Corresponding author and reprint requests: Jean-Pierre Bouvet, Unité d'Immunocytochimie, Institut Pasteur, 25 rue du Dr Roux F-75724 Paris 15, France Tel: +33 1 45 68 82 56 Fax: +33 1 45 68 86 39
Objective: To explain the high frequency of heterosexual transmission of AIDS in subSaharan countries, by investigating post-intercourse semen retention in the vagina for local pH neutralization as a possible interference factor with acidic inactivation of HIV virions.
Method: Two semen markers, prostate-specific antigen and prostatic acid phosphatase, were measured and compared with the pH values in the vaginal washes of 69 women from the Central African Republic. The capacity of semen to raise the vaginal pH was also investigated in vitro.
Results: Of 61 non-menstrual specimens, 74% contained at least one semen marker. The specimens with high levels of markers (group I) displayed an almost neutral pH (median 6.1), at variance with the semen-free group II (median: 3.7, P<0.003), and with group III (median: 4.0) corresponding to low or past semen retention. The in vitro study confirmed the high neutralization capacity of semen.
Conclusions: It is expected that post-intercourse neutralization of pH will both favor male-to-female transmission and prevent the acidity-associated loss of infectivity of the female-derived virions, thus allowing female-to-male transmission during further sexual contact.
Vaginal intercourse is the major mode of HIV transmission in developing countries, suggesting the presence of additional cofactors of infectivity. The acidity of the cervicovaginal fluid [1,2] normally provides protection against HIV, since both cell-free and cell-associated virions are inactivated by low pH values, as demonstrated by in vitro experiments [3–5]. In contrast, seminal plasma is slightly alkaline, leading to pH neutrality of the vagina during and after intercourse . Moreover, addition of human semen , or of phosphate-buffered saline , to the inoculate favors experimental intravaginal transmission of the simian virus to non-human primates. In a previous study we reported some cases of semen-borne antibodies to HIV in the cervicovaginal fluid, indicating the persistence of semen components in the vaginal tract after sexual intercourse . This finding led us to investigate, in an at-risk population, the persistence of semen as a possible cofactor in female-to-male transmission of HIV via an increase in the local pH. We selected a cohort of women from subSaharan Africa, where heterosexual transmission of HIV is very frequent.
Vaginal secretions were obtained from all (n=69) nonpregnant African women attending during a period of 3 days the Centre National de Référence pour les Maladies Sexuellement Transmissibles of Bangui (Central African Republic) and presenting no apparent infection by speculum examination. Ages ranged from 17 to 44 years (median value, 25; quartiles, 20 and 29). For cultural reasons, reliable information on the time that had elapsed since the last intercourse were not available in most cases. Serum antibodies to HIV were screened by ELISA (Diagnostics Pasteur, Marnes-la-Coquette, France) and confirmed by Western blot (New Lav blot I, DP).
