Assisted reproduction in HIV-1-serodifferent couples: the need for viral validation of processed semen

Background: Many HIV-infected men and women have a strong desire for a child. Assisted reproductive technologies (ART) are an option for HIV-serodifferent couples to reduce the risk of HIV transmission from an infected man to the woman. Potential HIV contamination of selected spermatozoa after semen processing is an important issue in this context.

Methods: HIV in processed semen obtained in our laboratory since 1995 were analysed. HIV RNA and DNA detection was performed in the selected spermatozoa of 125 men. HIV RNA was analysed in blood and semen plasma as well as HIV DNA in non-sperm cells.

Results: HIV RNA and DNA were detected in the selected spermatozoa of eight and two men (6.4% and 1.6%), respectively. HIV RNA was detected with a median load of 5 copies/10⁶ spermatozoa. Six of the eight men were untreated, one was taking nucleoside analogue therapy and one was on highly active antiretroviral treatment (HAART). HIV RNA detection was more likely to be positive in selected spermatozoa of men with high seminal plasma viral load. HIV RNA was detected in 26% and 11% of selected spermatozoa fractions when the seminal plasma load was > 10 000 copies/ml and 20-10 000 copies/ml, respectively, but in none when the seminal plasma tested negative.

Conclusion: Selected spermatozoa may be positive for HIV RNA detection even in treated patients. Viral validation of processed semen is necessary in ART programmes for serodifferent couples, particularly in men with only partially or poorly controlled HIV infection.

Introduction

The use of highly active antiretroviral therapy (HAART) has largely modified the prognosis of HIV infection. As a consequence of improved life expectancy and quality, many young HIV-infected men and women have a strong desire for a child. However, protected intercourse is strongly recommended for serodifferent couples in all circumstances. Indeed, compartmentalization of HIV replication in semen has been demonstrated for some men and, therefore, HIV blood viral load might not always reflect HIV replication levels in semen [^1-4]. Although HAART reduces HIV loads in both blood and seminal compartments [^5-9], low levels of HIV RNA can still be detected in seminal plasma and HIV-infected cells can be recovered in non-sperm cells (NSC) even in those who have undergone prolonged successful treatment [^8-10]. HIV-serodifferent couples are seeking assisted reproductive technologies (ART) to avoid the risk of HIV sexual transmission to the woman. In Italy and Spain, insemination of HIV-negative women with processed semen obtained from HIV-infected men has been performed since 1992. Two large series of 350 and 60 couples, respectively, reported no seroconversion in the women [^11,12]. In France, a research project evaluated ART for HIV-serodifferent couples using intracytoplasmic sperm injection technique (ICSI) and observed no seroconversion among 68 women [¹³].

In ART procedures, spermatozoa are usually recovered after semen processing on a density gradient, which may be followed by a 'swim-up' technique for the recovery of motile spermatozoa. HIV is present in the semen as free particles in the seminal plasma and as cell-associated provirus in CD4-carrying non-spermatozoa cells such as T lymphocytes and macrophages [¹⁴]. HIV infection of spermatozoa is controversial. Indeed, early publications reported HIV-1 DNA detection by polymerase chain reaction (PCR) in situ hybridization in spermatogonia, spermatocytes and spermatids of testes biopsies [¹⁵] and in spermatozoa selected by density gradients [¹⁶]. Moreover, intracellular virus-like particles were found in sperm cells by electron microscopy [¹⁷]. The presence of HIV proviral DNA or HIV particles in spermatozoa was not confirmed in more recent studies. A lack of CD4 and CCR5 receptors on the surface of spermatozoa has been reported, suggesting that these cells are unlikely to be readily infectable by HIV [^18-20]. However, HIV recombinant gp120 can bind to GalAAG glycolipid, which is expressed by 30-35% of human spermatozoa [^21,22]. Therefore, even if HIV is unlikely to penetrate and infect the spermatozoa, it is possible that gp120 particles may bind to the spermatozoa. Selected spermatozoa may consequently contain low residual amounts of HIV particles. Selected spermatozoa obtained from HIV-positive men have been assessed for the presence of both HIV free particles (HIV RNA) and HIV cell-associated virus (HIV DNA) with controversial results: HIV RNA or HIV DNA detected either in none [^20,23,24] or in up to 15-40% of sperm fractions tested [^4,12,25].

