As an immunosuppressant drug, cyclosporine A (CsA) has been widely used in transplantation since the early 1980s (Colombo and Ammirati, 2011). From then on, CsA remained a first-line therapy for patients with solid organ transplantation. However, CsA has a narrow therapeutic range and large interindividual pharmacokinetic variability. While underexposure might cause graft versus host disease (Rogosheske et al., 2014) and acute rejection episodes, overexposure might result in toxicity (Bardazzi et al., 2018). CsA is categorized as a biopharmaceutical classification system (BSC) class II drug due to low solubility and high permeability (Onoue et al., 2010). After oral administration, CsA is absorbed from the gastrointestinal tract with a bioavailability of approximately 30% (Drewe et al., 1992). As a lipophilic molecule, CsA has a high volume of distribution (3–5 L/kg); in blood, cyclosporine is extensively bound to erythrocytes. In plasma, approximately 90% is bound to proteins, primarily lipoproteins (U.S. Food and Drug Administration, 2015). The disposition of cyclosporine is generally biphasic, with a terminal half-life of approximately 8.4 h (range 5–18 h) (U.S. Food and Drug Administration, 2015). Cyclosporine is extensively metabolized by CYP3A4 and CYP3A5 and subject to efflux from renal tubular cells and other cells via P-glycoprotein (P-gp) (Hebert, 1997). Therefore, co-administration of CYP3A or P-gp inhibitors may alter the pharmacokinetics of CsA. For example, concomitant administration of ketoconazole has been reported to elevate CsA concentrations several-fold (Albengres and Tillement, 1992; Keogh et al., 1995). Imatinib, a potent inhibitor of CYP3A4 and P-gP, approximately doubled CsA exposure (Atiq et al., 2016).

Baicalin (baicalein 7-O-glucuronide, BG), the major bioactive compound from Scutellaria baicalensis (Shi et al., 2016), is widely applied in traditional Chinese medicine for the treatment of inflammation, hepatitis, various infections, and tumors (Xi et al., 2015; Zhao et al., 2016; Gong et al., 2017; Ming et al., 2018). In 2005, baicalin capsules (250 mg per capsule, approval no. H20158009) were approved by the state food and drug administration of China for the adjuvant therapy of hepatitis (2 capsules 3 times a day). After oral administration, BG is rapidly hydrolyzed to baicalein (B) by β-glucuronidase derived from intestinal bacteria (Huang et al., 2019). Both BG and its aglycone baicalein (B) have a low hydrophilicity (solubility of BG is 0.057 mg/ml in water) (Wu et al., 2011) and a relatively low permeability as determined in the Caco2 cell system [for BG, Papp = (0.275 ± 1.14) × 10⁻⁶ cm/s (Zhu et al., 2013); for B, Papp = 9.0 × 10⁻⁶ cm/s (Cai et al., 2016)], resulting in a very low oral bioavailability for both baicalin (2.2%) and baicalein (Wu et al., 2014). Compared to BG, B could be better absorbed in the gastrointestinal tract and then conjugated to BG in the gut wall and liver (Liu et al., 2009; Zhang et al., 2011). BG is extensively bound to proteins (86–92%) in human plasma (Tang et al., 2006), has a short elimination half-life (6.36 ± 5.85 h), and undergoes extensive metabolism (Noh et al., 2016).

Inflammation is a significant problem in organ transplant patients. With the anti-inflammatory and antioxidant properties of BG and its aglycone B, co-administration of BG might benefit the organ transplant patients treated with CsA (Shieh et al., 2000; Dinda et al., 2017). However, several lines of evidence suggest that BG may cause drug–drug interactions in humans. B is an inhibitor of CYP3A and P-gp in rats (Morisaki et al., 2013; Miao et al., 2016). For example, intravenous administration of high BG doses (0.23–0.90g/kg) to rats decreased the clearance of midazolam by up to 43% (Xin et al., 2013). In human liver microsomes, B was reported to potently inhibit CYP3A4 (Ki for mixed-type inhibition of bufalin 5′-hydroxylation 2.3 μM; IC₅₀ for inhibition of midazolam and nifedipine at their Km concentrations 13 and 15 μM, respectively) (Li et al., 2018). On the other hand, it has been reported that B (but not BG) may activate the human pregnane X receptors and the human constitutive androstane receptor (CAR) (Morisaki et al., 2013; Cheng et al., 2014; Miao et al., 2016), which could mediate CYP3A and P-gp induction. A recent study in rats indeed showed that single intravenous dose and multiple doses of BG had no effect on the pharmacokinetics of CsA, while oral administration significantly decreased C_max and AUC_0-∞ of CsA. This indicates that BG might affect intestinal absorption and/or secretion of CsA. Further study revealed that after multiple oral doses of BG treatment, the expression of P-gp of rats increased in the intestine, but was not changed in the liver (Tian et al., 2019). Beyond affecting CYP3A and P-gp, co-administration of BG also changed plasma protein binding and apparent volumes of distribution of nifedipine, another CYP3A probe, in rats (Cheng et al., 2014). An early study in rats which directly investigated the effects of Scutellaria radix decoction, BG, or B on CsA pharmacokinetics provided mixed findings: the decoction reduced exposure to oral but not to intravenous CsA, while BG and even more so B increased exposure to oral CsA (Lai et al., 2004).

