Nonalcoholic Steatohepatitis in Precision Medicine: Unraveling the Factors that Contribute to Individual Variability

Abstract

There are numerous factors in individual variability that make the development and implementation of precision medicine a challenge in the clinic. One of the main goals of precision medicine is to identify the correct dose for each individual in order to maximize therapeutic effect and minimize the occurrence of adverse drug reactions. Many promising advances have been made in identifying and understanding how factors such as genetic polymorphisms can influence drug pharmacokinetics (PK) and contribute to variable drug response (VDR), but it is clear that there remain many unidentified variables. Underlying liver diseases such as nonalcoholic steatohepatitis (NASH) alter absorption, distribution, metabolism, and excretion (ADME) processes and must be considered in the implementation of precision medicine. There is still a profound need for clinical investigation into how NASH-associated changes in ADME mediators, such as metabolism enzymes and transporters, affect the pharmacokinetics of individual drugs known to rely on these pathways for elimination. This review summarizes the key PK factors in individual variability and VDR and highlights NASH as an essential underlying factor that must be considered as the development of precision medicine advances. A multifactorial approach to precision medicine that considers the combination of two or more risk factors (e.g. genetics and NASH) will be required in our effort to provide a new era of benefit for patients.

1. Introduction

Precision medicine is an initiative to provide individualized medical care to each patient based on a set of predetermined factors in individual variability. This initiative has been embraced by many in academic, pharmaceutical, clinical, and governmental settings, and there are several prominent success stories where precision medicine has reached clinical practice (Ritchie, 2012). One of the main goals of precision medicine is to ensure that the correct dose of drug is given to each individual patient to maximize therapeutic benefit and minimize adverse drug reactions (ADRs) (Figure 1) (Ritchie, 2012). ADRs are classified into multiple types based on the underlying mechanism(s) for toxicity. The most common type of ADRs, the type A ADRs, are dose-related and are directly relevant to the goal of administering the correct dose to each patient (Edwards and Aronson, 2000). The number of adverse events reported to the FDA Adverse Events Reporting System has increased over the past decade, and since 2012 more than one million ADRs are reported annually. Because of the dose-related nature of ADRs, individual variables that impair any step in the absorption, metabolism, distribution, and excretion (ADME) of drugs can potentially influence drug response. This review summarizes these factors and emphasizes nonalcoholic steatohepatitis (NASH) as an emerging factor that has profound influence on pharmacokinetics (PK) that must be considered in the advancement of precision medicine.

Fig. 1. Precision medicine.

Shown is a population of patients who need treatment for a disease (top). Individuals in blue represent those who will have drug PK similar to what was observed in clinical trials and will be within the range of normal. Some individuals (green) harbor a single nucleotide polymorphism (SNP) in one of the genes responsible for the elimination of the drug resulting in decreased drug exposure and sub-therapeutic effect (A). Another population (yellow) has an underlying disease condition that impairs the elimination of the drug, increases exposure, and the risk of developing an adverse drug reaction and requires dose-adjustment (B).

2. Sources of individual variability in precision medicine

Pharmacogenetics

There are multiple factors that can influence individual variability in PK and ultimately drug exposure. Pharmacogenetics is the most well-known factor that has been a major driving force in the pursuit of precision medicine. Part of the reason genetics has received much of the focus in the early stages of precision medicine is due to the inherent clarity in a DNA sequence; there is little ambiguity in interpreting the genotype once the DNA has been sequenced. Even with this unequivocal nature of genotype, it is clear that SNPs can, but do not always, influence drug response. One explanation for this disconnect is that other factors in individual variability either counteract or overshadow the effect of the SNP in certain populations. This fact emphasizes a larger issue that the potential for genetics to describe the individual variability in drug response has been overstated, whereas the true complexity of the issues involved has been underestimated.

