What are the clinical implications for a drug with a narrow therapeutic index?

Warfarin

Warfarin is a narrow therapeutic index drug where individual daily dose requirements vary by at least 40-fold. Inability to maintain an INR between 2 and 3 can predispose to either thrombosis (INR<2) or haemorrhage (INR>3). Warfarin is metabolised by various P450 enzymes, the most important being CYP2C9. Two variants in the CYP2C9 gene (termed CYP2C9*2 and CYP2C9*3) reduce the activity of the enzyme and overall rate of the metabolic turnover of warfarin. The mode of action of warfarin is through interruption of the vitamin K cycle specifically by inhibiting the enzyme vitamin K epoxide reductase complex 1 (VKORC1) – variation in this gene can affect the daily requirements for warfarin. In most global populations, it has now been shown that age and body mass index, together with genetic variation in CYP2C9 and VKORC1, can account for at least 50% of the variation in daily dose requirements for warfarin. This has resulted in the development of dosing algorithms in an attempt to improve the accuracy and prediction of individual dose requirements for warfarin.

5 What are some of the common drug interactions with warfarin?

Because of its complex metabolism and relatively narrow therapeutic index, drug interactions involving warfarin are both common and clinically significant. The majority of the interactions occurs at the level of CYP metabolism, although other mechanisms may be possible. Table 66.2 lists some of the more common interactions.

For interactions listed as the causation being probable or highly probable, the strength of the clinical evidence is sufficient to expect a change in INR, requiring either adjusting the warfarin dose up or down by 25% to 50% in anticipation of the resulting change in INR or frequent INR monitoring to determine the dose adjustment necessary.

Foods and supplements may also alter the INR. Foods that increase the INR include garlic, mango, ginseng, grapefruit juice, and possibly cranberry juice. Ginger and fish oil have additive antithrombotic effects. Foods and supplements that reduce the INR include high vitamin K–content foods, enteral feeds, and soy milk.

46.4.2.2 Warfarin

Warfarin is the most widely prescribed anticoagulant for the prevention of thromboembolic events. Because it is a narrow therapeutic index drug, routine therapeutic drug monitoring (international normalized ratio, INR) is required for dose adjustments to avoid adverse drug reactions. Warfarin’s dose requirements exhibit interindividual variability and are influenced by both genetic and nongenetic factors. Pavani and coworkers developed an algorithm using a multiple linear regression model that includes Vitamin K intake, CYP2C9 (*2 and *3), and VKORC1 (*3, *4, D36Y and −1639 G>A). Using this algorithm, they explained the 44.9% variability in warfarin dose requirements [48]. Their following extended investigation explained the 61% variability [49].

Very recently, a new pharmacogenetic algorithm was developed to predict maintenance and starting warfarin indose in SI patients. This study explained a similar variability of ∼62.1% in dosing requirements. The investigators reported that CYP2C9*1/*2, *1/*3, and *2/*3 variant genotype carriers require a lower daily maintenance dose—2.8 mg, 2.3 mg, and 2.2 mg, respectively—than do carriers of the CYP2C9*1/*1 genotype (5.2mg). The CYP2C9, VKORC1, and GGCX gene polymorphisms were associated with reduced dose, excluding the polymorphisms rs7196161, rs7294, and rs2108622, which were associated with an increased dose [50]. In 145 NI individuals, VKORC1-1639 AA and CYP2C9 *3/*3 homozygous mutants were associated with risk of over anticoagulation and warfarin-related adverse events [51]. A significant effect of CYP2C9 (*2 and *3) and VKORC1 (C1173T and G-1639A) genotypes on warfarin dose was observed by Shalia and colleagues in 83 NI patients operated on for aortic/mitral valve replacement [52].

