Dosing of anticancer agents in adults

INTRODUCTION — Most anticancer agents have a steep dose response relationship and a narrow therapeutic index. Small variations in the administered dose can lead to severe and life-threatening toxicity in some individuals and underdosing in others, which may compromise cancer outcomes. Proper dose selection is of great importance, particularly in individuals with potentially curable diseases, such as lymphoma or testicular cancer, and in the setting of adjuvant treatment (eg, breast and colon cancer). Selection of the right dose is complicated by the fact that individuals have a highly variable capacity to metabolize and eliminate drugs.

The most relevant pharmacokinetic parameter for drug exposure is the area under the curve (AUC) of plasma concentration x time following a single dose. During drug development, drug level sampling at multiple time points helps define the relationship between drug administration and the AUC. AUC is influenced by drug dose and schedule, and patient-specific factors such as age, gender, height, weight, concomitant medications, inherited variations in drug metabolizing enzymes, drug transporters, and/or drug targets, and drug clearance (which depends on renal and hepatic function). As a result, there is much interindividual variation in the AUC following a single dose of a drug [1]. For most anticancer agents, attempts to minimize interindividual variation have been limited to normalizing doses based on body size (weight or body surface area [BSA]).

This topic will address issues related to dosing of anticancer agents in adults, including BSA-based dosing, which is used for most cytotoxic agents, weight-based dosing (eg, as is done for some cytotoxic agents such as melphalan and several therapeutic monoclonal antibodies), fixed dose prescribing (as is done for oral targeted agents such as tyrosine kinase inhibitors [TKIs]), AUC-based dosing (as is done for carboplatin), and pharmacogenetic as well as pharmacokinetic-guided dosing, including therapeutic drug monitoring. Dosing of anticancer agents in patients with renal or hepatic failure, and issues pertinent to dosing in the elderly are discussed in detail elsewhere. (See "Nephrotoxicity of molecularly targeted agents and immunotherapy" and "Chemotherapy hepatotoxicity and dose modification in patients with liver disease: Conventional cytotoxic agents" and "Chemotherapy hepatotoxicity and dose modification in patients with liver disease: Molecularly targeted agents" and "Systemic chemotherapy for cancer in older adults" and "Nephrotoxicity of chemotherapy and other cytotoxic agents", section on 'Dosing considerations for nephrotoxicity'.)

DEFINING OPTIMAL DOSE

Conventional cytotoxic agents — Appropriate dosing for cytotoxic anticancer agents has been largely determined from prospective and retrospective studies in which the goal was to maximize efficacy and minimize toxicity. The starting dose for conventional cytotoxic agents in phase I studies has generally been based on animal studies, where doses are usually escalated until the LD10 is reached (the dose that results in lethality in 10 percent of the treated animals). By convention, in human phase I studies, the first dose employed has been one-tenth of the LD10.

Based on the theory and intuitive belief that larger patients have a larger volume of distribution and a higher metabolizing capacity, it has been assumed that they require more drug to induce the same effects. In an attempt to minimize interindividual variation, dosing for most anticancer agents has generally been normalized using mg of drug per m² of body surface area (BSA), which is calculated using a patient's height and weight.

However, normalization of doses based on BSA does not account for most of the interindividual variation in drug exposure. For cytotoxic agents, interindividual variability in drug clearance based on the area under the curve (AUC) is expressed as the percent coefficient of variation (which is the standard deviation divided by the mean, times 100). Values of 25 to 70 percent are very common despite the use of BSA for individual dose calculation. (See 'Body surface area (BSA)-based dosing' below.)

More recently, given the trajectory towards "personalized medicine," pharmacogenetic and pharmacokinetic-based dosing strategies have received more attention, though these approaches are not yet in widespread use. (See 'Pharmacogenetics, pharmacokinetics, and therapeutic drug monitoring' below and "Treatment of adrenocortical carcinoma", section on 'Suggested regimen'.)

Importance of dose intensity — Historical data on optimal dosing for most existing cytotoxic agents have been generated using dose considerations that are derived from the "maximum tolerated dose" concept, irrespective of an individual's genetic constitution. The concept is that cytotoxic chemotherapy agents are administered at the maximum dose that an individual can tolerate before the onset of severe and even life-threatening toxicity. This dose, which is derived from phase I trials, is one dose level below that associated with dose-limiting toxicity and is termed the "maximum tolerated dose." This dose is used for subsequent phase II and III trials testing antitumor efficacy.

This approach is supported by a series of retrospective analyses indicating that the greater the dose intensity of an individual drug, the better the outcome. Dose intensity, defined as the amount of chemotherapy delivered per unit time, has been recognized as an important determinant of efficacy of cytotoxic chemotherapy in theoretical models, in vitro studies, and clinical trials [2,3]. The clinical data are strongest for breast cancer, in which prospective studies have demonstrated inferior outcomes for patients who receive lower than intended dose intensity [4-7] and better outcomes with dose-dense as compared with non-dose-dense therapy [8]. (See "Selection and administration of adjuvant chemotherapy for HER2-negative breast cancer", section on 'Importance of chemotherapy schedule'.)

Another line of evidence supporting the importance of dose intensity in oncologic outcomes is derived from the use of hematologic toxicity as a surrogate marker for efficacy and the delivery of effective chemotherapy doses. Retrospective analyses of clinical trials in lung [9], breast [7], and ovarian [10] cancer demonstrate inferior outcomes in patients who lack significant hematologic toxicity from myelosuppressive chemotherapy. The practice of giving higher drug doses to individuals who lack significant treatment-related toxicity seems logical given the correlation between lack of toxicity and low drug concentrations [11-14]. However, despite its logical appeal, there have been limited trials testing this concept, and it has not been widely adopted or endorsed in clinical practice. In contrast, the practice of reducing doses based on excess hematologic or other toxicity (eg, neurologic, gastrointestinal, or dermatologic) is widely accepted. In general, dose reduction for individual agents has been based on criteria that were used in clinical trials, often an arbitrarily selected percentage of the initial starting dose.

Given the limited amount of high-quality evidence in this area, additional prospective trials are needed to assess whether drug dosing guided by the occurrence of toxic effects (pharmacodynamics) could improve efficacy of standard cytotoxic regimens.

Newer targeted therapies and immunotherapy — There has been a paradigm shift in oncologic treatment away from the development of classic intravenously administered cytotoxic chemotherapy drugs toward immune checkpoint inhibitor immunotherapy, and the so-called "targeted therapies," such as kinase inhibitors, which are often administered orally on an ongoing daily basis, therapeutic monoclonal antibodies, which are dosed parenterally or subcutaneously.

When compared with classic cytotoxic agents, these novel agents show different relationships between levels of exposure, particularly exposure over time, and pharmacologic effects on the molecular drug target. Particularly with regard to molecularly targeted therapy, these agents are characterized by unique mechanisms of action, and many are highly specific for single or multiple key cellular biological pathways implicated in carcinogenesis. Anticancer activity levels for therapies targeted against defined cellular and molecular markers significantly improve if patient pool enrichment is carried out based on the presence of that specific marker in tumor tissue. Examples include tumor HER2 expression to select for treatment with trastuzumab, and the use of imatinib for patients with chronic myelogenous leukemia (CML) harboring an oncogenic BCR-ABL translocation. (See "Systemic treatment for HER2-positive metastatic breast cancer", section on 'Rationale for HER2-directed therapy and available agents' and "Cellular and molecular biology of chronic myeloid leukemia", section on 'The BCR-ABL1 fusion protein'.)

