Decreased pregnancy rate after in-vitro fertilization in HIV-infected women receiving HAART

A study on in-vitro fertilization (IVF) was conducted among HIV-infected women. In these patients, a reduced pregnancy rate after IVF was observed if the patient's own oocytes were used. However, no significant reduction in the pregnancy rate was found if donated oocytes were used. The CD4 lymphocyte count was independently associated with ovarian resistance to hyperstimulation. Subclinical hypogonadism mediated by immunosuppression may explain these observations, suggesting the need to optimize the immunological status of the patient before considering assisted reproduction treatments.

The introduction of HAART has led to an almost negligible rate of mother-to-child transmission of HIV infection ^[1], and therefore an increasing number of HIV-infected women have decided to become pregnant. Some preliminary studies suggested that HIV-infected women may have a decreased fertility rate ^[2], and a higher frequency of menstruation disturbances associated with low CD4 cell counts ^[3] and upper genital tract infections ^[4]. In addition, severe ovarian dysfunction, such as premature ovarian failure or ovarian resistance to stimulation, has also been described ^[5,6].

Little is known about the outcome of in-vitro fertilization (IVF) in HIV-infected women. The results of a small study suggested low pregnancy rates in HIV-infected women after IVF ^[7], but matched controlled studies are lacking. We aimed to assess whether HIV-infected women have a lower IVF success rate than non-HIV-infected women.

From April 2000 to March 2004, a cohort consisting of 50 IVF cycles in 35 HIV-infected women (group 1), 50 age-matched cycles in 37 non-HIV-infected women with HIV-infected male partners (serodiscordant couples, group 2), and 100 age-matched cycles in 82 non-HIV-infected women with non-HIV-infected male partners (group 3) was prospectively followed. In all cases, IVF treatment was indicated and only fresh cycles were analysed. In our institution HIV-serodiscordant couples were initially treated with intrauterine insemination, and only patients with a primary indication for IVF or after four failed intrauterine insemination attempts were allocated to undergo IVF. A parallel oocyte donation (OD) cohort consisting of 25 OD transfers in 20 HIV-infected women (group 4), 16 non-matched OD transfers in 14 HIV-serodiscordant couples (group 5) and 75 age-matched OD transfers performed in 68 non-HIV-infected couples (group 6) was also prospectively studied. Matching was performed by age (± 0.5 years) and the chronologically closest cycle was chosen. Ovarian stimulation, IVF procedures, the assessment of fertilization, embryo cleavage and establishment of pregnancy were performed according to standard procedures ^[8]. In all HIV-infected patients, the HIV plasma viral load and CD4 cell counts were measured at least within one month before reproductive treatment. Antiretroviral treatment was prescribed at the discretion of the treating physician according to updated Spanish guidelines ^[9].

In the IVF cohort, 43 out of 157 (21.5%) of the cycles were cancelled (Table 1). No differences were observed between cancelled and non-cancelled cycles regarding the baseline and IVF-related variables. There were no cancellations in the OD cohort. Table 1 shows the pregnancy rate among the study groups. HIV-infected women (compared with HIV-serodiscordant couples and controls) had a crude OR of 0.45 [95% confidence interval (CI) 0.23-0.85; Mantel-Haenszel χ², P = 0.012) for clinical pregnancy in all non-cancelled cycles (IVF and OD). A stratified analysis for the type of antiretroviral therapy showed a 0.30 odds ratio (OR) (95% CI 0.11-0.78; Mantel-Haenszel χ², P = 0.01) within the IVF cohort and a 0.68 OR (95% CI 0.27-1.71; Mantel-Haenszel χ², P = 0.42) within the OD cohort, for clinical pregnancy.

Table 1

Standard prognostic IVF-related variables did not differ significantly among the different IVF subgroups. Only follicle-stimulating hormone (FSH) required units (3721 ± 1218 for group 1, 3136 ± 1392 for group 2 and 3743 ± 19349 for group 3, one-way analysis of variance, P = 0.05) was significantly different between the study groups. Adjusting for maternal age, semen status, used ovarian suppression drug and technical transfer difficulty, the OR for the occurrence of clinical pregnancy in HIV women versus controls was 0.37 (95% CI 0.13-0.99). The adjusted OR for HIV-serodiscordant couples versus controls was 1.23 (95% CI 0.55-2.8).

Regression analysis showed that from standard IVF variables only the FSH units required during ovarian stimulation added significant explicative value to the model. For an increase of 100 required FSH units, the OR for clinical pregnancy decreased by 0.97 (95% CI 0.95-0.99).

The logistic regression analysis in the HIV patients on IVF (group 1) showed that among all the HIV-related variables (CD4 cell count before inclusion, CD4 cell nadir, viral load, months from HIV diagnosis, and months on HAART) only the CD4 cell count before inclusion (adjusted OR 0.994; 95% CI 0.991-0.997; P = 0.043) had a significant effect on the occurrence of ovarian resistance.

The results of this study support the finding that HIV-infected women undergoing IVF have an adjusted lower pregnancy rate. In order to evaluate whether the reduction in pregnancy rate was caused by the maternal environment or oocyte exposure, either HIV infection or antiretroviral treatment, a parallel study was conducted among patients who required oocyte donation. The effect of HIV infection on the likelihood of clinical pregnancy remained significant only in those women on IVF but not after OD. As no effect of HIV infection in this group was observed, one can hypothesize that the absence of uterine or implantative differences may explain the adverse outcome. The further exposure to HIV or HAART at the time of implantation does not seem to have an initial deleterious effect.

Among HIV-infected patients treated with IVF, the usual reproductive outcome parameters failed to detect any underlying mechanism that could explain the lower pregnancy rate. It would appear that the oocyte of an HIV-infected women begins the IVF process with some further undetectable handicap.

Ovarian resistance to hyperstimulation may be involved in this effect because a greater number of units of gonadotrophins were needed to stimulate these patients adequately. This resistance may reflect an underlying subclinical (normal menses) and subanalytical (comparable basal FSH values) hypogonadism, and supervoluation may be considered as a functional stress test on the ovary.

The potential basis for this subclinical hypogonadism with altered oocyte quality is not readily evident. HIV should not have a direct impact on the human oocyte because no receptors for HIV have been described on either the cumulus cells or on the surface of the oocyte ^[10]. Neither do our results suggest any effect of viral load on the occurrence of ovarian resistance. A potential hypothesis might be mitochondrial dysfunction as a result of the use of antiretroviral drugs ^[11,12]. In our series, however, neither the type of treatment regimen used nor the total duration of HAART were associated with ovarian resistance.