Prior to the completion of the human genome project the NIH Pharmacogenomics Research Network was founded to advance the understanding of how genes influence drug response. The massive amounts of data produced by the PGRN and others led to formation of the Pharmacogenetics and Pharmacogenomics Knowledge Base to help organize these data and make them readily available. Most recently, the Clinical Pharmacogenetics Implementation Consortium was formed to help advance pharmacogenetic data into clinical practice. This effort has been a testament to the power of focused scientific initiatives to rapidly advance a particular field into clinical practice. Several prominent genome-wide association studies (GWAS) have linked particular single nucleotide polymorphisms (SNPs) to the occurrence of certain ADRs (Ritchie, 2012). For example, the GWAS performed by the SEARCH Collaborative Group linked simvastatin-induced myopathy to a SNP in SLCO1B1, the gene encoding the hepatic uptake transporter organic anion transporting polypeptide-1B1 (OATP1B1), is a well-known and often cited success story in the pharmacogenetics era (Link et al., 2008). These data led to a specific recommendation for simvastatin prescriptions by the CPIC and a reduction in the maximum dose allowed (Wilke et al., 2012). Importantly, although pharmacogenetics and GWAS have received considerable attention, it is also clear from these data that genetics can only account for a small portion of the individual variability in PK and drug response (Baye et al., 2011; Sadee, 2013; Wei et al., 2012). Thus, other factors must be contributing, and/or the problem requires a more complex multifactorial approach.

One of the major concepts that can be taken from the success of pharmacogenetics is the central role of drug metabolism and transport genes in VDR (Wei et al., 2012). The large number of enzymes and proteins involved in the ADME of drugs, and the frequent overlap in substrate specificity among these ADME mediators requires that a detailed understanding of the ADME pathway(s) for each drug be examined. This understanding has driven the science of precision medicine towards seeking out and accounting for the various factors in ADME in order to determine individual variability.

Environment

Environmental factors that contribute to individual variability in drug response include variables such as age, diet, hormones, exposure to an exogenous compound (e.g. drug or toxicant), and underlying disease. Since the liver is a major organ in drug ADME processes, liver disease has the capacity to dramatically change the metabolic phenotype with little to no change in genotype. This is evident in the FDA Guidance for Industry and the many published reports that recommend dose adjustments for certain drugs in patients with hepatic impairment (Gandhi et al., 2012; Verbeeck, 2008). There are several well-known examples of drugs with moderate to high hepatic extraction ratio that have increased oral bioavailability in patients with liver cirrhosis such as midazolam, morphine, and verapamil (Verbeeck, 2008). According to the FDA, “despite extensive efforts, no single measure or group of measures has gained widespread clinical use to allow estimation in a given patient of how hepatic impairment will affect the PK and/or pharmacodynamics of a drug”. This is partially due to the varied etiology of different liver diseases that can impact liver blood flow and/or liver dysfunction at the level of drug metabolism and transport. Therefore, it is imperative that accurate diagnoses of disease be made so that correct liver phenotype can be either assessed empirically or attributed to the patient so that the correct drug dose is prescribed. In recent years, NASH is one liver disease for which there has been a considerable accrual of evidence to support its inclusion as an environmental factor in precision medicine (Clarke, et al., 2014a, 2014b; Fisher et al., 2009, 2008; Hardwick et al., 2011, 2012, 2014; N. Li et al., 2004; Lickteig et al., 2007).

3. Why is NASH currently not identified as a key player in individual variability?

NASH is the most severe form of a spectrum of diseases that fall under the umbrella classification of nonalcoholic fatty liver disease (NAFLD). NAFLD has quickly become the most common chronic liver disease in Western society (Ali and Cusi, 2009). In the US adult population, current estimates for the prevalence of NAFLD are between 20% and 40%, with NASH between 5.7% and 17% (Ali and Cusi, 2009; Lazo et al., 2013; McCullough, 2006; Williams et al., 2011). These estimates can vary dramatically based on the method of diagnosis and how the disease is defined (Chalasani et al., 2012; Lazo et al., 2013). A liver biopsy is currently the only reliable method to diagnose NASH and, despite current guidelines that recommend biopsy in NAFLD patients who are at increased risk of steatohepatitis and fibrosis, this procedure is often not performed because it is expensive and carries with it inherent risks of morbidity and mortality (Chalasani et al., 2012; Williams et al., 2011). A recently published study in middle-aged US adults which employed liver biopsy estimated the prevalence of NASH to be 12.2% and the authors suggest that there are millions of asymptomatic Americans with advanced NASH who are unware of the underlying liver condition (Williams et al., 2011). These facts suggest that with such a low percentage of definitive NASH diagnosis, many if not all epidemiology studies are simply insufficiently powered to identify NASH as a factor in the occurrence of adverse drug reactions. Importantly, although a diagnosis of NAFLD may include some patients with NASH, the mechanistic changes in ADME features that are specific to NASH complicate the predictive value of studies that only consider NAFLD without specifying NASH patients (Clarke, et al., 2014b; Hardwick et al., 2010, 2011; Lake et al., 2011).