The present study has analysed the results of HIV detection in seminal plasma, in NSC and in selected spermatozoa obtained in our laboratory since 1995 in order to improve the viral validation procedures in ART clinical programmes for HIV-serodifferent couples.

Methods

Patients

The semen of 125 HIV-1-infected men was investigated within three different studies between October 1995 and February 2000. The first study was cross-sectional and was designed to assess HIV burden in the semen of untreated HIV-infected men [⁴]. The second study, an 18-month longitudinal study, aimed at providing a dynamic evaluation of HIV cell-free and cell-associated virus shedding in the male genital compartment in response to prolonged HAART [⁸]. The third study included HIV-1-infected men wishing to have children with non-infected women and who were enrolled in a clinical trial of ICSI with processed semen. All three studies were conducted at Cochin and Necker Hospitals, in Paris, France and were supported by the National Agency for AIDS Research (ANRS). The Cochin Hospital's ethics committee (CCPPRB) approved the three studies. All patients investigated within these protocols provided written informed consent. All virological tests were performed in our laboratory at Necker Hospital.

Among these 125 HIV-infected men, 34 were not treated, 25 were on two or three reverse transcriptase inhibitors and 66 received HAART.

Sample collection and processing

Paired blood and semen samples were obtained for each of the 125 HIV-1 infected men. Blood plasma was frozen in portions at -80°C. Semen samples were obtained by masturbation at the Hospital Cochin laboratory, after a recommended 3 day abstinence. Samples were collected in sterile containers and processed within 1 h after collection, according to the World Health Organization recommendations. Standard measurements of spermatozoa and NSC concentrations and spermatozoa motility were performed. Serum samples as 2 ml undiluted portions or 1:1 diluted in RPMI (Gibco-BRL, Cergy-Pontoise, France) containing 5 mg/ml bromelin (Sigma-Aldrich Chimie, Saint-Quentin, France) were then centrifuged for 20 min at 300 × g on a two-layer discontinuous gradient of 45% and 90% PureSperm (J.C.D., Lyon, France) or of 47.5% and 95% Percoll (Sigma-Aldrich Chimie). Seminal plasma, NSC and spermatozoa were separately recovered. Selected spermatozoa and NSC were washed twice with medium [Ferticult (J.C.D) or RPMI] and centrifuged at 600 × g and 200 × g, respectively. Cells were counted and kept as dry pellets at -80°C. Seminal plasma fractions were centrifuged for 10 min at 6000 × g and supernatants were stored in portions at -80°C.

HIV RNA in blood plasma

HIV RNA in blood plasma was measured by the HIV-1 Monitor 1.5 (Roche SA, Neuilly sur Seine, France) according to the manufacturer's instructions, with a detection limit of 200 copies/ml. The samples containing < 200 copies/ml were tested with an ultrasensitive protocol that provided a detection limit of 20 copies/ml [²⁶].

HIV RNA in seminal plasma

HIV RNA in seminal plasma was measured as previously described [⁴]. The HIV RNA was extracted from 250 μl seminal plasma according to the Nuclisens kit procedure (Organon Teknika, Fresnes, France). HIV RNA amplification was performed with the HIV-1 Monitor 1.5. The internal control provided with the HIV-1 Monitor 1.5 kit was systematically added to the Nuclisens lysis buffer prior to extraction. This allowed validation of both extraction and amplification steps. The detection threshold for seminal plasma specimens varied from 20 to 340 copies/ml depending on the ability of RNA extraction to suppress PCR inhibitors. Each batch included one positive and one negative control, consisting of seminal plasma from HIV-seronegative subjects, spiked or not with a predetermined number of HIV RNA copies to obtain a concentration of 1000 copies/ml.

HIV RNA and HIV DNA in selected spermatozoa

RNA from HIV particles adherent to the spermatozoa was extracted from 4 × 10⁶ selected spermatozoa according to the Nuclisens kit procedure and the extracted HIV RNA was amplified with the HIV-1 Monitor 1.5 kit. The internal control provided with the HIV-1 Monitor 1.5 kit was systematically added to the Nuclisens lysis buffer prior to extraction to allow validation of both the extraction and amplification steps. The detection threshold was 5 copies/10⁶ spermatozoa. Each batch included one positive and one negative control, consisting of selected spermatozoa from HIV-seronegative subjects, spiked or not with a predetermined number of HIV RNA copies.

