`2
` 2
`
`2
`
`¡9¢99£9¤
`
`2 ¥¦¦¥
`
`23456789:59;<48=49;>?@7:59A:5B:569
`34C79<:@@<79D<:5:D4<9>7<7C45D7
`
`EFGHIFJKLJMFNFOPJQRSPJTNUJMFOOVWTNNJXYFNFZPJQRSJ[\]^_`\]abcade^f\gbh
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`/
`
`
`
`
`
`k
`j
`
` #
`
`
`lm n
`
`p q
`
`
`
`
`k
`, # / , (%% *%%)
`$
` uxyzy{|}{~~y
`, j
`
`, r**r
`
`
`
`***((r )r**r ***
`u**)0
`r**rr1r(
`
`
`#
` $ %
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`&
`
`
`
`
`
`()*
`
`
`
`
`
` -
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`MYLAN - EXHIBIT 1047
`
`
`
`I 16 Benet and Homer
`Ì ÍÎ]ÎÏ^\]Ð^ÑdÎ]Î`
`
`binding of warfarin alone and in the presence of
`
`phenylbutazone. In this in vitro experiment, they clearly
`
`
`showed that phenylbutazone displaced warfarin from
`
`
`its albumin binding sites. From these two experiments,
`
`
`
`
`an in vivo drug interaction observation and an in vitro
`
`protein-binding experiment, they proposed a cause for
`
`
`a clinical observation; that is, an increase in fraction
`
`unbound in plasma (fu) caused the changes in pro-
`
`§¨
`thrombin times by increasing the concentration of
`
`
`
`unbound drug. The problem is in the extrapolation from
`
`
`
`an in vitro observation to an in vivo effect.
`
`In reality, the clinical interactions observed with
`+
`the anticoagulant and antidiabetic drugs described, as
`
`well as a number of others reviewed by MacKichan,2
`
`
`Rolan,3 and Sansom and Evans,4 result from changes
` #
` $ %
`
`in drug metabolic clearance and not from changes in
`
`
`
`protein binding. However, because the protein-bind-
`
`ing changes do cause changes in pharmacokinetic
`
`ª
`parameters in certain cases, the belief in the clinical
`
`
`importance of these changes has persisted. We first
`
`
`review the basis for the pharmacokinetic parameter
`
`ª
`
`changes.
`
`PHARMACOKlNETIC PARAMETERS
`«¬®¯°±²³´µ¶³°·«®¯µ¶µ®¸
`There is clear evidence that plasma protein binding
`
`
`is relevant in the pharmacokinetic modeling of drugs,
`
`ª
`
`as has been primarily emphasized in terms of individ-
`
`
` -
`
`ual pharmacokinetic parameters9; that is, the volume of
`
`ª
`
`
`distribution (V)
` §¹¨
`
`(1)
`V = [fu/fuT] VT + VP
`§¨
`¹º»¼¹¹j
`depends on the fraction unbound in plasma (fu), the
`
`§¨
`fraction unbound in tissue (fuT), the volume of tissue
`
`
`(VT), and the volume of plasma (VP).10 For all drugs
`§¹¨
`
`§¹j¨*/
`with a V value 230 L (when VP has only a minor effect
` ¹
`
`on V), changes in fu therefore translate directly into
` ¹¨
`changes in V. 10
`
`All organ clearance models (here we use the simplest
`
`
`
`well-stirred venous equilibration modelloall) incorpo-
`
`
`rate a protein-binding term
`
`
`CL = [Qorgan - fu - CLint1/[Qorgan + fu - CLint]
`(2)
`,vº»¾ ¿¿,v ¼»¾ ¿,v ¼ §r¨
`in which CL is organ clearance, Qorgan is blood flow to
` ,v
`the clearing (eliminating) organ, and CLint is the intrin-
`
` ¨ ,v
`sic organ clearance of the unbound drug. High extrac-
`
`tion ratio drugs (Qorgan << fu - CLim) exhibit organ
` §¾ ÀÀ¿,v ¨
`clearance independent of fu (ie, CL E Qorgan), but for
`
`low extraction ratio drugs (Qorgan >> fu - CLim)
`
`(3)
`CL 5 fuCLint
`§ ¨
`,vÁ,v
`clearance depends on fu and the intrinsic ability of the
`
`organ to clear the drug (CLim).9a10
`
`
`CLINICAL PHARMACOLOGY 8r THERAPEUTICS
`MARCH 2002
`22929Ò92
`29¥¦¦¥
`
`Again, with use of the well-stirred model as an exam-
`
`
`
`ple, the hepatic bioavailability (FH) is given by the fol-
`
`lowing:
` u
`
`(4)
`FH = QH/[QH + fll ' CLint]
`§%¨
`/!º¾!»¾!¿,v ¼
`Then F E 1 for a low extraction ratio drug, but for a
`
`high extraction ratio drug
`
`
`§¨
`/!Á¾!»¿,v ¼
`in which FH is inversely related to fu and CLint and is
` /!