Dose-limiting toxicities for immunotherapy and targeted therapies are often markedly different from those of cytotoxic chemotherapy agents. Therefore, defining the optimal dosing of new targeted therapies, such as monoclonal antibodies and immunotherapies, may require alternative strategies since the dose-limiting toxicity of such agents may exceed the dose required for full therapeutic effect. In such cases, dosing to a measurable level of adverse events may result in greater toxicity without a measurable increase in benefit above lower doses. Whether the "maximum tolerated dose" paradigm is valid for anticancer agents with selectivity for a cancer-specific target (targeted therapies) is unclear. For these drugs, multiple approaches are being explored to develop alternative strategies for the identification of an optimal dose as opposed to the "maximum tolerated dose" [15]. At present, all of the orally active kinase inhibitors (most of which are tyrosine kinase inhibitors [TKIs]), as well as vemurafenib, dabrafenib, and trametinib (inhibitors of the BRAF serine/threonine protein kinase pathway), are dosed using a fixed dose schedule for all patients regardless of weight or BSA. By contrast, some therapeutic monoclonal antibodies are given using a fixed dose (eg, alemtuzumab, ofatumumab, pertuzumab), while others are dosed on a mg/kg basis (ipilimumab, bevacizumab, trastuzumab, panitumumab, brentuximab, ramucirumab), and still others are dosed according to BSA (rituximab, cetuximab). (See 'Fixed-dose prescribing' below and 'Weight-based dosing' below and 'Orally active small molecule kinase inhibitors' below.)

Although these drugs offer specific important advantages over conventional cytotoxic agents, they are still afflicted by some of the same problems, including extensive interindividual variability in clearance (figure 1) [16] and a narrow therapeutic window. Although it is imperative to ensure that sufficiently high local drug concentrations are reached so as to maximize efficacy without exacerbating toxicity, the optimal way to achieve this goal is not yet apparent [15]. For some drugs, biologic activity at the intended target can be used to select an appropriate dose. As an example, with the poly-adenosine diphosphate-ribose polymerase (PARP) inhibitor olaparib, biologic activity can be demonstrated by inhibition of the enzyme in tumor tissue. In phase I studies, biologic activity of olaparib was demonstrated with continuous twice-daily dosing at dose levels above 100 mg, demonstrating a maximum tolerated dose of 600 mg twice daily [17,18]. The final US Food and Drug Administration (FDA)-approved dosing is 400 mg twice daily.

For other drugs in which a biomarker of activity is not available or easily accessible, the best way to optimize dosing is not established. Therapeutic drug monitoring has not been widely studied or endorsed, but there is renewed interest in this area [19,20]. (See 'Therapeutic drug monitoring' below.)

BODY SURFACE AREA (BSA)-BASED DOSING

Methodology and clinical use — In an attempt to minimize the amount of interindividual variation in drug exposure in the calculation of effective drug doses, various methods have been developed, primarily based on body size descriptors. Body surface area (BSA)-based dosing has been widely adopted for most cytotoxic agents and some therapeutic monoclonal antibodies (rituximab, cetuximab), despite the lack of rigorous validation and even though its ability to reduce interindividual variation in drug clearance is limited. No superior dosing scheme has been demonstrated, with the exception of AUC-based dosing for carboplatin. (See 'AUC-based dosing' below.)

In the 1950s, investigators originally proposed normalizing doses of chemotherapeutic agents by using BSA, which had been noted to explain the variation in metabolic rates of animals across a range of species, including humans, and to correlate linearly with blood volume [21-23]. In an early retrospective study conducted in the 1950s, doses of mechlorethamine and methotrexate per unit of body weight were higher for smaller animals than for larger animals, and for children than in adults, while doses per unit of body surface area were nearly similar for all species and for humans of all ages [24]. Subsequently, despite the lack of rigorous validation (and an increasing number of publications that question the validity of BSA-based dosing for a number of conventional cytotoxic agents [25-35]), BSA-based chemotherapy dosing has become a de facto standard for determining chemotherapy dosing for most cytotoxic agents [36-38] and some therapeutic monoclonal antibodies (rituximab, cetuximab).

Unfortunately, for many anticancer agents, normalizing doses according to the BSA has only a limited ability to reduce interindividual variability in drug clearance after a single dose of an anticancer agent [21]. In a study of 1012 adult patients with cancer receiving one of six different cytotoxic chemotherapy drugs (cisplatin, docetaxel, paclitaxel, doxorubicin, topotecan, and irinotecan), clearance of all six drugs was poorly correlated with BSA (figure 2) [39]. The correlation coefficients (r values) of the regression curves representing the contribution of BSA to interindividual variation in clearance ranged from 0.17 to 0.44 (r² = 0.03 to 0.19). However, the precision for reaching the target AUC was significantly better for BSA dosing compared with fixed dosing for four of the six drugs.

Many oncologists have a false sense of precision based on BSA dosing when, in fact, interindividual variation for most drugs (both conventional cytotoxic, as well as novel oral agents) exceeds 30 percent (figure 1) [16], and the range of clearance among individuals can vary 4- to 10-fold [11].

No alternate body-size measures (including lean body mass [LBM], ideal body weight, body mass index [BMI]) have reliably performed better than the BSA in reducing the interindividual variability [40,41]. One study found that for a variety of therapeutic drugs, LBM was the best size descriptor for clearance of chronically dosed drugs [42]. The LBM has also been proposed as a potentially superior body size measure for anticancer agents [43], but a proper scientific rationale for the use of LBM or any other body size measure other than BSA is lacking [21,44].

Algorithms for calculating body surface area — A firm recommendation cannot be made for any of the available formulae for calculating a BSA-based dose of anticancer agents, and all are acceptable, both for obese and normal-weight individuals. The Mosteller formula is our preferred option as it is the easiest to calculate (calculator 1), remember, and provides estimates approximately in the middle of other estimates. It is also the most commonly used formula in practice and in clinical trials. (See 'Dosing for overweight/obese patients' below.)

In 1916, the first formula (the Du Bois method) for estimating BSA in humans was published [45]. Despite only nine subjects being studied, and 17 cadavers in the validation study [46], this formula has been considered to be relatively accurate for normal weight individuals and is still in use (calculator 2). Various formulae have subsequently been developed to estimate BSA using height and weight. These are listed in the table (table 1) and compared graphically in the figure (figure 3). The Mosteller formula is the easiest to remember and calculate (calculator 1) [47].

Over the range of typical weights and heights, commonly used BSA formulae do not vary significantly from each other (figure 3). As an example, even for overweight patients, the Mosteller and Du Bois formulae differ by only 3 to 5 percent [48]. Given this fact and the absence of trial data suggesting superiority of any specific formula, an expert panel convened by the American Society of Clinical Oncology (ASCO) concluded that any of the formulae is acceptable when calculating doses of anticancer agents according to BSA [49]; this position was upheld in the 2021 update of these guidelines [50].

Notably, individuals with very high BMI have been poorly represented in the cohorts used to generate these formulae. Despite the limited data, the BSA estimates from these formulae seem to differ more substantially in obese (BMI >30 kg/m²) and morbidly obese (BMI >40 kg/m²) individuals (figure 3) [51,52].

Dosing for overweight/obese patients — Obesity is associated with worse cancer outcomes, which may be due, in part, to underdosing through the practice of using less than actual body weight to calculate BSA or capping doses of cytotoxic chemotherapy agents that are dosed according to weight or BSA. There is no evidence to suggest increased toxicity for obese patients when dosed according to their actual body weight, while there are data indicating that underdosing is associated with inferior outcomes. ASCO guidelines recommend that actual body weight be used for calculating BSA-based dosing of systemic anticancer drugs (with the notable exception of bleomycin), especially when the intent of therapy is cure [49,50]; however, UpToDate hematology contributors also favor capping the doses of vincristine when used for treatment of lymphoma. Other recommendations for managing systemic therapy dosing in obese patients from the ASCO expert panel are outlined in the table (table 2).

Over the past several decades, the obesity epidemic has made its impact in oncology. Overweight and obese individuals (as defined according to BMI (calculator 3)) have higher rates of both cancer incidence and cancer-related mortality [53]. The reasons for this are multifactorial, but one likely contributor to the higher mortality rates described in overweight/obese patients with cancer is the systematic underdosing of anticancer agents, with a substantial proportion of obese patients receiving reduced doses because of the use of ideal or adjusted ideal body weight or capped doses rather than doses calculated according to actual weight [54-59]. As the prevalence of obesity has increased (according to projections from the World Health Organization [WHO], by 2015 more than 700 million adults worldwide will be obese [60]), awareness of this problem has grown, as reflected in practice guidelines addressing this issue [49,61,62].

Overweight and obese patients often receive chemotherapy doses that are less than recommended because drug doses are based on a weight that is below their actual body weight [58]. A nationwide survey of chemotherapy dosing in Australia found that only 6 percent of respondents routinely used actual body weight in BSA calculations for obese patients, with the majority capping the dose or using ideal body weight [56].

