Skip Navigation

Correlation of Human Immunodeficiency Virus Type 1 RNA Levels in Blood and the Female Genital Tract

  1. Clyde E. Hart,
  2. Jeffrey L. Lennox,
  3. Melody Pratt-Palmore,
  4. Thomas C. Wright,
  5. Raymond F. Schinazi,
  6. Tammy Evans-Strickfaden,
  7. Timothy J. Bush,
  8. Cathy Schnell,
  9. Lois J. Conley,
  10. Kelly A. Clancy and
  11. Tedd V. Ellerbrock
  1. Division of AIDS, STD, and TB Laboratory Research, National Center for Infectious Diseases, and Division of AIDS Prevention, Surveillance, and Epidemiology, National Center for HIV, STD, and TB Prevention, Centers for Disease Control and Prevention, Public Health Service, US Department of Health and Human Services, and Department of Medicine, Emory University School of Medicine, and Georgia VA Research Center for AIDS and HIV Infections, VA Medical Center and Department of Pediatrics, Emory University, Atlanta, Georgia; Department of Pathology, College of Physicians and Surgeons of Columbia University, New York, New York
  1. Reprints or correspondence: Dr. Clyde E. Hart, Centers for Disease Control and Prevention, 1600 Clifton Rd., N.E., Mailstop G19, Atlanta, Georgia 30333 (ceh4{at}cdc.gov).

  1. Presented in part: Fourth Conference on Retroviruses and Opportunistic Infections, Washington, DC, January 1997.

 
Next Section

Abstract

In this study, the correlations of human immunodeficiency virus type 1 (HIV-1) RNA levels in blood plasma, vaginal secretions, and cervical mucus of 52 HIV-1-infected women were determined. The amount of cell-free HIV-1 RNA in blood plasma was correlated with that in vaginal secretions (Spearman's rank correlation coefficient (r) = 0.64 P < .001 In both blood plasma and vaginal secretions, the amounts of cell-free and cell-associated HIV-1 RNA were highly correlated (r = 0.76, P < .01 and r = 0.85, P < .01, respectively). Cell-free HIV-1 RNA levels in blood plasma and vaginal secretions were negatively correlated with CD4+ T lymphocyte count (r = −0.44, P < .01 and r = −0.40, P < .01, respectively). Similar to the effect observed in blood plasma, initiation of antiretroviral therapy significantly reduced the amount of HIV-1 RNA in vaginal secretions. These findings suggest that factors that lower blood plasma virus load may also reduce the risk of perinatal and female-to-male heterosexual transmission by lowering vaginal virus load.

Worldwide, most human immunodeficiency virus type 1 (HIV-1) infections in adults have been transmitted sexually through exposure to infected seminal fluid and cervicovaginal secretions [1]. Moreover, most HIV-1 infections in children have been transmitted perinatally, and recent studies suggest that many of these infections have occurred during labor and delivery, when the infant is exposed to infected blood and cervicovaginal secretions [2]. Although the modes of HIV-1 transmission and factors associated with transmission have been studied extensively and clearly defined during the past decade, information about HIV-1 virus load in cervicovaginal secretions has only recently become available.

The initial investigations and some later studies of HIV-1 in the female genital tract used DNA-based polymerase chain reaction (PCR) or virus culture techniques to detect HIV-1-infected cells collected on vaginal or endocervical swabs [3–12]. In these studies, HIV-1 was found in about one-third or fewer of the specimens. For example, using DNA PCR, Clemetson et al. [8] found HIV-1 in 13 (17%) of 77 vaginal swabs and 28 (33%) of 84 endocervical swabs. Kreiss et al. [9] detected virus in 51 (37%) of 137 endocervical swabs. Detection of HIV-1 proviral DNA in the genital tract was correlated with genital tract inflammation, oral contraceptive use, cervical ectopy, and pregnancy. The presence of HIV-1 proviral DNA and the ability to isolate virus from cellular samples shows that infected cells capable of producing virus are present in the genital tract. However, these studies did not determine whether the virus was present in genital tract secretions or whether HIV-1-infected cells were producing virus in vivo.

More recent studies have reported the quantification of HIV-1 RNA as a measure of the amount of virus in cervical and vaginal secretions [13–17]. Cu Uvin et al. [14] detected HIV-1 RNA in 28 (39%) of 72 cervicovaginal lavage samples that consisted of both the cell-free and cell-associated fractions. Iversen et al. [16] studied 28 HIV-1-infected women and quantified cell-free virus in 29% of vaginal swab samples and 32% of initial cervical swabs. An immediate resampling of the cervix induced cervical bleeding, which was detected by visual inspection in 50% of the women and increased the proportion of samples with detectable HIV-1 RNA to 64%. Goulston et al. [17] also used cervical swabs to quantify cell-free virus in 24 (49%) of 49 HIV-1-infected women. In this study, the presence of blood in the genital tract was associated with increased HIV-1 RNA levels. In all three of these studies, a positive correlation was found between virus loads in blood plasma and female genital tract secretions [14, 16, 17]. In contrast, Rasheed et al. [13] found no correlation between virus loads in blood plasma and cell-free supernatant collected on cervicovaginal swabs. These studies show that HIV-1 RNA can be quantified in samples of cervicovaginal secretions, but the small number of participants in some of these studies reduces the power of analysis. Furthermore, the amount of blood in the cervicovaginal samples and the possibility that plasma virus may have significantly increased the virus load in the samples were not addressed in these studies.

In this study, we determined the correlation of virus loads in blood plasma and the female genital tract, using a quantitative-competitive reverse-transcriptase PCR (QC-PCR) assay. The QC-PCR assay was chosen because of its greater sensitivity in detecting and quantifying HIV-1 in genital tract samples from HIV-1-infected women than a commercially available assay. Using the QC-PCR assay, we were able to determine the proportion of virus load in the lower genital tract from cervical mucus and the cell-free and cell-associated fractions of vaginal secretions. We also examined the correlations of vaginal virus load with blood plasma virus load, CD4+ T lymphocyte count, antiretroviral therapy, and the presence or absence of a cervix. In addition, we calculated the proportions of HIV-1 RNA in vaginal secretions and cervical mucus that were derived from blood contamination.