`also directly dependent on QH.
`
`Because the half-life (tr/2) may be defined in terms of
`"
`Ãr¨
`
`
`the ratio of volume to clearance multiplied by ln2, it is
`
`
`
`recognized that for high extraction ratio drugs, when V
`
`230 L, this parameter will also depend on fu, as follows:
`½ *v
`
`tlz E [0.693(fu/fuT)VT]/Qorgan
`(6)
`
`ÃrÁ»*)( §¨¹¼¾
`§)¨
`However, tl/2 is independent of fu for low extraction ratio
`!
`Ãr
`drugs, for which V 230 L, as follows:
` ¹½ *v u
`(7)
`a: a [0.693(VT/fuT)]/CLint
`
`ÃrÁ»*)( §¹¨¼,v
`§0¨
`It is therefore correct that, depending on the pharma-
`+
`
`cokinetic parameters measured and the intrinsic clear-
`ª
`
`
`ance of the drug, certain pharmacokinetic parameters
`
`ª
`
`will change with protein binding but others will not.
`
`Furthermore, the changes in the individual pharmaco-
`/
`
`
`kinetic parameters may result
`in changes in the
`ª
`
`
`observed concentration—time profiles.12 However, the
`
`
`belief that the effective concentration of all drugs
`
`
`depends on protein binding is not correct, as we will
`
`show in the next section.
`
`EXPOSURE CONCEPTS
`µÅ«±¸Æ®µ·°±´°µ«¶¸
`The introduction of clearance concepts in the mid-
`
`
`1970s has had a major impact on recognizing the rele-
`(0*
`.
`
`vance of pharmacokinetics to the clinical practice of
`
`ª
`medicine. However, the pendulum has swung too far,
`
`
`
`
`and we now overemphasize the effect of disease states
`
` -
`and drug interactions on individual pharmacokinetic
`
`ª
`parameters rather than on the most relevant measure,
`
`
`
`
`drug exposure.
`
`Exposure is a term that reflects the drug levels to
`ÇÈÉÊ~Ëy
`
`which a patient is exposed after a dose or a series of
`
`doses. It is a measure of concentration integrated over
`
`
`time, commonly referred to as area under the curve
`
`
`
`
`(AUC). In some cases, particularly for toxicity issues,
`§ k,¨+
`
`the clinician may be concerned about the maximum
`
`
`
`
`
`exposure of drug at a particular time (Cmax) or that sys-
`
`
`¨
`temic concentrations be maintained above a threshold
`
`
`
`minimum effective concentration. However, we will
`
`
`
`
`consider the integrated exposure, AUC, because it is the
`
`parameter directly related to dose, as follows:
`
`
`
`
`
`CLINICAL PHARMACOLOGY 8: THERAPEUTICS
`VOLUME 71, NUMBER 3
`22929Ò92
`¡9¢×9£9¤
`
`AUC = [F - Dose]/CL
` k,º»/¿i
`
`(8)
`§1¨
`
`First, let us look at oral dosing. When we dose orally,
`/
`bioavailability (Feral) is the product of three availabil-
` §/ ¨
`
`ity factors (assuming negligible lung first-pass effects)
` §
`
`as follows:
` u
`
`(9)
`Feral = Fabs ' FG ' FH
`§(¨
`/ º/¿/n¿/!