The adverse impact of such underdosing on cancer outcomes can be illustrated by the following studies:

●An analysis of the effect of BMI on survival in operable breast cancer found that BMI >30 kg/m² adversely impacted overall survival (hazard ratio [HR] 1.25, p<0.01) [63]. Among obese patients, 96 percent had BSA calculated based on ideal weight, resulting in an average 12 percent reduction in chemotherapy dose.

●Retrospective analyses of other breast and ovarian cancer clinical trials have also demonstrated inferior clinical outcomes in patients who received arbitrarily reduced chemotherapy doses because of obesity [64-68].

However, it is not clear that the inferior outcomes associated with obesity can be entirely explained by lower chemotherapy doses. In a study of National Surgical Adjuvant Breast and Bowel Project (NSABP) colon cancer trials, obese individuals (BMI >35 kg/m²) had a poorer overall survival (HR 1.28, 95% CI 1.04-1.57) [69]. Dose capping at a BSA of 2 m² was common, with 55 percent of the obese and 73 percent of the very obese patients having a capped dose compared with 7 percent of normal weight patients. However, dose capping did not explain the association of high BMI with worse outcome.

One concern is that dosing of obese patients according to actual body weight might lead to excessive toxicity. This issue was addressed in a systematic review and meta-analysis of all studies that compared toxicity and efficacy of full weight-based chemotherapy in obese and normal weight cancer patients [70]. Hematologic treatment-related toxicity was significantly less in obese as compared with normal weight patients (odds ratio [OR] 0.73, 95% CI 0.55-0.98) while nonhematologic toxicity was similar (OR 0.98, 95% CI 0.76-1.26). Survival was not adversely affected when obese patients were dosed according to actual body weight.

Given the mounting data correlating underdosing with inferior outcomes in obese patients, ASCO convened an expert panel to study the appropriate dosing of anticancer agents [49]; the guidelines were subsequently updated in 2021 [50]. The expert panel recommended using actual body weight for BSA calculations in obese patients receiving systemic anticancer therapy that is dosed according to BSA particularly when the goal of treatment is cure. They recommended that clinicians follow the same recommendations for dose reduction, regardless of obesity status, for all patients and that consideration be given to resuming full weight-based doses in subsequent cycles, particularly if a possible cause of the toxicity (eg, renal or hepatic insufficiency) is identified. The panel recommended the use of fixed rather than BSA-based dosing only for bleomycin in treatment protocols for germ cell tumors (to limit pulmonary toxicity) and specifically recommended that doses of vincristine not be capped at 2 mg when used as a part of the CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) and CVP (cyclophosphamide, vincristine, prednisone) regimens, noting that the US FDA approved Prescribing Information for vincristine does not indicate that doses should be limited. However, many hematologic oncologists, including authors and editors associated with UpToDate, disagree with this recommendation. They, instead, continue to suggest capping vincristine at 2 mg total for lymphoma regimens and also capping the dose of brentuximab vedotin because of the high incidence of potentially disabling peripheral neuropathy when doses of these drugs are not capped [71]. (See "Treatment protocols for lymphoma" and "Treatment protocols for germ cell tumors".)

Issues related to brentuximab, which is typically dosed based on body weight, are discussed further below. (See 'Dosing for extremes' below.)

These recommendations for dosing based on actual BSA regardless of body weight also apply to BSA-based dosing for selected therapeutic monoclonal antibodies such as rituximab and cetuximab. These and other key recommendations of the expert panel are outlined in the table (table 2). Issues that are pertinent to drugs that are dosed by body weight rather than BSA are addressed below.

Underweight patients — Although less data are available than in the setting of obesity, we suggest using actual body weight for calculation of chemotherapy doses for underweight patients. However, among patients with weight loss and/or sarcopenia, organ function and performance status should be taken into consideration when dosing cytotoxic chemotherapy.

Underweight patients are relatively rare in the general population but are over-represented in the cancer population. Weight loss is a hallmark of cancer-associated cachexia, which is common among patients with advanced disease. (See "Pathogenesis, clinical features, and assessment of cancer cachexia".)

Weight loss is associated with a worse prognosis in many malignancies including colon, breast, and lung cancer:

●One study assessed gastrointestinal and lung cancer patients for cachexia by several methods (history of weight loss, muscle index, and computed tomography [CT] assessment) and found that cachexia was an adverse prognostic factor regardless of BMI [72].

●Another study of 141 underweight patients (BMI <18.5 kg/m²) from a cohort of 4288 patients treated in adjuvant colon cancer NSABP trials found excess non relapse-related mortality compared with normal weight patients (HR 2.23, 95% CI 1.50-3.31.) [69].

An underappreciated problem is that sarcopenia (defined as loss of skeletal muscle mass) is common among cancer patients even with normal- or obese-range BMI. In some cases, sarcopenia is treatment-related (eg, fluoropyrimidines, sorafenib, androgen-deprivation therapy). As an example, in a study of 55 women with metastatic breast cancer who were undergoing capecitabine treatment, 25 percent were sarcopenic by CT criteria [73]. Whereas most people with cachexia are sarcopenic, most sarcopenic individuals are not considered cachectic.

Sarcopenia is also associated with higher rates of treatment-related toxicity and inferior outcomes [73-77]. As examples:

●In the previously noted study of breast cancer patients receiving capecitabine, patients with sarcopenia were significantly more likely to have dose-limiting toxicity (sarcopenia was a significant predictor of dose-limiting toxicity (50 versus 20 percent, p = 0.04) and a shorter time to tumor progression [73].

●Another study of 55 patients with renal cancer treated with standard dose sorafenib also noted that rates of dose-limiting toxicity were highest in subjects with sarcopenia and a BMI <25 kg/m², and lowest in individuals who were not sarcopenic and/or overweight or obese (41 versus 13 percent) [74].

Because sarcopenia is an adverse prognostic feature in the absence of treatment, it is difficult to separate out the adverse impact of chemotherapy doses on outcomes.

In contrast to the situation for obese patients, there are few data on the optimal weight that should be used for dosing calculations. Proposals to prospectively use anthropomorphic measurements, electric impedance, and/or CT scans to measure body composition and chemotherapy dosing have not been rigorously studied. Lean body mass (LBM), when used as a scalar for dosing chemotherapy, is typically not measured but inferred from formulae, which may not be representative of the population of persons with cancer. Body composition is likely more important than BMI, but using LBM formulae derived from individuals without cancer may lead to erroneous conclusions.

In the absence of data for alternate dosing, our recommendation is to use actual body weight. Using the ideal body weight would result in a higher dose and would likely result in excess toxicity. However, among patients with weight loss and/or sarcopenia, organ function and performance status should be taken into consideration when dosing cytotoxic chemotherapy.

WEIGHT-BASED DOSING — In addition to some therapeutic monoclonal antibodies, weight-based dosing is used for a few cytotoxic agents, including cladribine, melphalan, and arsenic. In the absence of data suggesting increased toxicity for underweight or obese individuals receiving weight-based dosing, doses should be based on actual body weight.

Although uncommon, certain cytotoxic drugs are calculated based on weight rather than body surface area (BSA) in adult patients with cancer. This is largely based on how the drugs were initially developed. Examples include:

●Cladribine – Cladribine is dosed by weight in chronic lymphocytic leukemia and hairy cell leukemia, and it is dosed by BSA for other indications.

●Arsenic – Dosing in China was originally a flat dose, which changed to weight-based dosing when developed internationally.

●Melphalan – Both weight-based and BSA-based dosing are used for the treatment of patients with multiple myeloma.

In addition, several therapeutic monoclonal antibodies are also dosed according to weight, including ipilimumab, bevacizumab, trastuzumab, panitumumab, brentuximab, and ramucirumab.

Rationale — For conventional cytotoxics, there is no rationale for weight-based dosing other than historical precedent. For therapeutic monoclonal antibodies, however, there is rationale.

