A commentary on the use of pharmacoenhancers in the pharmaceutical industry and the implication for DMPK drug discovery strategies

Abstract

1. Paxlovid, a drug combining nirmatrelvir and ritonavir, was designed for the treatment of COVID-19 and its rapid development has led to emergency use approval by the FDA to reduce the impact of COVID-19 infection on patients.

2. In order to overcome potentially suboptimal therapeutic exposures, nirmatrelvir is dosed in combination with ritonavir to boost the pharmacokinetics of the active product.

3. Here we consider examples of drugs co-administered with pharmacoenhancers.

4. Pharmacoenhancers have been adopted for multiple purposes such as ensuring therapeutic exposure of the active product, reducing formation of toxic metabolites, changing the route of administration, and increasing the cost-effectiveness of a therapy.

5. We weigh the benefits and risks of this approach, examining the impact of technology developments on drug design and how enhanced integration between cross-discipline teams can improve the outcome of drug discovery.

Keywords:

• Pharmacoenhancer
• ADMET
• CYP inhibition
• P-glycoprotein inhibition
• pharmacokinetics

Introduction

The approaches and technology used for drug discovery and development continue to advance at pace. There have been significant advances in drug design leading to drug candidates which are highly tailored to the target protein of interest. Improvements in chemical libraries have been made in an attempt to improve the quality of hits following high-throughput screening (HTS) (Dahlin and Walters 2014). Fragment-based drug design (FBDD) is an alternative approach, where the final drug has been grown from small molecular weight compounds with weak target affinity (Davies et al. 2006; Congreve et al. 2007; Jhoti 2008; Johnson et al. 2018; Murray et al. 2012; Mortenson et al. 2014; Erlanson et al. 2016; de Esch et al. 2022). The process in which these molecules are grown is with detailed knowledge of the shape and properties of the binding site, fragment orientation and interactions with the amino acid residues within the binding site. Several marketed drugs have been developed using this approach (Tsai et al. 2008; Souers et al. 2013; Zhang et al. 2013; Murray et al. 2019). Chemical synthesis approaches have also advanced, to allow for the optimisation of chemical reactions to facilitate the synthesis of highly specific chemical modifications which can complement an FBDD approach (Grainger et al. 2019). Structure-based drug discovery (SBDD) is another approach where the binding of hits to the pocket of the target protein is experimentally determined. These advances have led to approaches that can deliver highly potent, selective and ligand efficient molecules. While the incorporation of drug-like properties are not specifically considered for these approaches, the insights gained can be applied to guide molecule modifications for improving pharmacokinetic (PK) properties without negatively impacting drug-target binding.

Controlling physicochemical properties continues to be cited as key strategy to successful discovery of oral drugs. Since Lipinski proposed five simple rules (MW ≤ 500, log P ≤ 5, H-bond donors ≤ 5, H-bond acceptors ≤ 10; the ‘rule of five’) that defined drug-likeness for orally delivered drugs (Lipinski et al. 1997), the scientific understanding of the influence of physicochemical descriptors on absorption, distribution, metabolism, excretion and toxicity (ADMET) has advanced. In some cases, a simple modification of the rules, such as determination of log D to account for ionisation at physiological pH instead of log P, can improve the structure–activity–relationship (SAR) between physicochemical and ADME properties (Bhal et al. 2007). In addition to lack of clinical efficacy and unmanageable toxicities, poor drug-like properties have been attributed to high clinical attrition rates (Sun et al. 2022), specifically inflation in drug size and lipophilicity (Hann 2011). Numerous factors can impact the final physicochemical properties of a drug, ranging from scientifically driven needs such as target class to the influence of historical organisational experiences and practices (Hann and Keserü 2012). Additionally, novel approaches, such as protein–protein interactions, targeting RNA, ‘undruggable targets’ and proteolytic-targeted degraders, have driven further consideration of the physicochemical property space and how drug-likeness can be achieved (Ursu et al. 2020; Wang et al. 2022; Zhang et al. 2022). Understanding the potential for chemical space beyond the rule of five to deliver successful drugs is evolving (Doak et al. 2014). Several approaches have been proposed such as intramolecular hydrogen bonding and macrocyclization. Others have employed facilitating oral exposure with sophisticated formulations or delivery systems. Another approach is the purposeful combination of two interacting drugs, known as pharmacoenhancement (Figure 1 and Table 1). The earliest example of this technique was the combination of penicillin with probenecid, which delays penicillin excretion (Marmell and Prigot 1957). The continued use of pharmacoenhancers decades later is indicative of a significant challenge that still remains in the drug discovery process, namely amongst others, optimisation of pharmacokinetic properties. In this commentary, we review pharmacoenhancers in clinical development and commercially approved, the risk–benefit ratio to this approach and the implication for current drug development strategies, particularly in the ADMET field.

