Introduction
Globally, the major HIV-1 transmission routes are through heterosexual contact and vertical transmission. Assessment of the presence and quantity of HIV-1 in genital fluids of women has become increasingly important in understanding transmission risks. Female genital secretions usually have been collected using cervicovaginal lavage (CVL) and swabbing techniques, methods which require pelvic examinations and are therefore labor intensive and expensive. Studies have documented the efficacy of using patient administered tampons for assessment of Chlamydia trachomatis, human papillomavirus, Trichomonas vaginalis, and Neisseria gonorrhea infections [1-7]. In 1997, a brief report suggested that tampons could be used to collect vaginal specimens for assessment of HIV-1 in the genital tract [8], although it did not compare this method with an alternate form of collection. Thus we undertook a study to assess the detection and quantification of HIV-1 recovered from self-administered tampons in comparison with recovery using CVL.
Methods
The Human Immunodeficiency Virus Epidemiology Research (HER) Study is a multi-site, prospective study of women with or at high risk of HIV-1 infection [9]. From March 1998 to February 1999, HIV-1-seropositive women enrolled in the HER Study from the Bronx, New York City site were invited to participate in a substudy if they had used tampons within the last year and had not yet reached menopause.
The standard HERS visit consisted of a semi-annual interview, a blood draw, and a gynecologic examination including a CVL, followed 2 weeks later by a psychosocial interview, as described elsewhere [10]. At a semi-annual visit, research staff explained the goals and requirements of the substudy. Women who agreed to participate were given several tampons (Tampax brand, regular absorbency) and told to return for their psychosocial interview wearing a tampon that they had inserted 6-8 h earlier. They were told that at the time of tampon insertion they should not be bleeding, and to abstain from douching and sexual intercourse for at least 24 h prior to tampon insertion. As per the HER Study protocol, each women was called to remind her of the appointment and if she had agreed to substudy participation, to abstain from douching and sexual intercourse , and to insert the tampon 6-8 h prior to her appointment.
Upon arriving for their research visit, women removed the tampons and placed them into a 50 mL centrifuge tube, provided by the research staff. Women were also requested to provide a urine specimen for toxicologic analyses, but were not required to do so. Participants were reimbursed an additional $15 for substudy participation. The study protocol was approved by the institutional review board of Montefiore Medical Center in the Bronx, New York City.
Laboratory analyses
Tampons in the 50 mL tube were refrigerated until processing later that day. During processing, 10 mL of phosphate-buffered saline was pipetted directly onto the tampon, which was then centrifuged for 10 min at 2000 r.p.m. (700 g). After centrifugation, the tampon was discarded and the eluded liquid refrigerated until shipment on ice packs via overnight carrier later that day to Johns Hopkins School of Public Health Virology Laboratory (Baltimore, Maryland, USA). Accompanying each specimen was a frozen 2 ml aliquot of CVL specimen obtained at the visit 2 weeks earlier. Specimens were stored at -70°C prior to testing.
The tampon eluents and CVL specimens were centrifuged at 17 000 r.p.m. (23 600 g) for 1 h. The pellet was used for viral quantitation by bDNA assay. Chiron-Bayer Quantiplex Version 3.0 assay (Bayer Corporation, Tarrytown, New York, USA) was used for viral quantitation in plasma, tampon eluent and CVL specimens, as per the manufacturer's instructions.
Statistical analyses
Data analysis used McNemar's test for comparison of paired proportions i.e., viral detection in paired CVL and tampon eluent specimens. Differences in quantity of viral load detected from paired samples were assessed using the Wilcoxon signed rank test for paired samples. Mantel-Haenszel chi-squared test and odds ratios (OR) with 95% confidence intervals (CI) were used for differences in proportions.
Results
A total of 130 women were approached for substudy participation, of whom 105 (81%) enrolled. Of those who refused participation, 16 (64%) reported they were not tampon users, seven (28%) had gynecologic symptoms including irregular bleeding, and two (8%) refused for other reasons, including one who was pregnant. Median age of participants was 39.5 years (range, 28-56 years). Except for Hispanic race/ethnicity, which was more common in our substudy (44% versus 17%; OR, 3.7; 95% CI, 2.4-5.6), participants were similar to other HIV-1-seropositive women enrolled in the HER Study.