Previous SectionNext Section

Materials and Methods

Eligibility criteria

HIV-1-infected women were eligible for enrollment in this study if they were 18–49 years of age, had a normal Pap smear within the previous 12 months, were expected to live at least 1 year, and either were not on antiretroviral therapy or had taken the same antiretroviral therapy for at least 3 months prior to enrollment. Women were excluded from the study if they (1) had an acute or chronic infection other than HIV-1 at their first study visit, (2) were currently using cytotoxic chemotherapy, systemic corticosteroids, cimetidine, or other immunosuppressive or immunostimulating medications, (3) were using medroxyprogesterone acetate or an intrauterine device, (4) had cervical intraepithelial neoplasia at their first visit, (5) had had cervical surgery during the previous 3 months, or (6) were pregnant. Participants were requested to refrain from vaginal intercourse and the use of intravaginal medications for 72 h before their examinations.

Specimen collection

At each clinical exam, vaginal secretions were tested for the presence of seminal fluid, using an acid-phosphatase assay [18]. Subsequently, a vaginal lavage sample was obtained by introducing 10 mL of PBS into the vagina and collecting the pooled fluid in the posterior vaginal fornix, taking care to avoid the cervix. A cervical mucus sample was then obtained by gently inserting a cytobrush into the cervical os and slowly rotating it through a 90° arc before removing it and placing it in 300 μL of PBS. Endocervical swabs were obtained to culture Neisseria gonorrhoeae and Chlamydia trachomatis and to prepare Gram's-stained smears to detect cervicitis. The vaginal lavage and cervical mucus samples were kept on ice until processed. In addition, ∼8 mL of venous blood was collected in a CPT vacutainer tube (Becton-Dickinson, Franklin Lakes, NJ) that contained acid citrate dextrose anticoagulant. Specimens were taken to the laboratory within 3 h for processing that day.

Blood plasma and peripheral blood mononuclear cells (PBMC) were isolated in CPT vacutainer tubes per the manufacturer's protocol, and PBMC were counted. PBMC were pelleted from the pooled plasma-PBMC mixture (200 g, 15 min), and the plasma was removed and stored as 1-mL aliquots that were frozen at −70°C. The PBMC pellet was washed and pelleted twice (200 g, 10 min) in PBS, and individual PBMC pellets (2 × 106 cells/pellet) were snap-frozen on dry ice and stored at −125°C.

Lavage samples were visually examined for blood and with the use of a reagent strip for urinalysis (Multistix-7; Bayer, Elkhart, IN) that measures whole blood and hemoglobin. Cells from lavage samples were counted and then pelleted (200 g, 15 min); the supernatant was removed and stored as 1-mL aliquots at −70°C. Cells from lavage samples were washed and frozen, using the procedure for PBMC described above.

Cervical mucus samples were examined for blood, using the procedure for lavages described above. Any portion of a cervical mucus sample that adhered to the cytobrush was removed and added to the 300-μPBS aliquot. The total volume of PBS plus cervical mucus was divided into two equal aliquots that were snap-frozen on dry ice and stored at −125°C.

RNA extraction

Total cellular RNA was extracted from PBMC and vaginal lavage frozen cell pellets, using RNAzol B (Biotecx Laboratories, Houston) per the manufacturer's protocol and adding 2 μg of polyadenosine RNA as a carrier during the precipitation step. Cell-free virion-associated HIV-1 RNA in blood plasma and vaginal lavage samples was isolated according to the procedure of Mulder et al. [19], with the exception that viral pellets were obtained by centrifugation (105 g) for 60 min. The resulting pellets were solubilized in 500 μL of a guanidinium isothiocyanate solubilization buffer (4.6 M guanidinium isothiocyanate, 40 mM Tris [pH 7.5], 80 mM β-mercaptoethanol, and polyadenosine RNA [3 μg]). An equal volume of isopropanol was added to the buffer solution, and the RNA was precipitated overnight at 4°C. The precipitated RNA was pelleted (16,000 g) for 30 min, rinsed with 75% ethanol, solubilized in H2O, and immediately assayed for HIV-1 RNA using QC-PCR. The remainder of the sample was stored at −70°C. Total RNA in the frozen cervical mucus sample was extracted with RNAzol B and 2 μg of polyadenosine RNA, as described above.

QC-PCR: HIV-1 internal standard and RNA synthesis

The plasmid DNA, pQP1δ50 (δ50D), consists of the original HIV-1 DNA internal standard for QC-PCR, pQP1δ80 [20], with a non-HIV-1, 31-bp insert [21]. The HIV-1 RNA internal standard (d50R) was transcribed in vitro from the T7 promoter of EcoRI-linearized δ50D with the use of an RNA transcription kit (Promega, Madison, WI), per the manufacturer's protocol. δ50R was purified with the use of RNAzol B, per the manufacturer's protocol, and RNA copy numbers were determined from optical density readings.

QC-PCR of HIV-1 RNA

Primers GAG04 and GAG06 (designed by Piatak et al. [20]) have inosines incorporated at those positions known for sequence divergence [22, 23]; GAG04 was biotinylated, and the resulting primer was designated GAG04-B. The GAG06/GAG04-B primer pair yields amplicons of 260 bp from wild-type HIV-1 and 211 bp from the δ50R internal standard. Reverse transcriptase PCR (RT-PCR) was performed, using four concentrations (101, 102, 103, and 104 copies) of δ50R internal standard (5 μL) and sample RNA (5 μL) in four separate RT-PCR reactions. Reverse transcription was done with the use of random hexamers and the GeneAmp RNA PCR kit (Roche Molecular Systems, Branchburg, NJ) in 25 μL, per the manufacturer's protocol. After completing the reverse transcription, 12.5 pmol of each primer (GAG04-B/GAG06) and 1 U of Taq enzyme in 25 μL were added in buffered conditions, following the manufacturer's protocol (Roche Molecular Systems). Cycling conditions were as previously reported [21]. PCR products were analyzed by electrophoresis on a 3% agarose gel, using ethidium-bromide staining, and by the microtiter plate assay described below.