Figure 1. Potential impact of pharmacoenhancers on the time-concentration profile of the co-administered drug. In all cases exposure of the co-administered drug, as measured by AUC, is increased by the pharmacoenhancer. (A) C_max increases whereas half-life (T_½) is unchanged, indicating a reduction in the first pass effect only, which could result from inhibition of gut metabolism (e.g. CYP3A) and/or inhibition of efflux transporters (e.g. P-glycoprotein). (B) C_max remains the same whilst T_½ is increased indicating an increase in systemic clearance due to inhibition of hepatic or extrahepatic metabolism (e.g. CYPs). The impact of a pharmacoenhancer will depend on the route(s) of clearance of the co-administered drug, the potency of the inhibitor and the free concentration of the inhibitor relative to the active drug achieved in the gut and/or liver.

Table 1. Pharmacoenhancer inhibitory potency against principal drug clearance mechanism.

Main mechanism	Clinical pharmacoenhancer	IC50 (µM)	Reference
CYP3A inhibition	Ritonavir	0.11	Xu et al. (2010)
	Cobicistat	0.15
	Itraconazole	0.006	Soherranen et al. (2004)
	Itraconazole metabolites	0.0004–0.0007
CYP2D6 inhibition	Bupropion	58	Hesse et al. (2000)
	Fluoxetine	0.98	Huang (2011)
	Quinidine	0.02–0.11
P-gp inhibition	Cyclosporin A	0.8–7	Morrissey et al. (2012)
	Elacridar	0.1–0.2
	Tariquidar	0.2
	Zosuquidar	0.024–0.10
	Encequidar	0.035	Kwak et al. (2010)
CDA inhibition	Cedazuridine	0.4	Ferraris et al. (2014)

Pharmacoenhancers in antiretroviral therapies

Nirmatrelvir

The recent successful development of Paxlovid by Pfizer in an accelerated time should be recognised as a testament to the scientific advancements in the drug discovery process. Several approaches have been cited as contributing to this success, including modelling for a more focussed clinical trial (Singh et al. 2022). Paxlovid is a combination of the orally active SARS-CoV-2 M^pro protease inhibitor, nirmatrelvir, and a concomitantly dosed drug, ritonavir, whose sole purpose is to enhance the pharmacokinetics of the active drug. Ritonavir, initially developed as a HIV protease inhibitor (Kempf et al. 1998), was subsequently repurposed as a pharmacokinetic booster and is currently utilised in many protease inhibitor multi-drug therapies, due to its potent inhibition of the drug metabolising cytochrome P450 (CYP) 3A4 isozyme (Hull and Montaner 2011). In addition, ritonavir is highly plasma protein bound, and it has been suggested that displacement of co-administered protease inhibitors can lead to improved efficacy by increasing the free fraction (Barry et al. 1997). It is generally accepted, however, that a concomitant increase in free clearance, would result in no appreciable change in free drug levels (Benet and Hoener 2002; Smith et al. 2010).

Nirmatrelvir was developed from an advanced compound, PF-00835231, which exhibited poor oral absorption. In the discovery of nirmatrelvir, a key objective was therefore to improve the preclinical oral absorption. This was achieved by the reduction of the number of hydrogen bond donors from five in PF-00835231 to three in nirmatrelvir using structure guided design (Eng et al. 2022). In preclinical studies nirmatrelvir was shown to be an orally active M^pro inhibitor, primarily undergoing CYP3A4-mediated metabolism. In the clinic, the required target plasma concentrations for efficacy were several fold above EC₉₀ and therefore co-administration with ritonavir was included in the clinical trial strategy, in order to increase nirmatrelvir exposures (Singh et al. 2022). The combination strategy resulted in an eight-fold increase in exposure (area under plasma concentration curve; AUC) compared to dosing nirmatrelvir alone, which was attributed to a decrease in gut metabolism (E_g), first pass hepatic extraction (E_h) and total clearance as a result of ritonavir induced CYP3A4 inhibition. Given that nirmatrelvir is a relatively polar compound (clogP < 1), metabolic inhibition by co-dosing with ritonavir switches the major route of clearance to renal excretion. The rapid development of Paxlovid demonstrates, at least in this case, that combination with a pharmacokinetic enhancer did not complicate or slow down clinical development. However, since Paxlovid’s conditional approval there have been multiple publications on caution in co-medication due to a significant risk of drug–drug interactions (DDIs). This highlights that complexities can arise with the employment of pharmacoenhancement strategies. Ritonavir also inhibits other CYP medicated metabolism (CYP2C8, CYP2C9, CYP2C19 and CYP2D6) and is an inducer of several metabolising enzymes such as CYP1A2 and the uridine 5′-diphospho-glucuronosyltransferase (UGT) family (Foisy et al. 2008). Although there have been advances in the development of physiologically based pharmacokinetic (PBPK) models for DDI risk assessments (Umehara et al. 2018; Montanha et al. 2022), the severities of drug interactions can be difficult to predict, particularly for drugs which are metabolised by multiple enzymes or have low CYP3A4 intrinsic clearance (Hsu et al. 1998). However, eligible COVID-19 patients with a high risk of developing severe disease are now benefiting from this accelerated drug development strategy, as in this case, the clinical benefit outweighs the risk of significant DDIs (Fishbane et al. 2022; Lamb 2022; Prikis and Cameron 2022; Shini Rubina et al. 2022; Waters et al. 2022).

