a high incidence of genital lesions. J Infect Dis
, 1995; 171:1611–1614.
5. Van Damme L, Ramjee G, Alary M, et al. COL-1492 Study Group. Effectiveness of COL-1492, a nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex workers: a randomized controlled trial. Lancet
6. Abusuwwa RF, Hoen TE, Cone RA, et al. Mouse model for quantifying acute cervicovaginal toxicity
caused by microbicides upon contact with columnar epithelium. Presented at: Microbicides. Antwerp, 2002 [Abstract A-176].
7. Phillips DM, Zacharopoulos VR. Nonoxynol-9 enhances rectal infection by herpes simplex virus in mice. Contraception
8. Achilles, SL, Shete PB, Whaley KJ, et al. Microbicide
efficacy and toxicity
tests in a mouse model for vaginal transmission of Chlamydia trachomatis. Sex Transm Dis
9. Guimaraes MD, Vlahov D, Castilho EA. Postcoital vaginal bleeding as a risk factor for transmission of the human immunodeficiency virus in a heterosexual partner study in Brazil. Rio de Janeiro Heterosexual Study Group. Arch Intern Med
10. Barnhart KT, Stolpen A, Pretorius ES, et al. Distribution of a spermicide containing Nonoxynol-9 in the vaginal canal and the upper female reproductive tract. Hum Reprod
11. Parsons AK, Cone RA, Moench TR. Uterine uptake of vaginal fluids: implications for microbicides. Presented at: Microbicides. Antwerp, 2002 [Abstract B-175].
12. Ivanov, IT. Low pH-induced hemolysis of erythrocytes is related to the entry of the acid into cytosol and oxidative stress on cellular membranes. Biochem Biophys Acta
Keywords:© 2004 Lippincott Williams & Wilkins, Inc.
Microbicide; toxicity; microhemorrhage; epithelial disruption; hemoglobin; menses