Non-menstrual (n=61) and blood-containing (n=8) vaginal fluids were collected by washing with 3 mL of distilled water, corresponding to an approximate 10-fold dilution of the native secretion . A microscopic analysis was carried out for the detection of Trichomonas vaginalis, followed by a Gram stain and by cultures on blood agar and Sabouraud's agar. After centrifugation at 1000g for 5 min, to remove cells and mucus, the specimens were immediately frozen and stored in hermetic vials at −20°C until use. Cell removal and these storage conditions prevented cell lysis and bacterial growth, which can induce protein degradation and pH changes. Two current prostatic semen markers [10,11], the human prostate-specific antigen (PSA) and prostatic acid phosphatase (PAP), were quantified in the vaginal samples by using ELISA kits with monoclonal antibodies (IMxTM system, Abbott Labs, Abbott Park, IL). Although other markers could be used , these two were chosen because commercial kits were easily available. The use of two markers increased the specificity of the detection. Indeed, antigenic properties of PSA are fragile, and PAP can be present at a very low level in the female urine and vaginal fluid, as discussed by Sensabaugh . Reference curves were drawn from dilutions of positive controls measured simultaneously, according to the instructions of the manufacturer. Preliminary investigations with diluted seminal plasma samples, acidified from pH 6 to 3.6, showed no influence of acidity on the level of the markers. However, a prolonged incubation of 24 h, at pH ≦4.5 and 37°C, significantly decreased the PSA value, whereas that of PAP remained unchanged. For this reason, discrepant results in favor of PAP were considered to be due to a prolonged presence of semen in the vagina, as indicated below. Vaginal pH measurements were carried out at room temperature (22°C) with a pH meter (Heito, Paris, France) calibrated with pH 4 and pH 8 reference solutions (Heito). This method is much more precise and reliable than the use of pH paper. Comparative measurement of neat and 10-fold diluted samples confirmed that this dilution with distilled water has negligible effects on the pH analysis of vaginal fluids. The capacity of the seminal plasma to neutralize the vaginal pH was investigated by addition of increasing volumes of a pool of 10-fold diluted seminal plasma samples, in distilled water, to a constant volume of a pool of semen-free vaginal washes. The genital secretion samples (10 in each group) were selected according to their pH value within the two median quartiles. Mixing of the pools and pH measurements were carried out immediately.
Statistical comparison of semen-containing and semen-free samples was carried out by the non-parametric U test of Mann and Whitney, and the probability was estimated by interpolation of a unidirectional table.
Various concentrations of PSA and PAP were observed in the 69 vaginal specimens. In the non-menstrual specimens, 74% contained at least one marker, and three different groups were delineated according to the levels of the semen components (Table 1). In group I (high semen retention), the vaginal levels of both semen markers were above 50 ng/mL (generally >100 ng/mL), whereas these levels were below 11 ng/mL in group II (semen-free). In group III (low semen retention), the levels were between 11 and 50 ng/mL and/or were discrepant, PAP being much higher than PSA in most cases. These breakpoints were chosen according to the discontinuity in the values observed. Blood-containing specimens were assigned to a separate group, group IV.
Table 1. Comparison of pH and semen markers in the vaginal washes of subjects with and without semen retention
Semen markers (ng/mL)a
aMedian value (quartiles).
cProstatic acid phosphatase.
(high semen retention)
(4.6 and 6.7)
(0.2 and 0.8)
(1.9 and 6.8)
(3.7 and 4.2)
(low semen retention)
(0.1 and 1.5)
(46 and >100)
(3.8 and 5.0)
(0.1 and 0.2)
(41 and >100)
(5.7 and 6.1)
Measurements of the vaginal pH showed a clearcut difference (P<0.003) between group I (median value: pH 6.1) and group II (median value: pH 3.7). Nevertheless, the pH range was wide in both cases, and the semen-free group II included some non-acidic values, whereas, conversely, some acidic values were observed in group I (Figure 1). In group III, the pH values were also scattered, but the median value (pH 4.0) was similar to that of the semen-free group. Among the specimens with highly discrepant marker levels (PAP >80 ng/mL and PSA <5 ng/mL), 10 of 14 had pH values ≦4.5. Overall analysis of these three groups showed that the vaginal pH of 33 of 61 subjects (54%) was ≦4.2. This proportion varied from 14/16 (87.5%) in the semen-free group II to only 3/16 (19%) in the semen-containing group I, with an intermediate value of 16/29 (55%) in group III. The pH of the blood-containing group IV was more homogeneous, ranging from 5.3 to 6.4 (median: pH 5.9), irrespective of the levels of the semen markers (Table 1). A comparison with the microbial and serologic data showed that individual pH variations correlated only with the level of semen markers and not with the presence of T. vaginalis (n=5, pH range 4.0–6.8) or of Candida albicans (n=14, pH range 3.5–7.2) or seropositivity for HIV (n=17, pH range 3.6–7.2).
Investigation of the buffering strength of seminal plasma showed that a small amount was sufficient to neutralize vaginal acidity. The linear portion of the pH neutralization curve ranged from 0.1 to 1 volume of the 10-fold diluted seminal pool for 1 volume of the pool of vaginal washes (Figure 2).