Although these data indicate that direct connection between NASH and VDR is difficult to assess based on the current data available, there are data linking NASH comorbidities to altered drug PK and the occurrence of ADRs (Hanley et al., 2010; Linares et al., 2009). For example, obesity has been suggested to be a risk factor for altered PK and the occurrence of ADRs associated with administration of morphine and docetaxel. It has been reported that an average of 37% of obese patients undergoing bariatric surgery have NASH, which is higher than the rate observed in the general US adult population (Lazo and Clark, 2008). For both morphine and docetaxel, hepatic dysfunction is well documented to alter PK and requires dose-adjustment to avoid the occurrence of ADRs, potentially due to alteration in hepatic metabolism and transporter function (Bruno et al., 2001; Eckmann et al., 2014; Hasselström et al., 1990; Linares et al., 2009; Ozawa et al., 2008, 2009; Yoon et al., 2012). For docetaxel, several studies have attempted to assess whether doses should be adjusted based on various measures of obesity and body surface area (Griggs et al., 2012; Rudek et al., 2004; Sparreboom et al., 2007). Ultimately, it has been concluded that there is a paucity of information to provide clinical guidelines for dose-adjustment of many anticancer drugs in obese patients (Griggs et al., 2012). Together, these data suggest that other factors besides obesity, BMI, or body surface area, such as a comorbidity (NASH), must be accounting for the difference in drug exposure. A recent study found that in pediatric patients receiving docetaxel, higher BMI was associated with increased renal toxicity and worse overall survival at 5 years compared to patients with normal BMI (Altaf et al., 2013). A letter to the editor in response to these results asserted that “being overweight or obese might merely have flagged other problems which disadvantaged their affected patients.” Importantly, although there is a higher prevalence of NASH in obese patients, obesity does not equal NASH and it is recognized that many other physiological changes are occurring in obesity that could affect drug disposition and toxicity (e.g. volume of distribution). These data are suggestive of an underlying condition such as NASH that can affect drug disposition and toxicity. It is important to point out that in order to gain a deeper understanding of factors such as NASH in precision medicine, there is great potential in the utilization of various approaches that do not require massive, multicenter clinical studies, such as physiologically based pharmacokinetics modeling. This type of modeling permits population level analysis without requiring the exorbitant cost of clinical trials.

4. NASH as another factor in precision medicine

The effects of NAFLD and NASH on the expression and function of drug metabolism enzymes and membrane transporters in rodent models and humans has been extensively reviewed previously (Buechler and Weiss, 2011; Merrell and Cherrington, 2011; Naik et al., 2013). The focus of this review is to highlight specific changes in hepatic and extrahepatic AMDE processes that occur in histologically verified NASH patients. These changes include hepatic phase I and phase II metabolism, hepatic membrane transporter expression, and renal function in NASH patients.

