Vincent, Kathleen L. MD*; Bell, Brent A. BS*; Rosenthal, Susan L. PhD*; Stanberry, Lawrence R. MD, PhD*; Bourne, Nigel PhD*; Sweeney, Yvonne T. Cosgrove BA†; Patton, Dorothy L. PhD†; Motamedi, Massoud PhD*
TOPICAL MICROBICIDES HAVE the potential to prevent the spread of HIV and other sexually transmitted diseases. However, safety is paramount because these products will be used repeatedly in healthy women. Colposcopy is currently the standard imaging technique for evaluation of safety of topical microbicides1; however, with limited sensitivity, it can only detect the presence of significant changes on the surface epithelium, such as focal bleeding, ulcers, peeling, color change, and vasculature changes. In recent years, concern has been raised about a topical vaginal product, nonoxynol-9 (N-9), that was believed to be safe, but that with frequent use, may actually increase transmission of HIV.2,3 The mechanism of increased transmission of HIV by N-9 is believed to involve epithelial injury and subsequent inflammatory response.4 Use of N-9 containing vaginal products does not induce epithelial changes that are detected by colposcopic examination,5 suggesting the need for a high-resolution method to image subepithelial structures in order to detect tissue changes and injury related to product use.
Optical coherence tomography (OCT) has potential for evaluating tissue disruption and inflammation because of its ability to visualize structural layers up to 1.5 mm below the surface of the epithelium, producing an image of tissue morphology.6 Correlations of OCT images and histology have been reported in clinical studies involving the evaluation of epithelial disorders, including neoplasms and inflammatory conditions of the skin,7 reproductive tract,8–11 gastrointestinal tract,12 and bladder.13 The unique features of the high-resolution OCT method could enable investigators to acquire multiple images of tissue response to microbicide exposure over time and following multiple applications of microbicides.
The purpose of the current study was to demonstrate the feasibility of OCT imaging applied to microbicide safety evaluations. High-resolution OCT visualization of macaque vaginal and cervical epithelial tissue under ex vivo and in vivo conditions was performed in parallel with colposcopy and histologic evaluation.
Materials and Methods
OCT offers micrometer resolution and cross-sectional imaging capabilities. It operates on the principle of low-coherence interferometry, with light split into sample and reference beams. The tissue is illuminated by the light from the sample beam, and the back-scattered light is collected and recombined with the reference beam to generate an image. Differences in size, composition, and distribution of particles result in variations in reflective properties of internal tissue microstructures. These variations in intensity of back-scattered light are used to create maps of tissue morphology. Organized stromal collagen14 and a high density of inflammatory cells15,16 have intensely backscattering properties, creating a whiter or brighter pattern on the OCT image. Tissue with high lipid15 or water content14 has less intense backscattering properties, mapped as darker areas on the OCT image.
The OCT system (Imalux Corp., Ohio) utilized low coherence light at 1310 nm wavelength, with a power of 6 mW, known to be safe for tissue imaging. The OCT imaging probe consisted of a 2.7-mm diameter flexible cable, which measured 3 m in length. The probe was sterilized by immersion in Cidex after each use.
In the in vivo studies, OCT imaging was performed after conventional colposcopy. OCT images of the vagina and cervix were obtained by an operator at predetermined anatomical sites as well as at colposcopically abnormal areas of the cervix and vagina. After the experiments were completed, the OCT images were graded by a skilled OCT grader masked to the treatments.
A Leisegang colposcope was used with digital camera and video recorder attachments. Skilled, cross-trained colposcopists graded cervical and vaginal tissues during the exam. Digital photography (CoolPix 990, 3.34 mega pixels, mounted to the colposcope) was used to document the appearance of the face of the cervix and the anterior and posterior vaginal fornices. Distinct areas within these anatomical sites were preselected for OCT imaging. In this way, colposcopic images corresponding to areas of OCT imaging were obtained.