For large proteins (>100,000 Daltons), the US Food and Drug Administration (FDA) recommends that starting dose be calculated by weight and not BSA [78], as these drugs are not cleared through the normal renal or hepatic mechanisms but through intracellular catabolism with nonlinear clearance. Allometric scaling models based on body size have not generally been successful in predicting clearance of high molecular weight proteins in humans. At least some data suggest that fixed dosing and weight-based dosing perform similarly across a range of therapeutic monoclonal antibodies, with fixed dosing being better for some antibodies, and weight-based dosing better for others [79]. These authors suggest that first in human studies be performed using fixed doses with a full assessment of body size effect on pharmacokinetic and pharmacodynamic variability thereafter to determine the optimal dose for phase III trials. The trend toward fixed-dose prescribing of monoclonal antibodies is discussed below.

Dosing for extremes — There are no data suggesting increased toxicity for underweight or obese individuals receiving weight-based oncologic therapeutics dosed by their actual weight. Consistent with updated guidelines from ASCO [50], doses for most agents should be based on actual body weight. However, many hematologic oncologists still recommend capping doses of brentuximab vedotin, as indicated in the FDA-approved United States Prescribing Information.

FIXED-DOSE PRESCRIBING — Fixed-dose (ie, "flat-dose") prescribing does not take body size into account. While this is normative for non-oncologic drugs, it is rarely used in cytotoxic chemotherapy agents. One notable exception is that bleomycin is dosed according to body surface area (BSA) in regimens used to treat Hodgkin lymphoma, but dosed at a fixed level in testicular cancer due to concern about pulmonary toxicity. (See 'Dosing for overweight/obese patients' above.)

However, fixed-dose prescribing is more commonly used for molecularly targeted agents and immune checkpoint inhibitor immunotherapy.

Monoclonal antibodies — There is a trend to dose monoclonal antibodies as a fixed dose that does not vary based on body weight. Monoclonal antibodies typically distribute in the blood plasma and extracellular fluid compartments, which increase less than proportionately with body weight, suggesting that weight-based dosing may not be necessary. Furthermore, monoclonal antibodies typically have a wider therapeutic window and would therefore be potentially amenable to fixed dosing. Pharmacokinetic data from clinical trials have been analyzed and indicate that interpatient variation in exposure is comparable for fixed and body-weight dosing [80].

The addition of hyaluronidase as a dispersion enhancer to therapeutic antibodies has allowed for subcutaneous administration, which significantly reduces administration time, does not require intravenous access, has been proven to be safe and effective, and is generally preferred by patients [81,82]. A number of monoclonal antibodies are now available in subcutaneous formulations and like the intravenous formulations, all have fixed dosing formulations (eg, rituximab and hyaluronidase, trastuzumab and hyaluronidase).

Updated guidelines for dosing of systemic therapy agents in obese cancer patients recommend that FDA-approved prescribing information for targeted therapies should be used in all patients, regardless of obesity status [50].

Orally active small molecule kinase inhibitors — For anticancer agents with an apparent selectivity for a cancer-specific target (such as tyrosine kinase inhibitors [TKIs]; BRAF serine threonine kinase inhibitors such as vemurafenib, dabrafenib, and trametinib; and inhibitors of the mammalian target of rapamycin [mTOR]), the best way to determine the optimal dose is unclear. At present, fixed doses are used for all patients regardless of weight or BSA, even though this approach is associated with a wide spread of plasma concentrations following standard dose regimens and substantial interindividual variability at the end of the dosing interval (trough concentration).

Given the lack of reduction of interindividual variation with weight- or BSA-based dosing and high interindividual variation in early clinical studies, as well as the convenience and simplicity of fixed dosing, all orally active targeted therapies, including TKIs, other kinase inhibitors, and mTOR inhibitors, have used fixed-dosing schemes. Updated guidelines for dosing of systemic therapy agents in obese cancer patients recommend that FDA-approved prescribing information for oral small molecule kinase inhibitors should be used in all patients, regardless of obesity status [50].

Is fixed dosing optimal? — Standard fixed-dose dosing regimens rarely result in comparable circulating concentrations of the active drug in all patients. It is increasingly apparent that variability in response to newer targeted drugs can be influenced not only by genetic heterogeneity of drug targets that determine tumor sensitivity, but also by the pharmacokinetic background of the patient (eg, cytochrome P450 enzymes, oral absorption, and environmental factors that influence pharmacokinetics), as well as adherence to the regimen [83]. The vast majority of targeted drugs are characterized by a wide spread of plasma concentrations following standard dose regimens, with interindividual variability at the end of the dosing interval (trough concentration) of up to 23-fold [19]. It is estimated that as many as 45 percent of patients are underdosed [84], yet there is no accepted a priori way to assign an optimized dose to an individual patient.

Adherence to treatment and bioavailability are major issues:

●Although oral administration provides the convenience of self-administration at home, the evidence suggests that adherence to oral cancer therapies is far from optimal, putting patients at risk for adverse outcomes [85-88]. As an example, for patients with chronic myelogenous leukemia who are treated with imatinib for several years, poor adherence is associated with a poor response, and it may be the predominant reason for an inadequate molecular response [87,88].

●The solubility of many TKIs is pH-dependent, and an elevated gastric pH may decrease bioavailability. Strong interactions are noted with antacids and other drugs that alter the pH of the stomach (particularly for pazopanib, dasatinib, and nilotinib).

●Drugs with low water solubility and high cell membrane permeability, such as TKIs, are particularly susceptible to food effects, especially when a high fat meal (HF) is consumed. The most striking example is lapatinib, which has a 150 percent increase in exposure (area under the curve [AUC]) with food [89,90]. Other kinase inhibitors with significant food effects include erlotinib, pazopanib, and nilotinib [91-93]. The US Food and Drug Administration (FDA)-approved manufacturer's product information recommends that these drugs be taken on an empty stomach.

There are also data to suggest that limited dosing options (due to pill sizes, which force large dose reductions for toxicity) have affected outcomes of comparative clinical trials and also compromise efficacy of TKIs in clinical practice [94], although in many cases the reduced dose levels have remained above the threshold of biologic activity for the drug [95,96].

Investigators have experimented with titrating doses to toxicity, building upon the observation that patients with target-related toxicity from TKIs, such as skin rash for epidermal growth factor receptor (EGFR) TKIs and hypertension for vascular endothelial growth factor (VEGF) TKIs, might have improved outcomes. Despite the rationale for this in patients treated with some EGFR inhibitors such as the anti-EGFR therapeutic monoclonal antibody cetuximab for metastatic colorectal cancer, trials suggest that this dosing strategy does not significantly improve outcomes with other targeted drugs, such as anti-EGFR TKIs (eg, erlotinib [97-99]) or TKIs targeting the vascular endothelial growth factor, compared with historical controls. (See "Systemic therapy for nonoperable metastatic colorectal cancer: Selecting the initial therapeutic approach", section on 'Adverse effects' and "Cardiovascular toxicities of molecularly targeted antiangiogenic agents", section on 'Association with antitumor efficacy'.)

Checkpoint inhibitors — Many of the immune checkpoint inhibitors (ICIs) that target the programmed death 1 receptor (PD-1) were initially approved based on weight-based dosing, and several observations noted a better antitumor response in individuals with a high body mass index (BMI) [100,101]. Subsequent evaluation of population pharmacokinetic data from clinical trials indicated that fixed dosing provided similar exposure distributions to weight-based dosing and had equivalent pharmacokinetic variability [102-104]. As a result, dosing recommendations, as reflected in the FDA-approved United States Prescribing Information for nivolumab and the United States Prescribing Information for pembrolizumab were changed from weight-based to fixed dosing for most indications in adults. Updated guidelines for dosing of systemic therapy agents in obese cancer patients recommend that FDA-approved prescribing information for immunotherapies should be used in all patients, regardless of obesity status [50].

However, this issue has been brought to the forefront by a new single center retrospective analysis of 297 patients receiving ICIs for treatment of cancer, in which 59 percent were overweight (BMI ≥25 kg/m²) [105]. Fixed dosing was used in 204 patients, and it was weight-based in 93 patients. As had been previously reported, higher BMI was associated with better overall and progression-free survival. However, the improved outcomes were limited to those who received weight-based ICI dosing. Patients with a BMI <25 kg/m² tended to have better outcomes with fixed-dosing while patients with BMI ≥25 kg/m² tended to have better outcomes with weight-based dosing, although these differences were not statistically significant.

