To determine what length of HAART administration during pregnancy was needed to achieve undetectable VL, we analyzed the kinetics of the virologic response in women who started HAART during pregnancy. The time to undetectable VL did not differ from that previously reported in HIV-infected nonpregnant adults [17
], with a majority of pregnant women achieving VL < 50
copies/mL after 12 to 24 weeks of therapy. The corollary of this observation is that pregnant women who start HAART during or before the 2nd trimester are likely to achieve VL < 50
copies/mL at delivery. In contrast, women starting HAART less than 4 weeks before the anticipated date of delivery are likely candidates to scheduled Cesarean sections.
In our patient population, the rate of Cesarean section was similar to that reported for other areas in the US and lower than that reported in Europe, where the use of scheduled Cesarean section is twice as common as in the US [1
]. However, MTCT in this study was lower than that documented in contemporaneous studies in the US [4
] and similar to that reported in Europe [1
]. These differences may be explained by the fact that 94% of HIV-infected pregnant women in this study received HAART for PMTCT, which is a higher proportion than the rate of 50% to 70% documented in other studies [2
]. Taken together, these observations suggest that high uptake of HAART may decrease the need of scheduled Cesarean sections for PMTCT [2
] without compromising protection of the neonate against HIV infection.
The analysis of the kinetics of CD4+ T-cells in pregnancy revealed a delay in the response to HAART compared with the kinetics of VL. In pregnancy, the most significant increases in CD4+ T-cell numbers occurred in the last 6 weeks of pregnancy and first 6 weeks of postpartum. This pattern was observed in women who started HAART before or during pregnancy and was independent of continuing HAART post-partum. This pattern is most likely due to physiologic changes of pregnancy, including the increased volume of distribution of white blood cells and/or hormonal changes depressing CD4+ T-cells. The implication of this observation is that a lack of increase of CD4+ T-cells in pregnant women may not indicate virologic failure.
An important question in the care of HIV-infected pregnant women is whether there is an ARV regimen more likely than others to rapidly decrease the VL in women diagnosed with HIV late in pregnancy. We did not find any significant differences in the kfinetics of VL as a function of the PI used in HAART. Previous studies showed that nonpregnant HIV-infected adults on NFV-containing regimens were less likely to achieve or maintain viral suppression compared with individuals receiving LPVr-containing regimens [20
]. However, during the limited duration of therapy for PMTCT, this difference may not be relevant. These data suggest that although HAART is critical in order to achieve a satisfactory virologic response during pregnancy, the choice of PI has secondary importance with respect to virologic response. Patel et al. showed that NVP-based regimens are associated with faster decay of VL compared with NFV-based regimens [10
]. In this study, the number of women who intitiated NVP during pregnancy was too low for a formal analysis. Further studies using NNRTIs that do not carry the same high risk of serious adverse events as NVP are warranted to determine whether an NNRTI-based regimen may be advantageous when therapy is initiated late in pregnancy.
With respect to plasma concentrations achieved by various PIs during pregnancy, we found that LPVr levels were below target in 20% to 25% of pregnant women on adult recommended doses across all trimesters, whereas for another PI, 35% to 80% of the levels were below target. Others have shown that up to 72% of women on LPVr do not reach target concentrations of LPV during the 3rd trimester [11
]. Acosta et al. found SAQ plasma levels uniformly above target [22
] during pregnancy, whereas in our study 50% of subjects had levels below target. Our findings on NFV plasma concentrations are in accordance with those of Nellen et al. who also found that 51% of pregnant women did not reach target concentrations [23
]. Differences in drug concentrations across studies may be accounted by technical differences, effects of race and ethnicity on drug metabolism, and compliance with therapy. A novel contribution of our study is to reveal that plasma PI concentrations are often below target not only in the last trimester of pregnancy but also earlier in pregnancy. This suggests that other factors, in addition to the increased volume of distribution of the drugs, may contribute to this effect. The practical implication of our findings is to underscore the importance of performing TDM or other pharmacokinetic studies during the 1st and 2nd trimesters of pregnancy and also after each dose modification, since 30% of the pregnant women on increased PI doses had plasma drug levels below target.
The use of TDM did not significantly affect the virologic response to therapy in this study. Some studies in nonpregnant HIV-infected adults showed improved virologic response to PI-containing regimens in subjects whose doses were adjusted based on TDM results [24
], but one study failed to show any benefit [27
]. This last study had a period of followup of only 12 weeks. Likewise, it may be difficult to assess the benefit of TDM in pregnancy due to the limited duration of therapy. Nevertheless, it is reasonable to assume that it would be undesirable to administer drugs at subtherapeutic levels to control a virus that has the ability to quickly develop drug resistance mutations [28
], particularly if ARV continues to be administered after pregnancy. Further studies are warranted to determine the effect of TDM on the development of resistance to ARV.
Among multiple factors that could potentially affect virologic response to therapy, suppression of viral replication at delivery was positively associated with adherence, confirming previous observations [29
]. We did not find significant differences in adherence between patients receiving once daily versus twice daily regimens or PI-containing versus sparing regimens, which is a novel finding, perhaps specific of pregnancy and has important implications for the clinical practice. Overall, ≥95% adherence was recorded in 82% of the women in this study, which is consistent with other reports [30
Utilization of psychosocial support services was significantly associated with virologic response to therapy during pregnancy, which makes intuitive sense, but has not been previously demonstrated in a formal analysis. Although the association between utilization of psychosocial services and virologic success may indicate that women who are committed to medical treatment for PMTCT are also likely to seek resources to improve other aspects of their lives, it may also indicate that support services facilitate adherence to therapy [30
In summary, undetectable VL at delivery was associated with lack of vertical transmission of HIV in this group of women who predominantly delivered by vaginal route. Undetectable VL was achieved as soon as 4 weeks after initiation of HAART in 50% of the patients and at 12 weeks in the vast majority. Achieving undetectable VL at delivery did not depend on the specific PIs or NNRTIs included in the treatment regimen. Adherence, which was important to reach undetectable VL at delivery, did not vary with the drugs included in the treatment regimen, either. The pattern that emerges is that critical factors for PMTCT without high utilization of Cesarean sections are initiating therapy ≥12 weeks prior to delivery with a regimen that is well tolerated and conducive to ≥95% adherence.