Discussion
To our knowledge, this report is the first description of the exposure of bedaquiline in pregnant women. We found bedaquiline and M2 exposure in pregnant women to be approximately 50% lower than expected in non-pregnant patients (14). Although we were underpowered, we found no significant difference between ante- and postpartum exposures.
There are several possible reasons for the low bedaquiline exposures we observed in the third trimester. First, increased metabolism of bedaquiline is a possible explanation - pregnancy is known to induce CYP3A4, which is the major route of bedaquiline metabolism (21). The increase in CYP3A4 expression would lead to higher clearance and lower bioavailability of bedaquiline, since it is present in both entero- and hepatocytes. Second, pregnancy reduces plasma albumin concentrations, to which bedaquiline is highly bound (22). The unbound fraction of bedaquiline may therefore increase, subsequently increasing its clearance and tissue distribution. In such a scenario, the total (bound+unbound) concentrations of bedaquiline in plasma would decrease, but this effect could be counter balanced by the large unbound fraction, thus maintaining relatively unchanged unbound levels. However, exploration of free bedaquiline exposure is required before a recommendation for a dose adjustment can be made. Finally, changes in body size (and possibly composition) may have affected bedaquiline disposition, but it is unlikely that the increased weight in pregnancy affected the exposure of bedaquiline as we used allometric scaling to account for this in the model, and changes in body size alone are therefore unlikely to explain the decreased bedaquiline concentrations we observed.
Similarly, we observed lower-than-expected bedaquiline levels at the postpartum visit. While it is generally accepted that pharmacokinetic sampling approximately six weeks postpartum is a reasonable time-point to allow the physiologic changes related to pregnancy to subside (23), there are some limitations in using this timeline as a control when exploring the effect of pregnancy on drugs with a long half-life such as bedaquiline. Given that the terminal half-life of bedaquiline is more than five months, (4) any change in pharmacokinetic parameters may only become apparent on drug exposure after a considerable amount of time, possibly months. Thus, even if most of the pregnancy effects (if any) had reversed in the first weeks after delivery, there may not have been sufficient time for the exposure of bedaquiline to reach a new equilibrium before the scheduled postpartum pharmacokinetic visit. An alternative explanation is that adherence could have decreased in the postpartum period; a systematic review reported poor postpartum adherence in patients on ART (24). Sub-therapeutic bedaquiline exposures could affect clinical outcomes and increase the risk of selecting for drug resistance.
We observed concerningly high concentrations of bedaquiline in the human milk samples we analysed, markedly higher than the maternal bedaquiline plasma concentrations, in keeping with the findings of an animal study. (10) The breastfeeding infant had a plasma bedaquiline concentration similar to maternal plasma (Figure 2), which could have implications for infant safety. In a previous animal study, rat pups who were breastfed to mothers treated with bedaquiline were reported to have low body weight. (10) In contrast, therapeutic concentrations of bedaquiline in infants (possible with long half-life drugs, which accumulate slowly, such as bedaquiline) could potentially be protective in infants exposed to RR-TB, obviating the need for TB preventive therapy. The three infants who were not breastfed had sub-therapeutic bedaquiline concentrations, likely from transplacental exposure, which could select for drug resistance should the infants develop RR-TB.
The main factors determining the transfer of a drug into human milk are its physicochemical characteristics (such as lipid solubility and degree of ionisation at different pH conditions) and its plasma pharmacokinetics (25). Fat-soluble drugs like bedaquiline cross lipid-protein cell membranes easily, hence transferring readily into human milk (25). The ease, with which drug molecules cross cellular membranes, depends on the drug’s degree of ionisation, which may vary in different pH conditions. Weak bases like bedaquiline (pKa = 8.9) (26) tend to be slightly less ionised in plasma compared to milk. This means that unionised plasma bedaquiline will transfer into human milk, were it is more likely to be ionised, favouring milk accumulation of the drug (27). Transfer of drugs into human milk may also be greater in drugs with a low affinity for maternal plasma proteins, but bedaquiline is highly protein-bound (>99.9%). (4) An additional factor is molecular weight, as drugs with low weight (<200 Da) reach human milk more easily, but the molecular weight of bedaquiline is 555.504 Da (28). Drugs, which have a long plasma half-life and therefore accumulate, such as bedaquiline, are prone to transfer into human milk compared with molecules which are cleared rapidly. The high concentration of bedaquiline in human milk suggests that the mammary glands could be a clearing site for bedaquiline. Excretion could be significant, since, on average, a baby consumes about 0.15 L/kg/day of human milk (18). Moreover, bedaquiline metabolism in breast tissue cannot be excluded, as there are contradictory reports on the expression of CYP3A4 in human breast tissue. (29–31)
Our study has several limitations. First, we did not measure unbound bedaquiline concentrations or albumin levels, so we are unable to conclusively determine if the reasons for the low concentrations observed are related to protein binding. Second, there was a high rate of participant loss to follow up, which limited our sample size, as many participants were unable for logistical reasons, to complete the postpartum PK sampling day. Third, PK sampling was not always performed on a day when bedaquiline was scheduled to be administered (dosing is three times a week). Although this was accounted for in our modelling, considering we did not use an adherence measure, the date and time of the last bedaquiline dose was obtained via participant self-report, which could be unreliable.
We report low exposures of bedaquiline in this series of pregnant women treated for RR-TB. Future studies should analyse bound and unbound bedaquiline concentrations with adherence measures to better understand the effect of pregnancy on bedaquiline exposure,  and assess whether a different dosing recommendation for bedaquiline in pregnancy is warranted. Bedaquiline appeared to significantly accumulate into human milk, which could be an exposure risk for breastfeeding babies and should therefore be investigated.
Acknowledgments
The authors would like to acknowledge the participants who volunteered for the study, and Sindisiwe Hlangu for her assistance with study implementation.
Conflicts of interest
The authors declare no conflicts of interest
Funding
The University of Cape Town Clinical PK Laboratory is supported in part via the Adult Clinical Trial Group (ACTG), by the National Institute of Allergy and Infectious Diseases (NIAID) of the National Institutes of Health under award numbers UM1 AI068634, UM1 AI068636, and UM1 AI106701; as well as the Infant Maternal Pediatric Adolescent AIDS Clinical Trials Group (IMPAACT), funding provided by National Institute of Allergy and Infectious Diseases (U01 AI068632), The Eunice Kennedy Shriver National Institute of Child Health and Human Development, and National Institute of Mental Health grant AI068632. The content is solely the responsibility of the authors and does not necessarily represent the official views of the sponsors. Catriona Waitt is supported by a Wellcome Clinical Research Career Development Fellowship 222075/Z/20/Z.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.