From: GREAT ORMOND STREET HOSPITAL FOR CHILDREN NHS
FOUNDATION TRUST <*joseph.standing *>

Date: Thu, 7 Dec 2017 13:00:51 +0000

Dear Rob,

Why do you want to use a model to predict the value of a covariate to add i=

nto your model? Apart from glomerular filtration rate, what other situatio=

ns would you do this?

Unless I was trying to do some fancy separation of renal and non-renal clea=

rance, I would simply ignore the fact there is a model to predict glomerula=

r fltration, and include its component parts e.g.

CL = THETA * (WT/70)**0.75 * FCREAT * FAGE

where FCREAT and FAGE are covariate functions for creatinine (e.g. (SECR/me=

dian value)**THETA ) and age.

Some examples of not using a model to predict GFR still gave an acceptable =

model of CL (what we were interested in):

Creatinie e.g.

Hennig S. Population pharmacokinetics of tobramycin in patients with and wi=

thout cystic fibrosis. Clin Pharmacokinet. 2013 Apr;52(4):289-301

Cystatin C example:

De Cock PA. Augmented renal clearance implies a need for increased amoxici=

llin-clavulanic acid dosing in critically ill children. Antimicrob Agents C=

hemother. 2015 Nov;59(11):7027-35.

BW,

Joe

Joseph F Standing

MRC Fellow, UCL Institute of Child Health

Antimicrobial Pharmacist, Great Ormond Street Hospital

Honorary Senior Lecturer, St George's University of London

Tel: +44(0)207 905 2370

Mobile: +44(0)7970 572435

________________________________________

From: owner-nmusers

of R.terHeine

Sent: 07 December 2017 11:56

To: nmusers

Cc: n.holford

com; j.h.proost

Subject: RE: [NMusers] Allometric scaling of renal clearance with estimated=

glomerular filtration rate

Dear Nick, Hans, Ruben, Max,

Great to hear several approaches and opinions on the use of glomerular filt=

ration approximations in PK modeling and scaling to body size. Thank you!

I have a hard time ignoring the CKD-EPI equations (with or without cystatin=

C), as they are well established and proven better predictors for GFR, whe=

n compared to the Cockroft-Gault. In general, sample sizes of pharmacokinet=

ic studies are smaller than those where the CKD-EPI and MDRD equations were=

developed. I am not convinced that developing a new creatinine/cystatin c =

equation for GFR for each PK analysis is the right approach. Then again, I =

also have a hard time scaling to BSA, as in, for example, obese patients th=

is is likely a poor body size descriptor to scale renal function.

Also, depending on the population and drug one may choose one equation abov=

e another. For example: if a drug is completely filtrated (no active secret=

ion), a cystatin C based equation is likely better explain variability in c=

learance of completely filtrated drugs (e.g. carboplatin). Another example:=

in cachectic patients one may argue that there is not enough muscle mass (=

and thus serum creatinine) to provide accurate GFR estimations and then cre=

atinine-independent equations may provide better equations.

Thinking about this the last couple of days and with your feedback, I am in=

clined to choose the equation based on population (e.g. cachectic or not?) =

and drug (e.g. filtration/active secretion) and, if the equation scales ren=

al function to BSA, convert it to scaling to FFM. Nonetheless, open to any =

other suggestion or discuss cons and pro's anytime!

Sincerely,

Rob

-----Oorspronkelijk bericht-----

Van: owner-nmusers

ens Nick Holford

Verzonden: woensdag 6 december 2017 20:06

Aan: nmusers

Onderwerp: Re: [FORGED] [NMusers] Allometric scaling of renal clearance wit=

h estimated glomerular filtration rate

Hi Rob,

Thanks for bringing this up again. I don't think much has changed since I w=

rote this in 2013

(http://cognigencorp.com/nonmem/current/2013-August/4697.html)

1. Theory Based Allometry or Surface Area

"Note that using surface area as a form of size standardization forglomerul=

ar filtration rate has no theoretical nor experimental support when compare=

d to theory based allometry (Rhodin et al. 2009). So I donot agree with sta=

ndardizing CLCR to 1.73 m^2. I know this is frequently done but in fact thi=

s is just based on tradition and an out of datetheory of scaling based on s=

urface area (see Anderson & Holford 2008)."