Highly active antiretroviral therapies

Originally treatments for HIV were based on large pill burden and complex administration instructions for patients. An advancement of HIV therapies was achieved when ritonavir was repurposed as a pharmacokinetic enhancer of retroviral drugs (Hull and Montaner 2011), as applied and discussed above in the case of nirmatrelvir. Ritonavir is now established as a key pharmacoenhancer underpinning highly active antiretroviral therapies. Ritonavir doses have been optimised for effective CYP3A4 inhibition alone, being substantially lower than doses required to elicit HIV protease inhibitor activity (Figure 2). As in the case of nirmatrelvir, the optimal dose for boosting the pharmacokinetics of other protease inhibitors is typically 100 mg once daily (Zelin and Petruschke 2014). Higher doses have been evaluated however, these have caused adverse events such as increased cholesterol triglycerides and liver enzymes, nephrotoxicity and other side-effects. There is evidence that, even at doses of 100 mg, ritonavir is responsible for some of the side-effects in patients on multidrug therapy and the addition of ritonavir can increase the incidence of side-effects in a dose dependant manner (Zelin and Petruschke 2014). Clinical trials have therefore evaluated the opportunities for even lower doses to deliver sufficient CYP3A4 inhibition whilst further reducing the potential of ritonavir driven side-effects. Whilst it is generally accepted that ritonavir predominantly exerts its pharmacoenhancer activity by inhibition of CYP3A4, the drug also inhibits the efflux transporter p-glycoprotein (P-gp) and as described above, other CYP isozymes and UGTs, therefore due to potential DDIs, careful clinical management is required. However, the benefit of simplified dosing regimens and reduced pill burden, which ultimately improves patient compliance and consequently efficacy, outweighs the management of DDI induced toxicities.

Figure 2. The structure, and M^pro, CYP3A and CYP2D6 inhibitory activity of cobicistat and ritonavir (data from Xu et al. 2010).

The successful application of ritonavir as a pharmacoenhancer in turn led to the pharmaceutical industry developing more selective CYP3A4 inhibitors, as opposed to optimising active drugs. This indicates the challenges certain drug classes have in optimisation of pharmacokinetics where key structural motifs for activity are inextricably linked to their pharmacokinetic properties. Cobicistat was developed as a more selective CYP3A4 inhibitor, devoid of protease inhibition or problematic side-effects (Figure 2). As cobicistat has no HIV antiviral activity, the potential risk of selection pressure for protease resistant virus when subtherapeutic doses of ritonavir are used in the absence of a therapeutically active protease inhibitor are removed (Xu et al. 2010). Inhibition of other CYP isozymes has also been reduced or removed with cobicistat, compared to ritonavir. During the discovery phase, efforts were made to improve the physicochemical properties, resulting in cobicistat’s high aqueous solubility as co-formulation with other drugs was desirable (Xu et al. 2010). Clinical trials have shown that cobicistat is as effective as ritonavir and now the pharmacoenhancer is employed in a number of multidrug antiviral therapies (Squillace et al. 2018).

