In this study, 8% of pregnancies were complicated by grade ≥3 AE probably or possibly due to ARV which is similar to previous reports [27
]. SAE probably or possibly related to the use of ARV included 3 hematologic, 4 hepatic, and 2
OC. One woman had multiple SAE during 2 pregnancies. The incidence of ARV-associated SAE and the rate of drug substitutions did not significantly differ across classes of drugs, suggesting that they were equally safe and well tolerated during pregnancy.
Among NRTI, AZT was the most poorly tolerated drug. In the first AZT trial for PMTCT [30
], the incidence of AE in mothers receiving AZT monotherapy was similar to that in placebo recipients. However, HIV replication was poorly controlled in that study and may have contributed to the overall incidence of AE. Other studies using combination therapy during pregnancy showed frequent hematologic toxicities in mothers and children who received AZT [27
]. In our study, although mothers who developed hematologic SAE ascribed to AZT were receiving combination ART at the time of the event, laboratory values improved after AZT substitution, suggesting that AZT was responsible for the AE. Despite its marginal safety and tolerability during pregnancy, AZT is the main drug recommended by USPHS and WHO for PMTCT. This is partly due to the reluctance to substitute a drug with proven efficacy. However, CART has higher or equal efficacy for PMTCT than AZT alone, making it possible to substitute AZT with other drugs to avoid undesirable side effects.
Drugs that could potentially substitute AZT in combination ART for PMTCT are TNV, ABC, and D4T, all of which synergize with 3TC. The Antiretroviral Pregnancy Drug Registry includes data on ≥628 pregnancies for each of these drugs with a rate of congenital birth defects similar to that of the general population [6
]. Although these drugs appear nonteratogenic, they face other limitations. TNV interferes with bone formation in experimental animals [13
]. In small numbers of reports in humans, results were variable [34
], and, therefore, providers tend to avoid TNV in pregnancy. The use of ABC is limited by its potential allergic reactions. This risk can be mitigated by HLA B5701 detection, which defines the likelihood of ABC hypersensitivity [36
]. In pregnancy, drug changes have to be quickly implemented, which may not be compatible with the delay required for HLA typing. Finally, D4T has been associated with peripheral neuropathy, lactic acidosis and other metabolic abnormalities including lipodystrophy [37
]. Nevertheless, this drug continues to be widely used in resource-limited countries. In our experience, D4T administered for a limited period of time during pregnancy was well tolerated. The average duration of treatment with D4T and AZT were similar in this study, but patients did not have to discontinue D4T during pregnancy. These findings are consistent with other studies that showed a lower rate of substitution of D4T compared to AZT in nonpregnant adults [39
]. Furthermore, D4T crosses the placenta and achieves sufficient levels in the fetus for pre-exposure prophylaxis. Although based on a limited number of observations (N
= 23), our data suggest that D4T may be a viable alternative to AZT during pregnancy.
NNRTI are uncommonly used during pregnancy other than single-dose NVP at delivery. EFV is contraindicated in the first trimester due its potential teratogenicity [41
]. EFV has recently been included in the WHO recommendations for combination ART after ≥14 weeks of gestations, but its use is still limited. Delavirdine and etravirine have been insufficiently studied during pregnancy. NVP, which was widely used for PMTCt in the late 1990s, is currently contraindicated in pregnant women with ≥250
L due to potential hepatic and cutaneous toxicity. In this study, NVP was used in 19 women with a median first visit CD4 of 419 (IQR
205–588). Three (16%) required NVP substitution due to mild or moderate AE.
NFV, the PI most commonly used in this study was poorly tolerated in 6 of 64 women (9%). The second most commonly used PI was LPV/RTV, which was well tolerated by 26 of 28 women. Three women developed 4 episodes of grade ≥3 hepatic AE, 3 of which were associated with PI (NFV, SAQ/RTV and indinavir). In addition, two cases of OC were diagnosed in women receiving PI. The relationship between OC and ARV is not clear. The ARV Registry includes very few episodes of OC, but the registry, which was designed for birth defects, does not systematically collect other AE, which may underestimate their incidence, including OC. OC resulting from a pregnancy-specific accumulation of bile acids is associated with fetal demise [42
]. The risk of OC is increased by chronic hepatitis C infection [43
], which was present in one of our study women, and by other chronic liver conditions. Otherwise, OC is quite uncommon in Europeans, with an incidence of 0.1 to 2%, but quite common in Chile (9 to 16%), possibly related to the genetic background of the population [44
]. PI and other ARV have hepatotoxic potential [11
] that may contribute to OC. This hypothesis deserves to be further studied.
In conclusion, the safety and tolerability of CART in pregnancy did not differ by class of ARV, but there were differences among individual drugs. Drugs with the poorest safety and tolerability were AZT, NVP, and NFV. Our findings support the need to devise new CART regimens for PMTCT that will avoid the use of poorly tolerated drugs.