`in which FabS is the fraction of administered drug that
` /
`
`is absorbed into the gut wall and does not flow back
`
`into the lumen, F6 is the fraction that gets through the
`
`
`gut wall unchanged, and FH is that fraction that passes
`
`through the liver and into the systemic circulation
`
`
`unchanged.13 Therefore, when a drug that is eliminated
`
`
`primarily by the liver is given orally, systemic AUC can
`
`
` k,
`be calculated by inserting equations 2, 4, and 9 into
`
`
`equation 8 to give the following:
`
`AUCoral = [Fabs - FG - Dose]/[fu - CLim]
`(10)
` k, º»/¿/n¿i
`Equation 10 is general and holds for both high and low
`$ *
`extraction ratio drugs that are cleared by the liver and
`
`given orally. It is the general consensus that pharmaco-
`
`
`logic effect is related to exposure to unbound drug con-
`
`centrations (AUCu). Therefore, for oral dosing when
`
`systemic elimination occurs from the liver, AUCu will
`
`
`
`
`be given by the following equation:
`
`AUCuoral = fu - AUCoral = [Fabs - FG - Dose1/CLint (11)
`
` k, º¿ k, º»/¿/n¿i
`Note that in equation 11 changes in fu have no effect
`t
`on unbound drug exposure; therefore no changes in
`
`pharmacologic effect would be expected for drugs that
`
`
`are administered orally and eliminated hepatically
`
`
`
`(Table I).
`§
`For any drug given orally when systemic elimination
`/
`
`
`is not hepatic (FH = 1) or for any drug given intra-
`
`venously (F = 1), then substituting the definition of
`
`organ clearance (equation 2) into equation 8 yields the
`
`following:
` u
`AUCuoral,nonhepatic = AUCIV = [Fabs . FG . DOSC
`
` k,
`(12)
`(QH + fu ‘ CLintH/[QH ‘ fu ‘ CLint]
`§¾!Ó,v ¨¼»¾!¿¿,v ¼
`§r¨
`in which FabS and F6 for intravenous dosing equal 1.
` / /n
`For a low extraction ratio drug (Qorgan >> fu - CLim)
`/
`after intravenous dosing
`
`(13)
`AUCuIV = fu - AUCIV E Dose/CLint
`
` k,+¹º¿ k,+¹Ái
`§ ¨
`and changes in fu will not affect unbound exposure.
`
`This will also be the case for low extraction ratio drugs
`
`
`cleared nonhepatically when they are administered
`
`
`orally, although FabS and FG will need to be considered,
` / /n
`as follows:
` u
`AUCuoral,nonhepatic E [Fabs ' FG ' DOSC]/CLint
`(14)
`
` k,
`
`Benet and Homer
`1 17
`ÍÎ]ÎÏ^\]Ð^ÑdÎ]Î` ¢
`
`Table I. Summary of types of drugs and routes of
`¶ØÙÚÛ·³Ü·#
`
`
`administration for which protein-binding changes may
`
`
`
`be clinically relevant
`
`
`Low
`High
`extraction
`extraction
`áÊâ
`Ý}Þßà
`ratio
`ratio
`yÈ{Ë{}Êzà yÈ{Ë{}Êzà
`Ë{}Ê
`Ë{}Ê
`
`Intravenous administration
`+
`
`Yes*
`Hepatic clearance
`ã
`!
`Yes*
`Nonhepatic clearance
`Oral administration
`ã
`t
`q
`
`No
`Hepatic clearance
`t
`!
`Yes?
`Nonhepatic clearance
`ã
`t
`*See Table H for drugs that meet these criteria.
`TNo drugs from a list of 456 dmgs14v15 met these criteria.
`ä#
`
`åt
` %) %
`
`
`No
`t
`No
`t
`No
`t
`No
`t
`
`In summary, for all
`low extraction ratio drugs,
`+
`
`
`regardless of route of administration, and for all drugs
`
`
`administered orally and eliminated primarily by the
`
`
`
`
`liver, total exposure is independent of protein binding
`
`and no dosing adjustments will need to be made for real
` .
`
`
`or anticipated changes in fu (Table I). Only high extrac-
`
`tion ratio drugs given intravenously and oral drugs
`
`eliminated by nonhepatic high extraction ratio routes
`
`
`will exhibit changes in unbound drug exposure when
`
`protein binding changes; that is, for a high extraction
`
`ratio drug (Qorgan << fu - CLim), equation 12 becomes
` §¾ ÀÀ¿,v ¨
`
`the following:
`
`
`(15)
`AUCIV E Dose/QH
`§¨
` k,+¹Ái
`and unbound AUC is obtained by multiplying both sides
` k,
`
`of equation 15 by fu, as follows:
`
`(16)
`AUCuIV = fu - AUCIV a [fu - Dose]/QH
`
` k,+¹º¿ k,+¹Á»¿i
`§)¨
`This will also be the case for high extraction ratio drugs
`
`
`not cleared by the liver when they are administered
`
`
`orally, once FabS and F6 are considered as factors.
