>>William Douglas Figg:

Our next speaker today is Dr. Amit Pai, who is an Associate Professor of Clinical Pharmacy,

and Deputy Director of the Pharmacokinetic Core Laboratory at the University of Michigan. Dr. Pai’s research focus is on optimizing

drug dose selection in special populations, such as obesity. He earned his Doctorate of Pharmacy degree

from the University of Texas Health Science Center, and completed a pharmacy practice

residency at Bassett Health care followed by an Infectious Disease and Pharmacokinetic

Fellowship at the University of Illinois at Chicago. We know you will enjoy today’s lecture.>>Manjunath Pai:

Hello. My name is Amit Pai. I’m an Associate Professor of Clinical Pharmacy

at the University of Michigan, and the Deputy Director of the Pharmacokinetics Core. It’s my pleasure today to present on the Pharmacokinetics

of obesity. This slide contains my disclosures. What I’d like to cover today is reviewing

the definition and epidemiology of obesity, explain the effects of body size on drug exposure,

with a focus on anti-microbials, demonstrate the similarity and differences in body size

scalars for drug dosing, and explain the rationale for alternate approaches to dose selection

in obese adult patients. This graphic gives a global distribution and

prevalence of obesity in adult females across the world, and in different parts of the world. As you can see from this graph obesity has

more than double since 1980, from almost less than 20 percent to over 35 percent in the

United States, and similar trends are seen in the developing countries, represented by

the brick nations, Brazil, Russia, India, and China. Based on current definitions in 2014, over

600 million adults were classified as obese, and this is really important because the consequences

of overweight and obesity kill more individuals in the word today for the first time in history,

than underweight individuals — than the effects in underweight individuals. This is the prevalence of obesity across the

United States, by county, and clearly there is diversity in the prevalence of obesity

in the United States, with the lowest rate seen in Routt County, and the highest in Greene

County, which — again this area here represents the southeast part of the United States, is

also associated with a high prevalence of diabetes. When we think about obesity and important

point is the definition and prevalence of obesity, and when you think of obesity this

definition — the original definition was derived by Adolphe Quetelet in 1835. He came up with the Body Mass Index definition

which is basically the weight of an individual divide by height in meters squared. And the purpose that Adolphe Quetelet sought

was basically to find a correlation between body size parameters and different events,

such as probability of criminal behavior. And so, when you think of this metric it was

really not designed for a pharmacologic purpose, or for other classification purposes. And not until 1972 was this index transformed

into what we call today as the Body Mass Index, a term defined by Dr. Ancel Keyes, who did

a lot of the initial work on the effects of starvation in humans. Based on this definition that was adopted

by the National Institutes of Health in 1985 body mass index is categorized into three

major categories for obesity, and those are obese class one, between 30 and 34.99 kilograms

per meters squared to obese class three, which is greater than or equal to 40 kilograms per

meters squared. And so again, this classification is broken

down into units of five in a way for simplicity, but these may not necessarily correlate with

predictors or prognostic variables such as incidents of diabetes. So, if you look at specific populations — Asian

versus Caucasian, the risk for diabetes actually increases when the body mass index is greater

than 27 kilograms per meters squared. Irrespective of that current definitions are

based on an obese categorization of 30 kilograms per meters squared or greater. Based on that definition one out of three

adults in the United States meet that definition, and one in six children meet that definition. Now when you think about body mass index it’s

clearly not a perfect index, because it doesn’t represent the true body weight composition

when you’re thinking of extremes of individuals and height. And so, here’s a graphical representation

of two individuals from the movie “Twins,” demonstrating Arnold Schwarzenegger with a

body mass index of 34 kilograms per meters squared, and Danny DeVito with a 35 kilograms

per meters squared, who clearly have very, very different body composition, but have

very similar BMI’s. And this is important because BMI again was

adopted in 1985, but several of our pharmacologic studies were based on another metric known

as the ideal body weight. This is based on height and gender. It’s a very simple rule, and I’m going to

go over that specific equation and explain the origins of that equation. That equation has been used in several pharmacokinetic

studies, and in general terms individuals who are 20 to 30 percent above ideal body

weight are categorized as being obese. Despite these definitions of obesity there

really hasn’t been much progress in modifications of product labels to provide specific dosage

recommendations for obese individuals. This is important because this drives a lot

of dosage selection in our current paradigms, and product labels are based on three scenarios. One is dosing individuals based on their body

surface area. This is a common practice in oncology; weight

based dosing, a common practice in anti-microbial chemotherapy, and when it’s based dosed this

is often based on total body weight dosing, or in some labels as lean body weight dosing. But the predominant approach to dosing drugs

is using a fixed dosing strategy. This is using a dose — for example, 500 milligrams,

irrespective of body size. But this always raises the question that if

you use a fixed dose across a population of adults is using the same dose sufficient for

a 65 versus a 130-kilogram individual, for example. This dosing controversy has also been seen

in oncology, despite body mass index dosing. A study by Griggs and colleagues showed that

37 percent of women with a body mass index over 35 kilograms per meters squared received

lower standard first cycle doses. Patients were also seen to have lower risks

of grade three and four toxicities, despite using higher doses, and this in part may be

related to under dosing these individuals. A survey by Fields and colleagues showed that

80 percent of oncologists dose capped at two meters squared. Again, the idea here is individuals felt that

they were overdosing individuals when they used total body weight to compute body surface

area, and have in the past arbitrarily dose capped at two meters squared. And this is important because this can lead

to under dosing individuals, especially those individuals who are at greater risk for cancers

that are associated with obesity, such as breast cancer and colon cancer. Because of this discrepancy efforts by Griggs

and colleagues and several investigators around the United States has led to issuance of guidance

by the American Society of Clinical Oncology against capping these doses and actually using

total body weight to compute body surface area. And what I will do today is explain the mathematical

rationale for why that is the right approach of computing body surface area. This has also been seen with anti-coagulants. Looking at the CRUSADE trial, which did include

about one out of five individuals over a hundred kilograms, 80 percent of patients over 150

kilograms received below the standard dose, and those individuals that were likely to

receive the standard dose actually were at a higher risk for bleeding. And so, this is an example of a drug that

has been historically dosed on body weight, and a scenario where using body weight may

lead to over dosing the drug and lead to events such as bleeding. Other groups have shown that individualized

dosing lean body weight equation, which I’ll also review, is actually less likely to cause

this problem of bleeding and bruising in obese patients, and so may serve as a better metric

of dosing patients on Enoxaparin. So, the question at hand is really do bigger

adults need bigger doses? And so conceptually over dosing is likely

when you use weight based doses, because using the same milligram per kilogram dose across

a body distribution can lead to computation of much higher doses. The opposite is true with fixed dosing. Using the same dose across a population can

have the opposite effect of leading to lower exposures in larger individuals. And so, it’s really balancing these two dosing

paradigms that we have to consider, really, on a drug by drug basis. So, let’s think of that dose selection that