There is no biological or experimental support for using surface area to sc=

ale renal function markers such as GFR and CLcr. In contrast, there is str=

ong biological based theory and experimental support for using theory based=

allometry (see Holford & Anderson 2017 for a recent review).

2. Mechanism Based Models for CLcr

I also wrote in 2013:

"The MDRD method of predicting glomerular filtration rate is astatistical a=

bsurdity which does not include any measurement of size for its prediction.=

I would certainly not recommend using it for anyscientific purpose. "

This applies equally well to the CKD-EPI method. Let me explain why it is a=

absurdity generated by a naive statistician using CLcr as an example.

CLcr can be calculated from the creatinine excretion rate (CER) and the ser=

um creatiniine. This is based on the definition of clearance and is true wi=

thout any assumptions.

CLcr=CER/Scr

If we then assume Scr is at steady state then CER will be equal to creatini=

ne production rate (CPR) and we can use this:

CLcr=CPR/Scr

All rational models for predicting CLcr without measurement of CER use mode=

ls to predict CPR e.g.

CPR=(140-Age)*Weight/72 use Cockcroft & Gault to predict CPR in males th=

en CLcr=CPR/Scr is Cockcroft & Gault CLcr ml/min

Dividing CPR by Scr gives the CLcr. This can be written equivalently but le=

ss clearly:

CLcr=CPR*Scr^-1

The empirical models such as MDRD and CKI-EPI (see below) involve the absur=

dity of estimating the known exponent for Scr of -1. These estimates must b=

e wrong based on the theory I have outlined above (unless the estimate is e=

xactly -1). The reported estimates are -1.209 for CKI-EPI and -1.154 for MD=

RD.

In addition, and more importantly,they have no direct measure of body size =

which seriously limits the value outside the typical weight distribution an=

d they are only applicable to adults. GFR can be described from premature n=

eonates to adults using theory based allometry and maturation based on post=

-menstrual age so GFR predicttions should try to follow the concepts used t=

here (Rhodin 2008).

So what to do?

First -- don't use MDRD or CKI-EPI unless you are sure you are applying the=

m to a population similar to that used to develop these empirical predictio=

ns. You could add allometric scaling to the eGFR by assuming the 1.72m^2 va=

lue is equivalent to 70 kg with a fat free mass (FFM) of

56.1 kg. Then scaling the eGFR by (WT/70)^(3/4) or (FFM/56.1)^(3/4).

I use the Schwartz (1992) equations for neonates, children and teenagers th=

en the Matthews (2004) equation for adults. I am working on an integrated m=

ethod for CPR prediction which was presented as a work in progress at PAGE =

this year. Watch this space...

Best wishes,

Nick

MDRD

eGFR =175 x (SCr)^-1.154 x (age)-0.203 x 0.742 [if female] x

1.212 [if Black]

CKI-EPI

eGFR = 141 x min(SCr/k, 1)^alpha x max(SCr /kappa, 1)^-1.209 x 0.9=

93^Age x 1.018 [if female] x 1.159 [if Black]

kappa = 0.7 (females) or 0.9 (males)

alpha = -0.329 (females) or -0.411 (males)

eGFR (estimated glomerular filtration rate) = mL/min/1.73 m2; SCr (standa=

rdized serum creatinine) = mg/dL

Holford NHG, Anderson BJ. Allometric size: The scientific theory and

extension to normal fat mass. Eur J Pharm Sci. 2017;109(Supplement):S59-S64=

.

Rhodin MM, Anderson BJ, Peters AM, Coulthard MG, Wilkins B, Cole M,

Chatelut E, Grubb A, Veal GJ, Keir MJ, Holford NH

Human renal function maturation – a quantitative description using

weight and postmenstrual age. Pediatr Nephrol. 2008

Schwartz GJ. Does kL/PCr estimate GFR, or does GFR determine k? Pediatr

Nephrol. 1992;6(6):512-5.

