Exclusive breastfeeding is recommended by the World Health Organization (WHO) ¹ up to 6 months of age and partial breastfeeding with complimentary food up to 2 years of age, because of the advantages on maternal and infant health. Knowledge regarding medicine use during breastfeeding is expanding, however, insights into medicine safety and transfer to human milk is still limited. Lack of knowledge can result in uncertainty and discontinuation of breastfeeding, postponing of maternal pharmacotherapy, or off-label use of medicines.

Questions regarding the safety of clopidogrel (CLP) exposure during breastfeeding are prevalent. To put this statement in perspective, the InfantRisk Call Center ² in the United States explored their requests received over a 12-year period. Based on their database, 306 calls regarding CLP use and safety while breastfeeding were received during this period, of which the majority came from patients. Furthermore, they found that CLP is the fifth most prevalent reason to recommend cessation of breastfeeding from 2021–2023 (personal communication, Kaytlin Krutsch).

CLP is used as secondary prevention of atherothrombotic events, such as a myocardial infarction or a cerebrovascular accident and is considered an essential medicine by the WHO. ³ CLP (molecular weight: 321.82 g/mol) is a prodrug that is metabolized in the liver into the inactive clopidogrel carboxylic acid (CCA) and the clopidogrel active metabolite (CAM). CAM binds irreversibly to the adenosine diphosphate (ADP) purigenic P2Y₁₂ platelet receptor, which prevents platelet aggregation. CLP occurs rapidly absorbed and metabolized (oral bioavailability: 50%) as the maximum concentration (C_max) of CAM is 30–60 min after intake. Both metabolites are highly bound to plasma proteins (98%) and the apparent volume of distribution in adults is reported to be 39.2 ± 33.5 L. CCA is formed by esterases and comprises 85% of CLP metabolism. Multiple cytochrome P450 (CYP) enzymes metabolize CLP to an intermediate metabolite 2-oxo-clopidogrel and further to CAM, mainly by CYP2C19, but also by CYP1A2, CYP2B6 and CYP3A4. CLP is excreted both in urine (50%) and in feces (46%). The half-life of CLP is circa 6 h in adults, whereas the half-life of CCA and CAM is approximately 8 h and 30 min, respectively (European Medicines Agency, 2023; Drugbank online, 2024).

The extent of metabolism from CLP to CAM is variable due to the different genotypes of the activating CYP2C19 enzyme. Four different phenotypes have been described according to genotype: ultra-rapid, extensive, intermediate and poor metabolizers. CYP2C19 has different alleles with two prominent alleles being nonfunctional (CYP2C19*2 and *3), resulting in a poor metabolizer status. The poor metabolizer phenotype is prevalent in 2% of the Caucasian population, in 4% of the Black population and in 14% of the Chinese population. Contrarily, gain-of-function alleles (CYP2C19*17) have also been reported and result in the ultra-rapid metabolizer phenotype. Pharmacogenetic tests to determine the CYP2C19 genotype are available but are not routinely recommended, even though subsequent dose-adjustment was proven as cost-effective (European Medicines Agency, 2023; Morris et al., 2022; Hertz et al., 2024).

CLP is used off-label in neonates with certain heart diseases at risk for thrombosis (e.g., after systemic-to-pulmonary artery shunt). However, dose finding has been challenging in neonates due to maturation of metabolizing enzymes, further modulated by CYP2C19 polymorphisms (Van den Anker et al., 2020). Generally, CLP intake is not recommended during pregnancy and breastfeeding (Nana et al., 2021). To our knowledge, no cases regarding CLP use during breastfeeding have been described yet. While acetylsalicylic acid is an alternative antiplatelet drug, there is no adequate information on its safety during breastfeeding either (Bell et al., 2011; Drugs and Lactation Database, 2021).

This study was conducted within the UmbrelLACT study and is a part of the Innovative Medicine Initiative (IMI) ConcePTION project ⁴ , intending to reduce the knowledge gap regarding medicine use and safety during pregnancy and lactation (Van Neste et al., 2024). These two case studies aimed to determine the concentration of CLP and its main metabolites in human milk and the estimated infant exposure, as well as safety data. This article was written according to the CARE guidelines and ‘Guidelines for reporting cases of medication use during lactation’ (Gagnier et al., 2013; Anderson, 2022).

This case series describes 2 European mother-infant pairs included in the UmbrelLACT study (NCT06042803), the protocol of which has been published (Van Neste et al., 2024). Shortly, both mothers provided a signed informed consent form for themselves and their infant prior to enrolment. Our study team was not involved in the two mothers’ decision to start clopidogrel treatment while breastfeeding, nor in any other aspect of their clinical care.

After inclusion, the individual patients completed a self-reporting questionnaire on maternal clinical data, such as biometry, medicine intake, and lactation-related information (Van Neste et al., 2024).