— from a pharmacokinetic perspective. And when you think of it from a pharmacokinetic

perspective a key concept is that of bi-equivalence. That is, trying to ensure that individuals

across the population have the same or have similar exposures, and that exposure can be

quantified by the area end of the curve, that is assuming having a similar AUC across the

population, or having similar peak concentrations of the Cmax concentration. And so, when this is administered — when

the drug is administered orally we also think of other parameters such as time to peak concentrations

of the Tmax, or in some multiple dose studies we may also consider the minimum concentration

or trough concentration as a metric to assess bi-equivalence. Now I think it’s really important to separate

the concepts of dose and exposure, because they’re certainly related, but can vary across

a population. So, if we give a one-gram dose to three individuals,

for example, those three individuals can have very different concentration time profiles. And so, when we simulate that across a population

we may get a distribution represented here in this histogram as a distribution of AUC’s

for that population. And that’s — that happens because we may

have a lot of inter-individual variability in the volume of distribution that impacts

the peak concentration, and inter-individual variability in the clearance of a drug that

can impact the AUC or area under of the curve. So, when we think of these pharmacokinetic

parameters the volume of distribution is often a term that’s easily confused. It is not physiological space, but rather

a proportionality constant that represents the apparent size of a compartment that the

drug will fill. And when you review the literature for a specific

drug what you may see is a lot of different v-terms [spelled phonetically], and so this

really depends on the method, the mathematical approach that was used to compute the volume

of distribution. So, if you use a non-compartmental method

you may see the value reported as VD, or VDSS. If you use a two-compartment approach you

may see the value reported as VC and VP, which represent the central and peripheral compartment. And if you have larger compartments describing

this you may see this numbered as V1, V2, and V3, and so on and so forth. These volumes essentially are used to help

fit the shape of the concentration curve, and again they do not represent an actual

physical space. They’re helpful because they can help compute

the Cmax or the peak concentration, and here represented by this equation is a very simple

way of thinking of this, which is if a dose was administered as a bolus, that does divide

by the central compartment value gives you some estimate of what that peak concentration

would be. The clearance is then just the mathematical

representation of volume that’s being cleared of drug in unit time. And that parameter is useful because it can

help compute the AUC or area under of the curve, represented again very simply as the

dose divided by clearance. Our objective most often when we’re thinking

about bi-equivalence is to achieve isometric AUC’s across a population. And to do that what we really need to do is

ensure that clearance scales with body size. So, if you’re going to dose a drug based on

body size we want to ensure that that scalar is really representative of clearance. Another perspective when we’re thinking about

optimal dose selection is the pharmacokinetic-pharmacodynamic, or the PK-PD perspective. And what’s done there is essentially looking

at the concentration time curve and breaking it up into different parameters. So often what’s done is to look at the peak

concentration of the Cmax, the area under of the curve, or the AUC, and the trough concentration,

which is the Cmin. These values can then be scaled to a potency

measure, in this case looking at anti-microbials we rely on something knows as the MIC, or

minimum inhibitory concentrations. And we can then index the pharmacokinetic

term with a potency measure such as Cmax or MIC over MIC, or the time above that MIC. Once we have those parameters we can then

look and see which of those parameters best correlates with the effect of the drug, the

safety of the drug, or some other measure. It doesn’t not have to be affected. It could also be emergence of resistance,

for example. Once we do that the parameters that best predict

effect help us classify that drug as either being concentration dependent, or time dependent. And that helps us decide whether we should

dose a drug once a day, or multiple times a day. When we use that principle, we can in essence

bucked drugs into different categories. And so, when we see drugs that are correlated

by their Cmax and AUC/MIC, we see we tend to refer those drugs as being concentration

dependent. When we see drugs that are really predicted

by the time above MIC we refer to those drugs as being time dependent. And when we look at antibiotics specifically

what we see is the AUC to MIC really categorizes most of the drugs. And the reason for this is AUC is really a

mathematical representation of concentration and time. And because it has both of those parameters

this parameter often correlates with effect. Beyond just looking at PK-PD we can also look

at how drugs are currently dosed, and when we look at anti-microbials again they fit

the pattern that we would expect; that is most of them are dosed on a fixed basis, this

is using the same dose across a population. A few of them, like the aminoglycosides, and

polymyxins, and other drugs similar, that are dose on body weight, usually on a milligram

per kilogram basis; and then some drugs actually have recommendations for both a fixed dosing

strategy, and a weigh based strategy. And so, let’s review again why this may or

may not be appropriate. Again, going back to the pharmacokinetic term

of volume of distribution, what you have here is two individuals who are represented roughly

twice in body weight, having very different body compositions. But what volume of distribution really represents

again is nothing physiologic. It simply represents a proportionality constant. The value is then indexed to body weight,

and so is often reported as liter per kilogram. That parameter helps us define what the peak

concentration is, and also as that value gets larger, it impacts the terminal half-life

of the drug. But again, this term does not represent bio-distribution,

but just represents a physical space, which is again very simply represented here, and

does not reflect that physiologic space. Now let’s simulate what would happen in two

individuals based on their body weight if volume increased with weight, but the clearance

of the drug did not increase with weight. So, what you have here in blue is an individual

that’s 60 kilograms versus 120 kilograms, or roughly twice the body weight. What you see in these profiles is because

the volume of distribution is smaller in the smaller individual you see higher concentrations. The effect of this is going to be lower concentrations