Matthews I, Kirkpatrick C, Holford N. Quantitative justification for

target concentration intervention -- parameter variability and

predictive performance using population pharmacokinetic models for

aminoglycosides. Br J Clin Pharmacol. 2004;58(1):8-19.

Holford N, Sherwin CM. Scaling renal function in neonates and infants to

describe the pharmacodynamics of antibiotic nephrotoxicity. PAGE 26

Abstr 7208 [wwwpage-meetingorg/?abstract=7208]. 2017.

On 06-Dec-17 23:52, R.terHeine

*>
*

*> Hi Ruben,
*

*>
*

*> Interesting work, Ruben. One may indeed question the validity of
*

*> glomerular filtration rate markers like cystatin C (that is only
*

*> filtrated and not actively secreted) to predict PK of drugs that
*

*> undergo active tubular secretion in patients with decreased renal
*

*> function. When glomerular filtration rate drops, the relative
*

*> contribution of active tubular secretion to renal clearance increases.
*

*> To me, it appears logical that creatinine is a better marker for
*

*> clearance drugs that are actively secreted, as creatinine also
*

*> undergoes active tubular secretion.
*

*>
*

*> Nonetheless, I’m also interested whether other people have considered
*

*> allometric scaling of MDRD/CKD-EPI derived GFR’s?
*

*>
*

*> Cheers,
*

*>
*

*> Rob
*

*>
*

*> *Van: *Ruben Faelens <ruben.faelens *

*> *Datum: *dinsdag 5 december 2017 om 7:13 PM
*

*> *Aan: *"Heine, Rob ter" <R.terHeine *

*> *CC: *"nmusers *

*> *Onderwerp: *Re: [NMusers] Allometric scaling of renal clearance with
*

*> estimated glomerular filtration rate
*

*>
*

*> Dear Rob,
*

*>
*

*> At PMX Benelux, there was an interesting talk about the correlation
*

*> between different metrics describing renal function by Stijn Jonckheere.
*

*> A part of the work presented was published:
*

*> https://academic.oup.com/jac/article/71/9/2538/1750427
*

*> <https://academic.oup.com/jac/article/71/9/2538/1750427>
*

*>
*

*> This may provide some perspective, or rather complicate things even
*

*> more, depending on your viewpoint.
*

*>
*

*> Best regards
*

*> Ruben Faelens
*

*>
*

On 06-Dec-17 06:17, R.terHeine

*>
*

*> Dear all,
*

*>
*

*> I am wondering what your thoughts are on the allometric scaling of
*

*> clearance of renally extreted drugs, where we have estimations renal
*

*> function.
*

*>
*

*> Simply scaling the predicted glomerular filtration rate from, for
*

*> example, the Cockroft-gault equation seems inappropriate, since weight
*

*> is already a part of the equation. Standardizing this to weight in the
*

*> Cockroft-gault equation can be done, a solution has been discussed
*

*> here: http://cognigencorp.com/nonmem/current/2013-August/4697.html
*

*>
*

*> However, in the recent years some new equations to calculate
*

*> glomerular filtration rate from endogenous markers have emerged. For
*

*> example the CKD-EPI CREATININE CYSTATIN C equation
*

*> https://www.kidney.org/content/ckd-epi-creatinine-cystatin-equation-2012
*

*> . As the addition of a muscle mass independent endogenous marker like
*

*> cystatin C is known to provide better estimations of GFR in, for
*

*> example, cachectic patients, it is likely that this equation may
*

*> outperform to predict renally filtrated compounds in this patient
*

*> group. It is rather odd that this CKD-EPI equation does not contain
*

*> any measure of body size. The outcome of this equation is a GFR scaled
*

*> to a BSA of 1.73m^2.
*

*>
*

*> I am wondering how you would allometrically scale the eGFRs from these
*

*> CKD EPI equations to, for example, fat-free mass.
*

*>
*

*> Cheers!
*

*>
*

*> Rob
*

*>
*

*> R. ter Heine, PhD, PharmD
*

*>
