Puisque l’auto-rapportage des comportements sexuels est sujet à de nombreux biais, les biomarqueurs de l’exposition au sperme sont de plus en plus utilisés pour détecter les rapports sexuels non protégés (RSNP). Cette étude visait à comparer la performance d’un essai PCR niché maison ciblant les gènes de la famille des protéines spécifiques testiculaires Y (TSPY) à celles de six autres méthodes de détection d’une exposition récente au sperme.
Suite à l’analyse de 45 échantillons, la sensibilité, la spécificité et la valeur prédictive ont été calculées pour chaque méthode en comparaison avec la PCR nichée ciblant TSPY. Les résultats suggèrent que les méthodes courantes de détection de l’exposition récente au sperme manquent de sensibilité comparativement à la PCR nichée ciblant TSPY.
Cet essai pourrait être d’une grande utilité pour détecter les RSNP dans des études observationnelles où plusieurs facteurs peuvent accélérer la clairance des biomarqueurs.
4.2 Abstract
Objectives: Because self-report of sexual behaviours is prone to biases, biomarkers of
recent semen exposure are increasingly used to assess unprotected sex. This study was undertaken to compare the performance of an in-house nested PCR assay targeting testis- specific protein Y-encoded (TSPY) genes to those of six other commonly used methods to detect recent semen exposure.
Methods: A subset of 45 vaginal samples was semi-randomly selected at baseline of a
prospective demonstration study aiming to assess the feasibility and usefulness of early antiretroviral treatment and pre-exposure prophylaxis among female sex workers in Cotonou, Benin. Unprotected sex was assessed with: self-report in the last two days, self- report in the last 14 days, a rapid prostate-specific antigen (PSA) detection test, a quantitative PCR assay targeting the sex-determining region (SRY), a standard PCR assay targeting SRY, a standard PCR assay targeting TSPY, and a nested PCR assay targeting TSPY (n-TSPY). Sensitivity, specificity, and predictive values were calculated for each method in comparison with the n-TSPY.
Results: A total of 70.5% (31/44) of vaginal samples tested positive with n-TSPY. Using
n-TSPY as reference, all other methods had poor sensitivity to detect unprotected sex, especially in low target samples. PSA had the best performance to detect unprotected sex with a sensitivity of 64.0%, a specificity of 100%, a positive predictive value of 100%, and a negative predictive value of 77.2%.
Conclusions: Results show that commonly used methods to detect recent semen exposure
lack sensitivity to detect unprotected sex compared to n-TSPY assay. The n-TSPY is an affordable assay that may have great utility in assessing unprotected sex in observational studies where many factors are expected to accelerate biomarkers’ clearance.
4.3 Introduction
Unprotected sex being a major risk factor for HIV and other sexually transmitted infections (STI), it must be accurately measured in HIV/STI surveillance, treatment, and prevention research. Although the most common method used to monitor condom use is self-report, this method is prone to social desirability and recall biases.1 To circumvent those biases,
biomarkers of recent semen exposure have been used to detect unprotected sex among women. Among the most characterized and widely used biomarkers are the prostate- specific antigen (PSA) and Y-chromosomal DNA (Yc-DNA).
PSA is a protein that can be found at mean concentrations >1 mg/mL in seminal fluid of healthy men154 and that can be detected in vaginal samples for a mean period of 20 to 27
hours and up to two days after exposure to semen.8-11 Yc-DNA is unique to males and is
found in every cells of a man, including spermatozoa. Using a polymerase chain reaction (PCR), Yc-DNA can be detected in vaginal samples up to two weeks after exposure to semen with a half-life for clearance of 3.8 days.12,13
The most commonly targeted gene to detect Yc-DNA by the use of PCR is the sex- determining region Y (SRY) gene.164 Most recently, the testis-specific protein Y-encoded
family of homologous genes has been identified as a novel target to detect Yc-DNA in vaginal samples.165 Both standard PCR followed by gel electrophoresis detection and
quantitative PCR have been used to detect SRY and TSPY genes in vaginal samples. With the use of either a standard or a quantitative PCR, a TSPY gene has been detected at higher rates and over a longer period of time than SRY in vaginal samples, while the use of a quantitative PCR increased sensitivity and window of detection of both SRY and TSPY genes compared to the use of a standard PCR.165,167
Other factors were shown to affect the sensitivity of Yc-DNA detection. Vaginal inoculation with low amount of semen has led to lower sensitivity of Yc-DNA detection compared to inoculation with higher amounts of semen.9 Post-coital Yc-DNA levels were shown to be
lower during menses compared to non-menses periods.173 Vaginal douching, a highly
prevalent feminine hygiene practice,174 is expected to accelerate the biomarkers’ clearance.