of the drug early on, but because the clearance is not changing the area under curve actually

will be identical for these two simulations. What you again will see is a lower Cmax. But you’ll also see over time a higher Cmin. So, when you’re thinking about a drug where

the volume increases with body weight, but clearance does not, this may be beneficial

for drugs that are time dependent, because what you’ll essentially have is increasing

time above a threshold concentration. Where this may be risky is if that time above

a threshold is associated with toxicity or some sort of adverse event. Now the solution of the effect is needed to

be early on — for example, if you’re using a drug for surgical prophylaxis — then this

can be overcome by using a higher initial dose to achieve the same concentration profile

later in the regimen. When we think about clearance this again is

the — is taking that volume and computing how much of that volume is clearing drug over

time, so it’s represented by liter per hour. What we know is this parameter does not really

increase in proportion to body size. So, what you saw is a reduction that is not

proportionate to body size, but often represents about a 50 percent increase as body weight

doubles, for example. The terminal half-life is inversely proportional

to this parameter, and so this parameter also affects the half-life of the drug. Here’s another simulation using the same principles

again, a 120 kilogram versus a 60-kilogram individual. Because the clearance increases, but the volume

does not increase, what essentially happens is you’ll see a lower AUC. You will see a slight lower Cmax over time

because the Cmin and trough concentration’s also decreasing. But the principle change is a reduction in

the AUC. And so, in this scenario what you’ll see is

the potential for failure if lower AUC’s are associated with lower effect, and that’s often

a case that’s seen. You’ll also see a propensity for emergence

of resistance if the dose is not modified with increasing body size. So, the solution in this scenario is the need

for a higher dose, and really the need for a higher maintenance dose. So, when you think about all these different

scenarios that have been presented so far what are the actual physiologic changes that

we’ve seen? One of the things — one of the studies that’s

really done this well is study done by Jeffrey Young and colleagues at the National Toxicology

Center, that have looked at autopsy data in over 300 individuals to sort of quantify what

happens to tissue as body weight increases, and have created mathematical models to explain

this. What you see in this graphical representation

is an expectation that when individuals are over 100 kilograms what you really see a gain

in is a gain in Adipose mass — relative to muscle mass. When you think of the specific organs what

you see is most organs will increase in size, such as the heart, lung, kidneys, and liver,

but they’ll usually reach a max of no more than two-fold change in size, and perhaps

a threefold change when you’re thinking of muscle mass, but again these organs responsible

for clearance of drugs do not increase by more than two-fold in size across an almost

five-fold distribution of body weight. Now this is important because again when we’re

scaling this information we really need to consider how does the change in body weight

impact the change in drug clearance? When you’re thinking about distribution and

metabolism in general there are not a lot of great studies that have done this, that’s

not really well characterized, but we can have general trends when we’re thinking about

drug distribution that can be based on the physiochemical properties of the drug. Typically, what we see is we we’re dealing

with drugs that are more acidic. We tend to see smaller volumes of distributions,

compared to drugs that are more basic that can be sequestered within tissue, and tend

to have larger volumes of distribution. When we’re looking at metabolism, specifically

looking at the cytochrome P450 system, what has been evaluated has been specific probes

of cytochrome P450 3A4, 2C9/19, and CYP2D6. And these studies that have looked at specific

probes of those pathways have seen limited effects of increasing body size and metabolism. An area where there has been a 1.5 to threefold

increase in clearance has been the cytochrome P450 2E1 isoenzyme system, that’s really not

responsible for the metabolism of a lot of drugs. This isoenzyme system is responsible for the

metabolism of more lipophilic compounds that are typically less than 200 daltons in weight. And so, this can impact some anesthetics. It can impact a key probe substrate knows

as chlorzoxazine — chlorzoxazone, sorry. And the metabolic profile of acetaminophen

into its more toxic metabolites. Again, this pathway does not influence the

metabolism of most drugs, so when we think about cyt-metabolism in general we do not

anticipate an increase in the metabolism as individuals increase in body size. Now when we’re accounting for these parameters

clearly weight, that we’ve spent quite of bit of time so far discussing, is really a

parameter that tends to correlate with the volume of distribution. When we look at clearance there are several

factors that are usually accounted for in population PK studies that look at race, height,

weight, age, sex, serum creatinine. There are also several intrinsic variables

that I won’t have an opportunity to cover today, but discuss — that include pharmacogenetic

variation. There could also be extrinsic variables such

as the impact of smoking, or dietary changes that can impact clearance. But typically, these are the parameters that

are looked it, and often these parameters are consolidated into a variable such as kidney

function that incorporates some of these parameters into a compositive parameter that correlates

with clearance. Again, we tend to treat in population PK analyses

these terms to be independent, but they can be inter-related. For example, height clearly correlates with

sex, and so that may not necessarily be an independent parameter when incorporated into

these models. What I did earlier is show changes in body

size, how those body size changes relate to tissues changes, and organ weights. But how do those actually relate to function? And so, this is a — one of the few studies

that’s done this really well. And this is a study done by Avry Chagnac and

colleagues, who have looked at healthy obese individuals. These are — again defining healthy obese

individuals is quite complicated. These are individuals who did not have hypertension

and diabetes, but were very large in body weight. And so, looking at these individuals: similar

age, they looked at glomerular filtration rates. And when you look at these again — although

individuals were almost twice in body size, what you see is only a 62 percent increase

in the glomerular filtration rate. And this is again consistent with several

animal model experiments that have also shown similar trends for increases in GFR relative

to body size. Another study by Friedman and colleagues has

looked at the population of pharmacokinetics of Iohexol, which is another marker that can

be used to compute glomerular filtration. And if you do a simulation using that population

model what you also see is if you look at the change in body weight between 50 and 200

kilograms you would not anticipate the glomerular filtration rate to increase by more than 60

percent. And so, there are several data sets that essentially

show that we do not expect body weight to increase mechanisms that would be associated

with drug clearance by more than 50 to 60 percent on average. Now in the clinical practice we often cannot

measure the creatinine clearance, or estimate the glomerular filtration rate. And so, we often rely on equations. And so, what I’d like to point out here is

that there are really two broad ways of doing this. One is using the EGFR, or Estimated GFR equations,