In our view, these data are interesting and hypothesis-generating, but not practice changing. The tumor populations were heterogeneous, there was "time period" bias given that patients treated in the era of weight-based and fixed dosing were likely different, and a similar efficacy for PD-1 pathway blockers has been shown over a wide range of weight-based dosing (from 2 mg/kg to 10 mg/kg).

AUC-BASED DOSING — As described above, the area under the curve (AUC) of plasma concentration versus time is the most relevant measure of drug exposure. AUC-based dosing is applicable for drugs that are cleared through glomerular filtration, like carboplatin, because there is a strong correlation between carboplatin clearance and creatinine clearance [106-108]. AUC-based dosing is not applicable to most other anticancer agents (with the possible exception of pemetrexed) because there are no characteristics (either alone or in combination) that can be used to predict drug clearance because elimination of the drug involves several pathways [109,110].

Carboplatin — The importance of glomerular filtration to the metabolism and excretion of carboplatin is emphasized by its usual dosing schema, which is based on an estimate of the glomerular filtration rate (GFR) and the desired level of drug exposure, according to the AUC of concentration x time (AUC, mg/mL x min).

Using the desired target AUC (which typically varies between 5 and 7 mg/mL x min) and the estimated GFR, the dose of carboplatin is then calculated by use of the Calvert formula: Total carboplatin dose, mg = Target AUC x (estimated creatinine clearance + 25). Because of potential changes in weight or renal function, this calculation should be repeated prior to each administered course of carboplatin.

Given the poor correlation between GFR and body surface area (BSA) [111], AUC-based dosing for carboplatin represented a significant success in reducing interindividual variation. This dosing scheme was developed eight years after the introduction of carboplatin, and in this study, the r = 0.886 (r² = 0.785) was much better than previous BSA-based dosing schemes, representing a true advancement in reducing interindividual variability for this drug [106,112].

The Calvert formula was developed based on accurate measurement of GFR using the clearance of 51-Chromium-labeled EDTA (51-Cr EDTA) [106]. In most cases, GFR is estimated using the serum creatinine, and calculated according to the Cockcroft-Gault equation (calculator 4). However, others suggest use of the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation due to greater accuracy [113], although this has not been seen in all studies [114,115]. (See "Assessment of kidney function", section on 'Measurement of GFR with plasma clearance'.)

Some clinicians (particularly in the field of gynecologic oncology) have preferred to estimate the creatinine clearance using the Jelliffe formula, which is not based on weight or BSA [116-119]. This is because carboplatin dosing was based on the Jelliffe formula in many of the trials that established carboplatin treatment for gynecologic cancers. However, in 2011, the Gynecologic Oncology Group (GOG) changed their policies for carboplatin dose calculation to recommend that estimated creatinine clearance be calculated by the Cockcroft-Gault method in all patients. A more extensive discussion of GFR estimation equations is provided elsewhere. (See "Assessment of kidney function", section on 'Measurement of GFR with plasma clearance'.)

The following issues are relevant to the carboplatin dose calculation:

●Total carboplatin dose – The total carboplatin dose is calculated in mg, not mg/m².

●Maximum glomerular filtration rate – The accuracy of dosing using a calculated rather than a directly measured GFR has been questioned [120]. Nevertheless, this is a standard approach.

When using an estimate instead of direct measurement of the GFR to calculate the carboplatin dose using the Calvert formula, calibration of the creatinine assay may impact dosing. In the United States, as of the end of 2010, all laboratories must use a creatinine method that has calibration that is traceable to an Isotope Dilution Mass Spectrometry (IDMS) reference measurement procedure [121]. However, this gives lower creatinine results than the older measurement procedures used to derive the Calvert formula and, hence, higher estimates of the GFR. This could result in higher calculated carboplatin doses and may potentially result in increased toxicities.

The GOG now recommends that creatinine clearance be estimated using a minimum value for serum creatinine of 0.7 mg/dL. Furthermore, if an estimated GFR based on measured serum creatinine is used in the Calvert formula, the US Food and Drug Administration (FDA) recommends limiting the maximal GFR for the calculation to 125 mL/min [122]. This recommendation does not apply if the GFR is directly measured. The GOG has now mandated a similar cap for patients on their carboplatin trials (including trials conducted in the setting of potentially curable disease).

For practical purposes, this means that the maximal allowed doses of carboplatin are as follows:

•AUC 6 = 900 mg

•AUC 5 = 750 mg

•AUC 4 = 600 mg

●Obese patients – Another point of unresolved controversy is the appropriate weight to use when calculating the estimated GFR [123,124]. The original Cockcroft-Gault formula to estimate GFR used actual body weight, but none of the patients was obese. Most clinicians use actual body weight in the Cockcroft-Gault formula for non-obese patients, although institutional practice varies. However, the use of actual body weight in the Cockcroft-Gault calculation can result in an overestimate of the GFR and a higher than needed carboplatin dose in obese individuals [124].

Among the methods suggested to calculate the appropriate dose of carboplatin in obese patients are the use of adjusted rather than actual body weight to calculate the GFR (as recommended by the GOG) [123,125], use of the Jelliffe formula (which is not based on weight or BSA), and the use of a flat dosing strategy based on an estimate of the population carboplatin clearance (140 mL/min) for overweight patients with normal renal function [123]. As an example, if a carboplatin AUC of 5 mg min/mL is desired, an appropriate dose would be 5 x 140 = 700 mg.

Guidance from expert groups is conflicting, and institutional practice varies:

•The optimal weight to use for calculating carboplatin by the Calvert formula was not specifically addressed in the American Society of Clinical Oncology (ASCO) guideline [49,50]. However, they did state that full weight-based dosing of cytotoxic chemotherapy agents should be offered regardless of obesity status.

•On the other hand the GOG recommends that actual weight be used for estimation of the GFR when using the Cockcroft-Gault equation as long as patients have a body mass index (BMI) of <25 kg/m² (calculator 5). For other patients, use of an adjusted weight is suggested (adjusted weight [kg] = [(actual weight - ideal weight) x 0.40] + ideal weight). A calculator for ideal body weight is available (calculator 6).

PHARMACOGENETICS, PHARMACOKINETICS, AND THERAPEUTIC DRUG MONITORING — The combined use of therapeutic drug monitoring (TDM; as a phenotypic approach) and genotyping of drug metabolic capacity is currently considered to be the most sophisticated way to individualize dose for drugs in which the clinical effects are difficult to evaluate, such as anticancer agents. However, the use of pharmacogenetic testing to select initial doses has not been widely endorsed or adopted, and at present, the only cytotoxic agent for which pharmacokinetic-guided dosing is in widespread use is mitotane. (See 'Therapeutic drug monitoring' below.)

Variation in body size does not adequately explain the majority of interindividual pharmacokinetic variability for most anticancer agents. The term pharmacokinetics (PK) refers to the processes by which a drug is handled by the body, which are grouped into the following phases (known as ADME):

●Absorption (for oral drugs)

●Distribution (to different organs/body compartment)

●Metabolism (activation/inactivation by enzymes)

●Elimination (renal, hepatic)

The most relevant pharmacokinetic parameter used to characterize drug exposure is the area under the curve (AUC) of plasma concentration x time, which can be estimated from drug level sampling at multiple time points. Although TDM is an accepted way to guide dosing of several agents with large interindividual pharmacokinetic variation in the noncancer setting (notably, antibiotics such as aminoglycosides and vancomycin, anticonvulsants, immunosuppressant drugs, lithium), this approach has not been established in general oncology practice (with the only exception being high-dose methotrexate, and even then, TDM is used to manage toxicity and not individualize drug doses). (See 'Therapeutic drug monitoring' below and "Therapeutic use and toxicity of high-dose methotrexate", section on 'Leucovorin administration'.)

Drug clearance is inversely proportional to the AUC and may be influenced by the following factors:

●Renal clearance is glomerular filtration rate (GFR) + secretion – absorption. As GFR can be easily estimated, dose selection for drugs that are cleared only by glomerular filtration, such as carboplatin, is possible. (See 'AUC-based dosing' above.)

●The effect of varying hepatic function is less clearly quantifiable, as the pathways of hepatic transformation and elimination are substrate dependent, and easily measurable liver function tests cannot be used to quantitatively characterize function. (See "Drugs and the liver: Metabolism and mechanisms of injury".)