Colposcopy was conducted following the WHO/CONRAD guidelines for colposcopic evaluation of vaginal products.17 The mucosal tissues of the cervix and vagina were assessed by low power magnification under white light. Vaginal and ectocervical tissues were evaluated for erythema, vasculature pattern, epithelial integrity, and any exceptional findings. Tissue erythema was described as general erythema (reddening) or further detailed by vasculature noted. Findings of intact and disrupted vasculature were commonly noted in baseline, untreated tissues. Indicators of tissue injury included areas of friability, abrasion, or ulceration, which were noted as “disrupted epithelium.” Colposcopic observations were noted on daily examination records, and documented by digital photography.
The studies were conducted in sexually mature female pigtailed macaques (Macaca nemestrina), housed at the Washington National Primate Research Center. Protocol approval was obtained from the University of Washington’s Animal Care Committee. All experiments were in compliance with National Institutes of Health Animal Use Guidelines.
Ex vivo Study
The ex vivo study was conducted to characterize the OCT images and validate them with histologic findings in 5 excised reproductive tracts. The tissue was received cold, wrapped in saline-soaked gauze, at the University of Texas Medical Branch Center for Biomedical Engineering’s laboratory for evaluation within 24 hours of excision at the Washington National Primate Research Center. OCT and gross images of sites with normal and abnormal appearances were collected. Biopsies of the corresponding OCT sites were taken for histologic analysis. Macaque cervical and vaginal morphologic features of OCT images were correlated to histologic findings.
In Vivo Studies
In Vivo Feasibility Study With Mechanical Injury
Four macaques were evaluated in vivo to determine the feasibility of utilizing OCT in conjunction with colposcopy. The animals were sedated according to protocol and evaluated with colposcopy18,19 and OCT daily for 3 days. Two animals received no intervention, while 2 animals received mechanical injury. One animal received cervical and vaginal biopsies and vaginal vault mechanical abrasion and the other received biopsies only. OCT images of normal areas as well as the biopsy and abrasion sites were obtained. OCT images were compared to colposcopic images and histologic findings on biopsy specimens. Imaging of the endocervix was performed; however, access for collection of images was difficult in the narrow, tortuous endocervical canal.
In Vivo Masked Controlled Study With N-9
In another study, 3 macaques were evaluated according to protocol in vivo. Each animal received a different treatment. One animal was given 1.5 mL of Conceptrol (4% N-9), and another was given 1.5 mL of hydroxyethyl cellulose (HEC) placebo gel. Both of these were administered intravaginally with a slip tip 3 mL syringe after colposcopy and OCT imaging were completed. The third animal received no treatment. The treatment condition of each animal was masked to the OCT and colposcopic graders.
The animals had baseline colposcopy and OCT image recordings on day 1, then were treated daily on days 1–3, and rested over the weekend on days 4 and 5. Daily treatment and image collection resumed on day 6, with a twice-daily dose of the N-9 and placebo gels on days 7 and 8. OCT and colposcopic images were collected on days 1–3 and 6–9.
To standardize the image collection process and to sample similar areas on subsequent days, the cervix and vault were divided into 4 anatomical sites, anterior cervix, posterior cervix, anterior vagina, and posterior vagina. Each anatomical site had 3 predetermined areas including left, middle, and right. Repeated samples of each area were taken thrice in a predetermined rotating pattern, for a total of 36 images per animal per day. This grid pattern was followed in order to maintain a consistent data sampling protocol across subjects.
In the in vivo studies, digital colposcopic images were recorded and later graded by experienced, cross-trained colposcopists who were masked to the treatment. Colposcopic scores were determined for anatomical sites approximately corresponding to the sites where OCT images were collected. Using the same grid pattern as the collection of OCT images, each colposcopic image was scored as 1, healthy appearing tissue; 2, tissue with findings such as vasculature and petechiae considered nonsignificant for the safety evaluation of topical microbicides; or 3, tissue with significantly abnormal findings such as epithelial disruption and friability.
To obtain quantitative results from the OCT images, a scoring system was developed. OCT images were compared to corresponding histologic findings from H&E stained tissue sections from ex vivo and in vivo macaque cervicovaginal biopsies. The observed changes were used to categorize images.
Biopsies from the ex vivo and in vivo feasibility studies were fixed in formalin and prepared with H&E staining. The presence or absence of the epithelial layer and inflammation were compared with features of corresponding OCT images.