and this method currently incorporates serum creatinine, age, and race, but does not incorporate

body weight. Instead that parameter is sometimes modified

in individuals by converting this term from a scalar that’s benchmarked to body surface

area, to a non-normalized term. This equation also has been calibrated using

the isotopic dilution mass spectrometry traceable creatinine, which is currently the standard

to ensure the creatinine measurements across institutions are similar. The more classical equation that is actually

incorporated in clinical trials over the last 40 years has been the Cockcroft and Gault

equation. This equation that was introduced in 1976

is a simple equation, and is often taught in many schools, and incorporates the use

of age, weight, serum creatinine. This study actually did not include any females

in its design. And so, this was an assumption that was placed

into the equation with the expectation that women on average have 10 to 20 percent less

lean body weight, and so a term of a 15 percent lower weight was used in females as an expression. Again, this model was never really validated

in females, but has been used over time, and it’s shown to be useful for several drugs. Another equation that’s emerged is the chronic

kidney disease and epidemiology equation. This equation was developed primarily to resolve

the issue of the former equation knows as the modified diet and renal disease equation,

or MDRD equation, that was really restricted to individuals who have GFR estimates less

than 60 mil per 1.37 meters squared. So, what this equation accomplishes is it

permits the calculation of GFR across the GFR distribution, so values below 60 and above

60 mil per 1.37 meters squared. As seen by this equation, again it is much

more complicated because it includes several equations that account for sex, and race of

individuals, but clearly there has been controversy in the use of race as a factor in this equation,

especially in multi-ethnic societies. This equation has not necessarily worked out

in those populations. Now this brings us back to again use of this

term, because clearance often correlates with several kidney function markers, but we often

dose the drug on body weight for certain drugs. And if we’re going to dose the drug on body

weight a scenario that’s expected is that the clearance of the drug should increase

in a linear and proportionate weight. Instead what we see is that the relationship

is really non-linear. And the reason why this error may occur is

early phase clinical trials tend to include individuals within a relatively narrow bandwidth

of body weight. And so, if you include individuals that are

for example between 50 and 80 kilograms you may consider those individuals to have a linear

change in clearance. But when you include the larger body weight

distribution in your pharmacokinetic studies you’re more likely to see this curvature and

non-linearity. Now this phenomenon is not and old phenomenon. This phenomenon has been evaluated in several

disciplines, and so the principles that are used in pharmacology really relate to principles

that were generated in the early understandings of resting metabolism. And so, what we have here are two major paradigms

that develop — that have developed over time. And so, this is what I often refer to as the

battle of the Maxes. This is Max Rubner’s work, from the early

1880’s, to almost 50 years later by work by Max Kleiber. Max Rubner’s work was a pivotal understanding

of the relationship of heat production relative to body surface area. And so, his work included experiments that

obviously would not be conducted today, but involved the use of animals that were placed

in chambers and allowed to starve over time. And what was seen in dogs is heat production

declined as a function of their body weight scaled to 0.6 Max Rubner’s work really looked it scaling

resting metabolism across species, and in so — in those experiments what was shown

is a — when you plot log heat production relative to log body weight production — log

body weight, I’m sorry — what you see is a relationship that has a slow parameter of

0.75. And his original work actually was around

0.73, but again for simplification most literature reports it as .75. Now this is relevant when we’re thinking about

Allometry because we’re thinking about relationships between body size, shape, and physiology. And so often we think of disparate comparisons

as apples and oranges. In this scenario I’m giving you the example

of an apple versus a Romanesco. And this is relevant because when we’re thinking

about computation of surface area we have to think of the world as either being smooth

surfaces, or the true phenomenon which is more rough surfaces. So, when you think of smooth surfaces if we’re

going to compute the surface area of this apple we would have to think of this apple

in three dimensions. And the area would simply be the volume of

this cube to the two-third power, or .67. What’s been shown more eloquently now by West

and colleagues [spelled phonetically] is that our computation of surface and really physiology

are — these relationships are really driven by fractal geometry. And the area is better represented by the

volume of tissue or other spaces to the three-quarters power, or basically .75. And so, the work by Rubner and by Kleiber

basically exists within these two-expected paradigm in science. Now this is again relevant because our approaches

to computation of body surface area that are used for drug dosing also rely on Euclidian

geometry. And so, what I have here are basically graphics

from about a hundred years ago, that used simple computation of body surface area by

— one example here which is — this is an image from the work of DuBois and Du Bois. So, DuBois and Du Bois and colleagues essentially

took a small number of individuals and paper-mâché’d them. And after paper mâché-ing them they removed

the paper mâché placed them on the ground, and took photographs of that paper mâché. After taking those photographs, they then

computed the surface are and derived and equation. So again, that process would have required

a computation based on smooth surfaces, and the Dubois and Dubois equation is represented

here by this equation, which is basically body weight to an exponent, and height to

an exponent. And other individuals over time have thought

this equation to not be representative because again it was based on nine individuals, and

then basically included an additional 23 individuals in their model. So, Gehan and George tried to expand on this

model by studying a larger sample of individuals, but really came up with similar exponents. And I’m going to explain why that’s the case. Haycock and colleagues tried to expand this

to include pediatric populations; and then we have Mosteller, who wrote a simple letter

to the editor in the New England Journal of Medicine that simplified a lot of these equations. And this is the equation, again because of

its simplicity, is included in a lot of textbooks, and simply includes the weight times height

of individual, divided by thirty-six hundred, and is essentially the square root of that. Now I’m going to show you that all these equations

have similar answers because they all rely on Euclidian geometry. So, with Euclidian geometry what you have

is weight is a function of volume times density, and height as you can imagine is just a single

dimension. So, if you take the exponent over weight,

which is a three-dimensional term, and add it to height, which is a one-dimensional term,

what you will see with all equations that have been constructed today is all those values

add up to two. And the reason they add up to two, again,

it’s based on Euclidian geometry that has integer based dimensions, and that’s essentially

what you see with body surface area, which is meters squared — which is a meters squared

term. And so, these body surface equations — essentially

several of them have been developed over time, but really are very similar because they’re

scaling the information the same way. The DuBois and Du Bois equation will lead

to computation of slightly lower values, and the reason that occurs is because the weight