Phase I metabolism

Approximately 12 of the 57 identified cytochrome P-450 (CYP) enzymes are known to metabolize 70–80% of all drugs in clinical use (Zanger and Schwab, 2013). CYP3A4/5, CYP2D6, and CYP2C9, alone account for the metabolism of 30.2%, 20%, and 12.8% of clinically used drugs, respectively (Zanger and Schwab, 2013). The reported NASH-associated changes in the mRNA and protein expressions and enzymatic activity of six of the human CYPs important to drug metabolism are shown in Table 1. In NASH livers the enzymatic activities of CYP2A6 and CYP2C9 are reported to increase, the activities of CYP1A2, CYP2C19, and CYP2D6 all decrease in NASH, and the activity of CYP3A4 does not change (Fisher et al., 2009). For CYP2E1, one study reported that the ex vivo enzymatic activity was not changed in NASH livers, whereas two separate studies using chlorzoxazone as an in vivo probe substrate reported increased enzymatic activity in biopsy-verified NASH patients (Chalasani et al., 2003; Fisher et al., 2009; Orellana et al., 2006). Collectively, these data suggest that the PK of drugs that are dependent on CYP-mediated hepatic metabolic clearance are altered in patients with NASH. Importantly, with the exception of CYP2E1, the enzymatic activity of each CYP isoform has been determined ex vivo from NASH livers. Before clear clinical guidelines can be made for NASH patients, there is a need to perform clinical studies to determine the enzymatic activity of each CYP isoform in vivo.

Table 1.

Cytochrome P450 mRNA expression, protein expression, and enzymatic activity changes in NASH

CYP	mRNA	Protein	Activity	Ref
CYP1A2	↔	↓	↓ (ex)	(Fisher et al., 2009)
CYP2A6	↑	↑	↑ (ex)	(Fisher et al., 2009)
	↓	-	-	(Rubio et al., 2007)
CYP2C9	↔	↔	↑ (ex)	(Fisher et al., 2009)
CYP2C19	↔	↓	↓ (ex)	(Fisher et al., 2009)
CYP2D6	↔	↓	↓ (ex)	(Fisher et al., 2009)
CYP2E1	↓	↓	↔ (ex)	(Fisher et al., 2009)
	-	↑	↑ (in)	(Orellana et al., 2006)
	-	↑	-	(Weltman et al., 1998)
	↑	-	-	(Baker et al., 2010)
	-	-	↑ (in)	(Chalasani et al., 2003)
CYP3A	↔	↔	↔ (ex)	(Fisher et al., 2009)
	-	↓	-	(Weltman et al., 1998)

No change (↔), increase (↑), decrease (↓), not measured (-), ex vivo (ex), in vivo (in)

Phase II metabolism

The conjugation of drugs to charged species such as glutathione, sulfate, and glucuronic acid increases drug solubility and facilitates elimination from the body. These metabolic reactions are mediated by the glutathione S-transferase (GST), sulfotransferase (SULT), and UDP-glucuronosyltransferase (UGT) enzyme families in multiple organs including the human liver. Several groups have investigated the expression and function of these enzymes in NASH and reported some dysfunction of several pathways. The summary of changes in the GST family at the mRNA and protein levels is shown in Table 2. For the GST family, it has been shown that the mRNA and protein expression of GSTA1 is increased in NASH livers (Bell et al., 2011; Hardwick et al., 2010). For the GSTM family the results are variable depending on the isoform. GSTM5 has been reported to have decreased mRNA expression in NASH, whereas GSTM3 mRNA expression was shown to be increased in NASH (Hardwick et al., 2010; Yoneda et al., 2008). For GSTM1 and M4 the results have been contradictory, with one study reporting no change in mRNA expression and others reporting decreased expression (Hardwick et al., 2010; Rubio et al., 2007; Yoneda et al., 2008). One study examined the mRNA expression of GSTM2, M4, and M5 in African American versus Caucasian patients with NASH and found over-expression of the enzymes specifically in the African American NASH patients (Stepanova et al., 2010). The protein expression of the GSTM family is reported to be decreased in NASH livers (Hardwick et al., 2010). Lastly, the mRNA and protein expressions of GSTP1 is increased in livers from NASH patients (Hardwick et al., 2010). To date, only one study has examined the overall GST activity in NASH livers and found an overall decrease in enzymatic activity (Hardwick et al., 2010). For the SULT and UGT family of enzymes there has only been one study showing that hepatic UGT activity was not changed in NASH, whereas SULT activity decreased (Hardwick et al., 2013). Although these data are complicated and difficult to clearly translate into clinical populations, they provide the impetus for further investigation into the GST-and SULT-mediated metabolic capacity in NASH patients.