These results indicate that indeed co-medication with BG might alter the pharmacokinetics of CsA in humans and also indicate that any respective DDIs may be mediated by several mechanisms. So far, no clinical studies have been reported on drug–drug interactions between BG and CsA. The aim of the current study, therefore, was to explore a potential effect of BG on CsA exposure in healthy volunteers and to assess possible effects on individual pharmacokinetic processes in detail.

In this study, non-compartmental analysis of the blood concentration vs. time profiles of CsA indicated that co-administration of a single 500 mg BG dose has no clinically relevant effect on the exposure of CsA in healthy volunteers. The compartmental population pharmacokinetic analysis further confirmed this result for underlying pharmacokinetic processes.

After oral administration, BG, due to low lipophilicity, may either be directly absorbed by the action of uptake transporters, or undergo hydrolysis by intestinal glucuronidase or intestinal microflora to release its aglycone B (Akao et al., 2010; Kang et al., 2014; Noh et al., 2016; Kalapos-Kovács et al., 2018). B is probably better absorbed and then efficiently conjugated to BG in the gut wall and the liver and thus restored to its original form BG (baicalein 7-O-glucuronide) as well as to baicalein 6-O-glucuronide (Liu et al., 2009; Zhang et al., 2011). Thus, both BG and B would be present at the various locations of CYP3A4 and P-gp in the gut wall and the liver, with the potential to modify their activity and/or expression (Morisaki et al., 2013; Miao et al., 2016), if sufficient concentrations were reached.

However, an effect of BG on the pharmacokinetics of CsA was not observed in this study based on all but two of the assessments. The observation in the population pharmacokinetic analysis that 95% CIs for Ka and Q exceeded the “no relevant effect” range probably just reflects that the study was not powered to quantify an effect on these parameters. There are three potential explanations for the finding that BG did not cause a drug–drug interaction (DDI). First, formation of B, the much more active moiety to cause DDIs (Morisaki et al., 2013; Xin et al., 2013; Tian et al., 2019), from BG may be very limited in humans in vivo. Second, even if B would be generated extensively, it is subject to (re-)glucuronidation in the gut, the gut wall, and/or the liver (Akao et al., 2010), thus losing (most of the) effects on CYP3A4 and P-gp. Indeed, when B single doses were administered directly to healthy volunteers, exposure to BG exceeded that to B more than 10-fold (Li et al., 2014), indicating that glucuronidation of B in humans is rapid and extensive. Third, it cannot be excluded that competing mechanisms of inhibition would cancel each other out, but the different time courses of inhibition of drug metabolizing enzymes, transporter, or protein binding vs. induction by increased protein synthesis make this explanation unlikely. In summary, independent of the underlying mechanism, it appears that the systemic exposure of B produced after a single oral administration of BG in this study was too low to affect the activities of CYP3A and P-gp.

The present study is the first to study any DDI of BG in humans. One study in rats reported that single dose of BG (136.6 mg/kg) markedly elevated the C_max and AUC of CsA to about 4 times and 6 times, respectively, compared with CsA administered alone (Lai et al., 2004). However, our recent study in rats showed that multiple intravenous doses of BG did not affect the exposure of CsA but multiple oral administrations of BG (80 mg/kg) could decrease the C_max and AUC_0-∞ of CsA by 38 and 25%, respectively (Tian et al., 2019). Further study resulted in that after multiple oral doses of BG treatment, the expression of P-gp of rats increased in the intestine, but was not changed in the liver (Tian et al., 2019). The different results of the two studies in rats might be related to the dosage, and the different results between rats and humans might be caused by the species differences, particularly in the formation and re-glucuronidation of B.

