Free dug concentrations in pregnancy: Bound to measure unbound?

“…the interpretation of total drug concentrations is complicated by the fact that pregnancy decreases albumin binding, which is expected to translate into lower total drug concentrations despite mostly unaffected unbound concentrations (pharmacologically active).”

With this statement Marzolini et al. 1 underline a phenomenon of major importance when interpreting drug exposure during pregnancy. In the case presented, the authors find subtherapeutic total and free (unbound) elvitegravir concentrations during pregnancy. This is an important finding because subtherapeutic exposure of this antiretroviral drug puts women at an increased risk of virological failure and/or development of antiretroviral resistance, as well as an increased risk of perinatal HIV transmission during pregnancy 2, 3. Unexpectedly, the authors also observe an increased protein binding during pregnancy. The statement above, combined with the findings of this case report, highlight that we cannot always rely on total concentrations for examining exposure during pregnancy. Here, we will first elaborate on the pharmacological mechanisms and physiology underlying the alterations in free and total concentrations during pregnancy. Next, we will present an overview of total and free concentrations for several other antiretroviral drugs during pregnancy. Finally, we will make some inferences based on the theory and the examples presented, and propose a rule of thumb for conducting clinical pharmacokinetic studies in pregnancy.

As mentioned in the opening statement, an important and well‐described pharmacological process that changes during pregnancy is drug protein binding 4, 5. The main binding proteins in plasma are albumin and α₁‐acid glycoprotein (AAG) 6. The concentrations of both are known to decrease during pregnancy 7. When a drug binds to plasma proteins, changes in the protein concentrations may affect the fraction of the drug bound to proteins. The drug fraction that is not bound to protein is referred to as the fraction unbound (f_u). The f_u is defined as the free concentration (C_free) divided by the total concentration (C_tot = C_free + C_bound), equation (1).

$$f_u=\frac{C_{\mathit{\text{free}}}}{\left(C_{\mathit{\text{free}}}+C_{\mathit{\text{bound}}}\right)}$$ (1)

Although this equation may be illustrative when calculating the f_u after measuring free and bound drug concentrations in plasma, it does not provide any information on how these parameters relate to each other from a mechanistic point of view. The f_u is controlled by the maximal protein binding capacity (B_max; proportional to the concentration of the binding protein, assuming one or more binding sites per protein with similar and constant binding affinity), the equilibrium dissociation constant (K_d; reflective of the binding affinity of the drug to the protein), and the free concentration of the drug, C_free (equation (2)).

$$f_u=\frac{K_d+C_{\mathit{\text{free}}}}{\left(B_{\mathit{max}}+K_d+C_{\mathit{\text{free}}}\right)}$$ (2)

When C_free is much lower than K_d, which is usually the case, f_u depends on B_max and K_d only, according to equation (3) (adapted from 6), and hence remains constant (i.e. linear binding).

$$f_u=\frac{K_d}{\left(B_{\mathit{max}}+K_d\right)}$$ (3)

Consequently, when protein concentrations decrease during pregnancy, the maximal protein binding capacity, B_max, decreases and hence f_u increases. During pregnancy K_d is expected to remain unchanged. Again, f_u does not depend on C_free nor C_tot (in case of linear binding). Instead, C_free and C_tot depend on the mechanistic pharmacological parameters K_d and B_max that define f_u.

Under steady‐state conditions, the impact of f_u on the C_tot and C_free depends on whether drug clearance is restricted by protein binding 8. When the rate of dissociation between drug and plasma protein is low compared to (intrinsic) elimination rate constants exhibited by the eliminating organs, total drug clearance will be restricted by protein binding. Note that intrinsic clearance is the intrinsic capacity of an eliminating organ to eliminate a drug and that the elimination rate constants characterizing the clearance processes are determined by physiology (e.g. enzyme and/or transporter abundance) and the affinity of the drug towards the metabolizing enzyme/transporter. When the dissociation rate of a drug from plasma proteins is high compared to the intrinsic clearance, total drug clearance will not be restricted by protein binding. In general, when drug clearance is restricted by protein binding, changes in f_u will affect C_tot but not C_free. When drug clearance is not restricted by protein binding, changes in f_u will affect C_free but not C_tot 9. Since the clearance of most drugs that are highly bound to plasma proteins is restricted by protein binding, we will only discuss this scenario. Nevertheless, it should be noted that in some cases (e.g. with alterations in intrinsic clearance during pregnancy) clearance may theoretically change from being restricted to not being restricted by protein binding, or vice versa.

For drugs with protein binding‐restricted clearance, steady‐state C_tot is inversely proportional to f_u, and C_free is independent of f_u, whatever the mechanisms of elimination are 9.

$$C_{\mathit{tot}}=\frac{C_{\mathit{\text{free}}}}{f_u}$$ (4)

Although equation (4) looks like a trivial rearrangement of equation (1), it actually represents a more physiological relationship between f_u, C_tot and C_free. Here, C_free is the independent variable and f_u is the parameter determined by the mechanistic parameters K_d and B_max. C_tot is the dependent variable. When f_u changes, this relationship can be used to investigate the impact on C_tot, given the independent variable C_free. This leads to an important implication: decreases in protein concentrations during pregnancy can decrease C_tot, but not C_free and hence the pharmacological effect remains unchanged (see opening quote). Note that the latter is related to the notion that only unbound drug is available to equilibrate across membranes towards the site of action.