Table 2.

Glutathione S-transferase mRNA and protein expression changes in NASH.

	mRNA	Protein	Ref
GSTA family		↑	(Hardwick et al., 2010)
GSTA1	↑	-	(Hardwick et al., 2010)
	-	↑	(Bell et al., 2011)
GSTA2	↑	-	(Hardwick et al., 2010)
GSTA4	↑	-	(Hardwick et al., 2010)
	↑	-	(Younossi et al., 2005)
GSTM family		↓	(Hardwick et al., 2010)
GSTM1	↔	-	(Hardwick et al., 2010)
	↓	-	(Yoneda et al., 2008)
GSTM2	↔	-	(Hardwick et al., 2010)
GSTM3	↑	-	(Hardwick et al., 2010)
GSTM4	↔	-	(Hardwick et al., 2010)
	↓	-	(Rubio et al., 2007)
GST M5	↓	-	(Rubio et al., 2007)
GSTP family		↑	(Hardwick et al., 2010)
GSTP1	↑	-	(Hardwick et al., 2010)

No change (↔), increase (↑), decrease (↓), not measured (-)

Hepatic membrane transporters

The membrane transporters located on the sinusoidal and canalicular membranes of hepatocytes are important mediators in the PK of many drugs. Work from our group has shown NASH-specific changes in many of these uptake and efflux transporters that may influence the ability of the liver to eliminate drugs (Table 3). For the uptake transporters, we found that the protein expression of OATP1B1 is increased, OATP1B3 is decreased, and OATP2B1 is not changed in human NASH livers (Clarke, et al., 2014b). Our recently published data in a rodent model of NASH showed that NASH-associated changes in hepatic Oatp transporters was insufficient to alter pravastatin PK but the combination of NASH and genetic loss of a single Oatp transporter caused a synergistic increase in pravastatin exposure (Clarke, et al., 2014b). Together, these data suggest that NASH alone may be insufficient to cause dramatic changes in PK and drug response for drugs dependent on hepatic OATP-mediated uptake. Rather, our data suggest that the combination of a genetic polymorphism in one uptake transporter in conjunction with NASH-associated changes in the expression of compensatory uptake transporters may have profound effects on drug disposition and potentially place these patients at increased risk of ADRs. For the hepatic efflux transporters our data clearly shows that the protein expression of many of the sinusoidal and canalicular transporters is increased in NASH, potentially decreasing hepatic exposure to some drugs, while increasing biliary and sinusoidal efflux (Hardwick et al., 2011). Interestingly, for MRP2 we identified a unique mechanism of regulation involving internalization from the canalicular membrane thus rendering this transporter non-functional for drug elimination into the bile. These changes in transporter expression could, both individually or in combination, influence the PK of drugs and drug response.

Table 3.

Membrane transporter changes in NASH.

	mRNA	Protein	Localization
Uptake (Clarke, et al., 2014b)
OATP1B1	↔	↑	-
OATP1B3	↓	↓	-
OATP2B1	↔	↔	-
Efflux (Hardwick et al., 2011)
MRP1	↑	↑	-
MRP2	↔	↑	Internalized
MRP3	↔	↑	Membrane
MRP4	↑	↑	-
MRP5	↑	↑	-
MRP6	↔	↑	-
Pgp	↑	↑	Membrane
BCRP	↑	↑	Membrane

No change (↔), increase (↑), decrease (↓), not measured (-)