In vitro assays have shown that some microbicides and lubricants could lead to partial or total inhibition of Yc-DNA detection.175,176 In observational settings where condom breakage
or slippage might lead to low exposure to semen, where menses or vaginal douching are expected to washout semen, and where women can use topical products that might interfere
with biomarkers detection, a test that allows the detection of low target concentrations is required.
Given the observations above, a quantitative PCR targeting TSPY genes might be the best option to detect unprotected sex in observational settings. However, quantitative PCR requires specialized and expensive equipment that is not always available in settings such as low to middle income countries.
Nested PCR is an affordable method that consists of two successive rounds of standard PCR with each round using different sets of primers. The double rounds of amplification increase the sensitivity of the assay while the double sets of primers increase its specificity.168 We hypothesised that a nested PCR assay targeting TSPY genes would be
more sensitive for the detection of recent semen exposure while still being specific. This study was thus undertaken to compare the performance of an in-house nested PCR assay targeting TSPY (n-TSPY) to six other used methods to detect recent semen exposure: self- reported unprotected sex in the last two days (SR-2d), self-reported unprotected sex in the last 14 days (SR-14d), a rapid PSA detection test (PSA), a commercial quantitative PCR assay targeting SRY (q-SRY), a standard PCR assay targeting SRY (s-SRY), and a standard PCR assay targeting TSPY (s-TSPY).
4.4 Methods
Samples analysed in this study were collected at baseline of a larger prospective observational demonstration study that aimed to assess the feasibility and usefulness of early antiretroviral treatment (E-ART) and pre-exposure prophylaxis (PrEP) among female sex workers (FSW) in Cotonou, Benin (ClinicalTrials.gov: NCT02237027).29,177 This study
was conducted from October 2014 to December 2016. At recruitment, participants provided a free and informed consent for study participation but were not informed of the specific purpose of semen detection before the end of the study to avoid further information bias. At the final visit, participants provided a free and informed consent for the explicit use of the vaginal samples to assess recent unprotected sex. Participants were free to decline or withdraw from the study at any time. The protocol, including the procedures for delayed information concerning the use of vaginal samples to assess unprotected sex, was approved by the ethics committee of the CHU de Québec-Université Laval and the Benin National Ethics Committee for Health Research.
At enrollment, a face-to-face interview was administered to the participants in a private setting by two trained interviewers to assess sexual behaviours from the past two or 14 days. For each of the two recall periods, participants were asked the number of vaginal sex acts with any type of sexual partners (client, regular partner, and non-paying and non-regular partner), the frequency of condom use (never, less than half of the time, at least half of the time, or always), and whether they experienced condom breakage or slippage. Participants who reported vaginal sex were classified as having had unprotected sex if condoms were not always used or if an episode of condom breakage or slippage occurred in the recall period. After the interview, a vaginal swab was collected by a physician for screening for PSA and Yc-DNA. Details about sample selection and preparation are presented in Supplementary Text 4-1.
Detection of PSA
PSA was detected using an ABAcard p30 detection cassette (Abacus Diagnostics, West Hills, CA) according to the manufacturer’s instructions. Briefly, 200 µL of the extract solution was added to the sample well and incubated for 10 min at room temperature. A pink line both at the test (T) and control (C) positions indicated a positive result while a pink line at the C position only indicated a negative result. The absence of a pink line at the C position indicated an inconclusive result.
Detection of Yc-DNA
Yc-DNA was detected using four different PCR assays: a commercial quantitative PCR targeting SRY (q-SRY), a standard PCR targeting SRY (s-SRY), a standard PCR targeting TSPY (s-TSPY), and a nested PCR targeting TSPY (n-TSPY). The q-SRY was performed using the Quantifiler Duo DNA Quantification kit (Applied Biosystems, ThermoFisher Scientific, Waltham, MA) according to the manufacturer’s guidelines with no modification.178
Briefly, 2 µL of total DNA extract was added to 23 µL of Quantifiler Duo Primer Mix. Amplification was performed using a 7500 real-time PCR system (Applied Biosystems, ThermoFisher Scientific, Waltham, MA) with the amplification settings reported in Table 4-1. Primers and probe sequences of the q-SRY being proprietary, this information could not be retrieved.
Primers for the s-SRY were synthesized according to previously published sequences.163
Primers for s-TSPY and n-TSPY were designed using Primer3,179 and their specificity was
confirmed by BLAST analysis. Primer sequences are reported in Table 4-1.