Estimation of cell viability

RT-PCR amplification of porphobilinogen deaminase (PBGD) RNA in cell-associated RNA extracts was used to semiquantitatively determine viable cell number. Nonerythryoid PBGD is a single-copy gene that is constitutively expressed in all mammalian tissues and cell types [24] and is used to estimate cell viability and the integrity of RNA extracts used for RT-PCR [25]. When the standard RT-PCR reaction described above was used and PCR was done for 30 cycles, the PBGD primers, PBGD5-B (biotin-TGTCTGGTAACGGCAATGCGGCTGCAAA-5′ and PBGD3 (GGCATGTTCAAGCTCCTTGG-3′), spanned an RNA splice junction and amplified a 250-bp product. A PBGD RNA external standard curve was generated from five RT-PCR reactions that contained 101, 102, 103, 104, and 105 cell equivalents of PBMC RNA. The linear range of PBGD RNA detection was 2 × 102 to 1 × 105 cell equivalents.

Quantification of QC-PCR products in a microtiter format

HIV-1 biotinylated PCR products were quantified in a microtiter format, as previously described [21]. Briefly, 5 μL of the QC-PCR reactions was added to duplicate wells of a streptavidin-coated, white microtiter plate (Boehringer Mannheim, Indianapolis) containing 200 μL of hybridization buffer (1 × SSC, 20 mM HEPES, 0.1% Tween 20, and 2 mM EGTA), and the plate was incubated for 30 min at 37°C. The microtiter wells were rinsed six times with 300 μL of wash buffer (PBS, 0.1% Tween 20, and 2 mM EGTA) before the streptavidin-bound biotinylated PCR products were denatured, using 250 μL of a 0.4 N NaOH, 0.6 M NaCl solution for 15 min at room temperature, and then rinsed as before. A digoxigenin (DIG) oligonucleotide probe (DIG-GGACATCAAGCAGCCATGCAAATGT; 40 ng/well) for detection of wild-type HIV-1 amplicons was added to one set of wells in 200 μL of hybridization buffer, and a δ50-specific DIG-oligonucleotide probe (DIG-TGTTGGGCGCCATCTCCTTGC; 40 ng/well) was added to the duplicate wells for 1 h at 37°C and rinsed as before. For quantitative luminometry, 5 ng of an anti-DIG antibody conjugated to the bioluminescent aequorin protein (Sealite Sciences, Bogart, GA) was added to each well in 200 μL of assay buffer (0.01 M PBS [pH 7.2], 0.5% gelatin, 0.15% Tween 20, and 2 mM EGTA), incubated at 37°C for 30 min, and rinsed as before. The well-bound aequorin/DIG-antibody was read with the use of a luminometer (ML3000; Dynex, Chantilly, VA). The relative luminescent units (RLU) generated from wild-type- and δ50-probed samples were used to calculate HIV-1 copy number, as previously described [21].

The QC-PCR technique in combination with the microtiter detection assay quantifies input HIV-1 RNA from 10 to 10,000 copies [21]. This equates to 100–100,000 HIV-1 RNA copies in 50-μL RNA extracts from cell-free and cell-associated vaginal lavage samples, cervical mucus, and PBMC samples and 800–800,000 HIV-1 RNA copies in 800-μL RNA extracts from blood plasma. Samples were determined to be HIV-1 RNA-positive, but unquantifiable, when their RLU value was less than that for the lowest copy number of internal standard (10 copies of>δ50 RNA) but at least 2.5-fold greater than the RLU value of an HIV-1-negative control.

Quantification of the PBGD PCR product (250 bp) was done in the microtiter format using a DIG-labeled probe (DIG-CAATGTTGCCACCACACTGTCCG), as described above. PBGD cell equivalents in a sample were calculated from the linear regression line of the external standard curve of PBMC RNA.

Calculation of blood plasma virus in genital tract samples

Hemoglobin (hgb) concentrations were used to calculate blood volumes in vaginal lavage and cervical mucus samples. The four hgb levels defined on the Multistyx-7 reagent strips were 0.015, 0.045, 0.135, and 0.405 mg hgb/100 mL. Samples with a signal of 0.405 mg hgb/100 mL were diluted and retested, so that the signal was within the range of the assay.

The amount of plasma virus in a vaginal lavage sample was determined with the use of the following series of equations. Total amount of hgb in a vaginal lavage sample (mg hgblavage) was determined, using the equation: mg hgblavage = (hgb signallavage × 0.01) × (volumelavage), where hgb signallavage is the hgb level (mg hgb/100 mL) estimated from the Multistyx-7 reagent strip, and volumelavage is the total volume of the lavage sample in milliliters. Since we could calculate the total amount of hgb in a vaginal lavage sample (mg hgblavage) and knew the hgb concentration in peripheral blood (mg hgbblood/100 mLblood), we could calculate the blood volume in vaginal lavage samples (volLblood), using the equation: volLblood = (mg hgblavage) ÷ (mg hbgblood/100 mlblood). The blood plasma volume (plasmalavage) in vaginal lavage samples was calculated using the volume of blood in the vaginal lavage samples (volLblood) and the peripheral blood hematocrit in the equation: plasmalavage = (1 − hematocrit) × (volLblood). Blood plasma virus in vaginal lavage samples (plasma HIVlavage) was calculated, using the volume of plasma in vaginal lavage samples (plasma vollavage) and the virus concentration in plasma (HIV RNA copiesplasma/plasma vol) in the equation: plasma HIVlavage = plasma vollavage × (HIV RNA copiesplasma/plasma vol). A similar series of equations was used to calculate the amount of blood plasma virus in cervical mucus.

Antiretroviral therapy

To examine the effects of antiretroviral therapy on virus load in vaginal secretions, we studied a subset of women who were beginning or changing drug regimens that included one or more new antiretroviral medications. These women must have had a detectable virus load in both blood plasma and vaginal lavage samples prior to changing or beginning therapy. After 2–10 weeks of therapy, blood plasma and vaginal lavage samples were obtained from the women, and the virus loads of the samples were determined. Another subset of study participants, who had measurable virus loads in blood plasma and vaginal lavage samples but had not begun or changed antiretroviral therapy, were selected as a control group.

Amplicor

This commercially available monitor test (Roche Molecular Systems) was used according to the manufacturer's protocol. The lower limit of quantification was 400 HIV-1 RNA copies/mL of sample.