*

*> Hospital Pharmacist-Clinical Pharmacologist
*

*>
*

*> Radboudumc, Nijmegen, The Netherlands
*

*>
*

*> Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het
*

*> handelsregister onder nummer 41055629.
*

*> The Radboud university medical center is listed in the Commercial
*

*> Register of the Chamber of Commerce under file number 41055629.
*

*>
*

--

Nick Holford, Professor Clinical Pharmacology

Dept Pharmacology & Clinical Pharmacology, Bldg 503 Room 302A

University of Auckland,85 Park Rd,Private Bag 92019,Auckland,New Zealand

office:+64(9)923-6730 mobile:NZ+64(21)46 23 53 FR+33(6)62 32 46 72

email:n.holford

http://holford.fmhs.auckland.ac.nz/

http://orcid.org/0000-0002-4031-2514

Read the question, answer the question, attempt all questions

Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het hande=

lsregister onder nummer 41055629.

The Radboud university medical center is listed in the Commercial Register =

of the Chamber of Commerce under file number 41055629.

***************************************************************************=

*****************************************

This message may contain confidential information. If you are not the inten=

ded recipient please inform the

sender that you have received the message in error before deleting it.

Please do not disclose, copy or distribute information in this e-mail or ta=

ke any action in relation to its contents. To do so is strictly prohibited =

and may be unlawful. Thank you for your co-operation.

NHSmail is the secure email and directory service available for all NHS sta=

ff in England and Scotland. NHSmail is approved for exchanging patient data=

and other sensitive information with NHSmail and other accredited email se=

rvices.

For more information and to find out how you can switch, https://portal.nhs=

.net/help/joiningnhsmail

Received on Thu Dec 07 2017 - 08:00:51 EST

Date: Thu, 7 Dec 2017 13:00:51 +0000

Dear Rob,

Why do you want to use a model to predict the value of a covariate to add i=

nto your model? Apart from glomerular filtration rate, what other situatio=

ns would you do this?

Unless I was trying to do some fancy separation of renal and non-renal clea=

rance, I would simply ignore the fact there is a model to predict glomerula=

r fltration, and include its component parts e.g.

CL = THETA * (WT/70)**0.75 * FCREAT * FAGE

where FCREAT and FAGE are covariate functions for creatinine (e.g. (SECR/me=

dian value)**THETA ) and age.

Some examples of not using a model to predict GFR still gave an acceptable =

model of CL (what we were interested in):

Creatinie e.g.

Hennig S. Population pharmacokinetics of tobramycin in patients with and wi=

thout cystic fibrosis. Clin Pharmacokinet. 2013 Apr;52(4):289-301

Cystatin C example:

De Cock PA. Augmented renal clearance implies a need for increased amoxici=

llin-clavulanic acid dosing in critically ill children. Antimicrob Agents C=

hemother. 2015 Nov;59(11):7027-35.

BW,

Joe

Joseph F Standing

MRC Fellow, UCL Institute of Child Health

Antimicrobial Pharmacist, Great Ormond Street Hospital

Honorary Senior Lecturer, St George's University of London

Tel: +44(0)207 905 2370

Mobile: +44(0)7970 572435

________________________________________

From: owner-nmusers

of R.terHeine

Sent: 07 December 2017 11:56

To: nmusers

Cc: n.holford

com; j.h.proost

Subject: RE: [NMusers] Allometric scaling of renal clearance with estimated=

glomerular filtration rate

Dear Nick, Hans, Ruben, Max,

Great to hear several approaches and opinions on the use of glomerular filt=

ration approximations in PK modeling and scaling to body size. Thank you!

