The Genetic Basis of ADRS

THE GENETIC BASIS OF ADRS

POLYMORPHISMS AFFECTING DRUG METABOLISM

Most drugs are degraded through a limited number of metabolic pathways, most of which involve micro- somal hepatic enzymes. Ingelman-Sundberg et al. (1999) reported that about 40% of this human cytochrome (CYP) P450-dependent drug metabolism is carried out by polymorphic enzymes capable of altering these metabolic pathways. The CYP P450 monooxygenase system of enzymes detoxifies xeno-biotics and activates procarcinogens and promutagens in the body through oxidative metabolic pathways. These enzymes play an important role in the elim-ination of endogenous substrates (such as choles-terol) and lipophilic compounds [such as central nervous system (CNS) drugs that cross the blood-brain barrier], which otherwise tend to accumulate to toxic concentrations. This very large and well-studied gene family consists of many isoforms – for example over 70 variant alleles of the CYP2D6 locus have been described (Ingelman-Sundberg et al., 1999). The distribution of variant alleles for these enzymes differs among ethnic and racial subpopula-tions, with significant implications for clinical practice in various areas (Table 51.1). Alleles causing altered (enhanced or diminished) rates of drug metabolism have been described for many of the P450 enzymes, and the underlying molecular mechanisms have been identified for some. Table 51.2 summarizes some clinically significant polymorphisms affecting drug metabolism and the drugs and drug effects associated with them; a comprehensive summary is available at http://www.hapmap.org/cgi-perl/gbrowse/gbrowse. Continuously updated descriptions of these alle-les and accompanying references can be found at http://www.imm.ki.se/CYPalleles/.

CYP2D6, which encodes debrisoquin hydroxylase, was the first of these enzyme-coding genes to be cloned and characterized, and it remains among the most studied. It is involved in the metabolism of many commonly used drugs, including tricyclic anti-depressants, neuroleptics, anti-arrhythmics and other cardiovascular drugs and opioids. Variant alleles may differ from the wild-type (normal) gene by one or more point mutations, gene deletions, duplications, multiduplications or amplification. These may have no effect on enzyme activity or may code for an enzyme with reduced, absent or increased activ-ity. The genetics and related biochemistry of these pathways are still being elucidated and are more complex than the following simplistic descriptions imply. Extensive metabolizers, representing 75%– 85% of the general population, are homozygous or heterozygous for the wild-type, normal activity enzyme. Intermediate (10%–15% of the population) and poor (5%–10%) metabolizers carry two reduced or loss-of-activity alleles. These individuals are likely to exhibit increased drug plasma concentrations when given standard doses of drugs are metabolized by this enzyme; this functional overdose results in increased risk of dose-dependent ADRs associated with these drugs. These individuals also are likely to experience lack of efficacy with prodrugs that require activa-tion by this enzyme; lack of morphine-related anal-gesic response to the prodrug codeine is one example. Ultrarapid metabolizers (1%–10%) carry duplicated or multiduplicated active genes; they will metabolize some drugs very rapidly, never achieving a therapeu-tic plasma drug concentration (and hence expected efficacy) at a standard dose. Alternately, an ultra-rapid metabolizer-given codeine may experience an ADR usually associated with morphine because of the increased conversion of prodrug to active drug; this often is true of active metabolites, as well.

Two variant alleles of CYP2C9, which result in reduced affinity for P450 oxidoreductase or altered substrate specificity, are associated with increased risk of haemorrhage with standard doses of the anti-coagulant warfarin. The clearance of S-warfarin in patients who are homozygous for one of the polymorphisms is reduced by 90% compared with patients who are homozygous for the wild-type allele (Ingleman-Sundberg et al. 1999). Similar reduc-tions in drug clearance related to one of these polymorphisms have been documented with other CYP2C9 substrates such as ibuprofen and naproxen (non-steroidal anti-inflammatories), phenytoin (anti-epileptic), tolbutamide (hypoglycemic/anti-diabetic) and losartan (angiotensin II receptor antagonist) (Daly, 1995). The high frequency of these polymor-phisms (up to 37% of one British population was heterozygous for one mutant CYP2C9 allele) and the severity of the potential ADR (haemorrhage with warfarin treatment) make this an important consider-ation in the selection and dose of warfarin and other affected drugs.

A second important polymorphism affecting the safety of warfarin was reported by Rieder et al. (2005). They reported that variants in the gene encoding Vitamin K epoxide reductase complex 1 (VKORC1) explained 25% of the variance in warfarin dose. The effect was three times that of CYP2C9.