The fact remains that C_free can be decreased during pregnancy, also illustrated in Table 1. This, however, is typically related to other physiological changes during pregnancy such as an increased intrinsic clearance. As stated by Marzolini et al. 1, the abundance of certain cytochrome P450 enzymes or UDP‐glucuronosyltransferases may be induced by alterations in hormone levels (e.g. progesterone) during pregnancy, resulting in higher hepatic intrinsic clearance. For drugs with protein binding‐restricted clearance, the eliminating organ is only able to eliminate C_free, as bound drug is not available to the organ for elimination. Consequently, an increased intrinsic clearance lowers C_free. Changes in C_free can also relate to other physiological changes during pregnancy. For example, decreased intestinal motility during pregnancy may alter drug absorption and hence change C_free. Furthermore, pregnancy‐induced changes in drug distribution may alter C_free (note that following an immediate phase of redistribution in plasma these changes also alter C_tot, the dependent variable).

Table 1.

Changes in total and free antiretroviral concentration during third trimester of pregnancy compared to postpartum with the corresponding fraction unbound

	Mean ratio Ctot (%) TT/PP	Mean ratio Cfree (%) TT/PP	Fu (%)		Ref
			TT	PP
Rilpivirine (n = 11)					10
C min	58	64	0.28b	0.23b
C max	80	90
Darunavir (twice daily; n = 5)					11
C 12	89	105a	12	10
C max	76	92a
Darunavir (once daily; n = 8)
C 24	59	67a
C max	78	94a
Darunavir (once daily; n = 12)					12
C min	50	62	23	18
C max	69	84
Etravirine (n = 10)					13
C min	111	92	0.075	0.083
C max	135	107
Lopinavir (n = 30)					14
C min	69	79	0.84	0.68

TT, third trimester; PP, postpartum; n, represents the number of paired observations

^aInferred from the reported fraction unbound during third trimester of pregnancy and postpartum.

^bInferred from reported secondary total and free PK parameters (C_max).

To further illustrate this, Table 1 provides an overview of changes in free and total concentrations for several antiretroviral drugs measured in clinical pharmacokinetic studies during third trimester of pregnancy and postpartum. Typically, in these studies the postpartum pharmacokinetics serve as a control representing the nonpregnant situation as pregnancy‐induced physiological processes are expected to have normalized 4–6 weeks postpartum. 15 All of these antiretroviral drugs are bound to plasma proteins and for most of them both C_tot and C_free decreased. For rilpivirine, the decreases in C_tot and C_free are rather similar, whereas for darunavir and lopinavir, the decrease in C_tot seems somewhat larger than the decrease in C_free. In all these cases fraction unbound was increased during pregnancy. Based on the theoretical considerations provided above, these results can be easily explained. Where the observed decreases in C_free are likely to be related to enzyme induction during pregnancy (or other factors that can affect C_free), the decreases in C_tot are most likely to be a result of the increased fraction unbound alongside enzyme induction. Differences between various drugs in the extent of the decrease in C_tot and C_free can have multiple reasons, including the fact that their pharmacokinetics rely on different physiological processes (e.g. binding to albumin vs. AAG) or differences in enzyme or protein affinity. Interestingly, etravirine C_tot was increased during pregnancy whereas C_free remained unchanged. Correspondingly, f_u was decreased during pregnancy, something also observed in the case presented by Marzolini et al. 1: a peculiar finding that is not easily explained 13. It may be related to a shift in (apparent) binding affinity, possibly because of pregnancy‐related changes in plasma composition (e.g. endogenous lipids or proteins), or increased levels of other binding proteins during pregnancy, as hypothesized by Marzolini et al. 1. Whatever the case, this indicates that during pregnancy total concentrations are not always a good surrogate for free concentrations, but also that despite our mechanistic knowledge of the pharmacokinetics during pregnancy, it still remains challenging to make inferences on C_free based on C_tot.

This leads us to the following rule of thumb: if we want to identify pregnancy‐induced changes (or for any other condition in which protein concentrations are substantially altered) in the pharmacologically active C_free, for highly protein bound drugs we need to measure C_free. This is by no means a new insight 5, 16. Still, for some clinical pharmacokinetic studies with highly bound drugs there is room for improvement as in general only C_tot is measured. This has a variety of reasons. The development of bioanalytical methods for measuring C_free can be challenging and validation is not straight‐forward. 17 Also, measuring C_free is more costly in terms of material, time and personnel. It can also be argued that pregnancy‐related decreases in C_free usually follow decreases in C_tot. Although this has been observed (most of the examples in Table 1), assuming this to be a reliable law is almost invariably wrong based on the considerations and examples presented. Overall, this underscores that as a community interested in optimizing pharmacotherapy during pregnancy, we need to push for measuring C_free when drugs are highly bound to plasma proteins.

Competing Interests

S.S., R.G. and D.B. declared no other relationships or activities that could appear to have influenced the submitted work.

We are grateful to Angela Colbers, Roeland Wasmann and Mette Benoist for their thorough revisions of the manuscript.

Schalkwijk, S. , Greupink, R. , and Burger, D. (2017) Free dug concentrations in pregnancy: Bound to measure unbound? Br J Clin Pharmacol, 83: 2595–2598. doi: 10.1111/bcp.13432.