Investigating the influence of vaginal retention of semen on the pH of the cervicovaginal fluid in African women with high risk of heterosexual HIV transmission, we have shown a frequent persistence of semen leading to a shift from acidity to neutrality. The semen-associated neutral pH can explain male-to-female transmission, and this persistent pH value after sexual intercourse could favor further female-to-male transmission.
The lack of reliable information on the time elapsed since the last intercourse was associated with the socio-cultural approach to sexual behavior in sub-Saharan countries. This impairment of the control of our study was predictable in this selected population and was in agreement with our hypothesis of educational differences between high- and low-risk populations being a major factor in the unusually frequent heterosexual transmission of HIV in these countries. Nevertheless, this missing information was not of critical importance, since the parameter considered was the level of retention, as measured by prostatic antigens and irrespective of the time of collection. Interference caused by individual variations of the original amount of these antigens in the ejaculate was avoided by comparing groups I and II, defined by very high or low values in the vagina.
The pH rise involves both seminal and vaginal parameters, which can vary according to the respective volume of each fluid and to the duration of the intravaginal semen retention. The cationic charge of semen is primarily carried by spermine—a polyamine having immunosuppressive activities —whereas the anionic charge comes from citric acid; both of these are synthesized by the prostate. The individual buffering capacity of semen must be related to independent variations of the concentrations of these two components. In contrast, the anionic charge of the vaginal fluid is carried by lactic acid produced by the local flora, which maintains low pH values . Our results demonstrate a post-intercourse rise of the vaginal pH—as shown both by the differences between group I and group II (Figure 1), and by the in vitro neutralization curve (Figure 2)—which can protect seminal HIV against acidity. During the following hours, the vaginal pH progressively returns to its normal acidic value, as observed in group III, the duration of this period depending probably on the initial amount of spermine. Prostatic proteins PAP and PSA are progressively eliminated  with the mucus stream, and/or degraded by acidity, as suggested by the discrepant PAP and PSA values in group III. Variations in the local flora could interfere with the pH value in cases of purulent infection but, in the absence of clinically apparent vaginitis, this effect was not critical here. In blood-containing specimens, the neutral pH values were probably related to the high buffering capacity of serum.
A complete and irreversible inactivation of free HIV virions has been demonstrated in vitro with pH values below 5.4 after a 20-min incubation at 37°C, and below 5.7 after a 2 h incubation . Inactivation is similar for cell-associated virions but is partially reversible unless acidity can degrade the cells themselves , a phenomenon usually occurring below a pH of 6 . Interestingly, our proportion of sperm-free subjects (2/16, 12.5%), with a vaginal pH above the upper limit (pH 4.2) defined for an almost complete and irreversible inactivation, is similar to the 12% figure recently reported for non-pregnant HIV-positive women excreting infective virus in their cervicovaginal secretions . Concerning the eight blood-containing fluids, the neutral pH confirms the multifactorial risk of transmission already observed during menses .
The neutral pH observed in most women with high levels of semen markers in their vaginal tract shows how local acidity is actually abolished by semen, favoring the male-to-female transmission of HIV by vaginal intercourse. Reciprocally, the intravaginal retention of semen, in the absence of vaginal toilet, can also prevent the acidity-associated loss of infectivity of the female-derived virions and favor a female-to-male transmission during a further sexual contact. These features may be relevant to the high HIV incidence in the developing countries where the transmission is primarily heterosexual.
We especially thank Dr Sylvio Iscaki and Professor Jacques Pillot (Institut Pasteur, Paris) for helpful discussion and support, the Institut Pasteur of Bangui (République Centrafricaine) for help in collection and storage of the samples, and Dr Frédéric Fieux (Abbott Laboratories, France) for kindly providing the PAP and PSA estimation kits. This study was supported by funds from the Institut Pasteur, the Agence Nationale de Recherche sur le SIDA (Grant No. 94130) and the French Ministère de la Coopération.