Patients who are homozygous for the null allele of CYP2C19 (poor metabolizers) are extremely sensitive to the effects of omeprazole (anti-ulcer), diazepam (anti-anxiolytic), propranolol (3-blocker), amitripty-line (tricyclic anti-depressant) and other drugs (Touw, 1997). CYP2C19 also is involved in the oxidation of the anti-malarial prodrug proguanil to cycloguanil, although it is unknown whether the polymorphism relates to its anti-malarial effects. The frequency of this polymorphism (3%–6% in Caucasians and 8%–23% in Asians) defines it as clinically signif-icant. Polymorphic alleles have been identified for several Phase II (conjugation) enzymes, and many of these are as important in drug metabolism as those associated with the Phase I (oxida-tion) enzymes discussed above. N -acetyltransferase 2, sulfotransferases, glucuronosyltransferases, catechol O-methyltransferase, dihydropyrimidine dehydroge-nase (DPyDH) and thiopurine methyltransferase (TPMT) are among the Phase II enzymes known to have clinically significant effects on drug metabolism (Mancinelli, Cronin and Sadee, 2000); some of these are summarized in Table 51.2. Polymorphisms of genes coding for these enzymes are particularly rele-vant in cancer chemotherapy (severe toxicity for homozygotes of null alleles of TPMT with thiogua-nine and azathioprine treatment and of DPyDH with 5-flourouracil treatment) and the treatment of Parkin-son’s disease with l-dopa (low methylators have an increased response to the drug).

POLYMORPHISMS AFFECTING DRUG TRANSPORT

Although cellular uptake of some drugs occurs through passive diffusion, membrane transporters also play a role in the absorption of medicines through the intestines, their excretion into bile and urine and their uptake into sites of action (such as brain, testes and cardiovascular tissue; tumour cells; synaptic cleft and infectious microorganisms) (Evans and Relling, 1999). Increasing attention is being focused on the possible role of polymorphisms of genes encoding drug transporters, some of which are summarized in Table 51.3.

One example of a transporter with relevance to drug response is p-glycoprotein (Pgp), an ATP-dependent transmembrane efflux pump that serves to extrude numerous drugs and other substances out of cells. Pgp is coded for by the multidrug resistance locus, MDR-1. Hoffmeyer et al. (2000) reported that a specific polymorphism, present in homozygous form in 24% of their Caucasian sample population, corre-lated with expression levels and function of MDR-1. Homozygous individuals had significantly lower MDR-1 expression and exhibited a 4-fold increase in plasma digoxin concentration after a single oral dose of the drug. Other substrates of Pgp include impor-tant drugs with narrow therapeutic indices, such as chemotherapeutic agents, cyclosporin A, verapamil, terfenadine, fexofenadine and most HIV-1 protease inhibitors (Meyer, 2000). In addition, over-expression of MDR-1 in cancer tumours has been associated with resistance to adriamycin, paclitaxel and other anti-neoplastic agents, and additional similar extrusion pumps are reported to contribute to drug resistance in various tumours (Sadee, 2000). Unfortunately, using Pgp to predict response has not been as successful as originally hoped.

Another potentially important gene family with a number of reported variants that may affect func-tion is that of the biogenic amine transporters, which play a role in the regulation of neurotrans-mitter concentrations (including serotonin, dopamine and GABA) in synaptic transmission (Jonsson et al., 1998). These transporters are the direct target recep-tors for many drugs such as anti-depressants and cocaine; polymorphisms of the serotonin transporter, in particular, have been associated with the modu-lation of complex behaviour (Heils, Teufel and Petri 1996) and may play a role in treatment with specific serotonin transporter inhibitors.

Mutations in other transporter-like proteins such as the sulfonylurea receptor (SUR) that regulates ATP-sensitive K+ channels and insulin secretion and nuclear factors such and hepatocyte nuclear factor-1 alpha and factor-1 beta are being studied both for their role in aetiology of disease and response to therapy. Pearson et al. (2004) reported on an elegant study to evaluate the metabolic picture and response to metformin in patients with type 2 diabetes and maturity onset of the young caused by mutations in either HNF-1 alpha and HNF-1 beta.