HIV DNA quantification was performed using a prototypic assay developed by Roche Molecular Systems [^27,28]. Lysis buffer (10 mmol/l Tris, pH 8.3, 50 mmol/l KCl, 2.5 mmol/l MgCl₂, 0.45% Tween-20, 40 mmol/l dithiothreitol and 400 μg/ml proteinase K) was added to 2 × 10⁶ selected spermatozoa. DNA was extracted from the lysis using the QIAamp DNA Blood Mini Kit (Qiagen, Courtaboeuf, France). A 2-3 μg sample of extracted DNA was amplified with a quantitative PCR method based on the Amplicor Monitor 1.5 kit. To validate HIV DNA amplification, a second PCR test was systematically performed after adding the equivalent of 125 copies HIV DNA extracted from a cell line containing 1 copy/cell HIV DNA (8E5 cell line). The detection limit was 5 copies/10⁶ spermatozoa.

HIV DNA in non-sperm cells

Quantification of HIV-infected cells (HIV DNA) was performed on 0.2-5 × 10⁶ NSC to which 100-400 μl lysis buffer from Whole Blood Specimen Preparation Kit (Amplicor; Roche SA) was added. DNA was extracted from this lysis preparation using the QIAamp DNA Blood Mini Kit and 20 μl of extracted DNA was amplified with an in-house real-time PCR method in ABI Prism 7700 Sequence Detection System (Perkin-Elmer Applied Biosystem, Courtaboeuf, France). This PCR amplified a conserved region of HIV-1 LTR, with a sensitivity of 5 copies/reaction [²⁹]. The detection threshold for HIV DNA in NSC specimens varied from 5 to 100 copies/10⁶ cells as it directly depended on the number of NSC available for testing.

Results

HIV RNA in blood plasma

HIV RNA was detectable in the blood of 50 of the 125 men (40%). Median HIV RNA loads in blood plasma were 18 000 copies/ml (range, < 200-570 000) for the 34 untreated men and 6000 copies/ml (range, < 200-450 000) for the 25 men treated by nucleoside analogues. All but 1 of the 66 patients taking HAART had a blood viral load < 200 copies/ml (Table 1). Whenever possible, plasma samples of the men treated by HAART were further tested with the ultrasensitive method (40 samples among the 65 were tested with this method). Seven of the forty (17.5%) patients had a residual blood viral load between 20 and 200 copies/ml, with a median of 56 copies/ml (range, 22-160).

Table 1

HIV RNA in seminal plasma

HIV RNA was detected in 46 of the 113 seminal plasma fractions tested (40.7%). It was detected in the seminal plasma of 26 (78.8%) of the 33 untreated men and in 15 (62.5%) of the 24 men treated by nucleoside analogues, with a median viral load of 5500 copies/ ml (range, < 20-1 000 000) and 700 (range, < 20-12 700), respectively. Seminal plasma was tested for 56 of the 66 men taking HAART. HIV RNA was detectable in the seminal plasma of five of these patients (8.9%) (Table 1). HIV viral loads in the seminal plasma of these five patients were 3224, 1796, 1419, 134 and 84 copies/ml, respectively. All five patients had < 20 copies/ml HIV RNA in their paired blood sample.

HIV RNA loads in blood and seminal compartments were significantly correlated (Spearman rank correlation, r = 0.75; P < 0.0001).

HIV DNA in non-sperm cells

HIV DNA was detected in 31 of the 118 NSC fractions tested (26.2%). Positive results were found in 18 of the 34 untreated men (52.9%) and 8 of the 22 men treated with nucleoside analogues (36.4%). NSC were available for testing in 62 of the 66 men taking HAART. HIV DNA was detectable in the NSC of five of these (8.1%). Seminal plasma HIV RNA was detectable for three of these patients, undetectable for one, and not tested for one.

Median seminal viral load was significantly higher [3.7 log₁₀ copies/ml; (range, 1.2-6.5)] in men presenting with a positive HIV detection in NSC than in men presenting with a negative HIV detection in semen (1.8 log₁₀ copies/ml; range, 1.0-5.0; Wilcoxon test, P < 0.0001).