Analyses were performed on 97 tampon eluents, 105 CVL and 104 plasma specimens from 105 women. Data from eight tampon specimens were unavailable because three women substituted a different brand of tampon, the laboratory reported technical difficulties that precluded analyses of two specimens, two women failed to return tampons, and one tampon specimen had visible traces of blood. One plasma specimen could not be drawn due to poor venous access.
In the 104 women for whom both plasma and genital specimens were available, HIV-1 was detected less frequently in genital specimens: [tampons (33/97, 34%) and CVL (48/104, 46%)] than plasma specimens (86/104, 83%;P < 0.001 for both plasma versus tampon and for plasma versus CVL). However, the proportion of genital specimens with detectable HIV-1 virus did not differ significantly by collection method. In women who had both genital specimen collection methods performed, there were 68 concordant and 29 discordant pairs (P = 0.14) (Table 1). Among 23 women with measurable genital HIV-1 viral load by both methods, the viral load from tampon eluents (median, 355 copies/ml; range, 52-120 898 copies/ml) was not significantly different from the viral load from CVL (median, 265 copies/ml; range, 61-35 637 copies/ml;P = 0.88).
Among the 86 women with measurable plasma HIV-1, the median plasma viral load was 2514 copies/ml (range, 51-481 037 copies/ml). Detectable genital viral load was associated with higher plasma viral load. Among the 43 women who had equal to or greater than the median plasma viral load level, 21 out of 40 (53%) had detectable viral load in the tampon eluent and 31 out of 43 (72%) had detectable viral load in CVL. For the 43 women with below the median plasma viral level, 10 out of 39 (26%) had detectable HIV-1 viral load in the tampon eluents and 16 out of 43 (37%) had detectable viral load in CVL. Measurable genital HIV-1 was found in two women with undetectable plasma HIV-1 viral load.
All women (n = 105) provided urine for toxicology. Urine toxicology revealed recent drug use in 34 (32%) women, and 12 (11%) others reported illicit drug use in the last 6 months, despite negative toxicology. Twenty-one of 42 (50%), 25 out of 46 (54%), and 40 out of 45 (89%) of tampon eluent, CVL, and plasma specimens in drug users had detectable viral load levels compared with 12 out of 55 (22%), 24 out of 59 (41%), and 46 out of 59 (78%) in non-users, respectively. The association between recent drug use and detectable HIV-1 reached statistical significance for tampon eluents (OR, 3.6; 95%CI, 1.5-8.6). Overall, only 51% of women (52/101) reported any current antiretroviral drug use, with illicit drug users somewhat less likely to report any antiretroviral drug use (48% versus 54% of non-drug users) although the difference was not statistically significant (P = 0.32). Detectable levels of HIV-1 in plasma, CVL and tampon specimens were associated with no antiretroviral drug use (Table 2).
Discussion
Self-administered sampling methods for the detection of sexually transmitted diseases have previously been reported as useful. We report that tampons are also an effective means of collecting genital secretions for the detection and quantification of HIV-1 at a time when women are not menstruating.
We found high patient compliance, with a substudy acceptance rate of 81%. The single most common reason for refusal was that women didn't like or weren't comfortable using tampons. Some may never have learned to use tampons properly, a possibility underscored by one participant returning her worn tampon still encased in its cardboard applicator. If tampons become a more common method for the collection of genital secretions, efforts to assist women in learning how to comfortably insert and use tampons are critical.
Tampon specimens were obtained, on average, 2 weeks later than CVL and plasma specimens. We rejected obtaining sequential genital samples for the comparison of collection methods in specimens obtained on the same day, which has been done in the detection of Chlamydia trachomatis, Neisseria gonorrheae, and Trichomonas vaginalis[3,5]. In those studies, tampon insertion and removal occurred prior to speculum examination for the collection of endocervical and vaginal swabs. Our substudy was restricted from interfering with the established HER study protocol, which included the collection of semi-annual CVL specimens. Therefore, sampling by CVL would have had to be done first. This was of concern because we wanted to assess the sensitivity of tampons for the collection of HIV-1 from women's usual vaginal environment. In addition, an earlier study using tampons to detect HIV-1 demonstrated low detection rates in eluents from tampons worn for less than 1 h [8]. Thus, sequential sampling would have required women to return the following day, which was impractical. These factors convinced us to obtain eluent from tampons that had been worn for 6-8 hours, albeit about 2 weeks after the collection of CVL and plasma specimens.
