Using Relative Infant Dose to Assess Drug Exposure in Breastfed Infants

The use of relative infant dose (RID) as a measure of drug exposure in breastfed infants has gained popularity in recent years. This trend is largely due to the accessibility of RID data in the literature, as its calculation typically requires only breast milk sampling, thereby eliminating the need for venous blood sampling from the mother or the infant. Additionally, the existence of cut-off values distinguishing ‘safe’ maternal drug use from use that might not be safe for the infant makes it relatively easy to interpret. If these assumptions really hold true, RID could be viewed as a most valuable tool for guiding drug use in lactating mothers in clinical practice.

However, as Flis et al. highlight in this issue of BCPT [1], there are important caveats. In their article, they demonstrate that RID values for monoclonal antibodies are often miscalculated. They also argue that, due to some specific properties of this drug class, RID may not be a suitable metric. This raises the question: Could this be a more general challenge?

For drugs administered daily, the process of RID calculation is relatively straightforward using the formula presented by Fris et al. [1]. Nonetheless, pitfalls remain. Ideally, milk samples should be collected throughout the dosing interval at sufficient frequency to estimate the area under the concentration–time curve (AUC) in milk and thereby also the average drug concentration in milk. Relying on a single or a few samples near trough levels will underestimate RID, whereas sampling near peak concentrations (C_max) may lead to overestimation. For drugs with extended dosing intervals (e.g., weekly or monthly), the situation becomes more complex [1], but still, the core principle remains, that is, using the average milk concentration over the dose interval derived from AUC data. If sampling is sparse but involves many subjects, population pharmacokinetic (PopPK) modelling is a valuable alternative to traditional AUC calculations [2]. However, when frequent sampling is available, the two methods yield close to identical RID values, as illustrated in our studies on cetirizine [3, 4].

For drugs with active metabolites, these should be included in RID calculations, that is, the pharmacologically active infant dose comprises the sum of the parent compound and its active metabolite(s). Prodrugs present an even more complex challenge from a theoretical point of view, although inactive parent substances are most often not included in the calculations.

Milk sampling should take place during steady-state conditions, which means that at least five elimination half-lives should pass after initiating or adjusting treatment before sampling. Sampling prior to steady state will underestimate RID. This is particularly relevant for drugs with long half-lives, such as monoclonal antibodies, where this condition may more often not be met [1].

Given that published RID values may not always be based on optimal milk sampling methods and calculation principles, it would be beneficial if textbooks and databases provided more information on the quality of the underlying data. This would help clinicians assess whether RID values are reliable or should be viewed with scepticism.

In 1988, a WHO working group suggested that drugs with an RID above 10% will ‘usually be unacceptable’ for breastfeeding mothers [5]. This limit was later reaffirmed based on RID data from 205 drugs [6], with additional evidence suggesting that drugs with RIDs above 25% pose an even greater risk. More recently, it has been proposed, for example, by a Danish working group on psychotropic drugs, to use a general cut-off of 5% for breastfeeding acceptability [7]. Further refinements have also been suggested, classifying infant exposure as minimal when the relative dose is below 2%, small when the relative dose is 2%–5%, moderate when the relative dose is 5%–10% and high when the relative dose is above 10% [8].

These thresholds, however, are not universally applicable. For highly toxic substances, such as cytotoxic agents and radiopharmaceuticals, breastfeeding should generally be avoided regardless of RID. Additionally, hypersensitivity reactions in the infant may occur independently of the RID [9].

There are also scenarios where infant exposure would be higher than expected based on what the standard RID risk classification would imply. For instance, if maternal drug clearance is unusually low due to genetic factors or drug interactions, the individual RID would be higher than the average reported in the literature, and the infant would be exposed to correspondingly higher doses [2]. When a large group of mothers have been included in the studies underlying available RID data, it could be expected that also some mothers with reduced elimination rates have been included. In such cases, using the highest individual RID from study data as a ‘worst-case’ estimate can help assess safety. If this value remains (well) below 10%, the safety margin is likely sufficient.

Moreover, RID thresholds assume standard maternal dosing. If the mother is treated with unusually high doses, the absolute infant exposure and also the associated risk will be higher than the RID suggests [10].

Finally, if the infant has impaired drug clearance (e.g., due to prematurity or concomitant illness), even drugs with modest RID values may cause adverse effects, particularly during prolonged maternal drug use. A review of published cases of adverse drug reactions (ADRs) in breastfed infants found that low infant clearance was the most likely cause in cases where the RID of the implicated drug was below 10% [9].

Conversely, RID values and thresholds assume that the infant is fully breastfed. If the infant is partially formula-fed, the actual drug exposure will be lower than the RID indicates, thereby increasing the safety margin [8, 11].

The safety margin will also increase with increasing infant age. During the first 3 months or so after birth, hepatic metabolism and renal function gradually mature, reducing the risk of ADRs compared to the neonatal period. A review found that approximately two-thirds of all reported ADRs in breastfed infants occurred within the first month and nearly 80% within the first 2 months post-partum [10]. In infants aged above 4–6 months, particularly those who are no longer not fully breastfed, any infant risk could be expected to be very low, even for drugs with high RID values.

RID thresholds are not relevant for drugs that are not absorbed systemically or are completely inactivated in the infant's gastrointestinal tract. However, it is often unclear whether certain types of drugs, such as those being protein-based, are inactivated in the infant gut to the same extent as in adults and also whether absorption from the infant gut may take place even when this does not occur in adults [1]. In such cases, guidance should rather be based on more complex clinical studies. For example, although infant risk cannot be completely excluded for monoclonal antibodies even though they are protein-based [1], the protein insulin is considered safe during lactation [12] and oral insulin might even offer beneficial effects for the infant [13].

Given the limitations outlined above, a universal, clear-cut and practically feasible ‘safe’ RID limit does not exist. Nevertheless, the 10% threshold can serve as a general rule of thumb for most drugs, although the 5% limit provides a higher safety margin and might therefore be preferable. For drugs with reported RIDs below 1%–2%, the safety margin is typically high, even in scenarios involving reduced maternal clearance, high maternal doses or impaired infant clearance.

In conclusion, there are important limitations and pitfalls to consider both related to the calculation and the interpretation of RID values. Nevertheless, being aware of these caveats and with some specific exceptions, its widespread availability, especially compared to metrics requiring maternal or infant blood sampling, makes RID a highly valuable tool in clinical decision-making for drug treatment in breastfeeding mothers.

The author declares no conflicts of interest.