I have a hard time ignoring the CKD-EPI equations (with or without cystatin=

C), as they are well established and proven better predictors for GFR, whe=

n compared to the Cockroft-Gault. In general, sample sizes of pharmacokinet=

ic studies are smaller than those where the CKD-EPI and MDRD equations were=

developed. I am not convinced that developing a new creatinine/cystatin c =

equation for GFR for each PK analysis is the right approach. Then again, I =

also have a hard time scaling to BSA, as in, for example, obese patients th=

is is likely a poor body size descriptor to scale renal function.

Also, depending on the population and drug one may choose one equation abov=

e another. For example: if a drug is completely filtrated (no active secret=

ion), a cystatin C based equation is likely better explain variability in c=

learance of completely filtrated drugs (e.g. carboplatin). Another example:=

in cachectic patients one may argue that there is not enough muscle mass (=

and thus serum creatinine) to provide accurate GFR estimations and then cre=

atinine-independent equations may provide better equations.

Thinking about this the last couple of days and with your feedback, I am in=

clined to choose the equation based on population (e.g. cachectic or not?) =

and drug (e.g. filtration/active secretion) and, if the equation scales ren=

al function to BSA, convert it to scaling to FFM. Nonetheless, open to any =

other suggestion or discuss cons and pro's anytime!

Sincerely,

Rob

-----Oorspronkelijk bericht-----

Van: owner-nmusers

ens Nick Holford

Verzonden: woensdag 6 december 2017 20:06

Aan: nmusers

Onderwerp: Re: [FORGED] [NMusers] Allometric scaling of renal clearance wit=

h estimated glomerular filtration rate

Hi Rob,

Thanks for bringing this up again. I don't think much has changed since I w=

rote this in 2013

(http://cognigencorp.com/nonmem/current/2013-August/4697.html)

1. Theory Based Allometry or Surface Area

"Note that using surface area as a form of size standardization forglomerul=

ar filtration rate has no theoretical nor experimental support when compare=

d to theory based allometry (Rhodin et al. 2009). So I donot agree with sta=

ndardizing CLCR to 1.73 m^2. I know this is frequently done but in fact thi=

s is just based on tradition and an out of datetheory of scaling based on s=

urface area (see Anderson & Holford 2008)."

There is no biological or experimental support for using surface area to sc=

ale renal function markers such as GFR and CLcr. In contrast, there is str=

ong biological based theory and experimental support for using theory based=

allometry (see Holford & Anderson 2017 for a recent review).

2. Mechanism Based Models for CLcr

I also wrote in 2013:

"The MDRD method of predicting glomerular filtration rate is astatistical a=

bsurdity which does not include any measurement of size for its prediction.=

I would certainly not recommend using it for anyscientific purpose. "

This applies equally well to the CKD-EPI method. Let me explain why it is a=

absurdity generated by a naive statistician using CLcr as an example.

CLcr can be calculated from the creatinine excretion rate (CER) and the ser=

um creatiniine. This is based on the definition of clearance and is true wi=

thout any assumptions.

CLcr=CER/Scr

If we then assume Scr is at steady state then CER will be equal to creatini=

ne production rate (CPR) and we can use this:

CLcr=CPR/Scr

All rational models for predicting CLcr without measurement of CER use mode=

ls to predict CPR e.g.

CPR=(140-Age)*Weight/72 use Cockcroft & Gault to predict CPR in males th=

en CLcr=CPR/Scr is Cockcroft & Gault CLcr ml/min

Dividing CPR by Scr gives the CLcr. This can be written equivalently but le=

ss clearly:

CLcr=CPR*Scr^-1

The empirical models such as MDRD and CKI-EPI (see below) involve the absur=

dity of estimating the known exponent for Scr of -1. These estimates must b=

e wrong based on the theory I have outlined above (unless the estimate is e=

xactly -1). The reported estimates are -1.209 for CKI-EPI and -1.154 for MD=

RD.

In addition, and more importantly,they have no direct measure of body size =

which seriously limits the value outside the typical weight distribution an=

d they are only applicable to adults. GFR can be described from premature n=

eonates to adults using theory based allometry and maturation based on post=

-menstrual age so GFR predicttions should try to follow the concepts used t=

here (Rhodin 2008).