Previous SectionNext Section

Results

Study Participants

This analysis includes data from the initial examination of the first 52 women enrolled in the Emory Vaginal Ecology Study of HIV Infection. None of the women had N. gonorrhoeae or C. trachomatis infections or cervicitis. Ten (19%) of the women had had a hysterectomy and thus did not have a cervix, and 2 (4%) tested positive for seminal fluid (patient nos. 002 and 034; table 1).

Evaluation of Virus Load Assays

Little information is available about the accuracy of different nucleic acid amplification techniques in quantifying HIV-1 RNA in the female genital tract. Consequently, we compared results of the QC-PCR assay to those obtained with the commercially available test (Roche Molecular Systems), using 31 blood plasma and 30 vaginal lavage samples from HIV-infected women. When the results from the commercial assay and the QC-PCR for each blood plasma sample were compared, 29 (90%) had a <4-fold difference, whereas 2 (10%) had 15- and 18-fold differences (figure 1A). Overall, blood plasma virus loads measured by the commercial assay and QC-PCR were highly correlated (Spearman's rank correlation coefficient =0.87, P < .01).

Figure 1
View larger version:
Figure 1

Comparison of HIV-1 virus loads in plasma (A) and vaginal lavage (B) samples determined by commercially available (Amplicor) and quantitative-competitive reverse-transcriptase polymerase chain reaction assays. Filled circle: virus was detected by both assays; open circle: virus was not detected by either assay; half-filled circle: virus was detected by 1 assay.

A comparison of virus loads from vaginal lavage samples from 30 HIV-1-infected women showed that 14 (47%) were positive by the commercial test and 22 (73%) were positive by QC-PCR (figure 1B). Vaginal virus loads were expressed as number of HIV-1 RNA copies/total lavage volume. The number of HIV-1 RNA copies/total lavage volume ranged from 1200 to 26,000 (median, 4500) by the commercial test, and <1000 but detectable to 120,000 (median, 3000) by QC-PCR. Virus load in lavage samples measured by the commercial and QC-PCR assays were correlated (Spearman's rank correlation coefficient = 0.55, P = .002), but the correlation was less than that for blood plasma. Moreover, of the 24 samples with detectable HIV-1 RNA by either assay, 14 (58%) were detected by the commercial test, compared with 22 (92%) by the QC-PCR assay (P = .03).

These results indicate that the commercial and QC-PCR assays have similar capabilities in quantifying virus loads in blood plasma. However, the QC-PCR assay detected HIV-1 in significantly more vaginal lavage samples than did the commercial test. Consequently, in this study, we used the QC-PCR assay to determine the virus loads in blood plasma and genital tract secretions.

Cell-Free HIV-1 RNA in Blood Plasma and Vaginal Secretions

Forty-six (88%) of 52 blood plasma specimens were positive for HIV-1 RNA (table 1). Plasma virus loads ranged from <800 to 270,000 copies of HIV-1 RNA/mL (median, 21,000 copies/mL). Cell-free vaginal HIV-1 RNA was detected in 32 (62%) of 52 lavage samples with a range of <1000 to 122,000 copies/lavage and a median of 2600 copies/lavage (table 1). Twenty-eight (54%) women had a virus load in vaginal lavage samples that was undetectable or detectable but <1000 HIV-1 RNA copies. Twenty (71%) of these 28 women had blood plasma virus loads <10,000 copies/mL, 7 (25%) had between 10,000 and 50,000 copies/mL, and 1(4%) had 77,000 copies/mL. Thus, most women with low vaginal virus loads had low blood plasma virus loads.

HIV-1 RNA virus loads in matched blood plasma and vaginal lavage samples of the 52 women had a correlation of 0.64 (Spearman's rank correlation coefficient, P < .001; figure 2A). Despite this correlation, 8 (29%) of 28 women with blood plasma virus loads >10,000 copies/mL had undetectable or detectable but unquantifiable (<1000 copies/lavage) levels of HIV-1 in their vaginal lavage samples. Virus spiking experiments indicated that inhibitors of the QC-PCR assay were not present in the vaginal secretions of women with low or undetectable vaginal virus loads (data not shown).

Figure 2
View larger version:
Figure 2

Comparison of virus loads in cell-free plasma and cell-free vaginal lavage samples (A), cell-free vaginal lavage and cervical mucus samples (B), cell-free plasma and cervical mucus samples (C), cell-free plasma and cell-associated plasma samples (D), and cell-free and cell-associated vaginal lavage samples (E). Filled circle: virus was detected by both assays; open circle: virus was not detected by either assay; half-filled circle: virus was detected by 1 assay.

HIV-1 was detected in vaginal lavage samples from 27 (64%) of 42 women with a cervix and 5 (50%) of 10 women without a cervix (P = .4). Detectable virus loads ranged from <1000 to 120,000 copies per lavage sample (median, 2600) for women with a cervix and from <1000 to 14,000 copies per lavage sample (median, 2000) for those without a cervix.

Total HIV-1 RNA in Cervical Mucus

HIV-1 RNA was detected in cervical mucus from 28 (67%) of 42 women. Cervical virus load was expressed as total HIV-1 RNA copies per cervical mucus sample. HIV-1 RNA copies per cervical mucus sample ranged from <200 to 12,000 (median, 400) (table 1). The HIV-1 virus loads in cervical mucus and vaginal lavage samples of the 42 women had a correlation of 0.59 (Spearman's rank correlation coefficient, P < .01; figure 2B), whereas the cervical mucus and blood plasma virus loads had a correlation of 0.53 (Spearman's correlation coefficient, P < .01; figure 2C).

Blood Contamination of Genital Tract Secretions

To estimate the proportion of virus load in female genital tract secretions derived from HIV-1 virus in blood, we measured the amount of blood in vaginal lavage and cervical mucus samples with the use of a quantitative test for hgb. Thirty-seven (71%) of 52 vaginal lavage samples and all of the 31 cervical mucus samples that were tested had detectable hgb. Eighteen (49%) of 37 vaginal lavage samples and 18 (58%) of 31 cervical mucus samples that had blood contamination had measurable plasma virus. HIV-1 RNA from blood was 0.002%–4.6% (median, 0.13%) of the vaginal virus load and 0.1%–100% (median, 7.0%) of the cervical mucus virus load (figure 3). Among the cervical mucus samples, blood plasma virus was <5% of the cervical mucus virus load in 9 samples, 5%–25% in 5 samples, and >25% in 4 samples. These calculations suggest that blood plasma HIV-1 is present in vaginal secretions and cervical mucus but usually at low concentrations.