In the in vivo N-9 study, measurements of epithelial thickness were determined from OCT images of the posterior cervix. In a typical OCT image, the epithelial layer could be easily identified. Because of natural variable thickness within the tissue20 demonstrated in the images, 3 measurements were obtained for each image. Images were taken from 3 sites of the posterior cervix (left, middle, and right) and each site had 3 repeated images, for a total of 9 images per day for each macaque’s posterior cervix. For each day of the study, the mean of the measurements from the 9 posterior cervix images was calculated.
NCSS software (McGraw-Hill, 2004) was utilized for statistical analysis of the image scores in the masked study. The OCT scores (categories 1, 2, or 3) were compared across animals (N-9, HEC, and no treatment) for each day. Thus, 7 3 × 3 χ2 analyses were conducted. A two-sided P value of <0.05 was used to determine statistical significance.
OCT Image Morphology
The macaque cervix and vagina OCT images are shown with corresponding histology for normal and abnormal tissue in Figure 1. The topmost dark band correlated to the glass window of the OCT probe and was present on all images. The tissue was imaged in cross-section, with the moderately bright layer representing the surface epithelium. The next brighter layer represented the connective tissue stroma. The light penetration decreased with increasing depth of imaging, creating an increasingly darker stroma toward the bottom of the image.
OCT images of normal tissue had distinct layers representing the epithelium and stroma, which correlated with normal stratified squamous epithelium and underlying stroma shown on the corresponding histology image (Figs. 1a, b). An OCT image with epithelial disruption and a chronic inflammatory infiltrate had high-intensity backscattering, seen as a bright area, corresponding to the disruption and infiltrate (Figs. 1c, d). OCT images of tissue with loss of the epithelium had a single intensely backscattering, or bright, layer representing the stroma (Figs. 1e, f).
OCT Scoring System
The presence or absence of clearly distinguishable layers and the intensity (brightness) of the layers were used as major parameters to grade the images. The scoring system was comprised of 3 major categories, normal (OCT category 1), mild-to-moderately abnormal (OCT category 2), and severely abnormal (OCT category 3).
Images scored as normal, category 1, contained a distinct bilayered structure, with the top layer representing the epithelium and the second layer representing the stroma (Fig. 1a). This bilayer was present throughout the width of the image. Minor differences in the appearance of the images included the degree of contrast between the epithelium and stroma and the presence of a cornified epithelium, represented by a slightly brighter band at the surface of the epithelium.
Images scored as mild-to-moderately abnormal, category 2, were heterogeneous across the surface of the epithelium (Fig. 1c). The images contained both a normal bilayered structure as well an area with loss of the bilayered structure, corresponding to tissue with normal epithelium adjacent to an area of abnormality, respectively.
Images scored as severely abnormal, category 3, had either absence of distinct layered morphology (Fig. 1e) or a bilayered structure with thinned epithelium (epithelial thickness <100 μm). Single-layered images had variations in intensity levels, ranging from very bright to moderately dark. A number of images with thinned epithelium had the stromal finding of darkened bands and spots suggestive of stromal edema and enlarged vessels11,14. The appearance of the abnormal areas of the category 2 and 3 images was similar, with the category 3 images exhibiting abnormality across the width of the image, whereas the category 2 images represented both normal and abnormal tissue within an image.
In Vivo Studies
In Vivo Feasibility Study with Mechanical Injury
OCT and colposcopic images of control tissue and tissue that was mechanically injured by biopsy or mechanical abrasion were collected over 3 days. Colposcopic examinations documented largely normal healthy tissue, with some areas of disrupted vasculature such as petechiae (colposcopy category 2—nonsignificant findings) in all 4 animals. Two of the four animals underwent mechanical tissue disruption by cytobrush abrasion (1 animal) or biopsy (both animals). The affected areas were identified as injured tissue sites by colposcopy on each follow-up examination.