4.4.3 Herb-Drug Interaction

HDIs are significant safety concerns in clinic. Many literatures have reported that herbs could interact with several clinical narrow therapeutic index drugs, including methotrexate (Liu et al., 2014d; Wang et al., 2014), anticoagulants (e.g., warfarin), immunosuppressant drugs (e.g., tacrolimus and cyclosporin), anti-HIV agents (e.g., indinavir and saquinavir), cardiovascular drugs (e.g., digoxin), and anticancer agents (e.g., docetaxel, irinotecan, and imatinib). Experimental determination of the absorption and disposition properties of herbal medicine, especially TCM constituents, is attracting more research groups worldwide (Wu et al., 2012). Multicomponent herbs are subject to sequential metabolism, concurrent metabolism, and multiple metabolism in vivo. Data on the interaction between Ranunculaceae herbal medicine and DMEs/transporters are accumulating. It should be highlighted that the HDI is a double-edged sword, given that the mild HDI could alleviate the metabolic clearance of the coadministered drugs and increase their AUC and half-life (T1/2), which might be good for their in vivo therapeutic effects, especially those with a relatively wide therapeutic window (Meng and Liu, 2014; Wang and Liu, 2014).

Herbal supplements are broadly used in cancer patients, but how they affect the chemotherapy is frequently undisclosed. Black cohosh was a stronger inhibitor than St. John’s wort and ginger root extract, for both CYP- and carboxylesterase (CE)-mediated biotransformation of tamoxifen and irinotecan, respectively (Gorman et al., 2013). Eight triterpene glycosides of black cohosh proved to be competitive CYP3A4 inhibitors with IC50 values ranging from 2.3 to 5.1 μM (Li et al., 2011a), while the alkaloids protopine and allocryptopine of black cohosh, also abundant in other Ranunculaceae genera (e.g., Coptis, Thalictrum, and Aquilegia), were competitive CYP2D6 inhibitors.

The microsomal CE catalyzes the irinotecan bioactivation to SN-38, a topoisomerase I inhibitor. However, the role of multiple CE isozymes is not known. A highly selective ratiometric fluorescent probe of human CE1 has been developed for in vitro monitoring and cellular imaging (Liu et al., 2014e). Two highly selective fluorescent probes can be used for the detection of hCE2 (Feng et al., 2014a,b). These innovative specific probes are highly valuable for real-time monitoring of hCE activity in complex biological systems and provide novel solutions for HDI studies (Fig. 4.3).

Of note is that BBR-containing Ranunculaceae herbs are involved in HDIs. For instance, goldenseal (H. canadensis) sizably inhibited CYP3A (Gurley et al., 2008a). AUC0–∞(107.9 vs. 175.3 ng•h/mL), T1/2(2.01 vs. 3.15 h), and Cmax (50.6 vs. 71.2 ng/mL) of midazolam increased, while total body clearance (CL) of midazolam decreased (1.26 vs. 0.81 L/h/kg). The simultaneous intake of goldenseal and CYP3A substrates may result in noteworthy HDIs. Goldenseal, but not the black cohosh extracts, significantly inhibited (∼50%) CYP2D6 activity (Gurley et al., 2008b). Goldenseal inhibited CYP2E1 most potently, followed by 1A2, 2D6, and 3A (Yamaura et al., 2011). Since CYP2E1 metabolizes acetaminophen (APAP) to the highly active intermediate, goldenseal could ameliorate APAP-induced acute liver failure.

Various amounts of BBR did not significantly alter the hepatic function of mice (Guo et al., 2011), and repeated use of the lower doses of BBR for 2 weeks had no influence on the gene expression of more than 20 main Cyps. However, the highest dose of BBR (300 mg/kg) downregulated Cyp3a11 and 3a25 expression by 67.6% and 87.4%, respectively, while Cyp1a2 (for 7-ethoxyresorufin O-dealkylation) mRNA was increased by 43.2%, and Cyp3a11 (for testosterone 6β-hydroxylation) and 2d22 (for dextromethorphan O-demethylation) activities decreased by 67.9% and 32.4%, respectively. The gene expression and enzyme activity of Cyp2a4 (for testosterone15α-hydroxylation), 2b10, and 2c29 (both for testosterone 16β-hydroxylation) were not altered. Lower dose BBR might not result in DDIs. However, high-dose BBR may reduce Cyp activities and cause DDIs.