Over a period of 24 h, starting from CLP intake, human milk samples were collected at steady-state. Human milk was expressed at home using an electric pump by patient 1 and using a manual pump by patient 2 to collect the total milk volume from both breasts at each breastfeeding moment. This volume was noted and 2 samples (3 and 7 mL) were set aside in the refrigerator, in separate tubes with and without the addition of reagent 2-bromo-3′-methoxyacetophenone (MPB), respectively. This reagent was needed to stabilize the CAM by derivatization, as described for plasma (Takahashi et al., 2008). The human milk samples were stored in the patient’s refrigerator for maximum 24 h and were afterwards transferred for long-term storage.

Maternal blood samples were collected within an hour of the first and last expression session of the 24-h sampling day. After blood collection, the derivatization reagent MPB was immediately added to one of the 2 tubes. For the first blood sample of patient 2, the MPB was added too late which caused unreliable results of the derivatized CAM (CAMD) in this blood sample, so this plasma sample was excluded in further analysis. Additionally, we collected one plasma sample of the partially breastfed infant of patient 2.

On each sampling day, a standardized questionnaire on the general health outcomes of the infant was completed by both mothers. This questionnaire was repeated once again 2 months after the last sampling day (Van Neste et al., 2024).

All samples were transferred to KU Leuven, Belgium, on ice for long-term storage (−80°C). Liquid chromatography with tandem mass spectrometry (LC-MS/MS) was used for the analysis of the samples for CLP, CCA and CAM (see Supplementary Material). The lower limit of quantification (LLOQ) for CLP, CCA and CAM was 0.008, 0.38 and 0.009 ng/mL and the limit of detection (LOD) for CAM is LOD 0.005 ng/mL. Sample concentrations below the LLOQ were processed as LLOQ/2 (Beal, 2001), when calculating pharmacokinetic parameters, such as area under the curve (AUC_0–24h) and average steady-state concentration (C_av,ss) and infant exposure estimations, such as daily infant dosage (DID) and relative infant dose (RID). Data analysis and graphs were prepared in Excel (version 16.85).

This case series is to our knowledge the first to provide observations on the human milk concentration of clopidogrel (CLP) and its main metabolites, the estimated infant exposure, and the reported health outcomes of the breastfed infants.

The infant exposure to maternal CLP intake was determined in the plasma of 1 breastfed infant and was low. Similarly, no infant adverse events were reported as of 2 months after the last sampling day. Furthermore, the estimated RID was lower than 0.2%, even when pooling CLP and its metabolites and using a higher daily milk intake volume of 200 mL/kg/day to mimic early infancy intake. These results are far below the arbitrary RID cut-off value of 10% (Anderson, 2018). Even though this cut-off was defined without empiric evidence, the observed RIDs are highly unlikely to cause any pharmacological effect in the breastfed infant, as supported by the absence of short-term adverse outcomes of the infants (Anderson, 2018). Furthermore, comparing the DID to the therapeutic doses administered to infants, the amount of infant exposure to CLP through human milk was confirmed to be low (0.008%–0.041%) ⁶ (Li et al., 2008).

The levels of CAM are below the LOD in human milk, except for two human milk samples. Clopidogrel is a P-glycoprotein (P-gp/ABCB1) substrate, but this efflux pump transporter of the blood-milk barrier is described as downregulated during lactation (Sangkuhl et al., 2010; Ahmadzai, 2014). Additionally, the transporter multi-drug resistance-associated protein 3 (MRP3/ABCC3) might have an influence on clopidogrel exposure during breastfeeding (Luchessi et al., 2017). To the best of our knowledge, no other transporter mechanisms were found for clopidogrel in literature. Furthermore, the low half-life contributes to the low CAM level. While CLP and CCA have a half-life of 6 and 8 h, the half-life of CAM is only 30 min (European Medicines Agency, 2023; Drugbank online, 2024).

The concentrations of CLP in plasma and human milk samples were higher during the second sampling day of patient 1 compared to the first sampling day (Figure 2). This increase was not observed in the measured metabolites, neither in plasma nor human milk. This difference might be explained by interindividual variability or the concomitant intake of acetylsalicylic acid during the second sampling day. Interestingly, drug-drug interactions between proton pump inhibitors and CLP have been described (as in the case for patient 2), for which combination with lansoprazole is favorable but still results in a reduction in CYP2C19 activity and anti-thrombotic effect (European Medicines Agency, 2023; Collet et al., 2014). However, no interactions between CLP and acetylsalicylic acid have been identified yet (Serbin et al., 2016).

The enzyme activity in infants is not yet mature (ontogeny), which means that conversion of CLP into CAM in the infant is expected to be minimal. More specifically, there is only 20%–25% of CYP2C19 expression levels of adults at birth and 40%–50% of adult levels at 1 year of age (Van den Anker et al., 2020). In case of CYP3A4, a high variability in activity is reported during the first 12 months of life. Its activity increases from 8% of adult activity at birth to 43% from 3 months to 1 year of life and to 108% from 1 year to 15 years of age (Caruthers and Dorsch, 2008). Furthermore, the role of CYP3A5 and CYP3A7 in the metabolism of CLP in infants has not yet been explored. As the concentration of CLP is so low, we do not expect any adverse health effects, even in case of the higher CLP level when combined with acetylsalicylic acid.