Figure 3
View larger version:
Figure 3

Percentage of HIV-1 virus load in vaginal lavage and cervical mucus samples from peripheral blood. Filled circle: plasma virus was present; closed circle: plasma virus was absent.

Cell-Associated HIV-1 RNA in Blood Plasma and Vaginal Secretions

Cell-associated HIV-1 RNA was detected in vaginal lavage samples from 20 (59%) of 35 women tested, with a range of <100–134,000 copies per sample (table 1). In comparison, PBMC-associated HIV-1 RNA was detected in all 32 of these women, with a range of <100–130,000 copies per 2 million cell equivalents (table 1). PBMC- and lavage sample cell-associated virus loads were highly correlated with cell-free virus loads in blood plasma and lavage samples, respectively (figure 2D, 2E).

Number of Cells in Vaginal Lavage Samples

The number of cells in vaginal lavage samples varied greatly, ranging from 8.0 × 104 to 4.1 × 108 cells per sample. The vast majority of cells in lavage samples were superficial squamous epithelial cells that were determined to be nonviable by trypan blue staining. To better estimate the total number of viable cells in a vaginal lavage sample, PBGD RNA cell equivalents were determined by a semiquantitative RT-PCR assay. Only 6 (17%) of the 35 lavage cellular samples had a detectable signal by RT-PCR (defined as 12000 cell equivalents of PBGD RNA per 106 squamous epithelial cells). In contrast, RNA extracts from PBMC samples had comparatively high levels of cell viability as measured by PBGD RNA (7.5 × 105 to 2.9 × 106 cell equivalents per 2 × 106 trypan blue-viable cells). The correlation between the amount of cell-associated HIV-1 RNA and the number of viable cells (PBGD RNA cell equivalents) in lavage samples was not statistically significant (r = 0.25, P = .17). However, the cell-associated vaginal virus load was weakly correlated with the total number of cells per lavage sample (r = 0.38, P = .03).

Total Virus Load in the Lower Female Genital Tract

Because the amount of HIV-1 RNA in vaginal secretions and in cervical mucus was determined in this study, the contribution of each to the total virus load in the lower genital tract could be calculated. Twenty-four women had virus load measurements for cell-free and cell-associated vaginal lavage and cervical mucus samples. Total vaginal virus load, including cell-free and cell-associated HIV-1 RNA, accounted for the major proportion (>50%) of lower genital tract virus in 16 (67%) of the women. When cell-free and cell-associated vaginal virus load were analyzed separately, cell-free virus accounted for the major proportion of lower genital tract virus in 12 (50%) of the women, while cell-associated virus was not a major proportion in any of the women. Cervical mucus virus was the major proportion of lower genital tract virus in 7 (30%) of the women. However, the total vaginal virus load was undetectable in 4 of the 7 women.

Effect of CD4+ T Lymphocyte Count on Virus Load

The proportion of vaginal lavage samples that had detectable HIV-1 RNA increased significantly with decreasing CD4+ T lymphocyte levels (P = .02, table 2). However, the proportions of blood plasma and cervical mucus samples that had detectable HIV-1 RNA were not significantly associated with CD4+ T lymphocyte level (table 2).

Virus loads in cell-free blood plasma and vaginal lavage samples were negatively correlated with CD4+ T lymphocyte counts (r = −0.44, P < .01 and r = −0.40, P < .01, respectively; figures 4A, 4B). However, cervical mucus virus loads were not correlated with CD4+ T lymphocyte counts.

Figure 4
View larger version:
Figure 4

Comparison of CD4+ T lymphocyte level and HIV-1 virus loads in plasma (A) and vaginal lavage (B) samples. Closed circle: virus load was detected; open circle: virus load was not detected.

Effect of Antiretroviral Therapy on Virus Loads

Cross-sectional analysis

Thirty-one of the 52 women were receiving stable antiretroviral therapy for at least 3 months when blood and genital tract samples were obtained (table 1). The median virus load in blood plasma was 2.6-fold lower for women receiving therapy compared with those receiving no therapy (P = .09). In addition, the proportion of women with detectable blood plasma virus was lower for women on therapy (100% vs. 81%; P = .04). However, for vaginal lavage and cervical mucus samples, the median virus loads and the proportion of samples with detectable virus were not significantly different between the women receiving therapy and those receiving no therapy.

Figure 5
View larger version:
Figure 5

Effect of antiretroviral therapy on HIV-1 virus loads in plasma and cell-free vaginal lavage samples. HIV-1 virus loads were determined for women receiving no antiviral therapy (A, B) and women who were beginning or changing antiretroviral therapy (C, D). Each line represents 1 woman's virus load (log10 HIV-1 RNA copies) at baseline (visit A) and at follow-up visit (visit B).

Table 1
View larger version:
Table 1

CD4+ T lymphocyte cell count and HIV-1 virus load in plasma, vaginal lavage, and cervical mucus samples.

Table 2
View larger version:
Table 2

Number of samples with detectable HIV-1 RNA by CD4+ T lymphocyte level (n = 52).

Initiation of new therapy

Eleven women who were beginning or changing antiretroviral therapy had a total of 14 new treatment regimens. Eight women began new regimens that included lamivudine plus either zidovudine (3 women), indinavir (4 women), or indinavir with another nucleoside (1 woman). Three women had a nucleoside analogue or protease inhibitor added to an ongoing therapy for a total of 6 different treatment regimens. Seven women who were not receiving antiretroviral therapy served as a control group. Before the initiation of new or different antiretroviral therapies, the mean virus loads in blood plasma and vaginal lavage samples and the mean CD4+ T lymphocyte cell counts were not significantly different between the treatment and control groups.

Of the 7 women not receiving therapy, 4 had a <0.5 log10 change in blood plasma virus load at the follow-up visit (compared with data from the baseline visit) (figure 5A). The other 3 women in the control group had either an increase (1 women) or decrease (2 women) in blood plasma virus load at the follow-up visit. Virus loads in the vaginal lavage samples of these women were more variable between the baseline and follow-up visits (figure 5B). Three women had an increase and 3 had a decrease of virus load in vaginal lavage samples 10.5 log10. Only 1 woman in the control group had a decline in both blood plasma and vaginal virus loads (>0.5 log10) at the follow-up visit.