Compared with ciprofloxacin alone, comedication of BBR (50 mg/kg) and ciprofloxacin significantly decreased Cmax of ciprofloxacin (Hwang et al., 2012). The pretreatment of BBR (50 mg/kg/day) and BBR-containing Huang-Lian-Jie-Du-Tang (HR; 1.4 g/kg/day) significantly decreased Cmax and AUC0→∞ of ciprofloxacin, as compared with the control group. P-gp and OCT (organic cation transporter) could be involved in reduced oral bioavailability of ciprofloxacin by BBR and HR.

Jiao-Tai-Wan (JTW), consisting of CC and Cinnamomum cassia, efficiently guarded the pancreatic islet morphology, enhanced the activation of hepatic AMP-activated protein kinase (AMPK), and upregulated the expression of glucose transporter 4 (GLUT4) in white fat and skeletal muscle (Hu et al., 2013). Thus, BBR-involved DDIs might also be mediated by transporter superfamily members.

Nigella-related HDIs are also highlighted in recent studies. For instance, Nigella sativa dose-dependently inhibited the gene and enzyme expressions of rat CYP2C11 (Korashy et al., 2015), thus reducing the amount of 4-hyroxy-tolbutamide, a tolbutamide metabolite, in vitro. The inhibitory effect of Nigella on rat CYP2C11 was stronger than that of Trigonella foenum-graecum and Ferula asafoetida, which could result in the undesirable effect of CYP2C11 substrates.

CYP3A4 and to a lesser extent CYP2C9-mediated metabolism of sildenafil could be impacted by Nigella (Hyland et al., 2001). Oral administration of N. sativa resulted in reduction of AUC0–∞, Cmax, and T1/2 as compared with the control (Al-Mohizea et al., 2015). Concurrent use of Nigella alters the PK of sildenafil, which might result in a decrease in sildenafil bioavailability. In rabbits, the concurrent use of N. sativa significantly decreased the Cmax and AUC0–∞ of CsA, a commonly used immunosuppressant (Al-Jenoobi et al., 2013). The aqueous extract of N. sativa dose-dependently inhibited sodium-dependent glucose transport through rat jejunum (Meddah et al., 2009). On the contrary, methanol and hexane extracts of Nigella seeds enhanced amoxicillin availability in both in vivo and in vitro studies (Ali et al., 2012). Nigella might increase intestinal absorption of amoxicillin.

Does AC inhibit/induce CYP3A? In one study, the production of 1-(2-pyrimidinyl)piperazine (PP) and 6′-hydroxybuspirone from the probe substrate buspirone (BP) did not change, suggesting that the rat CYP3A activity was not impacted by the single and repeated use of 0.125 mg/kg AC (Zhu et al., 2013). In RLMs, one-week AC pretreatment did not affect CYP3A protein levels. Therefore, the authors claimed that AC does not inhibit or induce CYP3A in rats and might not lead to CYP3A-associated DDI in the liver. However, in another study, multiple AC exposure (0.125 mg/kg) increased the AUC0–∞ of BP by 110% (Lijun et al., 2014), and the amounts of 1-PP and 6′-OH-BP were increased by 229% and decreased by 95%, respectively. Single/multiple AC exposure did not alter the first-pass (intestinal and hepatic) CYP3A activity when using oral BP as a probe in rats. Nonetheless, whether multiple AC exposure prominently changes the production of BP metabolites warrants further in vivo studies.