terms has an exponent of .45, so it has a smaller term over weight, and will lead to

computation of a smaller surface area. But again, the scaling is essentially the

same. So, this has dosing implications because if

the dose increases with weight or body surface area, but clearance does not increase with

the parameter, then what you would expect to see is a higher AUC in larger individuals. So, is it really acceptable when we think

about his, when we’re thinking about a milligram per kilogram dose? So, if you’re thinking about a drug, and the

drug’s product label is reported as six milligram per kilogram — if you use that same milligram

per kilogram in 40-kilogram individual you’re going to collect — you’re going to calculate

a dose of 250 milligrams if you use that same — in a 160-kilogram person you would calculate

a dose of a thousand milligrams. On average, you would calculate a dose of

500 milligrams. So, the question often is: Is that okay? Again, this is a common thing that we do. In practice, we address this issue often in

obesity by using another weight parameter. And so now let’s discuss what those other

weight parameters might be. One of those is lean body weight, which we

often think of as a good metric of muscle mass. Now measuring lean body weight is not as simple

as we would think. One: when you think about measurement, even

acquiring total body weight in a very large or morbidly obese individual is not simple. It’s not simple because a very large individual

may not be ambulatory, and so it’s difficult to actually get a body weight. You may also not have the right scales in

your institution to compute body weight for very large individuals. Computation of height also may be compromised

if this is not done correctly. Measurement can also include use of other

modalities such as bi-electric impedance analysis. You can have underwater weighing. You can have dexer [spelled phonetically],

which is an x-ray based method, or you can have a more recent method which relies on

air displacement, plethysmography, which again these systems can be used in healthy individuals,

but often not a system that can be used in acutely ill individuals. Because of this we often rely on estimation,

and that estimation happens with several equations such as the ideal body weight all the way

to the predicted normal weight, and I’m going to review some of those equations with you. When you think about these descriptors total

body weight is often sometimes referred to as actual body weight, so there’s different

rems used in the literature. And this is measured, and as I mentioned not

always easy to do. We have the ideal body weight equation, and

this is an equation that was developed over time, and referenced for the first time in

1974 by Ben Devine without an actual source. And as a resident, this was one of the first

questions I had to tackle, through a literature search, was finding the origin of the ideal

body weight equation. And what I discovered was it was simply a

rule of thumb, a farmer’s rule of thumb, that was based on our principle of fives, because

again what you see in the literature is this love for the number five, because we have

five fingers. We believe that five kilograms for every inch

over five feet would represent an ideal individual. And so, this term again that’s been used in

pharmacology was based on a very simple rule that men start at 110 pounds, and gain five

pounds for every inch, and women also gain the same weight for ever inch, which again

is not a reasonable hypothesis. Because the ideal body weight term did not

work for several drugs a modification was done, and that modification was called the

adjusted body weight. This was tested primarily with the aminoglycosides,

and a correction factor was found. Now that correction factor in the literature

actually ranges between 0.2 and 0.94, but on average is between 0.3 and 0.5. And so, what you see in a lot of text books

is the average of these averages, which is 0.4. And so, the adjusted body weight is simply

saying 40 percent of the difference between total and ideal body weight, plus ideal body

weight, is what we would term the adjusted body weight, and then use that weight to compute

the dose of the drug. Again, whether or not these are truly accurate

ways of representing body weight, what’s been shown is that they help dosing of certain

drugs. We also have other weight descriptors that

have been defined using more scientific methods, such as bi-electrical impedance, and also

based on animal data, and the best representation of that is the most recent equation knows

as the lean body weight 2005, or the Janmahasatian equation. This was based on 300 individuals in Australia

where they relied on body mass — body electrical impedance analysis to compete lean body weight. And this equation has been used in the literature

in more recent times, and computes lean body weight as a function of total body weight,

and a body mass index with slightly different parameters based on males and females. But again, all of these equations are essentially

transforming height and weight in to different metrics. And so that’s an important thing to remember,

that all we’re essentially doing is taking height and weight, and transforming them,

mathematically into another term. So, I want to highlight again what’s been

done in the literature, just to show that there is some harmonization in the principles

that I’ve laid out in this lecture. One of the approaches that’s often used in

pharmacy is using a combination of total body weight, ideal body weight, and adjusted body

weight. Now you can imagine if you’re using total

body weight to dose a drug, as a person increases in body weight you would get a proportionate

increase in the dose of the drug. And clearly that’s not necessarily a good

idea, especially in the extremes of weight. If you were to use that weight distribution

— actually if you were to use the height distribution, because again ideal body weight

is based simply on height, you would get a distribution of weights computed in this manner. If you then used adjusted body weight you

would get another distribution of weight. And then if you used a combination based on

this metric of — that’s often used in the literature you would get this distribution

of dosing weights across a population. And then if you model that data what you essentially

show is using those distribution of weights essentially gives you a dosing weight function

that’s three times total body weight to the .72 power. So, in essence what you see in the literature

are really divergent methods of actually dosing drugs, but in reality — in mathematical reality

these are really congruent approaches of scaling doses, and really the objective again, whether

you use body surface area, or you use an alternate body size descriptor, is you’re essentially

preventing someone from getting twice the dose as would a normal weight individual. So again, this brings, you know, the question

of who is right? Should we be scaling to this power or that

power? And the reality is that there really — the

answer exists somewhere in between that. When you look at this study by McLeay and

colleagues what they’ve demonstrated is it’s really drug dependent. And so, on average what you find is several

drugs will scale based on body surface area, but some drugs where the exponent is closer

to zero would imply that a fixed dosing strategy would be better, and other drugs using actual

body weight or total body weight may actually be beneficial for the dosing of the drug. But on average you would expect most drugs

to be dosed on a parameter such as body surface area. This is relevant because when we think about

drug development current paradigms include more physiologic based pharmacokinetic modeling

systems, and often what’s seen in the literature is the reporting of values based on a kilogram

basis, and that kilogram basis is often used to scale information. And so, it’s important to ensure that the

methods that are being used to scale information in obese individuals are scaling them using

some sort of factor, and not in a linear way. So, essentially what needs to be done is ensuring