We did not collect reliable information on the women's menstrual cycles, other than the fact that they were not menstruating at the time of collection of either genital specimen. Other evidence from a report using tampons for the detection of human papillomavirus (HPV) suggested that different quartiles of the menstrual cycle yielded changes in the pellet volume, but that these changes did not affect the detection of HPV [11]. Thus we concluded that the delay in specimen collection likely would not seriously affect our findings.
As expected, we found that the use of one or more antiretroviral medications was associated with a reduced risk of detectable HIV-1 viral load from plasma and both genital specimens, although the association was of borderline significance for tampon eluents (P = 0.06). However, only 51% of participants reported any current antiretroviral drug use, some reporting only a single medication. We collected urine samples for toxicologic analysis because we hypothesized that drug-using women would be less likely to take antiretroviral medication and therefore be at risk for higher plasma and genital HIV-1 viral load. We did find an association between recent drug use and high viral load, but a weak association between drug use and current antiretroviral therapy, which was not statistically significant. This finding is hard to explain unless drug users were more likely than non-drug users to over-report their antiretroviral drug use. Another explanation might be lack of compliance with sexual abstinence in drug users, if HIV-1 infected semen added to their detectable genital HIV-1 burden.
Our findings suggest that tampons are a promising tool for the detection and quantification of genital HIV-1 in situations where performance of CVL is impractical or undesirable.
References
1. Fairley CK, Chen S, Tabrizi SN, Quinn MA, McNeil JJ, Garland SM. Tampons: a novel patient-administered method for the assessment of genital human papillomavirus infection. J Infect Dis 1992, 165: 1103-1106.
2. Fairley CK, Robinson PM, Chen S, Tabrizi SN, Garland SM. The detection of HPV DNA, the size of tampon specimens and the menstrual cycle. Genitourin Med 1994, 70: 171-174.
3. Tabrizi S, Chen S, Fairley C. et al. Tampon-collected genital cells in the detection of Chlamydia trachomatis by polymerase chain reaction [letter]. J Infect Dis 1993, 168: 796-797.
4. Tabrizi SN, Chen S, Borg AJ, Lees MI, Fairley CK, Garland SM. Patient-administered tampon-collected genital cells in the assessment of Chlamydia trachomatis infection using polymerase chain reaction. Sex Transm Dis 1996, 23: 494-497.
5. Tabrizi SN, Paterson B, Fairley CK, Bowden FJ, Garland SM. Self-administered technique for the detection of sexually transmitted diseases in remote communities. J Infect Dis 1997, 176: 289-292.
6. Tabrizi SN, Paterson B, Fairley CK, Bowden FJ, Garland SM. Comparison of tampon and urine as self-administered methods of specimen collection in detection of Chlamydia trachomatis, Neisseria gonorrhoeae and Trichomonas vaginalis in women. Int J STD AIDS 1998, 9: 347-349.
7. Tabrizi SN, Fairley CK, Chen S. et al. Evaluation of patient-administered tampon specimens for Chlamydia trachomatis and Neisseria gonorrhoeae. Sex Transm Dis 2000, 27: 133-137.
8. O'Shea S, de Ruiter A, Corbett MK, Chrystie I, Newell ML, Banatvala JE. Quantification of HIV-1 RNA in cervicovaginal secretions: an improved method of sample collection. AIDS 1997, 11: 1056-1057.
9. Smith DK, Warren DL, Vlahov D. et al. Design and baseline participant characteristics of the human immunodeficiency virus epidemiology research study: a prospective cohort study of human immunodeficiency virus infection in US women. Am J Epidemiol 1997, 146 (6): 459-469.
10. Klein RS, Ho GYF, Vermund SH, Fleming I, Burk RD. Risk factors for squamous intraepithelial lesions on pap smear in women at risk for human immunodeficiency virus infection. J Infect Dis 1994, 170: 1404-1409.
11. Fairley CK, Chen S, Tabrizi SN, Quinn MA, McNeil JJ, Garland SM. The influence of quartile of menstrual cycle on pellet volume of specimens from tampons and isolation of human papillomavirus [letter]. J Infect Dis 1992, 166: 1199-1200.
© 2001 Lippincott Williams & Wilkins, Inc.