HIV RNA in selected spermatozoa

HIV RNA was detected in the selected spermatozoa of 8 of 125 men tested (6.4%). Six of these eight men were untreated, one was treated with nucleoside analogue drugs and one was receiving HAART (Table 1). The rate of HIV RNA detection in selected spermatozoa was 17.6% (6/34) in untreated men, 4.0% (1/25) in nucleoside analogues treated men and 1.5% (1/66) in men taking HAART (Table 1). The HIV viral loads detected in selected spermatozoa were very low, with a median of 5 copies/10⁶ (range, 5-240; Table 2). For these eight men, median HIV RNA blood viral load was 73 000 copies/ml (range, < 20-390 000 copies/ml) and median HIV RNA in semen plasma was 43 000 copies/ml (range: 430-330 000 copies/ml). For six of these eight patients, HIV DNA was detected in NSC (Table 2).

Table 2

Among the 66 men on HAART, HIV RNA was detected in selected spermatozoa from one man, with a viral load of 5 copies/10⁶. In a second semen sample obtained 1 month later, the viral load was 17 copies/10⁶ selected spermatozoa. However, HIV RNA was undetectable in the selected spermatozoa from a third semen sample obtained 6 months later. In the three semen samples of this patient, HIV RNA was detected in the seminal plasma with viral loads of, respectively, 1,796, 464 and 238 copies/ml. In contrast, HIV RNA was always undetectable in the blood plasma of this patient (< 20 copies/ml).

Selected spermatozoa with positive HIV RNA were classified according to the levels of HIV RNA in seminal plasma and in blood plasma. HIV RNA was positive in 5 (26.3%) of the 19 men with a seminal viral load above 10 000 copies/ml, in 3 (11.1%) of the 27 men with a seminal viral load between 20 and 10 000 copies/ml and in none of the 67 men with a undetectable HIV RNA in seminal plasma (Table 3). Conversely, HIV RNA in selected spermatozoa was positive in 1 of the 75 men (1.3%) with undetectable blood viral load (Table 3).

Table 3

HIV DNA in selected spermatozoa

HIV DNA was detected in the selected spermatozoa in 2 of the 125 men tested (1.6%). The two positive fractions were obtained from one untreated man and from one man treated with nucleoside analogue drugs (Table 1). Both men had also a positive HIV DNA detection in NSC (Table 2). None of the selected spermatozoa fractions obtained from the 66 men treated with HAART was positive.

Discussion

Data on HIV shedding into semen of infected men was derived from semen samples obtained in our laboratory since 1995. HIV was detected in the selected spermatozoa fraction of 10 men among the 125 tested (8.0%). Eight selected spermatozoa fractions were HIV RNA positive (6.4%) and two were HIV DNA positive (1.6%). HIV RNA load in selected spermatozoa was low, with a median of 5 copies/10⁶ spermatozoa (range, 5-240). Seven of the ten men with a HIV-positive spermatozoa fraction were untreated, two were treated with nucleoside analogue drugs and one received HAART. HIV RNA was detected more frequently in selected spermatozoa in men with high concentrations of virus in blood, apart from one man, who had detectable HIV RNA in the selected sperm fraction with a viral load in blood plasma < 20 copies/ml. HIV RNA in selected spermatozoa was more likely to be positive when the HIV RNA load in the seminal plasma was high. Indeed, HIV RNA was detected in selected spermatozoa samples from 26% and 11% of the patients grouped as having a seminal plasma viral load of > 10 000 and 20-10 000 copies/ml respectively, but in none of those whose seminal plasma tested negative. It is unlikely that HIV detection in selected spermatozoa reported in our study was a false-positive result. Indeed, very few false-positive results are observed with the amplification techniques used (HIV-1 Monitor 1.5 and Amplicor Monitor; Roche Diagnostics System). Furthermore, the fact that the results here are true positive results is supported by the good correlation of HIV RNA detection in selected spermatozoa with the level of seminal plasma viral load and by the absence of positive selected spermatozoa fractions in men with undetectable HIV RNA in seminal plasma.