that when this physiologic based models are used to derive estimates in obese individuals

that they’re being scaled appropriately based on information that’s actually derived from

the drug in question. So, now that I’ve gone through a lot of theory,

what I’d like to do is actually go over some key examples to illustrate how this impacts

drug dosing. When you think about the source of these different

weight descriptors the aminoglycosides serve as the key example. These are drugs that have a volume of distribution

between .25 and .35 liter per kilogram. What we’ve recognized over the last 70 years

with these class of drugs, where they were first classified as anti-biotics, are that

there’s a higher risk for toxicity when you’re dosing them on total body weight, and that

you can adjust the doses of these drugs based on kidney function. And in this case, we actually also have therapeutic

drug monitoring available that can allow you to modify the maintenance dose. There are several alternate body size descriptors

that have been used. These drugs are relatively small in size,

they have low plasma protein binding. We also know that the clearance of this drug

correlates very well with glomerular filtration rate. The dosing of this drug is based on body weight,

and when you look at the literature there’s a range of doses between 1 and 7 milligram

per kilogram, based on the indication of the drug. And again, in most institutions the dosing

is individualized based on therapeutic drug monitoring. This is a study that I conducted over a decade

ago, looking at almost 20,000 individuals who were dosed on gentamicin and tobramycin,

across a very wide body weight distribution of 30 to almost 206 kilograms. What we showed in this data set is that if

you were to rely on total body weight to scale the volume of distribution of the drug what

happens is you get an unsteady estimate of the volume of distribution; that is the volume

of distribution parameter goes down as the body weight goes up. If you were to use ideal body weight you see

the opposite phenomenon. And again, this is because ideal body weight

as a function of height, and not weight. Instead, if you were to use the equation that

I mentioned, the lean body weight 2005 equation, what you get is similar estimates of lean

body — of volume of distribution across the body distribution. So, the implications of that is if you were

to define a dose based on lean body weight you’re more likely to have a predictable Cmax

concentration for this drug, that is considered to be a concentration dependent or Cmax, or

AUC driven drug, and I’ll show you how this actually also can affect the AUC of the drug. This finding also matches what was seen in

animal models. So, work done by Salazar and colleagues showed

a value that was also similar in rats, when they scaled the information to fat free mass

or lean mass. Now with aminoglycosides, when you think about

the pharmacodynamics, they’re driven both by the peak to MIC and also by the AUC to

MIC. And so, the area under the curve is also important,

and so that’s driven by the clearance of the drug. And so, what we did is we also evaluated all

the different equations that could exist to compute the clearance of the drug, and see

what the correlations are. And when you look at that what we found is

that the cracov [spelled phonetically] — the chronic kidney disease and epidemiology equations

actually gave us the best correlation. But it’s really important to show that even

for a drug class like the aminoglycosides, that are considered to be well correlated

to clearance, that an equation like the CKD equation only explains 50 percent of the inter-individual

variability in the clearance of the aminoglycosides. And so, this is again the rationale for using

therapeutic monitoring to modify the dose of this drug. So, when you think about initial dose selection

of aminoglycosides, where perhaps gearing it to a Cmax to MIC target, or an AUC to MIC

target, but if you’re thinking about it from an AUC target we would consider that milligram

per kilogram dose of this drug, and we may consider different approaches. So, when you’re thinking about this drug use

of tobramycin in cystic fibrosis patients, what’s published in the literature is use

of a higher milligram per kilogram dose, or 10 milligram per kilogram. And the reason this makes sense is cystic

fibrosis individuals tend to be leaner, and because they’re lower in body weight the expectation

is the need for a higher milligram per kilogram dose. In contrast, if you’re going to think about

dosing this drug across a weight distribution, what could be considered is if you’re using

higher — if you have individuals across a higher weight distribution what may be necessary

is using a lower milligram per kilogram term. And so, this again fits within the paradigm

because the paradigm is based on 5 to 7 milligram per kilogram, but you may make the decision

to use a lower milligram per kilogram in a larger individual. And then the third alternative is saying this

may be a confusing algorithm. So instead if you could use a fixed milligram

per kilogram across a population, and then use lean body weight you would then compute

again very similar doses, but have a simpler metric to dose across a weight distribution. Another approach could be consideration of

kidney function as the dosing strategy, and this is — if you believe the AUC to MIC to

be the driver of the relationship. In this scenario we’ve published equations

that demonstrate how this could be done, and essentially you would use the creatinine clearance

estimate using a Cockcroft-Gault equation, to compute an aminoglycoside clearance, to

compute an initial dose based on this target value. This article for reference also relays how

the information can then — if therapeutic drug monitoring is applied, equations are

provided that can be used to compute the AUC of the drug to modify the dosing of the drug. Another drug where this is seen to be quite

relevant is with the dosing of Vancomycin. This dosing historically has been thought

to be reasonable, based on actual or total body weight, because the volume of distribution

of this drug is thought to be very similar to total body water estimates in most individuals. Now there could be multiple approaches that

are used. So, one approach that’s used in current guidelines

is to dose the drug based on total body weight. And so, when you look at that, using total

body weight you were — if you were to use 25 milligram per kilogram you will clearly

compute a much higher dose in a larger individual. And so, for most clinicians this may lead

to consideration of too high of a dose. The alternate in other guidelines when you’re

looking at the methicillin resistant staph OREUS [spelled phonetically] guidelines the

recommendation is to use no more than 2,000 milligrams as the dose. And so, in this scenario in a larger individual

you may end up using a 13 milligram per kilogram dose, and so for some clinicians that may

be too low of a dose. So, what would be the alternative? One of the alternatives would be to use the

same milligram per kilogram dosage, but then use a different body weight descriptor, and

so as the individual gets larger instead of using total body weight, use the adjusted

body weight function, and you would compute much lower doses. A simpler alternative could be to scale the

doses, and so since most individuals have calculators that have square root function. What I’m showing here is if you take the weight

of the individual divided by the average weight, and take the square root of it, you’d basically

be able to replicate the dosing of this drug across a population that would match up to

adjusted body weight dosing. When you’re thinking about Vancomycin, again

this is a drug that also undergoes therapeutic drug monitoring. What has happened over time is guidelines

that have suggested that only trough concentrations are necessary. But a point that needs to be highlighted is

that trough concentrations do correlate with the AUC of this drug, but only predict about