Extrahepatic effects of NASH

The link between severe liver disease and kidney disease has been known for more than a hundred years (Fabrizi et al., 2013) but the recognition that non-end stage liver disease can influence kidney function and the development of chronic kidney disease (CKD) has only recently come to light. It is now accepted that NAFLD and NASH are associated with CKD independent of other risk factors such as age, sex, BMI, hypertension, diabetes, smoking, and hyperlipidemia (Armstrong et al., 2014; Fabrizi et al., 2013; Hamad et al., 2012; Y. Li et al., 2014; Mikolasevic et al., 2013; Targher et al., 2010, 2011; Yasui et al., 2011; Yilmaz et al., 2010). Two of these studies specifically examined this relationship in biopsy-verified NASH patients and found a much higher rate of CKD among NASH patients. Targher et. al. found that patients with NASH had lower glomerular filtration rate (75.3 ± 12 versus 87.5 ± 6 ml/min per 1.73 m²) and a greater frequency of CKD (25% versus 3.7%) compared to control patients matched for age, gender, and BMI (Targher et al., 2010). This group also found that increasing fibrosis stage was positively associated with albuminuria and negatively correlated with glomerular filtration rate (Targher et al., 2010). Yasui et. al. also reported that CKD was present in 21% of NASH patients versus only 6% in non-NASH control patients (Yasui et al., 2011). These studies suggest that, in addition to changes in hepatic ADME processes, changes in kidney function may also contribute to altered PK in NASH patients and should be considered in the context of drugs dependent on renal elimination.

5. Moving forward with precision medicine

Advancements in modern medicine have brought increased sophistication and precision in our ability to diagnose and treat diseases and, inherently, these advances have also increased the complexity and complications associated with pharmacotherapy. Precision medicine seeks to understand and account for these complexities in order to maximize the therapeutic benefits and minimize the adverse effects in medical practice. This review has highlighted the promise of precision medicine and emphasized the importance of identifying sources of individual variability in drug ADME processes to ensure accurate dosing. Of particular emphasis has been the data to support inclusion of NASH as a one of these factors that may be underlying cause of VDR observed in the clinic. As precision medicine moves forward with the inclusion of NASH as a risk factor there are several key points that need to be considered in order to advance this science into clinical practice.

The foremost point is the necessity for better, non-invasive diagnosis techniques to distinguish NASH from the early stages of NAFLD. Without a reliable non-invasive technique for diagnosis, assessment of the role of NASH as an underlying factor in VDR will continue to be difficult to determine and, if NASH patients are determined to be an at-risk population, it will be even more difficult to translate dose adjustment recommendations into clinical practice. This NASH diagnosis issue accentuates the deeper challenge in hepatology that liver diseases are often complex, multi-factorial phenotypes making it difficult to correctly diagnose and categorize patients into at-risk populations. The idea of deep phenotyping has been suggested as a way to improve precision medicine and may be an interesting avenue forward for understanding effects of liver diseases on ADME parameters (Robinson, 2012). As described above for NASH, these phenotypic characteristics would need to include assessment of the expression and function of the many mediators that can influence hepatic and extrahepatic drug metabolism and transport. These assessments can be particularly challenging in patient populations given the diversity in drug substrates and the complex overlap in substrate specificity among the drug metabolism enzymes and membrane transporters. This clearly shows that, in addition to the need for reliable diagnosis of NASH, the ADME phenotype of liver diseases needs to be determined for accurate translation of clinical observations into recommendations for precision medicine. In fact, these altered ADME phenotypes in NASH could potentially be used as reliable non-invasive diagnostic techniques for NASH. Despite these challenges, with the advancement of “omics” technologies these goals for NASH diagnosis and phenotyping are attainable and should be given high research priority.