The EconoTaq Plus Green 2X Master Mix (Lucigen, Middleton, Wisconsin, USA) was used for amplification with the standard and nested assays. For the s-SRY and the s-TSPY, 5 µL of total DNA extract was added to 20 µL of master mix containing 1 µM of primers and EconoTaq solution according to the manufacturer’s protocol.180 For the n-TSPY, 5 µL of a
1:10 dilution of the amplification product of the s-TSPY was added to 20 µL of master mix containing 1 µM of primers targeting TSPY inside the previously amplified sequence. Amplification was performed using a Bio-Rad iCycler (Bio-Rad Laboratories, Hercules, CA, USA). Amplifications settings are reported for each assay in Table 4-1. After amplification, the final PCR products of the s-SRY (229 base pairs (bp)), the s-TSPY (178 bp), and the n-TSPY (115 bp) were visualized under UV transillumination following electrophoresis on a 2% agarose gel and staining with Gel Red (Biotium Inc. Fremont, CA).
Each test sample was run in replicates of five for the q-SRY or in replicates of three for the s-SRY, the s-TSPY, and the n-TSPY. A sample was considered negative if all replicates had no amplification and positive if at least one replicate had amplification. For the q-SRY, a sample was also considered quantifiable if at least three replicates of five had amplification. In each PCR assay, DNA extracted from peripheral blood mononuclear cells from male and female donors were used as positive and negative controls, respectively, and nuclease-free water was used as a no template control (NTC). All laboratory procedures were performed by female technicians (unblinded) to avoid male DNA contamination.
Statistical analyses
The proportion of women who had unprotected sex according to each of the seven methods was calculated and reported with 95% confidence intervals (95%CI). As we hypothesized that n-TSPY would be the most sensitive of the seven methods, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for the six other individual methods compared to n-TSPY. To better define the capacity of the different methods to detect unprotected sex compared to n-TSPY, we planned to calculate sensitivity among all n-TSPY positive samples, but also among low-, moderate-, or high-
positive n-TSPY samples as defined by the number of replicates with amplification (one, two, or three out of three, respectively).
Contrary to sensitivity and specificity that do not depend on the outcome prevalence, predictive values are only meaningful if the prevalence of the outcome of interest as measured by the reference method represents the actual proportion of the positives in the relevant population. Because positive n-TSPY samples were over-represented in our sample (Supplementary Text 4-1) we estimated the predictive values by taking into account the expected proportion of positives in the target population using the formulas presented in Supplementary Text 4-2.181
Agreement of results between n-TSPY and each of the six other methods was tested using the McNemar’s test. Statistical analyses were conducted in SAS Studio, version 3.71 (SAS Institute Inc., Cary, NC, USA).
4.5 Results
Out of the 361 FSW recruited in the E-ART/PrEP study, a random subset of 45 women were screened for unprotected sex by the use of seven methods. Five women had missing data for at least one of the methods and for three other women, self-reported sexual behaviours and vaginal samples were not collected on the same day, which did not allow a reliable comparison of self-report with the biomarkers for these women. Thus, the proportion of women who had unprotected sex was calculated among all available data for each method, but comparison analyses were restricted to the 37 samples for which all seven unprotected sex measurements were available and could be reliably compared together. None of the PSA tests was inconclusive. No case of false-positive Yc-DNA was observed in any of the negative or NTC controls that were run with any of the PCR assays. All standard and nested PCR products were of expected size.
Baseline characteristics of the 45 participants are shown in Table 4-2. Mean age was estimated at 34.1 years (SD = 9.4). Most participants were Beninese, poorly educated, not married, and HIV negative.
Prevalence of unprotected sex
Among the selected 45 samples, the proportion of FSW who had unprotected sex according to the different methods varied from 22.2% (95%CI: 11.2%-37.1%) to 70.5% (95%CI: 54.8%-
83.2%), with the lowest prevalence having been measured by s-SRY and the highest, by n-TSPY (Table 4-3).
A total of 14 samples were positive with the q-SRY (31.1%; 95%CI: 18.2%-46.7%). Out of these, four samples (28.6%; 95%CI: 8.4-58.1%) were quantifiable, with a median concentration of 170 Yc-DNA copies/sample (IQR=97-319 copies/sample).
Performance of commonly used methods to detect recent semen exposure compared to the nested PCR TSPY assay
Among the 37 samples for which all seven unprotected sex measurements were available and could be reliably compared together, 25 (67.6%) were tested positive with n-TSPY. Of those 25 positive samples, nine were low-, two were moderate-, and 14 were high-positive. Due to the low number of moderate-positive samples, these were pooled together with low-positive samples for the calculation of sensitivities by subgroups.