40 percent of the inter-individual variability. And so, the trough concentration is a simple

metric. It’s easy because a single concentration can

be measured, but it may not represent the true AUC in a specific individual. What we’ve shown in a series of studies, is

the importance of getting a peak concentration measurement in obese individuals. And this is really important because basing

approaches that can be used to compute the AUC of a drug like Vancomycin in a population

really needs an accurate estimate of the volume of distribution. And so, you can imagine if you are missing

a peak concentration measurement you don’t actually know which concentration time profile

truly represents the individual. And so, you can have different scenarios represented

here, with the true scenario represented by the actual concentration measurement, and

a two-compartment model. And so again when we think about achieving

the right dose in an obese individual, a common issue is not being able to approximate what

the true volume of distribution is for the specific drug, and so in this scenario what

we demonstrate is the importance of peak concentration measurement. More recently we’ve published a study looking

at the pharmacokinetics of Vancomycin, and looking at alternate metrics. As I mentioned, with the several equations

what we’ve relied on for over a hundred years is using height and weight to compute alternate

body size descriptors. Stewart Wang and colleagues at the University

of Michigan have led a group known as the Morphomics Group, and these individuals have

developed mathematical algorithms that can take existing computer tomography data, and

convert them to different body size metrics. So, this graphical representation is taking

existing data from individuals who are in the hospital who may have had a CT scan done

for some medical reason. And then taking that data to compute parameters

such as body depth, facial area, total solus area, and several other parameters represented

in this slide. What we then did is look retrospectively at

the pharmacokinetic profile of Vancomycin, and assess the correlation of these pharmacokinetic

parameters to these newer body composition metrics. What we were able to clearly demonstrate is

that the volume of distribution of Vancomycin was poorly predicted by body weight, and was

really better predicted by T12 to T4, which is representing again the vertebral columns

T12 to T4 torso area, was a better correlant [spelled phonetically]. But again, looking at this you’d see a very

poor correlation, but in relative terms a better correlation to body weight. We were also able to demonstrate that using

total solus area that — which would be a metric representative of muscle mass, that

this metric was a better predictor of clearance than relying simply on body weight. And so, again this more recent study is clearly

not ready for prime-time use, but just represents a movement away from our simple measures of

height and weight to define drug dosing. For other drugs that are just dosed on a fixed

dosing basis, drugs like Ceftaroline, that — who’s pharmacodynamics are based on time

above MIC, which you’ll see for several drugs and shown earlier, is reductions is the peak

concentration, but really a convergence in the profile. And so, this study by Justo and colleagues

out of Keith Rodvold’s group at the University of Illinois at Chicago, have clearly shown

that for certain drugs like beta lactams you may need higher doses with the first dose,

but really because of this convergence in the profile maintenance doses probably do

not need to be adjusted for most beta lactams. For another drug like Levofloxacin, a drug

that’s concentration dependent, what’s shown in the label is use of a higher milligram

dose that allowed really shortening of the dosage regimen from 10 days to a shorter regimen

in patients with pneumonia. For this drug the AUC to MIC is predictive

of the response, and the observed AC of this drug is between 50 and 150 milligram per hour

liter. This drug also has a really good correlation

to kidney function, and because the clearance of this drug has a good correlation in theory

the dosing of this drug could be improved by computing the clearance of the drug relative

to creatinine clearance. Now in the United States we do not offer therapeutic

drug monitoring for drugs like Levofloxacin, but this study conducted in collaboration

with Dr. Federico Pea at the University of Udine in Italy; what we’re able to show is

therapeutic drug monitoring can be used to improve the dosing of drugs in individuals

across a much larger body weight distribution of 98 to 250 kilograms. What we’re able to show is those individuals

may need doses higher than 750 milligrams to achieve isometric exposures to those that

are smaller in size. But again, the critical piece here is that

therapeutic drug monitoring was available to ensure that we didn’t overdose individuals. And so, this recommendation of using higher

doses is truly off label, and cannot be recommended in clinical practice in the United States,

but for institutions that do have therapeutic drug monitoring it does create a mechanism

to consider uses of higher doses of drugs to achieve the exposures necessary to improve

outcome of a drug like Levofloxacin. Additional examples for other drugs that are

not anti-microbials include work by Dr. Katherine Neib’s [spelled phonetically] group at the

University of Leiden. This group has done several studies looking

at specific probes of CYP-metabolism. And they’ve done a really nice study more

recently with Midazolam, looking at a population of individuals who’ve undergone bariatric

surgery. And so, this is an important study because

this study allows an evaluation of individuals as their own controls. So, this would represent — this figure here

represents individuals who are larger, who are obese, who undergo bariatric surgery,

and over a one-year span lose quite a bit of weight. And because they lose weight, and you can

measure pharmacokinetics before and after weight loss, you can look at changes in drug

clearance. And what they’ve shown with Midazolam, which

is a probe of CYP3A4, is that the systemic clearance increased with this drug with weight

loss. But there’s really no alterations in oral

bioavailability. And so, this expectation is different again,

because we would think in terms of do we need to increase the dose for body size? In this scenario the idea is we perhaps may

need to lower certain doses of certain drugs if this is really true in obese individuals. For other drugs like acetaminophen this is

actually metabolized by multiple pathways, and so it’s not really a good probe for CYP2E1. But this is an important study because this

study shows a shift in the metabolic profile in morbidly obese individuals, represented

by the blue box plot versus the green box plot of non-obese individuals. What you see is an increase in the cystine

[spelled phonetically], and although not statistically significant an increase in the mercaptopurate