For women who began or changed antiretroviral therapy, there was a 0.5–1.6 log10 decrease in their blood plasma virus loads at the follow-up visit in 10 of 14 new treatment regimens (figure 5C). In the four new treatment regimens that did not result in a decrease in blood plasma virus load, 3 were a single antiretroviral treatment and 1 was didanosine and nevirapine. The 14 new treatment regimens resulted in a 0.5–2.1 log10 decrease in all of the virus loads in vaginal lavage samples at the follow-up visit (figure 5D). The decrease in vaginal virus load at follow-up was significantly different from the change in vaginal virus load at follow-up in the control group (P = .006). In addition, virus load in the vaginal lavage sample taken at the follow-up visit became undetectable (<1000 HIV-1 RNA copies/total lavage) in 10 of the new treatment regimens (P = .07 vs. control group). Moreover, in 12 treatment regimens, the matched blood plasma and vaginal lavage virus loads both decreased >0.5 log10, whereas in the control group, only 1 of 7 matched blood plasma and vaginal lavage virus loads decreased >0.5 log10 (P = .003).

Previous SectionNext Section

Discussion

In this study, we determined that HIV-1 RNA levels in blood plasma, vaginal secretions, and cervical mucus were correlated in women who did not have a genital tract infection. In these women, cell-free HIV-1 RNA levels in blood plasma and vaginal secretions were negatively correlated with CD4+ T lymphocyte count. Furthermore, the initiation of antiretroviral therapy significantly reduced the amount of HIV-1 RNA in vaginal secretions.

The percentages of women in our study with detectable HIV-1 RNA in cell-free vaginal lavage and cervical mucus samples were 62% and 67%, respectively. Previous studies have reported detecting HIV-1 RNA in 29%–58% of vaginal swab or cervicovaginal lavage samples [13, 14] and 32%–64% of cervical samples obtained by using swabs or cytobrushes [16, 17]. In each of these studies, except for that by Rasheed et al. [13], a positive correlation was found between the HIV-1 RNA levels in matched blood plasma and genital tract samples. The higher detection rates in our study may have been the result of several factors. Our RNA extraction procedure, which pellets cell-free virus by ultracentrifugation, may have increased the detection of low levels of virus in vaginal lavage samples by concentrating the virus or reducing the amount of potential inhibitors of the QC-PCR assay. Additionally, in the direct comparison study, the blood plasma virus loads were highly correlated between the QC-PCR and commercial assays and were within the range of variations reported for comparisons of commercially available assays [26, 27], but the QC-PCR assay detected HIV-1 RNA in significantly more vaginal lavage samples than did the commercial test.

In recent studies of men [28–30], the correlation between virus loads in blood plasma and semen and the effect of antiretroviral therapy on HIV-1 levels in semen were similar to those found in our investigation and in other studies of women [14, 16, 17]. Thus, most data currently available suggest that HIV-1 virus loads in blood plasma and genital tract secretions are correlated in both men and women who do not have genital tract infections. However, in women who have genital tract infections, HIV-1 blood plasma virus loads have not been correlated with the detection of HIV-1 in cervicovaginal lavage samples, when CD4+ lymphocyte levels are controlled for [31]. Studies that quantify genital tract HIV-1 in women during and after these infections are needed to understand how infection affects the correlation between virus loads in blood plasma and female genital secretions.

The correlations we observed between virus loads in blood plasma and female genital secretions were not the result of blood contamination. Although blood was detected in almost 75% of the vaginal lavage samples, <5% of vaginal virus loads was from blood-derived virus. Moreover, in 87% of the cervical mucus samples, virus from contaminating blood accounted for <25% of the virus load in the samples. In previous studies that have also found a correlation between blood plasma and female genital tract virus loads, the proportion of genital tract virus load derived from blood contamination was not determined [14, 16, 17]. The study by Cu Uvin et al. [14] did not report on the presence of blood in cervicovaginal lavage samples collected for virus load analysis. However, in the other two studies, the observation of blood in vaginal or cervical swab samples was associated with a significant increase in the levels of genital tract HIV-1 RNA [16, 17]. Furthermore, Ghys et al. [31] suggest that lavage sampling may be a better method for detecting HIV-1 in genital tract secretions than swabbing, since swabbing may provoke bleeding that could introduce blood plasma virus into genital secretions.

To improve the comparability of results between future studies of HIV-1 in the female genital tract, standard sampling techniques should be used. We recommend that vaginal lavage and cervical mucus samples be collected separately and that virus loads be reported as total number of HIV-1 RNA copies per sample. Also, since blood contamination may contribute a significant proportion of the virus load in certain samples, especially from the cervix, the amount of blood in each sample should be measured with the use of a quantitative technique. Vaginal secretions should also be tested for seminal fluid. Because of a lack of consensus about which component of female genital tract secretions is correlated with infectivity, both cell-free and cell-associated fractions should be obtained.

The high correlation between cell-free and cell-associated vaginal virus loads suggests that vaginal HIV-1 may be produced in situ. Previous studies have shown that cervical mucus and vaginal secretions of infected women contain HIV-1 proviral DNA, and cell-associated virus from the vagina and cervix can be propagated in cell culture [3–5, 7, 8]. In our study, the number of viable cells in >80% of vaginal lavage samples was below the level of detection. However, a small number of HIV-1-infected cells in vaginal secretions could be the source of cell-free virus. Additional information about viral replication in situ, such as data about multiply spliced HIV-1 RNA, are needed to determine if HIV-1-infected cells in vaginal secretions are a significant source of cell-free virus in the vagina.

The correlation of cervical and cell-free vaginal virus loads might be the result of cervical mucus draining into the vagina. This hypothesis is supported by the fact that in most of the women we studied, >50% of the combined lower genital tract virus load was found in the vaginal lavage sample. Also, 19 (90%) of 21 women with a vaginal virus load had detectable cervical HIV-1 RNA. However, 5 (50%) of 10 women without a cervix had detectable HIV-1 in their vaginal lavage samples, suggesting that HIV-1 shedding might also occur through the vaginal mucosa. Further investigations, including studies that compare blood plasma, vaginal, and cervical HIV-1 quasispecies, are needed to determine the origin of the virus in female genital tract secretions.