The additional use of population pharmacokinetic analysis might serve as a powerful tool to improve the understanding of potential DDI. First, the non-compartmental analysis might result in poorly estimated parameters such as clearance and volume(s) of distribution and confound sources of variability such as interindividual, intraindividual, and inter-occasion variability based on the actual observation instead of basic pharmacokinetic parameters as the dependent variable. Secondly, the population pharmacokinetic approach enables to assess PK processes underlying drug exposure separately. This way, it was possible to show that neither clearance nor volumes of distribution of CsA, which are the most important PK parameters to describe exposure, were affected by BG. As a limitation, it was not possible to separately describe intestinal and hepatic metabolism of CsA and thus to assess interaction at these two sites separately; to this end, both oral and intravenous administration of CsA would have been required (Gazzaz et al., 2018).

Pharmacokinetic data of BG in humans are sparse. Published data include the pharmacokinetics of BG after oral administration of BG to healthy subjects in a bioequivalence study (Wu et al., 2005), and pharmacokinetics of BG and B after single and multiple oral doses of B administered to healthy volunteers (Li et al., 2014; Pang et al., 2016). The AUC of BG after oral administration of 750 mg BG (613 ng h ml⁻¹) (Wu et al., 2005) was similar to that after oral administration of only 100 mg B (580 ng h ml⁻¹) (Li et al., 2014), which would be compatible with poor bioavailability of BG; unfortunately, in the study with BG administration, B concentrations were not quantified. Furthermore, BG was absorbed very slowly, with a T_max of 7.4 h (Wu et al., 2005), while B was absorbed much faster with a T_max of about 1 h for both BG and B (Li et al., 2014). These data suggest that both extent and temporal course of systemic concentrations of BG and probably its metabolite B in the present study were not sufficient to mediate an effect on CsA pharmacokinetics.

Although the current study would not suggest that dose adjustment might be warranted when BG and CsA are co-administered, because of the limited information on BG pharmacokinetics, a mechanistic extrapolation to other settings such as chronic BG dosing, higher BG doses, different timing of BG doses, or administration of B instead of BG would not be possible. Thus, in additional studies, BG and B plasma and/or blood concentrations should be quantified. Secondly, multiple dosages of BG should be used to treat the subjects before the administration of CsA to achieve maximal BG/B exposure as the single treatment of BG might not have been enough to modify CYP3A4 and/or P-gp. Furthermore, while in this study, the pharmacokinetic parameters of CsA were consistent with other healthy volunteer studies of CsA (Garg et al., 2011), clearance of CsA in healthy volunteers was two times higher than that of kidney or liver transplant patients (Bo et al., 2010), suggesting that DDI results from healthy volunteers might not be directly extrapolated to organ transplant patients. Thus, future investigations to evaluate the effect of chronic BG administration on CsA pharmacokinetics in patients would also be of interest. Finally, CYP3A5 genotype affects the clearance of CsA (Song et al., 2012). Thus, subjects with different CYP3A5 genotypes would be preferable to be recruited for subsequent studies.

Both the pharmacokinetic data and the limited safety data of the present study do not provide evidence that BG co-administration with CsA would cause an additional risk. The possibility of co-treatment with these drugs for transplant patients might be beneficial, because BG could exert antiviral, anti-inflammatory, anti-oxidation, and anti-proliferation effects (Lee et al., 2015; Ming et al., 2018). Indeed, in an observational study, it was reported that when BG was given with the antiviral drug telbivudine for the treatment of hepatitis in the clinic, the liver function (ALT) normalization rate, HBV DNA and hepatitis B virus markers (HBeAg) negative conversion rate, and anti-HBe serum conversion rate in the treatment group (BG and telbivudine, n = 64) were significantly higher compared to the reference group (telbivudine alone, n = 62) (Fang et al., 2011). Thus, these first results on a potential DDI between BG/B and CsA are encouraging and should be expanded by further studies as described above.

In summary, this is the first clinical study to investigate the effect of BG on CsA pharmacokinetics in humans. The current dosage of BG (500 mg single dose) and CsA (200 mg single dose) was generally safe and well tolerated in the adult subjects without serious adverse events observed. Both non-compartmental analysis and the population pharmacokinetic approach did not provide any evidence that the exposure to CsA and/or underlying pharmacokinetic processes of CsA were changed to a clinically relevant extent by BG. These results need to be confirmed in studies with maximal chronic exposure to BG and quantification of BG and B.

This study was carried out in accordance with the standards of Good Clinical Practice, all applicable regulations, specific legal requirements, and ethical principles as described in the Declaration of Helsinki with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Ethics Committee of at the first affiliated hospital of Zhengzhou University (Henan, China; approval no. SR201509).

XT and XZ designed and supervised the clinical trial. RL, JW, SL, YC, JiZ, and JiaZ carried out the clinical trial. RL, JW, and XL performed the non-compartmental analysis. XL conducted the population pharmacokinetic analysis. XL, RL, UF, and JW wrote the manuscript. UF, MT, and XT edited the manuscript.