The role of CYP2C19 is crucial in the formation of CAM, which means that the genotype of this enzyme plays an essential role in the activation of CLP as well. More specifically, poor metabolizers have a reduced exposure to CAM, which might cause a reduction in platelet responsiveness to CLP. On the other hand, ultrarapid metabolizers activate CLP more rapidly. This CYP2C19 polymorphism is subsequently an explanation for the variability seen in the effect of CLP (Van den Anker et al., 2020; Karaźniewicz-ŁAda et al., 2014; Karaźniewicz-Łada et al., 2012). However, in this case series the 2 patients were assumed to be extensive metabolizers based on their plasma levels corresponding to this phenotype, whereas the levels visually deviated from the intermediate or poor metabolizer type (Dorota, 2014). The maternal poor metabolizer phenotype should result in higher concentrations of CLP in maternal plasma and milk. This maternal phenotype would theoretically result in the highest infant systemic exposure after breastfeeding and poses the highest risks for the infant. However, since the dose ingested by the infant is estimated to be very low irrespective of the maternal polymorphism and infant conversion of CLP to CAM is expected to be minimal because of ontogeny, we do not expect a clinically significant effect of either the maternal or infant CYP2C19 metabolizer phenotype.

Some limitations were identified while conducting this study. First, since human milk sampling was performed by the patients at home, it is impossible to be certain whether the patients followed the sample instructions perfectly. Second, this specific study was complicated by the extra sample handling needed to stabilize CAM. Furthermore, our UmbrelLACT protocol describes human milk expression by an electric pump, according to the FDA guidance (FDAb). An electric pump was not available during the sampling day of patient 2, so human milk was expressed by a manual breast pump. Nevertheless, an electric pump can overestimate the human milk volume and fraction of lipid content due to complete emptying of the breasts (Anderson and Sauberan, 2016). As this study does not allow to determine the AUC in maternal plasma to calculate the M/P ratio, the M/P ratio was calculated using plasma AUC from non-breastfeeding patients from literature. As relevant characteristics of their physiology, such as body composition and enzyme activity, changes during pregnancy and lactation, the M/P ratio based on literature might differ from the real value (Van Neste et al., 2023). Moreover, CAM levels in human milk were almost all under the LOD. This implies that the AUC for CAM is very low and could not be accurately determined. Even though this is a limitation of this study, it indicates that the active component of CLP is only minimally present in human milk, emphasizing the estimated low risks of CLP intake during breastfeeding. Furthermore, the toxicity of CCA in infants has not yet been explored. Except for a slight increase in neurologic adverse events (i.e., seizure and stroke) in infants using clopidogrel (0.2 mg/kg/day) compared to placebo (7.8% vs. 2.3%, p < 0.001), clopidogrel and its metabolites are well-tolerated in infants (Wessel et al., 2013). Due to the low CCA concentrations in human milk, no toxic effects are expected from CCA. Last, the infants in this case series were over 6 months old which implies already more mature enzyme activity and were both already receiving complementary foods. Therefore, these results are less representative for medicines exposure during the first weeks or months of life.

This is the first paper that quantified CLP exposure of breastfeeding infants and provides relevant data to support shared-decision making in clinical practice. As highlighted by the numbers from the InfantRisk Call Center (unpublished data), the quantification of CLP and its metabolites in human milk is highly relevant as many clinical requests have been received regarding this compound. Discontinuation of breastfeeding had been advised in case of CLP use (European Medicines Agency, 2023; FDA, 2024), however, infant exposure seems to be low according to this first study. As breastfeeding itself has many benefits for mother and child, the low exposure is an argument pointing at compatibility of maternal CLP use and breastfeeding for these cases.

More research is required to assess the impact of genetic polymorphisms on the exposure through human milk and to confirm our results. Additionally, PBPK models of CLP and its metabolites are recommended to predict the exposure in breastfeeding infants by simulations, as only few cases or milk samples are needed for this approach (Nauwelaerts et al., 2023). These PBPK models could be extrapolated to explore challenging situations, such as different maturation of enzymes in infants or the changing physiology in lactating women during postpartum. Furthermore, this approach can investigate multiple scenarios related to a broader postnatal age range and all different CYP2C19 genotypes, similarly to the CYP2D6 genotypes study (Willmann et al., 2009).

In summary, an estimated low infant exposure to CLP and its metabolites in human milk was reflected by a low infant plasma concentration. This case series found no short-term adverse events from CLP use during breastfeeding. However, these two cases likely do not fully capture the exposure during early infancy and its pharmacokinetic variability, warranting additional data collection, as well as PBPK simulations on infant exposure and safety of maternal CLP during breastfeeding.