Most studies highlighted the impact of herbal medicine on Western drugs, but not vice versa. The possible reasons are: (1) the constituents of herbal medicine are too complicated and their effects are versatile; therefore, it is challenging to select the appropriate PK and PD markers of herbal medicine; and (2) the bioactivity and systemic exposure of the single ingredients of herbal medicine is very often moderate, and there is a lack of rapid and strong potency. In HDI studies, special attention should be given to the effects of dose, regimen, and mode of medication, since empirically more HDIs occur only under high-dose or long-term administration.

To date, information is still lacking for the main CYP and UGT enzymes in the less-studied medicinal plants. Information of P-gp and other drug transporters is also limited (section below, Phase III: Drug Transporter). The role of another phase I DME flavin-containing monooxygenase (Hao et al., 2007; Hao and Xiao, 2011) in HDIs remains elusive. ABC transporter and solute carrier (SLC) (Hao et al., 2013d) superfamilies have many other transporters, besides P-gp, OCT, and GLUT, which await future investigations. To date, there are very few systematic methods for quantitative forecast of the scale and probability of herb-drug interactions. Physiologically based pharmacokinetic (PBPK) modeling could be used to increase prediction correctness of possible HDIs (Brantley et al., 2014), but the premise is that abundant in vitro information for building such quantitative relationships is available in the near future (Azam et al., 2014). The DDI prediction of Ranunculaceae shares the same challenges and problems as that of other CMM. For instance, very often there is more than one inhibitor of the same DME. The integrative effects of the sum of the multiple weak inhibitions are sometimes considerably high. It also should be noted that HDIs are always complex, as each herb contains many ingredients that may simultaneously interact with multiple targets. To date, it is not realistic to predict the potential HDI by IVIVE (in vitro–in vivo extrapolation) and PBPK modeling, as many key parameters, for example, the plasma concentration of each component, the unbound fraction, the inhibitory activity of metabolites, and the half-life of each inhibitor, are absent.

3.5.5 Others Drug Classes

In the case of antibiotics, TDM is vital for vancomycin and three specific aminoglycosides (amikacin, gentamycin, and tobramycin) due to the nephrotoxicity and ototoxicity exhibited by these drugs and their narrow therapeutic drug windows [20,113]. Clinically, TDM is carried out by immunoassays on account of high throughput and automation. However, with antibiotic drug resistance becoming commonplace, patients are on multiple antibiotic regimens. In such cases, multiplexed LC-MS/MS methods are highly beneficial in comparison to immunoassays [113–115]. For antituberculosis drugs, TDM is typically not recommended because of the broad-spectrum nature of these drugs and prolonged treatment regimens [20]. Regarding antiretroviral drugs, although TDM is not currently in clinical practice, it is suggested that TDM can greatly benefit in studying patient adherence [21].

1.01.6.4 Avermectin

Avermectin fulfilled its promise as a cost-effective animal health anthelmintic. Like every other biologically active compound, it causes side effects. Had avermectin been a drug for humans, one could have launched a major effort to improve on the drug's therapeutic index, regardless of the cost. I remember discussing that issue with the late Dr Michael H. Fisher, an outstanding, softly spoken scientist who was responsible for the chemistry of animal health research at Merck. Given that avermectin has five double bonds, I was not very optimistic that we would be able to increase safety in a cost-effective way. Dr Fisher's more optimistic outlook was fully vindicated by subsequent developments. The medicinal chemists succeeded in selectively reducing the C 22–23 double bond in good yield. The resulting dihydroavermectin (ivermectin) was actually slightly less potent than avermectin B1, but, to our delight, it displayed a therapeutic index superior to that of its natural product precursor.