The in vivo OCT images of the cervix and vagina on the baseline day were generally normal. There were images that were abnormal in unmanipulated tissue, suggesting there were baseline abnormalities at these sites. The animals used in this study had previously participated in studies involving manipulation of the genital tract, increasing the potential for having abnormalities. However, abnormal findings have also been recognized by colposcopy at baseline in macaque and human studies.5,18,21,22
The differences in OCT images of a healing biopsy site and the adjacent tissue are shown in Figure 2. The tissue before the biopsy had a normal layered structure (Fig. 2a). After the biopsy removal of the epithelium, the image had an abnormal appearance demonstrating that OCT could detect loss of the epithelium (Fig. 2b). The junction of the biopsied and adjacent tissue could be clearly identified because of the presence of an abnormal single-layered structure located to the left of a normal layered structure (Fig. 2c). The tissue adjacent to the biopsy was graded as normal (Fig. 2d).
The differences in OCT images of an area that was mechanically abraded are shown in Figure 3. The tissue before abrasion appeared normal (Fig. 3a). After the abrasion, the epithelial layer was thinned (Fig. 3b), while adjacent tissue was normal (Fig. 3c).
In Vivo Masked Controlled Study with N-9
Colposcopic scores of images collected during the masked N-9 study with the treatments of N-9 gel, HEC placebo gel, or no treatment were within the range of normal findings, with primarily healthy tissue (colposcopy category 1) or nonsignificant findings (colposcopy category 2). A single colposcopic finding, identified outside of the predetermined areas for OCT imaging in the animal exposed to HEC gel, was captured by digital photography and later scored as 3 (significantly abnormal finding) on day 7.
The OCT images from each animal were graded as an OCT category 1, 2, or 3. All animals exhibited some abnormal images (categories 2 and 3) before treatment and on subsequent treatment days. For the treated animals, daily applications of N-9 or HEC placebo gel were performed on days 1–3 and 6, with twice daily treatments on days 7 and 8. Representative OCT images are shown in Figure 4 on days 1 and 7 with N-9, HEC, and no treatment. The OCT images of the posterior cervix were similar in all 3 animals at baseline. On day 7, the image from the animal treated with N-9 showed thinned epithelium with darkened areas in the stroma suggestive of edema and enlarged vessels.11,14 The epithelium of the animal treated with HEC placebo was thickened while maintaining its bilayered structure on day 7. The OCT images for the animal that received no treatment were similar throughout the course of the study.
The 3 animals were not significantly different from each other at baseline [χ2(DF = 4) = 6.86, P = 0.14]. It should be noted that there were abnormalities found in all animals at baseline. The percentage of scores that were 3s ranged from 11% to 14% at baseline. After 1 day of treatment (day 2), there were no significant differences between animals, [χ2(DF = 4) = 9.47, P = 0.05]; however beginning at day 3, the OCT scores were significantly different by animal and remained so for the rest of the observation period. The distribution of OCT scores by animal and day are presented in Table 1. The animal that received HEC had significantly greater 1s and fewer 2s and 3s as compared to the other animals on days 3, 6, 7, 8, and 9. On days 3, 7, and 8, the animal that received N-9 had fewer 1s and more 2s and 3s as compared to the other animals. At day 9, the animal that received no treatment had more 3s compared to the other animals. Thus, OCT distinguished between the animals based on their treatment group with generally better scores for the animal that received HEC and generally worse scores for the animal that received N-9.
OCT was used to measure the thickness of the epithelial layer of the posterior cervix in vivo for each day of the masked study. Baseline epithelial thickness measurements ranged from 192 to 248 μm. Changes in epithelial thickness over time are shown in Figure 5. N-9 treatment had little impact on epithelial thickness in this pilot study; however, the epithelial thickness measured in the HEC treated animal dramatically increased over the 9-day observation period from 238 to 447 μm.