Pharmacoenhancers in oncology and CNS disorders

Osimertinib

Osimertinib is a third-generation, irreversible tyrosine kinase inhibitor of L858R/T790M mutant epidermal growth factor receptor (mEGFR) (Finlay et al. 2014). The drug was approved for the treatment of L858R/T790M EGFR positive non-small cell lung cancer (NSCLC) and subsequently became the first-line treatment for all EGFR positive NSCLC patient cases. Osimertinib is well absorbed, highly bioavailable (approximately 70%) and has a long half-life following a single dose in healthy volunteers (Vishwanathan et al. 2019). The major routes of metabolism were determined to be CYP3A4/5 mediated and metabolites (AZ7550 and AZ5104) at ∼10% osimertinib total exposure were observed (Dickinson et al. 2016). Both of these major metabolites demonstrated activity against mEGFR, but with suboptimal profiles compared to osimertinib. AZ7550, an N-dealkylation product, demonstrated a similar selectivity window over wild-type (WT) to osimertinib, but with reduced activity against mEGFR. AZ5104, a demethylated metabolite, had improved potency against mEGFR but reduced selectivity against WT EGFR which was unfavourable due to dose-limiting toxicities associated with WT EGFR inhibition (Ward et al. 2013). Preclinical studies indicated that AZ5104 would contribute to, but not solely drive efficacy. However, post clinical commentaries suggested the potential of AZ5104 to contribute to observed drug induced toxicities (Finlay et al. 2014). In this scenario, inhibition of CYP3A-mediated metabolism could potentially present favourable toxicity profiles whilst maintaining efficacy. Proof-of-concept clinical trials were therefore initiated to investigate osimertinib pharmacokinetic properties following co-administration with CYP3A4 modifiers, investigating combinations with the inducer rifampicin and inhibitor itraconazole, which are typically utilised in DDI studies (NCT02197247 and NCT02157883, respectively). Co-administration with rifampicin caused a 78% decrease in AUC compared to osimertinib alone, with a similar decrease for AZ5104 but an insignificant increase for AZ7550 exposure (Vishwanathan et al. 2019). Though both metabolites are produced by CYP3A4, the increase of AZ7550 was attributed to an imbalance in the formation and elimination rates which was not observed for AZ5104. Following co-administration with itraconazole, no clinically relevant exposure changes were observed for osimertinib and its metabolites, indicating osimertinib is safe, but may not benefit from co-administration with CYP3A inhibitors in a clinical setting (Vishwanathan et al. 2018).

More recently, a further proof-of-concept trial examining the pharmacokinetic boosting of osimertinib (OSIBOOST, NCT03858491) in combination with another CYP3A inhibitor cobicistat was undertaken and represents an example of pharmacoenhancer employment as a cost saving strategy, particularly for newly approved treatments where costs are extremely high. The aim of the trial was to assess the ability of cobicistat to reduce the first-pass effect and clearance, thereby increasing osimertinib exposure and, hence reducing the efficacious dose. Pharmacokinetic boosting of osimertinib with cobicistat in patients with NSCLC was demonstrated to be feasible without increasing toxicity, however the degree of boosting exposure was reported to be variable (van Veelen et al. 2022). The use of cobicistat to boost anti-cancer drugs has only been described in two previous cases and in current trials with olaparib, in contrast to the extensive number of studies in antiretroviral settings (van Veelen et al. 2022), therefore further clinical trials are required to assess if this approach is viable in this disease population.

Olaparib

Olaparib is a further example of employing pharmacokinetic boosting in an to attempt to maintain affordable healthcare and reduce the cost of expensive drugs. Olaparib is a poly (ADP-ribose) polymerase (PARP) inhibitor and was first approved as a monotherapy for the treatment of BCRA1/2 mutant advanced ovarian cancer patients (Tutt et al. 2010). Pre-clinically, olaparib was highly bioavailable in dog with a relatively low clearance compared to hepatic blood flow. Interestingly, rat also showed high bioavailability of 100%, whereas the total plasma clearance was equivalent to 68% of hepatic blood flow, indicating the hepatic first-pass effect was much smaller than anticipated (Menear et al. 2008). Whilst clinical bioavailability data is unavailable, it has been suggested from mass balance studies and physiological-based pharmacokinetic modelling that olaparib also undergoes a limited first-pass effect in humans. As CYP3A mediated metabolism is considered to be the major elimination route of olaparib various clinical trials were undertaken to assess the impact of co-administration with pharmacoenhancers. Clinical trials demonstrated that co-administration with itraconazole resulted in a statistically small but significant increase in olaparib exposure (NCT01900028). In contrast, the exposure of olaparib decreased when dosed in combination rifampicin (NCT01929603). In both cases, co-administration of itraconazole or rifampicin affected exposure, however neither altered the terminal half-life of olaparib. This may indicate that the first-pass effect was predominantly altered by either induction or inhibition, rather than impacting systemic clearance (Dirix et al. 2016). Evaluation of clinical pharmacoenhancement of olaparib by co-administration with cobicistat is currently being undertaken (NCT05078671), to improve exposure, tolerance and cost-effectiveness.