POLYMORPHISMS AFFECTING DRUG RECEPTORS AND TARGETS

Many drugs interact with specific targets such as receptors, enzymes and other proteins involved with cell cycle control, signal transduction and other cellu-lar events. Genes encoding these targets occur in polymorphic forms that may alter their pharma-cologic response to specific medicines. For exam-ple, variants affecting β-adrenergic receptors are a major determinant of β-agonist bronchodilator (e.g. albuterol) response in asthmatic patients. A specific common polymorphism has been linked to increased β receptor down-regulation in response to treat-ment with albuterol, which may result in decreased drug efficacy and duration of action (Tan et al., 1997; Liggett, 2000). However, other studies have failed to show the expected correlation between the variant and clinical response (Lipworth et al., 1999).

Drysdale et al. (2000) suggested that specific haplo-types (the array of alleles on a given chromo-some) may have greater predictive value regarding response to β-agonist bronchodilators than the pres-ence of individual polymorphisms. They reported marked variation in the ethnic distribution of the most frequently observed haplotypes (>20-fold differences) and in the mean β-agonist responses by haplotype pair (>2-fold differences). These authors suggested that the interactions of multiple polymorphisms within a haplotype may affect biologic and therapeutic pheno-types and that haplotypes may be useful as pharma-cologically relevant predictive markers.

Arranz et al. (2000) completed a comprehen-sive study of variants in multiple neurotransmitters and receptors in 200 schizophrenic patients. They reported that a set of six sequence variants involv-ing the 5-hydroxytryptamine (serotonin) receptor, the histamine receptor (H2) and the promoter region of the serotonin transporter gene successfully predicted response to treatment with clozapine (a neuroleptic) in 76% of patients, with a sensitivity of 95% for satis-factory response. Several of these individual polymor-phisms had been previously studied in this context, but with inconsistent findings. If the results of this retrospective study are prospectively validated, then they will form the basis of a simple test to optimize the usefulness of this expensive drug in a heteroge-neously responsive group of patients.

The risk of drug-induced long QT syndrome, a cause of sudden cardiac death in individuals with-out structural heart disease, has been linked to five gene variants, each encoding structural subunits of cardiac ion channels that affect sodium or potas-sium transport and are affected by anti-arrhythmics and other drugs (Priori et al., 1999). Priori et al. (1999) reported that a significant number of indi-viduals carry ‘silent mutations’ of these genes; the resulting alterations are insufficient to prolong the QT interval at rest, but affected individuals may be especially sensitive to any drug that affects potassium currents. The combination of these silent mutations with even modest blockade induced by a variety of drugs used for many purposes can result in prolonga-tion in action potential that is sufficient to trigger the onset of a serious ventricular arrhythmia (torsade de pointes). Roden and his colleagues, however, found less than 10% of patients suffering from drug-induced long QT actually had any of the known mutations associated with familial long QT syndrome (Yang et al., 2002).

Polymorphisms affecting steroid hormone nuclear receptors may affect individual response to drugs and hormones. For example, glucocorticoid resis-tance in asthma patients has been associated with increased expression of the glucocorticoid recep-tor 3-isoform (Sousa et al., 2000); activating muta-tions of the mineralocorticoid receptor have been linked to hypertension exacerbated by pregnancy (Geller et al., 2000) and dominant negative muta-tions of peroxisome proliferator-activated receptor gamma (PPAR gamma) have been associated with severe insulin resistance, diabetes mellitus and hyper-tension. Huizenga et al. (1998) identified a poly-morphism affecting the glucocorticoid receptor that was present in 6% of their elderly study popula-tion. These individuals appeared healthy but exhibited increased sensitivity (reflected in cortisol suppres-sion and insulin response) to exogenously adminis-tered glucocorticoids. The authors postulated that this increased lifelong sensitivity to endogenous gluco-corticoids might be reflected in the observed trends towards increased body mass index and decreased bone mineral density in affected individuals. This polymorphism also may be related to the development of early or serious ADRS with exogenous glucocor-ticoid treatment in carriers, but this has not yet been established.

Some investigators have reported a relationship between variants in the angiotensin converting enzyme (ACE) gene and individual sensitivity to ACE inhibitors such as enalapril, lisinopril and captopril, but the results reported by other teams fail to show an association, so this finding remains to be confirmed (Navis et al., 1999).

The beta adrenergic receptor is the target for drugs used to treat asthma, hypertension, and heart failure. Two polymorphisms appear to have an effect on some drugs for the treatment of asthma as well ask risk of heart failure.

