Body surface area in dosing anticancer agents: scratch the surface!

Grochow et al. (1) questioned the use of body surface area in dose normalization more than a decade ago in this Journal. In this issue, Baker et al. (2) retrospectively assessed the pharmacokinetics of 33 investigational agents tested in phase I trials during the past decade as a function of body surface area in 1650 adult cancer patients. Based on their findings, they recommend that the practice of calculating starting doses on the basis of body surface area in phase I trials should be abandoned. Historically, we have initiated phase I dosing based on body surface area and then retained the practice, carrying it forward through phase II and III studies and ultimately to Food and Drug Administration labeling. Should we accept the recommendation by Baker et al. and abandon the use of body surface area for dose determination? To answer this question it may be useful to put body surface area-based dosing of anticancer agents into perspective. What is body surface area and how did it enter our practice? Body surface area is equivalent to the two-dimensional surface area of the skin. It is difficult to measure and therefore commonly estimated on the basis of formulas that use body weight and height in the calculation. The most commonly used formula was published by Du Bois and Du Bois in 1916 (3). Obviously, the objective at that time was not to develop a formula to dose anticancer agents. Du Bois and Du Bois were working on “clinical colorimetry” (now known as basal metabolic rate). The body surface area of mammals correlates with basal metabolic rate. As may be expected in warm-blooded animals, body surface area is also proportional to blood volume. But, as Baker et al. point out (2), body surface area is not well correlated with glomerular filtration rate (4). Body surface area is also not associated with liver function (5). The practice of using body surface area in scaling drug doses began with Freireich et al. (6), who quantitatively compared toxicity of anticancer agents in mouse, rat, hamster, dog, monkey, and humans. Thus began the use of body surface area in scaling a dose from a mouse or other laboratory animal to an initial starting dose for a phase I study in humans. The estimation of body surface area is based on nomograms or computer programs that contain the Du Bois and Du Bois formula or similar formulas. This practice has been handed down through generations of clinical oncologists. Why do we still use body surface area? Hippocrates is credited with instructing us to do no harm. It is one of our deepest desires to benefit our patients when we prescribe drugs and not to cause harm. This is exceedingly difficult with a group of drugs that have a very narrow therapeutic index, such as anticancer drugs. In our quest to reduce variability in drug response among patients, we seek to reduce the variation in drug exposure. Drug doses that are “calculated” on the basis of body surface area give us a sense of accuracy and safety. However, in the experience of any practicing oncologist, pharmacokinetic and pharmacodynamic variability among patients remains great. Is the continued use of body surface area based on scientific data and, therefore, rational? The work by Baker et al. (2) in this issue of the Journal suggests no on both counts. For the 33 investigational agents, body surface area-based dosing statistically significantly reduced the interpatient variability in drug clearance for only five drugs, namely docosahexaenoic acid– paclitaxel, 5-fluorouracil/eniluracil, paclitaxel, temozolomide, and troxacitabine. The authors point out that for the handful of drugs for which clearance was associated with body surface area, the relative reduction in variability of clearance was between 15% and 35%; thus, only up to one-third of the total variability could be explained by body surface area. For the agents for which body surface area was associated with clearance, the authors suggest a potential relationship with blood volume or glomerular filtration rate. Furthermore, in the case of paclitaxel, they point to the vehicle (Cremophor EL) that is used in the formulation of the drug as having an impact on the phar-