Our findings that blood plasma and vaginal virus loads are correlated and that a significant reduction in vaginal virus load occurs after antiretroviral therapy is initiated support the hypothesis that antiretroviral therapy, which lowers blood plasma virus load, reduces the risk of perinatal and heterosexual transmission by lowering vaginal virus load. For example, zidovudine reduces perinatal transmission from 23% to 8% when administered antepartum, intrapartum, and postpartum, and several studies suggest that most cases of perinatal transmission occur during delivery [32–34]. Thus, zidovudine may reduce perinatal transmission in part by lowering vaginal virus load. Moreover, antiretroviral therapy may similarly reduce female-to-male heterosexual transmission. Additional investigations are needed to determine if antiretroviral therapy reduces virus load to low or undetectable levels in vaginal secretions over long periods. If antiretroviral therapy has this effect, strategies to prevent heterosexual transmission will need to focus more on early identification and treatment of HIV-1-infected adults.

Previous SectionNext Section

Footnotes

  • Informed consent was obtained from patients in this study, and guidelines of the US Department of Health were followed in the conduct of clinical research.

  • Financial support: Centers for Disease Control and Prevention (collaborative agreement U64/CCU412279); Emory Medical Care Foundation (grant to J.L.L.).

  • Received April 9, 1998.
  • Revision received November 18, 1998.
Previous Section
 