Small et al. (2002) reported that African Americans were at significantly greater risk of developing heart failure if they carried a single copy of the 2c dele-tion of the adrenergic receptor. In animal models, this deletion results in an ineffective form of the recep-tor and higher norepinephrine levels. When combined with the ‘hyperfunctioning’ 389 mutation, the risk was multiplied several fold. The 2c deletion is more common in African Americans, and the authors hypothesize that this may be the reason for higher rates of heart failure in African Americans. The numbers in the study were smaller for Caucasians and did not result in a statistically significant risk. Hajjar and MacRae (2002) in their editorial accompanying this paper warn that the data must be replicated to be considered.

POLYMORPHISMS RELEVANT TO CANCER CHEMOTHERAPY

The basis of many forms of cancer chemotherapy involves the administration of maximum tolerated dosages with the goal of inflicting the greatest damage to malignant cells while causing the least damage to normal tissue. Genetic variations of drug-inactivating enzymes in normal tissues may increase the risk of severe toxicity or even death. As mentioned above, TMPT-deficient (homozygous; –0.3% of the popu-lation) individuals treated for acute lymphoblastic leukaemia with standard doses of mercaptopurine, thioguanine and azathioprine (immunosuppressant) may experience severe and potentially lethal bone marrow toxicity. A dose reduction of up to 15-fold may be needed to avoid haematotoxicity in these patients (Evans et al., 1991). TPMT genotyping or phenotyping (by assessing red blood cell enzyme levels) before the institution of therapy with any of these agents has become accepted practice at some medical centres (Sadee, 2000).

Several similar examples have been documented (Iyer and Ratain, 1998): patients with variant DPyDH cannot inactivate 5-fluorouracil, resulting in myelo-suppression and neurotoxicity, while overexpression of DPyDH in tumours is linked to resistance to that drug; N-acetyltransferase-2 rapid acetylators (30%– 60% of Caucasians and 80%–90% of Asians) are at risk of greater bone marrow toxicity with amon-afide treatment (topo-isomerase II inhibitor), and patients who have a genetic deficiency of glucuronida-tion because of a variant promoter of UGT-glucuronosyltransferase UGTIA1 are at increased risk of myelosuppression and diarrhoea when treated with the topoisomerase I inhibitor irinotecan. At least one example of an activating variant of a co-factor/enzyme has been reported: mutations of NAD(P)H (nicotinamide-adenine dinucleotide phos-phate, reduced form) : quinone oxidoreductase (which activates cytotoxic anti-tumour quinones such as mito-mycin C) protect against cytoxic metabolites but also may reduce anti-tumour efficacy (Gaedigk et al., 1998).

Growth factor receptors may be overexpressed in some tumours, potentially affecting the efficacy of chemotherapy. One example of this involves the humanized monoclonal antibody trastuzumab (Herceptin^TM , which was designed to target an onco-gene (HER2/neu) that is overexpressed in some breast cancers and other cancers with poor prognoses. Trastuzumab, when given with paclitaxel and doxoru-bicin, enhances the cytotoxic effects of the anti-neoplastic agents in breast cancer tissues with high HER2/neu expression. Some researchers suggest that an optimal approach to cancer chemotherapy would involve genotyping both malignant and normal cells when feasible (Sadee, 2000).

Unfortunately, the Epidermal Growth Factor Recep-tor (EGFR) story is not as clearcut, but the use of pharmacogenomics and drug probes are helping scientists to understand the redundant pathways of growth. Early enthusiasm about the effectiveness of EGFR inhibitors (erlotinib and gefitinib) was followed by studies that showed no benefit when combined with cytotoxic drugs. However, Lynch and associates reported activating mutations in the EGF receptor that appeared to underlay responsiveness of non-small cell lung cancer to gefitinib (2004). A subgroup of patients had impressive response: women, patients who had never smoked, patients with adenocarci-noma, and Asians. A majority of the tumours in these patients were found to have a mutation in the EGFR gene which increased the sensitivity of the tumour to anilinoquinazoline inhibitors of EGFR. This is a rapidly moving and potentially fruitful area of both basic and clinical research.

Adoption of predictive tests associated with drug treatment has been extremely high in oncology. The percentage of physicians using a predictive test prior to treatment with Herceptin has exceeded early esti-mates and sales of Herceptin have exceeded expecta-tions. Clearly the use of a predictive test was a benefit to doctors, patients, and the developers of Herceptin, Genentech.