Dextromethorphan

Dextromethorphan, a sigma-1 receptor agonist and uncompetitive N-methyl-d-aspartate (NMDA) receptor antagonist 2, has been used as an antitussive in the clinic since the late 1950s. Dextromethorphan was discovered to have neuroprotective activities in preclinical models and as such could be beneficial to patients in a number of CNS disorders (Werling et al. 2007). However, it is highly metabolised by CYP2D6-catalysed O-demethylation to dextrorphan (von Moltke et al. 1998). The selectivity and specificity of this route of metabolism has led to the adoption of dextromethorphan as a probe substrate for CYP2D6 activity both in vitro and in the clinic (Schadel et al. 1995). A drawback of employing pharmacoenhancers is the impact of genetic CYP polymorphisms, which is one of the factors that can contribute to pharmacokinetic variability in drug exposure in any given patient population. CYP2D6 is a well-known polymorphic isozyme of the CYP superfamily with individuals being categorised on a spectrum from poor to ultrarapid metabolisers (Rüdesheim et al. 2022; Gaedigk et al. 2008). Therefore, predicting the response of a patient to a given dose of dextromethorphan can be challenging unless the CYP2D6 polymorphic status of the patient is known pre-treatment. As such, extra consideration has to be given to phenotyping patient populations when employing these specific strategies. Furthermore, adverse events have been attributed to the metabolite, dextrorphan and as such inter-individual variability in the clinical tolerability of dextromethorphan can be observed.

Whilst CYP2D6-mediated metabolism is key to the clearance of dextromethorphan, other mechanisms such as CYP3A metabolism are involved in the drug’s clearance and may also contribute to patient population variability in efficacy and toxicological responses. Clinical interaction studies have shown dextromethorphan exposure can increase following grapefruit juice ingestion, which is indicative of the role in intestinal CYP3A in the first-pass effect (Strauch et al. 2009).

Thus far only CYP3A4 inhibitors as pharmacoenhancers have been discussed in this review, however other CYP inhibitors are employed. Co-administration with CYP2D6 inhibitors, such as bupropion and fluoxetine, to enhance exposure of dextromethorphan have been utilised in many combinations (Stahl 2019; Bent et al. 2021). Quinidine (Figure 3), an antiarrhythmic drug and potent CYP2D6 inhibitor, combined with dextromethorphan, was launched by Avanir Pharmaceutics as Nuedexta® for the treatment of Pseudobulbar affect (PBA) (Cruz 2013). A dose of 10 mg quinidine is sufficient to inhibit the CYP2D6-mediated metabolism of dextromethorphan and enhance exposure. Following single and repeated combination doses of dextromethorphan 30 mg/quinidine 10 mg, subjects had an approximately 20-fold increase in dextromethorphan exposure compared to dextromethorphan dosed alone (Cruz 2013). In addition, the combination resulted in the increased exposure of dextromethorphan only, as quinidine is metabolised by CYP3A4, and dextromethorphan has no CYP inhibitory activity. However, care must be taken when Nuedexta® is administered to patients on co-medication with other CYP2D6 substrates such as paroxetine to avoid potential adverse events, again highlighting the inherent complexities in multiple drug therapies. Whilst improvements in exposure have driven the co-administration of dextromethorphan with CYP2D6 inhibitors, blocking the formation of dextrorphan also improves tolerability, as this metabolite is responsible for the key side-effects of dextromethorphan, and is an example of the additional benefits of employing pharmacokinetic boosters.

Figure 3. The structure and inhibitory activity of quinidine on CYP2D6, CYP3A4 and P-glycoprotein (data from Galetin et al. 2002; Hutzler et al. 2003; Morrissey et al. 2012).

