Gao, Klumpen & Gurney Expert Opin. Drug Metab. Toxicol. (2008) 4(10) 1309 catalyzed by uridine diphosphate glucuronosyltransferases (UGTs) [9] . Many of the genes that encode these enzymes are polymorphic, which may be a cause of inter-individual pharmacokinetic variability for some anticancer drugs. For example, CYP3A4 has an inter-individual variation in activity as high as 20-fold, which accounts for the large inter-individual differences in the disposition of drugs that are metabolized by this enzyme [10,11] . 2.4 Excretion The main routes of drug excretion are through the biliary tract and kidneys. Most anticancer drugs, including anthracyclines, taxanes and vinca alkaloids, are excreted mainly through the biliary tract [12] . Several transport systems are involved in hepatocyte uptake and biliary excretion of drugs, including the solute carrier and ATP-binding cassette transporter families. Liver metastasis, concomitant medications and other disease can affect transporter systems and reduce biliary excretion of anticancer drugs. Methotrexate, carboplatin, bleomycin, etoposide and capecitabine [6] , on the other hand, are primarily excreted by renal filtration. Hence impaired renal function from any cause can potentially decrease clearance and increase the toxicity associated with these drugs. 2.5 Factors affecting anticancer disposition Multiple factors, including physiological variables, genetic characteristics and environmental factors, can affect anticancer drug disposition ( Table 1 ). It is important to realize that these factors are interrelated and that no drug is eliminated by one route or mechanism only. Even drugs that are predominantly eliminated by the liver usually have some renal elimination. Vinorelbine and fluorouracil (5FU) are examples of drugs for which hepatic elimination is thought to be paramount [13] . However, paradoxically, it is impairment in renal function that is associated with a significant proportion of interpatient variability in toxicity and systemic exposure of these two drugs [14-16] . Another example is sorafenib, a potent oral multikinase inhibitor that is metabolized by two pathways: oxidation mediated by CYP3A4 to sorafenib N -oxide (major pathway) and conjugation mediated by UGT1A9 (minor pathway). The concomitant use of ketoconazole, a potent CYP3A4 inhibitor, did not lead to the predicted increase in plasma concentration of sorafenib although the formation Table 1 . Some factors affecting anticancer drug disposition. Type of factor Factors Examples Effects Ref. Physiological Old age Most drugs No clear effects [92,93] Ethnicity Fluorouracil, gefi tinib Various effects [94,95] Performance status Most drugs Poor performance, ��� toxicity Obese Cisplatin ��� Clearance [96] Cachexia Etoposide ��� Albumin binding, ��� hepatic clearance, ��� toxicity [97] Infl ammation Docetaxel ��� CYP3A4, ��� toxicity [98] Organ functions Hepatic dysfunction Etoposide, epirubicin ��� Clearance, ��� toxicity [97,99] Renal function Cisplatin, etoposide ��� Clearance, ��� toxicity [100] Peritoneal/pleural effusion Methotrexate ��� Distribution, ��� toxicity [101] Genetics TPMT defi ciency 6-Mercaptopurine ��� Clearance, ��� toxicity [102] DPD defi ciency Fluorouracil ��� Clearance, ��� toxicity [103-105] UGT1A1 polymorphism Irinotecan ��� Clearance, ��� toxicity [106] Treatment schedule 3 weekly versus weekly Doxorubicin ��� Cardiac toxicity [107-109] Infusion versus bolus Fluorouracil ��� Myelotoxicity [110] p.o. versus i.v. Etoposide ��� Absorption, ��� effect [111] Environment Medication (Paroxetine) Tamoxifen ��� Active metabolites [112] Supplement (St John���s wort) Imatinib, anastrazole ��� Clearance, ��� concentration [113,114] Cigarettes Irinotecan ��� AUC [115] Complementary medicine Various Various effects [114] ��� : Decrease ��� : Increase ��� : No change. DPD: Dihydropyrimidine dehydrogenase i.v.: Intravenous p.o.: By mouth (per os) TPMT: Thiopurine methyltransferase UGT: Uridine diphosphate glucuronosyltransferase. For personal use only.