The sensitivity, specificity, and predictive values for each method compared to n-TSPY are reported in Table 4-4. When calculated among all positive n-TSPY samples, the sensitivity of the six other methods varied from 32.0% (95%CI: 15.0%-53.5%) to 64.0% (95%IC: 42.5%-82.0%), with s-SRY having the lowest sensitivity and PSA having the highest sensitivity to detect semen among positive n-TSPY samples. The s-TSPY (48.0%; 95%CI: 27.8%-68.7%) was more sensitive than s-SRY (32.0%; 95%CI: 15.0%-53.5%) to detect Yc-DNA and q-SRY (56.0%; 95%CI: 34.9%-75.6%) was more sensitive than either s-SRY or s-TSPY to detect Yc-DNA. Despite being the third method with the higher sensitivity, the q-SRY could only quantify 16.0% (95%IC: 4.5%-36.1%) of the 25 positive n-TSPY samples. When calculated among low/moderate-positive n-TSPY, the sensitivity was very low for all methods. The q-SRY and s-SRY could detect none of the 11 low/moderate-positive samples while the highest sensitivity was estimated at only 18.2% (95%CI: 2.3%-51.8%) with the SR-14d or PSA. Among high-positive samples, sensitivity was higher for all methods. Still, the s-SRY had the lowest sensitivity with 57.1% (95%CI: 28.9%-82.3%). PSA and q-SRY could detect 100% (95%CI: 76.8%-100%) of the 14 high-positive samples. However, the q-SRY could quantify only 28.6% (95%CI: 8.4%-58.1%) of the high-positive samples. The lowest PPV was observed with SR-2d with only 66.2% of FSW being positive for n-TSPY while reporting having had unprotected sex. The PSA, q-SRY, and s-TSPY had the
highest PPV with every positive test being also positive for n-TSPY (PPV=100%). The s-SRY had the lowest NPV with 62.2% of negative s-SRY being also negative for n-TSPY. The highest NPV was observed with PSA with 77.2% of negative PSA being negative for n-TSPY.
SR-14d was the only method for which the null hypothesis of agreement of results with n-TSPY could not be rejected (p=0.092).
4.6 Discussion
This study was undertaken to compare the performance of an in-house designed nested PCR assay targeting TSPY to six other other methods to detect recent semen exposure: self-reported unprotected sex in the last two days, self-reported unprotected sex in the last 14 days, a rapid PSA detection test, a commercial quantitative PCR assay targeting SRY, a standard PCR assay targeting SRY, and a standard PCR assay targeting TSPY. Our results confirm our hypothesis that n-TSPY is more sensitive for the detection of recent semen exposure than any of the compared methods. Indeed, among a semi-random subset of 45 women at baseline of the E-ART/PrEP study, the higher prevalence of unprotected sex was measured with n-TSPY. Compared to n-TSPY, all other methods have shown sensitivities <65% to detect positive n-TSPY. Sensitivities were particularly low to detect low/moderate- positive n-TSPY samples, suggesting that unprotected sex detection was impaired by low target concentrations. The low concentration of Yc-DNA in our subset of samples was confirmed by the quantification performed with q-SRY. Using the q-SRY, only 28% of the positive samples were quantifiable (16% of the n-TSPY positive samples), with a median concentration of 170 Yc-DNA copies/sample. In a previous study using the same commercially available q-SRY among sexually active women participating to a microbicide trial, sample concentrations were more than twice those we observed.24,182 The difference
could be explained by higher rates of vaginal douching habit in our population where, at baseline, 100% of a random subset of 221 FSW reported at least one vaginal douche in the last two days, while only about 30% of participants reported having douched after sex in the microbicide trial.182
Our results are consistent with previous studies showing higher detection of unprotected sex using PSA compared to SR-2d and suggest that some participants under-reported unprotected sex.14,17,25 Our results are also consistent with previous studies showing that a
TSPY gene is more sensitive than s-SRY to detect Yc-DNA and that quantitative PCR is more sensitive than standard PCR to detect Yc-DNA.165,167
The q-SRY, s-SRY, and s-TSPY were less sensitive than SR-14d to detect unprotected sex. These results are consistent with a previous study showing lower frequency of Yc-DNA detection compared to self-report in the last 14 days among female adolescents.26 Though
Yc-DNA can be detected up to two weeks after semen exposure, Yc-DNA concentration declines rapidly over this period of time,12,167 which could explains lower rates of unprotected
sex as measured by Yc-DNA detection compared to SR-14d. In our study, q-SRY, s-SRY, and s-TSPY were also less sensitive than PSA, a biomarker of semen exposure in the last two days, suggesting that the former tests lacked sensitivity to detect even very recent semen exposure.
After n-TSPY, PSA offered the best performance in detecting unprotected sex with a sensitivity of 64.0%, a specificity and PPV of 100%, and an NPV of 77.2%. Though SR-14d had the worse specificity of all compared-methods, it had the second-best sensitivity to detect unprotected sex either among low/moderate- or high-positive n-TSPY samples and