Chemotherapeutics

Efflux mechanisms, which are fundamental processes involved in the pharmacokinetics of drugs, were first identified as resistance mechanisms developed by tumours on exposure to chemotherapeutics (Kukal et al. 2021). The transporters responsible for drug efflux, in particular the multidrug resistance proteins (MDR), the breast cancer resistance proteins (BCRP) and p-glycoprotein (P-gp) have been extensively characterised. These transporters are expressed in multiple tissues and play a critical role in limiting the oral exposure of drugs and in controlling drug penetration of the blood–brain barrier (BBB). Chemotherapeutics, such as taxanes and anthracyclines, are substrates of efflux transporters. Therefore, oral exposure, development of resistance, and the brain acting as a sanctuary for metastases can be an issue for chemotherapies. The oral exposure of docetaxel and paclitaxel have shown to be enhanced in clinical trials on co-administration with cyclosporin A, a strong P-gp inhibitor (Meerum Terwogt et al. 1999; Malingré et al. 2001). However, clinical trials have had limited success with existing drugs repurposed as P-gp inhibitors, such as cyclosporin A and verapamil, due to tolerability of the combinations (Ferry et al. 1996). Whilst more recent clinical studies have employed the use of drugs specifically developed to inhibit P-gp, such as elacridar, tariquidar and zosuquidar, these have not resulted in combined drug approvals due to tolerability, limited improvements in exposure or efficacy (Sandler et al. 2004; Pusztai et al. 2005; Kelly et al. 2011). In an attempt to address tolerability issues, the development P-gp inhibitors with reduced permeability to limit systemic exposure resulted in the discovery of encequidar (Smolinski et al. 2021). Early clinical trials demonstrated increased oral exposure of P-gp substrate paclitaxel in combination with encequidar (Kwak et al. 2010; Ma et al. 2022). However, Phase 3 trials in patients with metastatic breast cancer have yet to lead to NDA approval (NCT02594371). In the face of significant challenges in adopting a pharmacoenhancer approach for taxanes, other groups have opted to develop semisynthetic oral taxanes, as new tools to overcome innate or acquired P-gp-mediated taxane resistance (Distefano et al. 1997). Additionally, other promising approaches are being undertaken such as the discovery of dual P-gp and CYP3A4 inhibitors, where targeting concurrent inhibitions may improve oral bioavailability of their common substrates, while reducing side-effects driven by pan inhibition. These novel dual inhibitors require further biological and physiochemical optimisation in order to achieve sufficient intracellular concentrations to elicit the desired pharmacological in vitro and in vivo response. (Urgaonkar et al. 2022).

Decitabine

Whilst many pharmacoenhancer examples involve modulation of CYP or efflux-mediated mechanisms, the application of PK boosters extends beyond CYP and P-gp inhibition. Decitabine, a cytidine antimetabolite DNA hypomethylating agent and an inhibitor of DNA methyltransferase, is used in the treatment of myelodysplastic syndromes (MDS) and acute myeloid leukaemia (AML), a heterogeneous group of haematopoietic cancers (Gore et al. 2006). Decitabine exhibits very low oral exposure as it undergoes a significant first-pass effect driven by rapid deamination catalysed by cytidine deaminase (CDA), an enzyme highly expressed in the gut and liver. Patients are therefore required to attend clinic to receive therapy as an intravenous (IV) infusion, typically on multiple days within a treatment cycle. Hence, the development of oral inhibitors has been an area of active interest. Combination therapy with the CDA inhibitor, cedazuridine, was developed by Astex Pharmaceuticals, to allow the oral delivery of decitabine. After preclinical proof-of-concept was demonstrated in vivo in mouse and non-human primate (Oganesian et al. 2013), the first-in-human dose escalation clinical trial was undertaken, assessing different dose combinations, guided by pharmacokinetic and pharmacodynamic observations (Savona et al. 2019). On concomitant administration at a relatively low fixed dose of 35 mg decitabine with 100 mg cedazuridine, the bioavailability of decitabine was greatly enhanced and enabled the oral combination to achieve AUC exposures similar to parenteral IV form of decitabine (20 mg/m², 1 h infusion). Based on these phase 1 results, a phase 2 trial was initiated at this dose combination in a randomised crossover design comparing total cycle AUC plasma–time course exposures and pharmacodynamic changes over two cycles between oral decitabine/cedazuridine and IV decitabine (Garcia-Manero et al. 2020). Decitabine/cedazuridine combinations achieved 93.5% and 97.6% of IV total cycle decitabine AUC with separate tablets and fixed-dose combination (FDC) tablets, respectively (Savona et al. 2019). A confirmatory phase 3 trial achieved 98.9% of IV decitabine AUC in a population of subjects with MDS or chromic myelomonocytic leukaemia (CMML) (Garcia-Manero et al. 2019). The combination therapy, now known as Inqovi®, was subsequently approved by the FDA (USA), for the treatment of myelodysplastic syndromes, based on pharmacokinetic AUC equivalence against IV decitabine as the primary endpoint, and with clinical efficacy and safety as secondary endpoints (Kim et al. 2022).

As cytidine deaminases have a limited role in xenobiotic metabolism, the risk of additional drug-drug interactions with this combination are low. The specificity and selectivity of cedazuridine means that tolerability of the combination was in line with that of IV decitabine alone, resulting in an orally administered decitabine drug that reduces patient burden. This case study highlights the potential and benefit of pharmacoenhancers to improve delivery systems to patients.