[spelled phonetically] metabolite of acetaminophen, metabolites that would be associated with

the hepatotoxicity potential of this drug. And so, in this scenario even though clearance

may increase with the drug like acetaminophen, recommendations cannot be made to increase

the dosage of a drug like acetaminophen, because this may lead to an increase in toxic metabolites

of the drug. And so, studies like this are really important

too, because they can give us good insights on whether doses should be changed in certain

populations, but really if metabolites are also changing in those — in obese individuals. Another important compound that’s been evaluated

is Propofol. And so, this study again be Neib and colleagues

looked at the glucuronidation profile of Propofol, and really were able to show that the clearance

of a compound like Propofol, which is a very low molecular weight, small compound, best

scales allometrically, and this is going to be again a phenomenon that’s seen with several

drugs again, is that for most of these drugs we would expect that the clearance of the

drug to scale to an exponent of .67 or .72. And that is essentially what was seen in this

study, implying that perhaps body surface area or a metric like that may be reasonable

to the dosing of Propofol. Now this has been actually investigated in

much more detail by Eleveld and colleagues, who’ve taken data sets across different weight

distributions, and across different populations of adults, children, and elderly, and have

come up with a much more complex and comprehensive model that not just deals with the issue of

obesity, but really deals with the overall profile. And this is a process that’s being heralded

through the open TCI Initiative, and is an important one to help improve the dosing of

compounds like Propofol that are used in anesthesia. So, what I’ve presented today is a lot of

different paradigms that are currently used to define the dosing of drugs in obesity,

but it’s important to consider that if we have some drugs where we shouldn’t be dosing

them in body weight — rather we should be dosing them on a fixed basis, how do we change

that dosing paradigm? And if we’re going to do that how do we pay

attention to improving the dosing of a drug once a drug is marketed? So, what I’d like to present in the last few

slides here is data with Daptomycins. This is a study that I performed about a decade

ago at the University of New Mexico, where we looked at morbidly obese versus normal

weight individuals, and dosed Daptomycin based on body weight. And what we saw is a very small change in

the volume of distribution of this drug. Even though that the weight of the individual

was almost twice as high in the morbidly obese versus the normal weight individuals the volume

of distribution did not increase proportionately to body weight. We also used bi-electrical impedance analysis

to compute the fat free mass of these individuals. And when we did that we actually were able

to normalize the volume of distribution, implying that total body weight is not the right metric

for this drug. When we looked at clearance of this drug we

were also able to show that the clearance of this drug does not increase proportionately

to body weight, and again this parameter was scaled better with fat free weight. And in this scenario, we also looked at measurement

of clearance because these individuals were matched on their sodium iothalamate glomerular

filtration rate. So, this is a well performed study, and we

were able to show again that neither of these parameters really scaled to body weight. And this is important because this drug is

currently dosed on body weight on a 4 milligram per kilogram and six milligram per kilogram

basis, with specific guidance to not modify the dose for obesity. There are also recent guidance that suggests

that the dose of this drug should be increased to 8 to 10 milligram per kilogram, which would

imply much higher exposures in obese individuals. More recently we published a study in collaboration

with Marco Falcone and colleagues at the University of Rome, and this study was based in individuals

that were critically ill. And in these 50 individuals what we were able

to show again is that body weight does not correlate with the clearance of this drug. Rather, there were certain individuals who

had augmented clearance of this drug. And in fact, what we were able to show is

individuals who had bacteremias, who were sicker individuals, tended to have higher

clearance. And that clearance was not related to body

weight. Again, we were able to discover that through

therapeutic drug monitoring, which is widely not available for this drug. Now if our findings are consistent with the

original findings from over a decade ago. So, if you look — were to look at the original

population PK model for a drug like Daptomycin what you would see is the clearance of this

drug was actually related to body temperature, which is — in most cases when you’re looking

at this you’d say that this is an odd term to incorporate into a population model. But what it really represents is this idea

of illness. If someone is — has a severe infection they’re

more like to be febrile, and if they’re likely to be febrile their temperature will be higher. And so, you would compute a higher clearance. They also incorporated renal clearance in

the term, but when you looked at the function most of that clearance is not really driven

my kidney function. There’s just a value of 0.807, which is the

same value that we saw in our healthy volunteers. Now if you look at this again, tabulate it,

and using a reference population what you would expect is, even if the kidney function

increased and body temperature increased, your maximal expectation for clearance across

that population would be no more than a 70 percent higher clearance, which would imply

that the dosing of the drug, or the absolute dosing of the drug does not need to be increased

by more than two-fold. So again, if you were going to dose a 50-kilogram

person a 150-kilogram person, you would dose that individual, the 150-kilogram person,

three times more if you used total body weight. And clearly the mathematical expectations

are that that would be too high. This has also been shown in febrile neutropenic

patients, when you’re thinking about the effects of — severity of illness, and clearance. Now this is problematic because how do we

solve this issue once a drug is on the market? So, the product label currently recommends

that the drug be dosed on total body weight. And so, some investigators have suggested

in larger individuals — in larger individuals to switch scalars. So instead of using total body weight, because

you would accidentally perhaps calculate too high of a dose, consider switching to lean

body weight. And why this is an issue is if you were to

look at the distribution of lean body weight across a population — this is a representation

of the lean body weight in males, and this is a representation in females — if you were

to use these distributions you would get this phenomenon. You would dose individuals between and 50

as 111 kilograms on total body weight, and so what would happen is you would compute

a dose between 300 and 666 milligrams. And then by switching scalars what you’d essentially

do is drop the dose for individuals that are larger. So, the effect of switching scalars for some

drugs is you would end up giving larger individuals much smaller doses than they need. And so, our group has proposed consideration

of fixed dosing strategy for this drug, to essentially give similar doses across a population,

but that recommendation also needs to be validated through a prospective steady. So, to summarize the key points that I’ve

laid out to day: obesity is associated with the changes in volume of distribution that

might require the use of higher initial doses relative to maintenance doses. Our expectation is obesity has limited changes

in drug clearance, and that for the majority of drugs those changes are likely explained

by Allometry. We also expect that total body weight may

be reasonable for an initial dose, but it’s really unlikely to be a useful metric for

defining the maintenance dose. And the loading dose of a drug could be used

to aid the dosing of drugs that are time dependent — pharmacokinetics. But we should really move toward consideration

of body size stratified or composition stratified dosing regimens for computation of the maintenance

dose of a drug. With that I’d like to thank you for your attention. I’d also like to thank Dr. William Figg, and

Dr. Lisa Cordes, the National Institutes of Health for this opportunity to present on

the principles of clinical pharmacology. If you have any questions, please direct them

to the coordinators of this course.