References

  1. 1.↵
    1. Mostad SB,
    2. Kreiss JK
    . Shedding of HIV in the genital tract. AIDS 1996;10:1305-15.
    MedlineWeb of Science
  2. 2.↵
    1. Kuhn L,
    2. Abrams EJ,
    3. Matheson PB,
    4. et al
    . Timing of maternal-infant HIV transmission: associations and early polymerase chain reaction results. AIDS 1997;11:429-31.
    MedlineWeb of Science
  3. 3.↵
    1. Vogt MW,
    2. Witt DJ,
    3. Craven DE,
    4. Byington R,
    5. Crawford DF,
    6. Schooley RT
    . Isolation of HTLV-III/LAV from cervical secretions of women at risk for AIDS. Lancet 1986;1:525-7.
    MedlineWeb of Science
  4. 4.
    1. Wofsy CB,
    2. Cohen JB,
    3. Hauer LB,
    4. et al
    . Isolation of AIDS-associated retrovirus from genital secretions of women with antibodies to the virus. Lancet 1986;1:527-9.
    MedlineWeb of Science
  5. 5.↵
    1. Vogt MW,
    2. Witt DJ,
    3. Craven DE,
    4. et al
    . Isolation patterns of the human immunodeficiency virus from cervical secretions during the menstrual cycle of women at risk for the acquired immunodeficiency syndrome. Ann Intern Med 1987;106:380-2.
    Abstract/FREE Full Text
  6. 6.
    1. Van de Perre P,
    2. DeClerq A,
    3. Cogniaux-Leclerc J,
    4. Nzaramba D,
    5. Butzer JP,
    6. Sprecher-Goldberger S
    . Detection of HIVp17 antigen in lymphocytes but not epithelial cells from cervicovaginal secretions of women seropositive for HIV: implications for heterosexual transmission of the virus. Genitourin Med 1988;64:30-3.
    MedlineWeb of Science
  7. 7.↵
    1. Henin Y,
    2. Mandelbrot L,
    3. Henrion R,
    4. Pradinaud R,
    5. Coulaud JP,
    6. Montagnier L
    . Virus excretion in the cervicovaginal secretions of pregnant and nonpregnant HIV-infected women. J Acquir Immune Defic Syndr 1993;6:72-5.
    MedlineWeb of Science
  8. 8.↵
    1. Clemetson DBA,
    2. Moss GB,
    3. Willerford DM,
    4. et al
    . Detection of HIV DNA in cervical and vaginal secretions. JAMA 1993;269:2860-4.
    Abstract/FREE Full Text
  9. 9.↵
    1. Kreiss J,
    2. Willerford DM,
    3. Hensel M,
    4. et al
    . Association between cervical inflammation and cervical shedding of human immunodeficiency virus DNA. J Infect Dis 1994;170:1597-601.
    Abstract/FREE Full Text
  10. 10.
    1. Zorr B,
    2. Schafer APA,
    3. Dilger I,
    4. Habermehl KO,
    5. Kosh M
    . HIV-1 detection in endocervical swabs and mode of HIV-1 infection. Lancet 1994;343:852.
    CrossRefMedlineWeb of Science
  11. 11.
    1. Nielsen K,
    2. Boyer P,
    3. Dillon M,
    4. et al
    . Presence of human immunodeficiency virus (HIV) type 1 and HIV-1-specific antibodies in cervicovaginal secretions of infected mothers and in gastric aspirate of their infants. J Infect Dis 1996;173:1001-4.
    Abstract/FREE Full Text
  12. 12.↵
    1. John GC,
    2. Nduati RW,
    3. Mbori-Ngacha D,
    4. et al
    . Genital shedding of human immunodeficiency virus type 1 DNA during pregnancy: association with immunosuppression, abnormal cervical or vaginal discharge, and severe vitamin A deficiency. J Infect Dis 1997;175:57-62.
    Abstract/FREE Full Text
  13. 13.↵
    1. Rasheed S,
    2. Zhiliang L,
    3. Dong X,
    4. Kovacs A
    . Presence of cell-free human immunodeficiency virus in cervicovaginal secretions is independent of viral load in the blood of human immunodeficiency virus-infected women. Am J Obstet Gynecol 1996;175:122-30.
    CrossRefMedlineWeb of Science
  14. 14.↵
    1. Uvin SC,
    2. Caliendo AM
    . Cervicovaginal human immunodeficiency virus secretion and plasma viral load in human immunodeficiency virus-seropositive women. Obstet Gynecol 1997;90:739-43.
    CrossRefMedlineWeb of Science
  15. 15.
    1. Loussert-Ajaka I,
    2. Mandelbrot L,
    3. Delmas MC,
    4. et al
    . HIV-1 detection in cervicovaginal secretions during pregnancy. AIDS 1997;11:1575-81.
    CrossRefMedlineWeb of Science
  16. 16.↵
    1. Iversen AKN,
    2. Larsen AR,
    3. Jensen T,
    4. et al
    . Distinct determinants of human immunodeficiency virus type 1 RNA and DNA loads in vaginal and cervical secretions. J Infect Dis 1998;177:1214-20.
    Abstract/FREE Full Text
  17. 17.↵
    1. Goulston C,
    2. McFarland W,
    3. Katzenstein D
    . Human immunodeficiency viruses type 1 RNA in the female genital tract. J Infect Dis 1998;177:1100-3.
    Abstract/FREE Full Text
  18. 18.↵
    1. Davies A
    . A preliminary investigation using p-nitrophenyl phosphate to quantitate acid phosphatase on swabs examined in cases of sexual assault. Med Sci Law 1978;18:174-8.
    MedlineWeb of Science
  19. 19.↵
    1. Mulder J,
    2. McKinney N,
    3. Christopherson C,
    4. Sninsky J,
    5. Greenfield L,
    6. Swok S,
    7. et al
    . A rapid and simple PCR assay for quantitation of human immunodeficiency virus type 1 RNA in plasma: application to acute retroviral infection. J Clin Microbiol 1994;32:292-300.
    Abstract/FREE Full Text
  20. 20.↵
    1. Piatak M,
    2. Luk KC,
    3. Williams B,
    4. Lifson JD
    . Quantitative competitive polymerase chain reaction for accurate quantitation of HIV DNA and RNA species. BioTechniques 1993;14:70-80.
    MedlineWeb of Science
  21. 21.↵
    1. Guenthner PC,
    2. Hart CE
    . A quantitative competitive PCR assay for HIV-1 using a microtiter-based detection system. BioTechniques 1998;24:810-6.
    MedlineWeb of Science
  22. 22.↵
    1. Cassol S,
    2. Salas T,
    3. Lapointe N,
    4. Arella M,
    5. Rudnik J,
    6. O'Shaughnessy M
    . Improved detection of HIV-1 envelope sequences using optimized PCR and inosine substituted primers. Mol Cell Probes 1991;5:157-60.
    CrossRefMedlineWeb of Science
  23. 23.↵
    1. Myers G,
    2. Korber B,
    3. Berzofsky JA,
    4. Smith RF,
    5. Pavlakis GN
    , editors. Human retroviruses and AIDS. Los Alamos, NM: Los Alamos National Laboratory; 1991. A compilation of analysis of nucleic acid and amino acid sequences.
  24. 24.↵
    1. Sassa S
    . Regulation of the genes for heme pathway enzymes in erythroid and in non-erythroid cell. Int J Cell Cloning 1990;8:10-26.
    Medline
  25. 25.↵
    1. Saltarelli MJ,
    2. Hadziyannis E,
    3. Hart CE,
    4. et al
    . Analysis of human immunodeficiency virus type 1 mRNA splicing patterns during disease progression in peripheral blood mononuclear cells from infected individuals. AIDS Res Hum Retroviruses 1996;12:1443-56.
    MedlineWeb of Science
  26. 26.↵
    1. Van damme AM,
    2. Schmit JC,
    3. VanDooren S,
    4. et al
    . Quantification of HIV-1 RNA in plasma: comparable results with the NASBA HIV-1 RNA QT and the Amplicor HIV Monitor test. J Acquir Immune Defic Syndr Hum Retrovirol 1996;13:127-39.
    MedlineWeb of Science
  27. 27.↵
    1. Weverling GJ,
    2. Lange JM,
    3. de Jong MD,
    4. de Weerd H,
    5. Goudsmit J,
    6. de Wolf F
    . A comparison of serum HIV-1 RNA levels as measured by two quantitative assays in zidovudine-treated, asymptomatic, HIV-infected individuals. Antiviral Ther 1996;1:255-63.
    Medline
  28. 28.↵
    1. Gupta P,
    2. Mellors J,
    3. Kingsley L,
    4. et al
    . High viral load in semen of human immunodeficiency virus type 1-infected men at all stages of disease and its reduction by therapy with protease and nonnucleoside reverse transcriptase inhibitors. J Virol 1997;71:6271-5.
    Abstract/FREE Full Text
  29. 29.
    1. Vernazza P,
    2. Gilliam BL,
    3. Dyer J,
    4. et al
    . Quantification of HIV in semen: correlation with antiviral treatment and immune status. AIDS 1997;11:987-93.
    CrossRefMedlineWeb of Science
  30. 30.↵
    1. Vernazza PL,
    2. Gilliam BL,
    3. Flepp M,
    4. et al
    . Effect of antiviral treatment on the shedding of HIV-1 in semen. AIDS 1997;11:1249-54.
    CrossRefMedlineWeb of Science
  31. 31.↵
    1. Ghys PD,
    2. Fransen K,
    3. Diallo MO,
    4. et al
    . The associations between cervicovaginal HIV shedding, sexually transmitted diseases and immunosuppression in female sex workers in Abidjan, Cote d'Ivoire. AIDS 1997;11:F85-93.
    MedlineWeb of Science
  32. 32.↵
    1. Sperling RS,
    2. Shapiro DE,
    3. Coombs RW,
    4. et al
    . Maternal viral load, zidovudine treatment, and the risk of transmission of human immunodeficiency virus type 1 from mother to infant. N Engl J Med 1996;335:1621-9.
    CrossRefMedlineWeb of Science
  33. 33.
    1. Fang G,
    2. Burger H,
    3. Grimson R,
    4. et al
    . Maternal plasma immunodeficiency virus type 1 RNA level: a determinant and projected threshold for mother-to-child transmission. Proc Nat Acad Sci USA 1995;92:12100-4.
    Abstract/FREE Full Text
  34. 34.↵
    1. Dickover RE,
    2. Garatty EM,
    3. Herman SA,
    4. et al
    . Identification of levels of maternal HIV-1 RNA associated with risk of perinatal transmission: effect of maternal zidovudine treatment on viral load. JAMA 1996;275:599-605.
    Abstract/FREE Full Text
| Table of Contents

This Article

  1. J Infect Dis. (1999) 179 (4): 871-882. doi: 10.1086/314656
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 205 (5)
  1. Journal of Infectious Diseases
  1. Alert me to new issues

Most