28.2.1 Drug Level Monitoring—Calcineurin Inhibitor Levels

CNIs are currently the standard of care for immunosuppression in most kidney transplant recipients. Cyclosporine (CsA) and tacrolimus have dramatically improved short-term allograft outcomes since their adoption into clinical use. Under-dosing has been associated with higher rates of acute rejection and over-dosing increases the risk of electrolyte disturbances, metabolic derangements, and nephrotoxicity. Given the narrow therapeutic index, drug monitoring is vital with CNIs. The pharmacokinetics of CNI can be inconsistent and dependent on a number of variables such as presence of meals, variability of gastrointestinal motility (i.e., diarrhea), concurrent usage of medications affecting CYP450 3A4 activity, and decreased renal function.12–15 Due to inter- and intrapatient variability, drug monitoring allows for individualization of drug dosing in order to ensure efficacy and limit toxicity. Although there have not been randomized control trials comparing the outcomes for monitoring and not monitoring, it is generally accepted that drug monitoring is considered favorable.

b Introduction to the Therapeutic Window

Therapeutic window and the closely related concept, therapeutic index, are sometimes taken into account when determining dosing. The therapeutic window refers to the drug dose needed, where there is a pressing need to maintain “exposure” within a range that is effective and yet avoids undue AEs. Therapeutic window is taken into account when assessing the need for dose modification under circumstances such as patient variability, the influence of food on exposure, and the influence of a coadministered drug on exposure. The term “exposure” usually refers to a parameter of drug concentration in blood, such as AUC, Cmax, tmax, or Cmin.

FDA’s Guidance for Industry on bioequivalence defines, “narrow therapeutic range drug products as those containing … drug substances that are subject to therapeutic drug concentration or pharmacodynamic monitoring … or where product labeling indicates a narrow therapeutic range designation. Examples include digoxin, lithium, phenytoin, theophylline, and warfarin.”10 FDA’s Guidance for Industry on drug interactions provides further information on drugs with a narrow therapeutic range or window.11 The cited references describe drugs with a narrow therapeutic window, where these are amiodarone (cardiac antiarrhythmic drug),12 sunitinib (anticancer drug),13 tacrolimus (immunosuppressant),14,15 and warfarin (anticoagulant).16

A dramatic account of the narrow therapeutic window for warfarin is provided by an account of the Yup’ik native Americans in Alaska, where their warfarin requirement is changed because of a naturally occurring variant in one of their cytochrome P450 enzymes (CYP4F2*3).17,18

Cytochrome P450 is abbreviated as “CYP” and the cytochrome P450 enzymes may be called CYP enzymes or CYP isozymes. The CYP enzymes occur as various isozymes, such as CYP1A2, CYP2B6, CYP2D6, and CYP4F2. CYP4F2 catalyzes the hydroxylation of the phytyl side chain of vitamin K, thus leading to the degradation of vitamin K and consequent impaired blood clotting. But a genetic variant of CYP4F2, namely, the variant known as CYP4F2*3, does not much catalyze the degradation of vitamin K. Thus, where a person’s genome encodes CYP4F2*3 instead of the wild-type CYP4F2, the consequence is elevated levels of vitamin K in the body. When a patient with CYP4F2*3 needs to be treated with the “blood thinner” warfarin, the warfarin dose must be increased to counteract the greater stores of vitamin K in the body. In short, because of the narrow therapeutic window of warfarin, any increase in the amount of warfarin's target (vitamin K) in the body, will need a corresponding change in the amount of warfarin administered to the patient. To provide scientific background, warfarin inhibits epoxide reductase and prevents the recycling of vitamin K in the body.

Klein19 provides an example of a drug with a narrow therapeutic window and of the goal of maintaining plasma concentration to achieve efficacy and to avoid AEs. The drug is theophylline, a drug for treating airway obstruction. Optimal therapeutic serum levels are 8–15 μg/mL. AEs, such as nausea, vomiting, abdominal cramps, cardiac arrhythmias, and convulsions, can occur where theophylline in serum is >25 μg/mL.

To ensure that the blood concentration is maintained within the narrow concentration range, theophylline is provided in an extended release form and not as an immediate release form.

What will happen if the therapeutic index value is smaller?

What does it mean when a drug has a narrow therapeutic index?

What is therapeutic drug monitoring and its clinical significance?

What is a major disadvantage of the therapeutic index?