OCT provides high-resolution imaging capabilities in vivo, where one can see layered structures and measure the thickness of the layers, contrasted to conventional white light imaging technology (colposcopy), which shows only superficial features. Before the development of OCT technology, a biopsy was required to determine the thickness or evaluate the presence or absence of epithelium. OCT provides a noninvasive surrogate biopsy to evaluate changes in the morphology of cervical and vaginal wall epithelium. We have shown in pigtailed macaques that OCT can be used to measure epithelial thickness and we have developed a simple 3-point scoring system to assess tissue morphology. The results of this pilot study in nonhuman primates suggest that OCT may be useful for evaluating the safety of vaginal products in women. Furthermore, OCT is a developing technology with continuing improvements to resolution and speed of image processing, which may translate to an enhanced ability to screen wider areas more rapidly and with more detail.16,23
The ex vivo study showed that OCT images of the female macaque lower genital tract are similar to those of humans.8–11 Markers of tissue injury including loss of epithelium and inflammation were detectable on OCT images. The in vivo mechanical injury study demonstrated that OCT is complementary to colposcopy, feasible in the macaque, and capable of detecting epithelial changes caused by mechanical injury. The biopsy sites showed a lack of epithelium while the abraded sites showed thinned epithelium. Comparison of histologic findings from biopsy specimens collected in the ex vivo and in vivo mechanical injury studies with the corresponding OCT images permitted the development of a 3-point scoring system. OCT images of the macaque were comparable to those of humans, with similar distinction between epithelial and stromal layers, suggesting that the scoring system developed in the macaque would be applicable to humans.
In the in vivo masked N-9 study, it was possible to detect differences between animals under different conditions for OCT but not for colposcopy. The OCT scores of the animal that received N-9 included more 3s and the scores of the animal that received HEC had more 1s. Colposcopy did not show differences in these animals, suggesting that continued evaluation of the usefulness of OCT in detecting subtle changes is warranted. However, all animals had baseline abnormalities noted by colposcopy and OCT. This is consistent with reported colposcopy findings at baseline in untreated animals and humans5, 18, 21–22. OCT detected abnormal findings overall throughout the study; for example, the percentage of scores of 3s for the untreated macaque ranged from 14% to 31% of her daily scores. This suggests that in the macaque model, abnormal findings by any method of evaluation may not be uncommon. This may be attributed in part to the use of the animals in previous studies involving manipulation of the reproductive tract as well as random variation across time.
OCT was used to measure the epithelial thickness over time. While the N-9 treatment did not affect epithelial thickness, there was an increase in epithelial thickness over time in the placebo treated animal. Since only one animal was given the HEC placebo treatment, these findings should be interpreted with caution. However, one possible explanation for this change includes normal cycling of reproductive hormones. Changes in genital tract epithelial thickness over time have been observed in rhesus macaques,20,24 where increases in the thickness of the epithelial layer were associated with increased estrogen levels.
The limitations of the current study include the small sample size and the inability to obtain biopsies in the masked study. The number of macaques used in the masked study was small, as is often the case in studies involving the macaque and other nonhuman primates. While obtaining biopsies at each site of OCT image collection would have improved the ability to validate the scoring system, it also would have caused tissue injury and might have affected the natural history of the treatment study.
This study describes the initial steps in the development of OCT for the evaluation of topical microbicide safety. The scoring system developed during this study is practical and useful for differentiating the degree of tissue damage without invasive biopsies. OCT has potential to become a valuable tool in the evaluation of safety of topical microbicides. As OCT is undergoing continued development for safety evaluation, consideration will need to be given to the cost of implementation in both animal and human studies and the feasibility of implementation across a variety of settings including international sites. Development of methods such as OCT will help to facilitate and expedite safety studies of topical microbicides in this public health effort to reduce the transmission of HIV and other sexually transmitted diseases.
1. WHO/CONRAD Manual for the standardization of colposcopy for the evaluation of vaginal products, Update 2004. Punta Cana, Dominican Republic: CONRAD/WHO, 2004.
2. Van Damme L, Ramjee G, Alary M, et al. COL-1492 Study Group. Effectiveness of COL-1792, a nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex workers: A randomised controlled trial. Lancet 2002;360:971–977.
3. Kreiss J, Ngugi E, Holmes K, et al. Efficacy of nonoxynol-9 contraceptive sponge use in preventing heterosexual acquisition of HIV in Nairobi prostitutes. JAMA 1992;268:477–482.