So what to do?

First -- don't use MDRD or CKI-EPI unless you are sure you are applying the=

m to a population similar to that used to develop these empirical predictio=

ns. You could add allometric scaling to the eGFR by assuming the 1.72m^2 va=

lue is equivalent to 70 kg with a fat free mass (FFM) of

56.1 kg. Then scaling the eGFR by (WT/70)^(3/4) or (FFM/56.1)^(3/4).

I use the Schwartz (1992) equations for neonates, children and teenagers th=

en the Matthews (2004) equation for adults. I am working on an integrated m=

ethod for CPR prediction which was presented as a work in progress at PAGE =

this year. Watch this space...

Best wishes,

Nick

MDRD

eGFR =175 x (SCr)^-1.154 x (age)-0.203 x 0.742 [if female] x

1.212 [if Black]

CKI-EPI

eGFR = 141 x min(SCr/k, 1)^alpha x max(SCr /kappa, 1)^-1.209 x 0.9=

93^Age x 1.018 [if female] x 1.159 [if Black]

kappa = 0.7 (females) or 0.9 (males)

alpha = -0.329 (females) or -0.411 (males)

eGFR (estimated glomerular filtration rate) = mL/min/1.73 m2; SCr (standa=

rdized serum creatinine) = mg/dL

Holford NHG, Anderson BJ. Allometric size: The scientific theory and

extension to normal fat mass. Eur J Pharm Sci. 2017;109(Supplement):S59-S64=

.

Rhodin MM, Anderson BJ, Peters AM, Coulthard MG, Wilkins B, Cole M,

Chatelut E, Grubb A, Veal GJ, Keir MJ, Holford NH

Human renal function maturation – a quantitative description using

weight and postmenstrual age. Pediatr Nephrol. 2008

Schwartz GJ. Does kL/PCr estimate GFR, or does GFR determine k? Pediatr

Nephrol. 1992;6(6):512-5.

Matthews I, Kirkpatrick C, Holford N. Quantitative justification for

target concentration intervention -- parameter variability and

predictive performance using population pharmacokinetic models for

aminoglycosides. Br J Clin Pharmacol. 2004;58(1):8-19.

Holford N, Sherwin CM. Scaling renal function in neonates and infants to

describe the pharmacodynamics of antibiotic nephrotoxicity. PAGE 26

Abstr 7208 [wwwpage-meetingorg/?abstract=7208]. 2017.

On 06-Dec-17 23:52, R.terHeine

On 06-Dec-17 06:17, R.terHeine

--

Nick Holford, Professor Clinical Pharmacology

Dept Pharmacology & Clinical Pharmacology, Bldg 503 Room 302A

University of Auckland,85 Park Rd,Private Bag 92019,Auckland,New Zealand

office:+64(9)923-6730 mobile:NZ+64(21)46 23 53 FR+33(6)62 32 46 72

email:n.holford

http://holford.fmhs.auckland.ac.nz/

http://orcid.org/0000-0002-4031-2514

Read the question, answer the question, attempt all questions

Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het hande=

lsregister onder nummer 41055629.

The Radboud university medical center is listed in the Commercial Register =

of the Chamber of Commerce under file number 41055629.

***************************************************************************=

*****************************************

This message may contain confidential information. If you are not the inten=

ded recipient please inform the

sender that you have received the message in error before deleting it.

Please do not disclose, copy or distribute information in this e-mail or ta=

ke any action in relation to its contents. To do so is strictly prohibited =

and may be unlawful. Thank you for your co-operation.

NHSmail is the secure email and directory service available for all NHS sta=

ff in England and Scotland. NHSmail is approved for exchanging patient data=

and other sensitive information with NHSmail and other accredited email se=

rvices.

For more information and to find out how you can switch, https://portal.nhs=

.net/help/joiningnhsmail

Received on Thu Dec 07 2017 - 08:00:51 EST