As discussed in the introduction, HIV free particles present in seminal plasma may bind to the surface of the spermatozoa [^21,22]; this may account for our findings that HIV RNA can be detected in selected spermatozoa and that spermatozoa fractions were more likely to be positive when the HIV load was high in the seminal plasma. There is a body of evidence suggesting that HIV does not penetrate and integrate the DNA of the spermatozoa and that HIV proviral DNA detected in semen is from the presence of infected leukocytes [¹⁴]. Contamination of the sperm pellet with non-sperm infected cells may account for the detection of HIV DNA in selected spermatozoa. In our study, semen processing with density gradient and pellet washing did not always completely eliminate HIV from the fraction of selected spermatozoa. In another study, HIV was detected in 6 of the 101 selected spermatozoa fractions tested (5.6%) [¹²]; however three other studies did not report the presence of either HIV DNA or HIV RNA in spermatozoa recovered after a density gradient followed by a swim-up technique [^20,23,24]. The discrepancy between these results and ours may be partially explained by the use of two successive techniques of selection in other studies. Indeed, the use of a swim-up step, as it allows for a more efficient separation of spermatozoa and non-sperm cells, might help to reduce the contamination of the selected spermatozoa fraction by infected leukocytes and, therefore, reduce the risk of HIV DNA detection on processed semen. The use of a swim-up step might also explain that previously published studies failed to detect HIV RNA on selected spermatozoa. However, semen-processing techniques with or without swim up have rarely been compared [^20-23]. In two studies, a total of 23 semen samples were tested and HIV RNA was detected in only one spermatozoa specimen without swim-up and in no spermatozoa specimen with swim-up [^20-23]. A single published case of discrepant results between HIV RNA detection in swim-up and non-swim-up spermatozoa does not offer strong evidence that the addition of a swim-up step is able to eliminate all residual HIV particles binding on the spermatozoa. We believe that the discrepancy on HIV RNA positive findings in selected spermatozoa between our study and others might be explained by the lower sensitivity of the HIV RNA detection method used in other series (20-80 copies/10⁶) [^20,23,24] and the larger size of our series. Indeed, HIV RNA load in positive selected spermatozoa fractions was low, with a median of 5 copies/10⁶, and only a highly sensitive technique is able to detect such low concentrations. Another explanation could be the relatively low number of patients tested in previous series (54 men including only four men with a seminal plasma viral load > 10 000 copies/ml). We did not add a swim-up step in our protocol and we do not wish to do it in the future, mainly because up to 50% of the spermatozoa might be lost with this technique and this might compromise the success of ART, particularly in men with poor spermatozoa counts or mobility. Indeed, semen alterations were frequent in our population of HIV-infected men and the proportion of rapid motile progressive spermatozoa was significantly reduced (unpublished data). Therefore, a systematic swim-up step may compromise intrauterine insemination in a large proportion of HIV-serodifferent couples.

Same day HIV testing of processed semen is difficult to achieve in most laboratories. Therefore, cryopreservation of spermatozoa is necessary in order to allow viral validation. This has major consequences for ART management in HIV-serodifferent couples. Indeed intrauterine insemination with thawed spermatozoa might not be highly successful and the use of in vitro fertilisation or ICSI may often be necessary. In order to simplify the management of HIV-serodifferent couples and because HIV detection in processed semen was negative, some groups are proposing the use of selected spermatozoa without prior viral validation. However, the three series on which this protocol is based were relatively small and the techniques used in these studies for HIV detection in selected spermatozoa were not as sensitive as the one used in our study. Our findings indicate that sensitive virological testing can detect low residual HIV RNA in selected spermatozoa. Although, the detection of low HIV RNA levels in selected spermatozoa may not reflect the presence of infectious virus, we believe that HIV-positive selected spermatozoa fractions should not be used for ART. HIV detection in selected spermatozoa was more likely to be positive for men with high HIV replication level in blood or seminal plasma. Therefore, we believe that if the use of processed semen without prior viral validation could be discussed as a possibility for men with well-controlled blood and/or seminal plasma viral loads, it should be prohibited for men with partially or poorly controlled HIV infection. In conclusion, collection of extensive virology data from larger series is still necessary to design a consensual algorithm for the management of HIV-serodifferent couples wishing to have children.

Acknowledgements

We thank the physicians from the Departments of Internal Medicine and Gynaecology and Obstetric of the Cochin Hospital for their contribution in enrolling the patients and Isabelle Da Silva and Nathalie Matignon for their technical support. We also gratefully acknowledge the contributions of all patients.

Sponsorship: This study was supported by the Agence Nationale de Recherche contre le SIDA (ANRS).