●Inherited variations in drug metabolizing enzymes, drug transport proteins, and drug targets can substantially alter drug exposure. The term pharmacogenetics (used interchangeably with the term "pharmacogenomics") refers to the study of how germline (inherited) genetic polymorphisms affect the pharmacokinetics of drug exposure.

Pharmacogenetics — The rationale for pharmacogenetic studies is to investigate genes that can predict responsiveness to a specific drug increase the number of responders and decrease the number of subjects affected by adverse drug reactions. Despite known polymorphisms in both drug metabolizing and transporting proteins that influence drug exposure and pharmacokinetics in patients receiving anticancer agents, and the availability of testing for many of these polymorphisms, genotyping has not become widespread or widely accepted for any of these drug classes. The causes are multifactorial and include the relative rarity of these conditions, the fact that factors other than inheritance of high-risk polymorphisms affect variability in drug response, the cost and inconvenience of testing, and most importantly, a lack of data from clinical trials demonstrating a convincing difference in outcomes from the use of pharmacogenetic testing.

Drug metabolism — Several inherited polymorphisms in drug metabolizing enzymes that impact oncologic therapeutics have been identified, as described below. However, to date, the translation of pharmacogenetic parameters into clinical practice of medical oncology has been surprisingly disappointing, and none of these tests is widely endorsed or used.

Fluoropyrimidines, dihydropyrimidine dehydrogenase, and thymidylate synthase variants — To avoid the risk of severe and potentially fatal reactions, the manufacturers of both fluorouracil (FU) and capecitabine recommend that the drugs are contraindicated in patients with known deficiency in the metabolizing enzyme dihydropyrimidine dehydrogenase (DPD). Testing is commercially available to detect the most common high-risk polymorphisms in DPD, as well as polymorphisms in a second enzyme associated with fluoropyrimidine toxicity, thymidylate synthase (TYMS). (See "Chemotherapy-associated diarrhea, constipation and intestinal perforation: pathogenesis, risk factors, and clinical presentation", section on 'Predictive markers'.)

Testing for these variants is appropriate in any patient experiencing severe toxicity after receiving a fluoropyrimidine-containing regimen. However, given the low frequency of finding a predictive allele and the fact that patients who lack one of these high-risk variants may still suffer grade 3 or 4 fluoropyrimidine-related toxicity, preemptive genetic testing of all patients due to receiving a fluoropyrimidine in order to identify those with DPD deficiency is controversial and not widely practiced. These assays are only useful to stratify patients into categories of risk for severe toxicity, and there are no data to suggest that they are of benefit in selecting the appropriate initial dose of a fluoropyrimidine. (See "Chemotherapy-associated diarrhea, constipation and intestinal perforation: pathogenesis, risk factors, and clinical presentation", section on 'Testing for DPYD and TYMS variants'.)

UGT1A1 polymorphisms and irinotecan — Preemptive testing for uridine diphospho-glucuronosyltransferase 1A1 (UGT1A1) genotypes that are associated with a poor metabolizer phenotype prior to initiating irinotecan is a controversial and evolving area, and experts differ. Although some institutions routinely screen all patients, in our view there is insufficient evidence to recommend preemptive testing in all patients beginning an irinotecan-containing regimen. For individuals with a known UGT1A1 allele that is associated with reduced enzyme activity (eg, those known to have Gilbert's syndrome), lower initial doses of unencapsulated irinotecan are indicated. Reduced initial dose of liposomal irinotecan are also appropriate for those who are known or suspected to be homozygous for *28, although the data supporting this recommendation are less certain. Whether these considerations apply to sacituzumab govitecan, an antibody drug conjugate that consists of a humanized antitrophoblast cell-surface antigen 2 (Trop-2) monoclonal antibody coupled to SN-38, is unknown. For patients who experience severe irinotecan toxicity dose reduction is required regardless of UGT1A1 genotype.

The most common toxicities associated with irinotecan are severe neutropenia and diarrhea, which are unpredictable, yet can lead to significant morbidity and mortality. Risk factors include age over 65, female sex, poor performance status, impaired liver function, how chemotherapy is administered (ie, short-term infusion versus bolus dosing), and reduced UGT1A1 enzyme activity [126-131]. Irinotecan is a prodrug that is hydrolyzed in the liver to SN-38, the active moiety; SN-38 is further metabolized to the inactive SN-38 glucuronide in the liver by the enzyme UGT1A1.

Individuals who inherit certain polymorphisms in the UGT1A1 gene or its promoter have reduced enzymatic activity and they are referred to as having a "poor metabolizer" phenotype, because of decreased clearance of SN-38, which increases the risk for severe irinotecan-related neutropenia, and to a lesser degree, diarrhea. Most of the reported data describing the excess toxicity experienced by these individuals are in those carrying one or more *28 alleles; other lower frequency alleles may also be associated with the poor metabolizer phenotype but there are fewer data on both population prevalence and clinical significance (table 3) [132,133]. In general, patients carrying two alleles conferring decreased expression of function are at a higher risk (eg, homozygotes with *28/*28), compared with those carrying one allele (eg, *28/*6, *1/*28).

Some poor metabolizers may be identified because they have Gilbert's syndrome, an inherited deficiency in UGT1A1 enzyme activity caused by polymorphisms in the UGT1A1 gene (typically the *28 allele), and characterized by increased unconjugated bilirubin in the blood, which is usually asymptomatic. An estimated 9 percent of individuals of the general population in the Western world are homozygous for variant *28/*28, and up to 42 percent are heterozygous [134-137]. (See "Gilbert syndrome".)

Otherwise, the identification of individuals who have a poor metabolizer phenotype requires genetic testing for high-risk alleles. Genetic testing for the presence of genotypes that are associated with a poor metabolizer phenotype is available from several laboratories [138], and the US Food and Drug Administration (FDA)-approved label for irinotecan recommends that clinicians "consider" testing, with reduced initial irinotecan doses for those with UGT1A1 *28 and *6 homozygotes and compound heterozygotes with *6/*28 variants to reduce the likelihood of dose-limiting neutropenia. Similarly, the United States Prescribing Information for liposomal irinotecan also recommends a lower starting dose for those known to be homozygous for *28. However, routine preemptive use of this assay in all patients who are to receive either unencapsulated or liposomal irinotecan is not widely accepted and the clinical context in which these tests should be used remains to be established.

●Unencapsulated irinotecan – The majority of the studies are in patients receiving unencapsulated irinotecan. Initial reports suggested that UGT1A1 *28 homozygotes (and heterozygotes to a lesser degree) were at high risk for both irinotecan-related diarrhea and neutropenia [132,135,139-142]. However, subsequent data indicated that the magnitude of the problem with diarrhea was not as great as initially suspected [132,136,140,143-146]. The magnitude of the risk of severe neutropenia and diarrhea is highly variable between reports. Most of the available data are in individuals of European ancestry with *28 alleles, and only limited data are available for other genotypes, including *6 and *93 [126,147], and for individuals of African or Asian ancestry.

The value of testing to predict severe toxicity has been addressed only for *28. In one study, considering *28/*28 to be a positive test result, the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for severe neutropenia were 11, 94, 30, and 82 percent, respectively [148]. Similar test performance was observed for severe diarrhea (sensitivity, specificity, PPV, and NPV of 13, 92, 22, and 85 percent, respectively).

The anticipated benefit of preemptive testing for UGT1A1 genotype prior to initiating treatment with irinotecan is that the risk of adverse drug-related side effects (especially severe neutropenia) among patents found to have a poor metabolizer phenotype can be reduced by lowering the initial (and subsequent) doses of irinotecan. The concomitant harm is that reduction in irinotecan dose may also reduce chemotherapy effectiveness in tumor suppression and long-term survival.

There are no prospective studies that have genotyped patients before the first treatment, modified starting doses, and then compared clinical outcomes (severe neutropenia, tumor response, survival) after modified dosage compared with outcomes of patients receiving standard dose. In one model-based clinical simulation study, if pretreatment genetic testing were performed in all patients initiating irinotecan treatment, and a 25 percent dose reduction of irinotecan was applied to patients carrying UGT1A1 *28/*28, the risk of severe neutropenia would be reduced from 45 to 18 percent, and severe diarrhea would be reduced from 19 to 9 percent, respectively [149]. However, oncologic outcomes were not addressed. Preemptive testing with upfront dose reduction resulted in only marginal increases in quality-adjusted life years but a cost reduction of €580 per patient.