Pharmacoenhancers in pre-clinical settings

Although this review is focussed on the clinical applications of pharmacoenhancers, the potential applications in pre-clinical settings are worth noting. Pharmacoenhancers can be utilised to achieve adequate exposure in order to explore the pharmacology of early non-optimised compounds in animal models (Amin 2013; Watanabe et al. 2016). As an example, 1-aminobenzotriazole (ABT) is a non-specific CYP inhibitor which has been widely employed to manipulate the in vivo exposure of co-administered compounds in pre-clinical studies. In addition to enhancing parent compound exposure, ABT reduces the exposure of metabolites and can therefore help to identify the contribution of parent versus metabolites to either the pharmacological activity (Capello et al. 1990) or toxicity (Leong et al. 1997). Co-administration of ABT can also indicate whether metabolism or absorption is limiting oral exposure (Caldwell et al. 2005). However, ABT also delays gastric emptying confounding the interpretation (Stringer et al. 2014). Thus, as in the clinical setting, pharmacoenhancers can be a useful but complex tool that needs to be carefully considered if required. Furthermore, due to the differences in CYP isozymes expression and substrate specificities in experimental animals compared to humans, the translatability of pharmacoenhancer effects from pre-clinical to clinical research can be challenging (Hammer et al. 2021).

Evolving DMPK strategies in drug discovery

Given the inherent risks in employing pharmacoehancement strategies, such as toxicities associated with DDIs and variability in patient responses, driven by factors such as genetic polymorphisms and gender differences in CYP expression, it is our opinion that the main focus within the pharmaceutical industry should remain the discovery and development of new chemical entities, with optimal drug-like properties for target engagement and efficacy. Instead of co-administration, for example, with CYP3A4 inhibitors, medicinal chemistry and drug metabolism and pharmacokinetic (DMPK) teams, should be striving to optimise structural design to reduce metabolic liabilities, with close inter-discipline discussions and rationalisation of screening cascades. A case example is the discovery of tolinapant, a dual cIAP/XIAP antagonist, which was influenced in part by the optimisation of CYP3A4 metabolism (Tamanini et al. 2017; Johnson et al. 2018). The species differences in the pharmacological response to cIAP antagonism prevented any pharmacokinetic or toxicological evaluation in dogs. Limiting the selection of higher species for preclinical evaluation to non-human primate highlighted the need for reduced CYP3A metabolism to avoid low exposure driven by intestinal and hepatic first pass effects. Introduction of an in vitro CYP3A4 phenotyping assay during lead optimisation was part of the strategy in the discovery of tolinapant (Johnson et al. 2018; Ward et al. 2018).

The ability of in vitro ADME assays to impact design will increase where the output has a direct relationship to the in vivo setting. Furthermore, low predictability of in vitro intrinsic clearance measurements to in vivo clearance is likely to result in an increased number of sub-optimal compounds being evaluated in nonclinical species. Numerous in vitro assays have been adopted for the determination of hepatic metabolic intrinsic clearance with microsomes and suspension hepatocytes being the most common in vitro systems employed. Hepatocytes are viewed as the ‘gold standard’, yet generic microsomal assays are run routinely as the primary screen due to their compatibility with high throughput screening and automation. Activity of CYP and UGT isozymes is impacted by microsomal incubation conditions, therefore, the predictability of this system will be dependent on the conditions employed and the substrate specificity of the compounds being evaluated (Rowland et al. 2007; Manevski et al. 2013; Palacharla et al. 2017). During lead optimisation, the key mechanisms which will contribute to the human clearance of a chemical series or individual compound are unknown, therefore fine-tuning experimental conditions to provide the most predictive IVIVE in preclinical species is an approach that can provide reassurance of the likelihood of human intrinsic clearance measured under similar conditions to predict the outcome in the first in human trial.

Drug discovery, in particular to ADMET, is constantly evolving and the improvements and innovations in ADMET screening have been extensively reported (Shou 2020). Whilst these changes have increased the ability to generate larger datasets with reduced turnaround times, it is our opinion that these developments could reduce the opportunities for serendipitous observations by a DMPK scientist that could lead to more detailed understanding of compound behaviour. The disconnection between data generation and the DMPK scientist could be, in part, responsible for the reported reduction of certain expertise within the DMPK community (Kramlinger et al. 2022). Automation has significantly changed how in vitro ADMET assays are conducted, with increased throughput and miniaturisation to reduce the supply burden on compound and reagents. Bioanalytical technologies, such as ballistic or rapid-fire chromatography and mass spectrometry detector technology, have also contributed to improvements in assay sensitivity and speed (Wu et al. 2012). Furthermore, the data transfer infrastructure allows for seamless raw data generation to parameter calculation, database reporting and analysis using sophisticated visualisation tools for SAR analysis. These developments have provided a vast array of data at the fingertips of medicinal chemists.

