Purpose of postpartum or lactation PK studies

In the f​​irst two weeks’ to two months’ postpartum, physiological changes from pregnancy can affect drug exposure. However, after this period, the pharmacokinetics in breastfeeding mothers should be similar to those of non-pregnant adult women. The main concern is the potential dosage that the infant may receive through breastmilk. 

Lactation PK Study Design

Breastmilk exposure to medication is influenced by several factors: 

1. The drug's physicochemical properties 
2. The mother’s pharmacokinetics 
3. The infant’s age and feeding patterns

Infant PK is further influenced by their own characteristics. Notably, the infant may have differing pharmacogenomics resulting in faster or slower drug metabolism, as compared with the mother. This is best outlined in a review of viral transmission through breastmilk (27)​​​​​​. These factors are summarized in figure 5.1.

For a drug to enter breastmilk, it must first achieve a certain concentration in the mother’s blood and then transfer into the milk through active transport or passive diffusion. The infant receives a "dose" during breastfeeding, which then needs to be absorbed from the gut into the systemic circulation to reach maximum concentration before metabolism and elimination. Typically, maximum concentrations in breastmilk are reached later than in plasma, and the elimination of the drug from breastmilk is often slower. This five-minute presentation explains this in more detail: 

Figure 5.2 demonstrates the milk/plasma concentrations over time. It is therefore critical to document when blood and milk samples are collected relative to the time the dose was taken.  

Lactation PK modelling

Physiologically based pharmacokinetic modelling (PBPK) is often used to ensure pharmacokinetic parameters can be estimated with an appropriate degree of precision. This ‘bottom up’ approach uses all prior data about a drug, including physicochemical properties, animal data, action on transporters, enzyme induction, etc. to predict PK parameter precision. In lactation, the following factors are most relevant: 

1. molecule size
2. protein building
3. lipid solubility
4. transporters are also likely relevant, though more research is needed (28).

​​​Algorithms to predict milk/plasma ratios from the physicochemical properties of a drug are available and can be integrated within a PBPK model to simulate plasma concentration-time profiles of drugs in mothers and breastfeeding infants (29). Factors affecting the milk/plasma ratio include the pH and fat content of the milk; these can be assessed in simulations.

PBPK models consider the time-variant complex interplay between physiological parameters and drug-related characteristics in nursing mothers and infants and can be used to simulate scenarios where clinical data are lacking or sparse that may lead to changes in FDA labelling. For example, t​​​​his approach was used to complement mother-only data captured in a clinical lactation study of moxidectin (used for treatment of onchocerciasis or river blindness), a highly lipophilic drug with extensive tissue distribution, complex disposition, and a long half-life of >20 days (30). Although an FDA-standard lactation pharmacokinetic study in 12 healthy breastfeeding women showed maternal plasma and milk exposure over time, concerning rat pup data initially led the FDA to recommend avoiding breastfeeding for six months, a restriction considered impractical and potentially unsafe in endemic settings. Using PBPK modelling based on these clinical data, investigators demonstrated that infant exposure via breastmilk was far lower than in rat pups, leading the FDA to revise its guidance to a seven-day breastfeeding interruption, now reflected in the moxidectin product label. In TB, this approach could be useful for bedaquiline, a drug with a long terminal half-life (approximately 5.5 months) and sustained tissue distribution that undergoes hepatic CYP3A4-mediated metabolism. ​​Various PBPK models available in the literature (31) will allow simulation of infant exposures of bedaquiline during breastfeeding, accounting for ​​time variant changes in CYP3A4 expression (32, 33), protein binding and other physiological parameters.

While this approach can be very powerful, there is a need to harmonize reporting and approaches in lactation PBPK modelling to ensure consistency in how PBPK lactation models are developed and reported​​​​​​ (34).

Population PK modelling can also be applied in breastmilk PK. For example, paired maternal plasma and breastmilk ​​levofloxacin concentrations were analysed using nonlinear mixed-effects modelling, demonstrating rapid plasma-breastmilk equilibration and a predictable milk-to-plasma relationship that enabled estimation of infant exposure (25). The analyses showed that levofloxacin readily distributes into breastmilk, with milk concentrations modestly higher than plasma, but that infant exposure via breastfeeding was only a small fraction of the maternal dose, highlighting the value of PopPK modelling for informing risk–benefit assessments for the infant during maternal TB treatment.

Lactation PK study sample size

Often, there is little or no prior data on drug concentrations in breastmilk, making it difficult to estimate a sample size for these studies. An interim analys​i​s to validate assumptions is often needed to ensure the study design remains appropriate.

A stochastic simulation exercise takes everything already known about drug exposure and runs experimental ​simulations ​to determine​ PK​ parameter estimates. In general, a sample size of between 15 and 30 individuals for an intensive pharmacokinetic study should be sufficient to estimate the lactation transfer of a drug. In situations where a drug is rarely used in breastfeeding, even a small number of well-characterized profiles from mother–infant dyads (and sometimes even a single well-written case study) can be of immense value.

Interpretation of breastmilk an infant PK

To assess an infant's drug intake during breastfeeding, a comparative measure is important. The preferred comparison is the therapeutic dose for neonates or infants, though this data is often unavailable. Instead, the relative infant dose (RID) is used: 

When adapting a plasma assay for breastmilk, partial validation is essential. Standards and quality controls should be included in the same donor breastmilk used for the test samples. It is vital that these donors are not on any other medications. One example includes the validation and application of a quantitative liquid chromatography tandem mass spectrometry assay for the analysis of rifapentine in human milk developed for IMPAACT 2001 (35)​​. The assay was developed, as per above, and then applied to clinical trial samples to evaluate rifapentine concentrations in breastmilk.

The FDA provides guidance on bioanalytical method development for industry.

Examples of Breastmilk Validation Assays and PK studies from TB clinical trials:

• Rifapentine
• Levofloxacin
• Bedaquiline
• Delamanid

Several registries, databases, and biobanks exist that can be leveraged by investigators looking to harvest data on medicines used in pregnant and lactating women.

• FDA Adverse Event Reporting System – post-marketing surveillance data including exposure and adverse drug events during pregnancy and lactation. 
• All of Us Research Program – NIH initiative that integrates survey, EHR, and biospecimen data, including for pregnant and lactating women. 
• NICHD Data and Specimen Hub – repository of deidentified biospecimens and clinical data
• PKRxiv – open-access platform hosted by University of Liverpool for sharing individual-level pharmacokinetic and pregnancy-related data.
• Pregnancy Risk Assessment Monitoring System – survey of maternal experiences and behaviour in the peripartum period.