Thus, while there seems to be a clear relationship between UGT1A1 genotype (especially *28) and severe neutropenia (with a more limited relationship with severe diarrhea), and testing appears to be cost-effective [137,149-151], there is a lack of direct evidence to support the clinical utility of preemptive genotyping and modifying initial irinotecan doses as a way to improve clinical outcomes.

In January 2022, the FDA modified the irinotecan prescribing information to specify that individuals who are homozygous for the UGT1A1 *28 or *6 allele and compound heterozygotes (*6/*28) are at risk for severe irinotecan neutropenia and both preemptive testing for these UGT1A1 high-risk genotypes and a reduced initial dose of irinotecan in poor metabolizers should be "considered".

Although some institutions now preemptively test all patients prior to the first dose of irinotecan [137,152,153], routine use of genotyping to select the appropriate dose of the drug in all patients who are to receive unencapsulated irinotecan, at least when dosed on the basis of body surface area (BSA), has not been widely accepted for several reasons:

•There are no prospective clinical studies evaluating the benefits (less treatment-related toxicity) versus risks (potential for inferior disease control) of preemptive testing. Even if adverse drug-related effects were reduced, this may come at the expense of a reduction in tumor responsiveness, leading to overall net harm.

•Only approximately 1 in 10 patients will be identified as being homozygous for *28/*28, and the excess risk of severe neutropenia that is attributable to the inheritance of this specific high-risk polymorphism seems to be small, particularly at doses <150 mg/m² per week [144]. However, newer data suggests that as many as 17 percent of the United States population may be identified as having a poor metabolizer phenotype based on finding two decreased function alleles on routine pharmacogenomic testing for *6, *28, or *80 (a variant which is high linkage disequilibrium with *28, and a proxy for *28 [154]) [155].

•The analytic validity of UGT1A1 testing in clinical practice is unknown. Laboratories offering such testing may include variants in addition to *28 for which little evidence is available. Data on the clinical utility of tests for UGT1A1 variants other than *28 are very limited.

•How much to reduce the irinotecan dose for known poor metabolizers is an unresolved issue, and there is no consensus on this point [132,156]. Some have recommended an initial 20 percent dose reduction [157], and others, a 30 percent initial dose reduction [137,158,159]. (See "Chemotherapy-associated diarrhea, constipation and intestinal perforation: pathogenesis, risk factors, and clinical presentation", section on 'UGT1A1 polymorphisms'.)

•Inheritance of UGT1A1 *28 polymorphisms seems to account for only a fraction of the observed variability in irinotecan toxicity [126,160,161]. It is likely that both inherited (eg, alternative UGT1A haplotypes or polymorphisms in other genes involved in irinotecan disposition [126,160,162,163]) and nongenetic factors (eg, pretreatment bilirubin levels, gender, smoking, co-medications [126,127,164-166]) contribute to a patient's risk of irinotecan-related toxicity.

Others have questioned whether individuals with a wild (*1) UGT1A1 genotype might be underdosed. Several authors have suggested that alternative dosing strategies, such as genotype-guided dosing, might reduce the marked interindividual variation in drug exposure that results when irinotecan is dosed according to BSA [161,167]. Patients who do not have the *28/*28 genotype might be able to tolerate much higher doses of irinotecan than are contained in standard regimens such as irinotecan plus leucovorin and bolus and short-term infusional fluorouracil (FOLFIRI) [161,168-170]. A proof of principle study showed that UGT1A1 genotype is a feasible alternative to BSA-based irinotecan dosing [161]. However, it is not yet clear how this will affect clinical practice, and this strategy cannot be recommended.

●Encapsulated (liposomal) irinotecan – Very few data are available on the association of UGT1A1 genotypes with toxicity from liposomal irinotecan. At least one pharmacokinetic study suggests no significant impact of the UGT1A1 *28 homozygous genotype on SN-38 clearance [171]. Nevertheless, the United States Prescribing Information for liposomal irinotecan also suggests reducing initial starting dose to 50 mg/m² in patients known to be homozygous for the UGT1A1 *28 allele.

●Sacituzumab govitecan – Few data are available on the association of UGT1A1 genotypes and toxicity from sacituzumab govitecan, an antibody drug conjugate that consists of an anti-Trop-2 monoclonal antibody coupled to SN-38. Although some have suggested that lower initial doses should be considered in those with a poor metabolizer phenotype based on UGT1A1 genotype [155], at least two studies (including a preliminary report from the ASCENT trial) failed to find a correlation between dose-limiting neutropenia after the first cycle and UGT1A1 genotype [172,173]. (See "ER/PR negative, HER2-negative (triple-negative) breast cancer", section on 'Sacituzumab govitecan'.)

●Recommendations from expert groups – Not surprisingly, there is disagreement among regulatory agencies, and expert groups regarding preemptive testing for UGT1A1 alleles prior to initiation of irinotecan:

•The Clinical Pharmacogenetics Implementation Consortium (CPIC) has assigned level "A" to the UGT1A1-irinotecan pair, indicating that genetic information should be used to change prescribing of the drug, but CPIC guidelines for irinotecan are not currently available [174].

•The National Comprehensive Cancer Network guidelines state that patients who are homozygous for the UGT1A1 *28 allele of who have a clinical diagnosis of Gilbert's syndrome are at increased risk of neutropenia with unencapsulated irinotecan and that clinicians refer to the United States Prescribing Information for clinical guidance [175]. Specific guidance is not given for liposomal irinotecan.

•The American Society of Clinical Oncology does not provide guidance on this matter.

•The European Society for Medical Oncology state that UGT1A1 genotyping is an option and should be carried out prior to starting unencapsulated irinotecan in patients with a suspicion of UGT1A1 deficiency as reflected by conjugated bilirubin <20 percent of total bilirubin, and if an irinotecan dose >180 mg/m² per dose is planned [176].

Thiopurine S-methyltransferase and thiopurines — Thiopurine S-methyltransferase (TPMT) is responsible for the metabolism of thiopurines, which includes 6-mercaptopurine [6-MP]. Polymorphisms in the TPMT gene can result in functional inactivation or markedly decreased activity of the enzyme, and an increased risk of treatment-related leukopenia. Over 24 low-functioning genetic variants have been identified, but the two most common (TPMT*2 and *3) account for more than 95 percent of defective TPMT.

TPMT testing is not specifically recommended by the FDA prior to treatment with 6-MP. Although dose reductions of up to 90 percent may be needed in individuals with low or absent TPMT activity, many clinicians treating acute leukemia with 6-MP only perform TPMT genotyping if there is unexpectedly severe or prolonged myelosuppression. This subject is discussed in more detail elsewhere. (See "Overview of pharmacogenomics", section on 'Thiopurines and polymorphisms in TPMT and NUDT15'.)

Drug transport

●Methotrexate pharmacokinetics and toxicity are influenced by polymorphisms in transporter proteins including solute carrier organic anion transporter 1B1 (SLCO1B1) and organic anion transporter protein 1B1 (OATP1B1). (See "Overview of pharmacogenomics", section on 'Drug transport'.)

●Sunitinib toxicity has been correlated with specific haplotypes of efflux transporter genes ATP-binding cassette (ABC)B1 and ABCB2. (See "Overview of pharmacogenomics", section on 'Drug transport'.)

Therapeutic drug monitoring — TDM, which involves sampling of plasma or serum drug levels to determine optimal drug dosing, is theoretically appealing. This technique is in widespread use for a number of therapeutic areas including antiseizure medication, antibiotics, and immunosuppressive drugs. These classes of drugs all employ continuous dosing whereas cytotoxics are generally dosed cyclically, making TDM logistically more difficult as it would typically require sampling at multiple time points. Furthermore, as chemotherapy regimens are commonly based on combinations of drugs, determination of AUC values for specific drugs would require multiple blood samples to be drawn. Other challenges are that TDM is expensive and labor intensive, and most clinical practices do not have the technical infrastructure to adequately and rapidly process blood samples for clinical pharmacokinetic analysis.