The second key point that needs to be considered is the combined effects of the multiple factors that influence individual variability in drug ADME. This combinatorial approach will include factors such as genetics (including rare variants), underlying disease states, and polypharmacy (drug-drug interactions) (Markert et al., 2013; Ritchie, 2012). Dr. Scott Friedman in his highlight of our recent article suggested that the confluence of multiple factors could explain some of the less common “idiosyncratic” or type B ADRs (Figure 2) (Friedman, 2014). Furthermore, some of the current ADRs classified as type B ADRs may be more accurately classified as type A ADRs if the various factors influencing PK and drug exposure are determined. This point is illustrated by our recent pravastatin PK data discussed above clearly showing that a single factor alone may not be sufficient to alter PK but, when combined with another risk factor, can exhibit a synergistic increase in drug exposure (Figure 3) (Clarke, et al., 2014b). This example accentuates the fact that the key mediators need to be known for each substrate because PK sits at the intersection of drug metabolizing enzymes, membrane transporters, and various other molecular players (Sadee, 2013). The statin drugs serve as a well-studied example of a class of drugs where a detailed determination of the various factors influencing PK has been determined. Although the statins all have similar structures, and hepatic uptake is the rate limiting step in statin PK, statins do not exhibit a class effect for the dose-dependent ADR of myopathy because each statin has varied metabolism and transporter affinities (Elsby et al., 2012; Ho et al., 2006, 2007; Jacobsen et al., 1999; Pasanen et al., 2006; Prueksaritanont et al., 1997; Wilke et al., 2012). The choice to focus on specific interactions between certain ADME mediators must be informed by mechanistic and clinical data and be selected with prudence because the number of possible interaction combinations is impossible to assess empirically (Baye et al., 2011; Sadee, 2013). Decisions for which interactions are worth pursuing will be based on the preclinical and clinical data for overlapping substrate specificities of metabolism enzymes and membrane transporters discussed previously. Lastly, although the focus of this review has been on adverse outcomes associated with individual variability in PK, it is important to recognize that these factors, alone or in combination, can also result in attenuated therapeutic effects. A good example of this is the potential for decreased ezetimibe efficacy in NASH due to altered disposition of ezetimibe-glucuronide, the pharmacologically active metabolite, away from its intestinal target site (Hardwick et al., 2012). Furthermore, there is potential for the combined effect of genetic and environmental factors to have counteracting effects that make dose-adjustment more complicated. For example, the SNPs CYP2C9*2 and *3 reduce the metabolism of warfarin by ~30–40% and ~80–90%, respectively, (Lee et al., 2002) and led the FDA to make several modifications to the warfarin label which include recommended dose adjustments based on CYP2C9 genotype (Johnson et al., 2011). CYP2C9 activity is reported to increase in NASH (Fisher et al., 2009), potentially offsetting the impact of impaired metabolic function of the SNPs by causing an unexpected increase in warfarin exposure.

Fig. 2. Intersection between two risk factors for an adverse drug reaction.

A genetic only screening approach (top row) captures single nucleotide polymorphisms (SNPs) (red star) in the major player(s) in the elimination of a drug that create a moderate increase in the risk of an adverse drug reaction (ADR). A phenotype only approach (middle row) that identifies at-risk patients based on a disease phenotype such as nonalcoholic steatohepatitis (black circle) causes changes multiple players in the elimination of the drug and may create a moderate risk of the ADR. A combined gene-X-environment approach (bottom row) captures both risk factors and will also identify the low frequency events where the overlap of the two risk factors creates a high risk of the ADR in a population (black oval).

Fig. 3. A multifactorial approach to precision medicine.

Pravastatin disposition illustrates how several factors in individual variability can work individually and in combination to increase the risk of myopathy. (A) Pravastatin disposition in healthy individuals carries with it minimal risk of myopathy. (B) The changes in uptake transporters that occur in NASH may slightly increase exposure to pravastatin and potentially the risk of myopathy. (C) The presence of a SNP in SLCO1B1 is known to increase exposure to pravastatin. (D) The combination of the SNP and NASH may combine to dramatically increase pravastatin exposure and the risk of myopathy.

Precision medicine will continue to be an important aspect of clinical practice that seeks to account for the various sources of individual variability in PK and VDR rather than focusing on a single factor in isolation. This review has emphasized NASH as a previously unrecognized factor in individual variability that must be considered in the implementation of precision medicine. Although the prospect of categorizing and cataloging the individual and combined factors in individual variability appears daunting, the positive results obtained thus far are encouraging for the future of the burgeoning field of precision medicine.

Abbreviations

ADME

absorption, distribution, metabolism, excretion

ADR

adverse drug reaction

BMI

body mass index

CKD

chronic kidney disease

CYP

cytochrome P-450

GST

glutathione S-transferase

GWAS

genome-wide association study

NAFD

nonalcoholic fatty liver disease

NASH

nonalcoholic steatohepatitis

OATP

organic anion transporting polypeptide

SNP

single nucleotide polymorphism

SULT

sulfotransferase

UGT

UDP-glucuronosyltransferase

VDR

variable drug response