4. Hillier SL, Moench T, Shattock R, et al. In Vitro and In Vivo: The story of nonxynol-9. J Acquir Immun Defic Syndr 2005;39:1–8.
5. Watts DH, Rabe L, Krohn MA, et al. The effects of three nonoxynol-9 preparations on the vaginal flora and epithelium. J Infect Dis 1999;180:426–437.
6. Tearney GJ, Brezinski ME, Bouma BE, et al. In vivo endoscopic optical biopsy with optical coherence tomography. Science 1997;276:2037–2039.
7. Gladkova ND, Petrova GA, Nikulin NK, et al. In vivo optical coherence tomography imaging of human skin: Normal and pathology. Skin Res Technol 2000:6:6–16.
8. Escobar PF, Belinson JL, White A, et al. Diagnostic efficacy of optical coherence tomography in the management of preinvasive and invasive cancer of uterine cervix and vulva. Int J Gynecol Cancer 2004;14:470–474.
9. Zuluaga AF, Follen M, Boiko I, et al. Optical coherence tomography: A pilot study of a new imaging technique for non-invasive examination of cervical tissue. Am J Obstet Gynecol 2005;193:83–88.
11. Escobar PF, Rojas-Espaillat L, Tisci S, et al. Optical coherence tomography as a diagnostic aid to visual inspection and colposcopy for preinvasive and invasive cancer of the uterine cervix. Int J Gynecol Cancer 2006;16:1815–1822.
12. Shen B, Zuccaro G Jr, Gramlich TL, et al. In vivo colonoscopic optical coherence tomography for transmural inflammation in inflammatory bowel disease. Clin Gastroenterol Hepatol 2004;2:1080–1087.
13. Zagaynova EV, Streltsova OS, Gladkova NK, et al. In vivo optical coherence tomography feasibility for bladder disease. J Urol 2002;167:1492–1496.
14. Welzel J, Bruhns M, Wolff HH. Optical coherence tomography in contact dermatitis and psoriasis. Arch Dermatol Res 2003;295:50–55.
15. Raffel OC, Tearney GJ, Gauthier DD, et al. Relationship between a systemic inflammatory marker, plaque inflammation, and plaque characteristics determined by intravascular optical coherence tomography. Arterioscler Thromb Vasc Biol 2007;27:1820–1827.
16. Hsiung P-L, Pantanowitz L, Aguirre AD, et al. Ultrahigh-resolution and 3-dimensional optical coherence tomography ex vivo imaging of the large and small intestines. Gastrointest Endosc 2005;62:561–574.
17. WHO/CONRAD Manual for the standardization of colposcopy for the evaluation of vaginal products, Update 2000. Geneva: CONRAD/WHO, 2000.
18. Patton DL, Kidder GG, Cosgrove-Sweeney Y, et al. Effects of multiple applications of benzalkonium chloride and nonoxynol 9 on the vaginal epithelium in the pigtailed macaque (Macaca nemestrina). Am J Obstet Gynecol 1999;180:1080–1087.
19. Patton DL, Cosgrove Sweeney YT, McCarthy TD, et al. Preclinical Safety and Efficacy Assessments of Dendrimer-Based (SPL7013) Microbicide Gel Formulations in a Nonhuman Primate Model. AAC 2006;50:1696–1700.
20. Poonia B, Walter L, Dufour J, et al. Cyclic changes in the vaginal epithelium of normal rhesus macaques. J Endocrinol 2006;190, 829–835.
22. Patton DL, Cosgrove-Sweeney YT, Balkus JE, et al. Vaginal and rectal topical microbicide development: Safety and efficacy of 1.0% Savvy (C31G) in the pigtailed macaque. Sex Transm Dis 2006;33:691–695.
23. Pitris C, Goodman A, Boppart SA, et al. High-resolution imaging of gynecologic neoplasms using optical coherence tomography. Obstet Gynecol 1999;93:135–139.
24. Ma Z, Lu FX, Torten M, et al. The number and distribution of immune cells in the cervicovaginal mucosa remain constant throughout the menstrual cycle of rhesus macaques. Clin Immunology. 2001;V100:240–249.