Given these logistic and economic challenges, the only cytotoxic agent for which pharmacokinetic-guided dosing is in widespread use is mitotane. Plasma monitoring is recommended to maintain plasma levels between 14 and 20 mg/L; levels ≥14 mg/L are associated with better outcomes, and a significant increase in neurotoxicity is reported when levels exceed 20 mg/L. (See "Treatment of adrenocortical carcinoma", section on 'Suggested regimen'.)

TDM is also used for patients receiving high-dose methotrexate; but in that case it is not used to select drug dose, but instead to guide the use of leucovorin rescue to mitigate excess toxicity. (See "Therapeutic use and toxicity of high-dose methotrexate".)

TDM may be emerging as a valid alternative to current strategies for dosing of some drugs:

●Infusional fluorouracil – Infusional FU is the sole anticancer agent for which TDM has been validated as a method to improve the therapeutic index in more than one randomized trial [177,178]. However, validation of the benefits of TDM in randomized trials for commonly used FU-containing regimens (particularly in the setting of metastatic colorectal cancer) is needed. Until such data are available, it is premature to conclude that pharmacokinetically guided dosing should be integrated into clinical practice.

Like most other cytotoxic drugs, FU dosing is based on BSA. However, a complete lack of association between BSA and FU clearance has been shown by two independent groups [179,180]. Pharmacokinetically guided dosing might improve the therapeutic index [178,181-183]:

•In an early trial, 122 patients with head and neck cancer undergoing induction chemotherapy with cisplatin plus FU were randomly assigned to standard FU dose (4 g/m² by 96 hour continuous infusion) or at a dose adjusted according to the FU AUC after the first dose [178]. FU doses were significantly less during cycles 2 and 3 in the TDM arm, and toxicity (myelosuppression, mucositis) was also less; objective response rates were similar (82 versus 77 percent in the standard dose arm).

•In a later trial, 208 patients with metastatic colorectal cancer received 1500 mg/m² FU over eight hours with leucovorin and were then randomly assigned to continue weekly BSA-based fixed dosing or individualized dosing based on a single measurement of the FU plasma concentration at steady state, calculated to achieve an AUC of 20 to 25mg/mL x hour [177]. Patients receiving AUC-based FU dosing had significantly higher response rates, longer median survival, and less toxicity, including diarrhea.

While these data are intriguing, the FU/leucovorin (LV) regimen used in this trial is not typical for any disease. There are no published data on the benefits of pharmacokinetically based dosing in patients receiving modern FU-containing combination regimens, such as those containing oxaliplatin or irinotecan, although one early report suggests promise [184]. Validation of the benefits of TDM in randomized trials for commonly used FU-containing regimens is needed [185]. (See "Systemic therapy for nonoperable metastatic colorectal cancer: Selecting the initial therapeutic approach".)

●Orally administered targeted therapies, including tyrosine kinase inhibitors – The majority of new orally administered targeted therapies are once- or twice-daily fixed-dose drugs. The efficacy and toxicity of many tyrosine kinase inhibitors (TKIs) have been shown to correlate with trough levels [19]. Even in tumors with driver mutations (eg, chronic myeloid leukemia [CML] and epidermal growth factor receptor [EGFR]-mutant lung cancer), which are very sensitive to TKIs, there is a minimum threshold below which the drug is inactive. Furthermore, underdosing may lead to ineffective treatment or acquired resistance. Although TDM is not yet in general clinical use, it may be a rational step for improving the efficacy of TKI therapies.

Although data on the benefits of TDM are emerging for a variety of agents [186], they are most compelling for imatinib, where specific trough concentration values have been proposed for CML and gastrointestinal stromal tumors based on clinical efficacy data [187,188]. Patients with the likeliest benefit for TDM include those with suboptimal response or treatment failure, adverse events, suspected drug interactions, or nonadherence to therapy [187,189]. Clinical practice has not yet been altered by these findings. (See "Tyrosine kinase inhibitor therapy for advanced gastrointestinal stromal tumors".)

SUMMARY AND RECOMMENDATIONS

●BSA-based dosing

•Body surface area (BSA)-based dosing is used for most cytotoxic agents and some therapeutic monoclonal antibodies (MoAbs; eg, rituximab, cetuximab). Over the range of typical weights and heights, commonly used BSA formulae give similar results (figure 3); all are acceptable. (See 'Body surface area (BSA)-based dosing' above.)

•Actual body weight should be used for the BSA calculation for cytotoxic chemotherapy drugs regardless of whether the patient is over or underweight, especially when the intent of therapy is cure. Even when doses are normalized according to body size, the capacity to metabolize and eliminate drugs is highly variable among individuals.

•Vincristine cap – We cap vincristine doses at 2mg total dose in lymphoma regimens in order to avoid peripheral neuropathy. However, an ASCO expert panel recommended not capping doses of vincristine. (See 'Dosing for overweight/obese patients' above.)

●Weight-based dosing

•Weight-based dosing (eg, cladribine, melphalan, and arsenic) should be based on actual body weight.

•Brentuximab cap – However, many hematologic oncologists still recommend capping doses of brentuximab, as recommended in the FDA-approved United States Prescribing Information. (See 'Weight-based dosing' above.)

●Molecularly targeted and immunotherapy agents

•Fixed dosing is used for tyrosine kinase inhibitors (TKIs), BRAF serine threonine kinase inhibitors (vemurafenib, dabrafenib, and trametinib), immune checkpoint inhibitors that target the PD-1 pathway, and mTOR inhibitors regardless of weight or BSA. (See 'Orally active small molecule kinase inhibitors' above and "Treatment protocols for lymphoma".)

•Dosing for therapeutic monoclonal antibodies (MoAbs) is variable. Some are dosed using a fixed-dose schedule (eg, alemtuzumab, ofatumumab, pertuzumab, pembrolizumab, nivolumab, dostarlimab, avelumab, atezolizumab), others are dosed on a mg/kg basis (ipilimumab, bevacizumab, durvalumab, trastuzumab, panitumumab, brentuximab, ramucirumab), and others are based on BSA (rituximab, cetuximab). (See 'Weight-based dosing' above.)

●Carboplatin – For most patients, carboplatin dosing uses the Calvert formula, which is based on desired exposure (AUC of concentration x time) and the glomerular filtration rate (GFR).

•When estimated GFR is based on measured serum creatinine, we limit the maximum GFR to 125 mL/min for this calculation. This suggestion does not apply if the GFR is directly measured. (See 'Carboplatin' above.)

•In the Calvert formula, we use actual weight for GFR estimation by the Cockcroft-Gault equation as long as patients have a body mass index of <25 kg/m² (calculator 5). For other patients, use of an adjusted weight is preferred.

●Other dosing considerations – Therapeutic drug monitoring (TDM) and preemptive pharmacogenetic testing are more sophisticated ways to individualize drug dosing. At present, these are only used for selected drugs:

•Mitotane – Plasma monitoring is used in mitotane regimens to improve outcomes and reduce toxicity. (See "Treatment of adrenocortical carcinoma", section on 'Suggested regimen'.)

•Irinotecan – There is insufficient evidence to recommend preemptive testing for uridine diphospho-glucuronosyltransferase 1A1 (UGT1A1) genotypes associated with a poor metabolizer phenotype in all patients beginning irinotecan.

For individuals with a known UGT1A1 allele that is associated with reduced enzyme activity (eg, those known to have Gilbert's syndrome), lower doses of irinotecan are indicated. Reduced initial dose of liposomal irinotecan are also appropriate for those who are known homozygous for *28, although the data supporting this recommendation are less certain.

Patients who experience severe irinotecan toxicity require a dose reduction regardless of the UGT1A1 genotype. (See 'UGT1A1 polymorphisms and irinotecan' above.)

•6-mercaptopurine (6-MP) – Testing for polymorphisms in the purine S-methyltransferase (TPMT) gene is not recommended prior to treatment with 6-MP. Many clinicians treating acute leukemia with 6-MP only perform TPMT genotyping if there is unexpectedly severe or prolonged myelosuppression. (See 'Thiopurine S-methyltransferase and thiopurines' above.)