Dose calculation of anticancer drugs 1310 Expert Opin. Drug Metab. Toxicol. (2008) 4(10) Individualized dosing BSA dosed Fixed dosed Systemic drug exposure Frequency in patient population Underdose Optimal range Overdose Figure 2 . Variations in systemic drug exposure using BSA-based dosing, fi xed dosing and the ���Holy Grail��� of truly individualized dosing. BSA: Body surface area. of the oxidative metabolite, sorafenib N -oxide, was decreased [17] . This suggested that conjugation mediated by UGT1A9 is able to compensate when the oxidation pathway is suppressed. Similarly, paclitaxel is predominantly metabolized in the liver by CYP2C8 to less active metabolite 6 �� - hydroxypaclitaxel. A second minor pathway by CYP3A4 produces metabolite B. However, marked inter-individual variability in the relative predominance of individual paclitaxel metabolites has been reported because of greater CYP3A upregulation than CYP2C8 in the presence of P450 inducers like dexamethasone, a drug commonly used in conjunction with paclitaxel [18] . 3. Initial dose selection options Anticancer drugs are characterized by a narrow therapeutic index and wide inter-individual variability. The systemic exposure after standard doses of cytotoxic drugs varies 4 ��� 10-fold between patients [19,20] . Different methods have been used to calculate the dose of anticancer drugs to reduce inter-individual variability. 3.1 Body surface area-based dosing The concept of body surface area (BSA equivalent to skin body surface area) was based on studies performed at the beginning of the 20th century, which established that the basal metabolic rate varied among various species as a function of a power less than 1 ( ��� 0.75) of weight, which approximately corresponds to the variation of BSA as a function of weight. BSA was originally calculated using a formula based on length and weight developed by Du Bois and Du Bois in 1916 from an investigation that involved only nine individuals [21] . Although validation of the derived formula was not performed initially [22] , it was introduced into clinical pharmacology in the 1950s to predict a safe starting dose in Phase I clinical trials from preclinical animal toxicology data [23] and soon after was adopted for dose calculation of cytotoxic drugs with little justification. However, the pharmacokinetic parameters of several anticancer drugs do not correlate with BSA [1] and BSA-based dosing failed to produce a consistent systemic drug exposure ( Figure 2 ) [1,24-28] . BSA is only associated with a statistically significant reduction in inter-individual variability of clearance in ��� 15% of anticancer drugs and for these agents, the relative reduction in variability of clearance is between 15 and 35% [28] . Moreover, BSA-based dosing does not take into account abnormal body habitus such as cachexia and morbid obesity [29] , nor does it correlate well with drug disposition processes. A recent survey of 181 medical oncologists in Australia demonstrates the confusion related to BSA-based dosing [30] . In the adjuvant disease setting, 12.7% of respondents used ideal rather than actual body weight (BW) to calculate BSA, or chose whichever was less. When treating obese patients, only 6.1% of respondents routinely used actual BW. Of the remainder, 69.5% of respondents either capped the BSA at 2 m 2 or used ideal BW. In underweight patients, 95% of respondents routinely calculated BSA using actual BW. However, body size is not completely without influence. Numerous studies have shown that obese patients with breast cancer should receive larger doses of chemotherapy [31,32] . However, we must find ways to move beyond using BSA as the sole method for dose calculation. Despite BSA-based dosing, some patients suffer dose-related severe toxicity. Just as importantly, many more patients have minimal or no toxicity and may be underdosed [26,33-35] . 3.2 Flat-fi xed dosing, fi xed-dose per BSA clusters and dose banding Flat-fixed dosing is very common in the medical practice outside medical oncology, where the ramifications of overdosing are not such an issue. It has also been adopted For personal use only.