The manner in which we perform drug discovery has not only changed in the technologies used but also in the business models. The use of contract research organisations (CROs) for studies has expanded significantly and they have an integral role in the drug discovery process (Steadman 2018; DeCorte 2020). In particular, in vitro ADME assays, namely intrinsic clearance in hepatic microsomes or hepatocytes and permeability/efflux assays using Caco-2 or MDCK cell monolayers, are staples offered by CROs. Provision of data from these assays has also required the development of accelerated generic LCMS methodologies appropriate for the throughput and diversity of compound chemical properties likely to arise from working with multiple organisations (Xu et al. 2002; Badman et al. 2010; Ye et al. 2011). Furthermore, whilst some CROs offer flexible assay conditions, typically each CRO has a standardised assay format to allow the most efficient delivery of data to a broad range of clients and collaborators. Standardisation of assays has also been driven internally within pharmaceutical companies with centralisation of resources for high-throughput assays which have provided opportunities to reduce the design-make-test-analyse (DMTA) cycle (Wernevik et al. 2020). Whilst this approach provides opportunities to reduce this cycle, the distancing of the data generation from the DMPK scientist has the potential to reduce the fundamental understanding of a chemical series properties and behaviour that are not reported in the transferred data. Discussions with the medicinal chemistry teams in interpretation and incorporation of the results into new design ideas are a critical part of the drug discovery process, and a detailed knowledge of the way a compound behaves in assays has the potential to enhance the input from DMPK scientists (Plowright et al. 2012). As in the case of tolinapant, the close inter-discipline interactions enabled modification of the cascade in response to data and rapid DMTA cycles.

Concluding remarks

Many publications have evaluated the changes in practices and technology developments that have led to DMTA cycle improvements including data systems to allow seamless data transfer between collaborators (pharma, CROs and academic groups) (Ballard et al. 2012). However, the direct interaction between medicinal chemist and DMPK scientist is scarcely mentioned. Indeed, it has been suggested that the responsibility to improve the pharmacokinetic properties and tolerability of a drug candidate falls on the shoulders of medicinal chemistry groups (Kerns 2013). However, the drug discovery industry acknowledges that success relies on multifaceted activities within multidisciplinary teams, with emphasis on inter-disciplinary communication as a key aspect to ensure a productive drug pipeline and the reduction of late-stage attrition.

The speed at which paxlovid has been developed in response to a global pandemic is a demonstration of how far the pharmaceutical industry has advanced and can react to immediate health needs in the face of a pandemic. However, the requirement of pharmacoenhancers is an indication of where drug discovery challenges remain. Recent applications of pharmacokinetic boosting combinations, such as in the case of osimertinib and olaparib, are intended to reduce the cost of the active drug required in the therapeutic regimen rather than improve suboptimal pharmacokinetics. While both cost reduction and improved pharmacokinetics are admirable goals, it is important to consider the impact of unintended drug–drug interactions from pharmacoenhancers. The potential clinical burden and cost of controlling drug–drug interaction risks of pharmacoenhancers with other co-medications should also be considered when a PK boosting strategy is employed. Additional complexities such as patient disease states, which can impact CYP gene expression and/or activity, may require consideration in the clinical setting (Kacevska et al. 2008; du Souich and Fradette 2011).

Polypharmacy is also inevitable with an ageing population and many diseases have common aetiologies and/or demographics, for example, smoking and obesity are risk factors for both cardiovascular disease and cancers. Patients are therefore likely to be on several medications for pre-existing conditions before presenting with new diseases. Clinical teams will be presented with the need to consider potential inadvertent drug–drug interactions alongside the deliberate effect of a pharmacoenhancer combination. An extensive review has shown that drug–drug interactions with chemotherapeutics can lead to a reduction in patient survival (Sharma et al. 2019). Furthermore, a retrospective analysis indicated that approximately 40% of drug-related deaths in a university hospital were attributable to drug–drug interactions (Montané et al. 2018). Clinical therapeutic drug monitoring has been established as an approach for dose tailoring to an individual patient, which is particularly important for drugs with a narrow therapeutic index. This activity requires a multidisciplinary team including DMPK scientists, clinicians, nurses and pharmacists to work in concert to ensure best practices are achieved (Kang and Lee 2009).

Pharmacoenhancers clearly have a role to play and have significantly benefitted patients with improved therapies and may be a favourable approach over novel drug design in those situations where urgent treatments are required, as in in the case of COVID19, or in disease populations with unmet clinical needs. However, considering the additional complexity in the clinical setting, DMPK scientists along with their medicinal chemistry colleagues should continue to evolve drug discovery strategies based on a detailed mechanistic understanding of chemical series properties and new technologies to deliver drugs with optimal human pharmacokinetics.

Disclosure statement

This commentary is based on the views of the authors and not necessarily the views of Astex Pharmaceuticals.