`
`APPLICATION OF PRINCIPLES
`««Ô³°¶³±´·±Õ·«®³´°³«Ôµ¸
`We can now explain why changes in protein binding
`&
`cannot be important for warfarin. Warfarin is eliminated
`
` &
`
`by hepatic metabolism but is a low extraction ratio
`
`
`
`drug, so FH E 1 and in fact For,“ = 1.14,15 Therefore its
` /!Ö / º%
`total systemic exposure is described by equation 10 and
`
`
`its unbound exposure by equation 11, which is inde-
`
`pendent of protein binding. Changes in fu caused by
`
`either disease effects or drug interactions will therefore
`
`not be expected to influence clinical outcome, and no
`
`
`adjustment of drug dosing should be required.3a4 The
`.
`
`effect of phenylbutazone on warfarin levels and effi-
`
`cacy can be explained by noting that phenylbutazone
`
`inhibits warfarin metabolism (CLint decreases).16 In
`
`
`§,v
`
`
`
`1 18 Benet and Homer
`ç ÍÎ]ÎÏ^\]Ð^ÑdÎ]Î`
`
`CLINICAL PHARMACOLOGY 8r THERAPEUTICS
`MARCH 2002
`22929Ò92
`29¥¦¦¥
`
`Table II. The 25 drugs in a list of 456 drugs14,15 for which protein binding may influence clinical drug exposure
`¶ØÙÚÛ·³³Ü·
`
`after nonoral administration, with use of cutoffs of >70% for protein binding (f11 < 0.3) and 20.28 Qorgan for clear-
`ance
`
`
`
`
`Protein binding (%)
`èËÊ{y}zàx}z}zÞàéêë
`92
`Alfentanil*
`(r
`
`95
`Amitriptylinefl:
`(
`
`
`96
`Buprenorphine*1‘
`()
`"
`so
`Butorphanoli‘i'l‘
`1*
`" äåô
`95
`Chlorpromazine*:l:
`(
`,
`-
`91
`Cocaine*|\
`(
`,
`78
`Diltiazemi‘i:
`01
`i -
`äó
`78
`Diphenhydraminefl:
`01
`i
`
`76
`Doxorubicin*
`0)
`i ä
`84
`Erythromycinfl:
`1%
`$
` äó
`84
`Fentany1*
`1%
`/
`95
`Gold sodium thiomalate (INN, sodium aurothiomalate)?
`(
`n
`
`
`
`
`92
`HaloperidolH:
`(r
`!
`97
`Idarubicinfl:
`(0
`+ äó
`99.8
`Itraconazole*:l:
`((1
`+ -
`70
`Lidocaine*
`0*
`v
`78
`Methylprednisolonei‘H:
`01
`©
`98
`Midazolam*H:
`(1
`©-
`äåó
`70
`Millinone*
`0*
`©
`99
`Nicardipinefl:
`((
`t
`70
`Pentamidine*
`0*
`j
`
`98
`Propofol*
`(1
`j ä
`87
`Propranolol*:!:
`92
`Remifentanil*
`10
`j äó
`(r
`
`
`
`93
`Sufentanil*
`(
`#
`90
`Verapamili‘zl:
`(*
`¹
` äó
`f“, Fraction unbound in plasma; ngm, blood flow to the clearing (eliminating) organ.
`*Intravenous administration.
`/
`s¾
` ¨
`TIntramuscular administration.
`ä+
`
`iDoes not apply for oral administration of this drug.
`å+
`
`
`§>0.28 QH; CLH 2 6 mllmin per kilogram; see text for explanation for this low cutoff for “high.”
`ói
`
`”Nasal delivery.
`òÂ*r1¾!s,v!÷)
`
`
`s
`1[>0.28 QR; CLR 2 4.8 mllmin per kilogram; see text for explanation for this low cutoff for “high.”
`út
`#Probably metabolized in blood by nonspecific esterases.
`õÂ*r1¾s,v÷%1
`
`
`s
`öj
`
`
`CL (ml/min - kg)
`ìáàéíîïí}zàðàñÞë
`10.6§
`*)ò
`l 1.5 §
`ò
`13.3§
` ò
`22§
`rrò
`8.6§
`1)ò
`32§
` rò
`11.4§
`%ò
`6.2§
`)rò
`l6.2§
`)rò
`8.0§
`1*ò
`12.3§
`r ò
`4. 8‘][
`%1õ
`11.8§
`1ò
`29§
`r(ò
`l2.7§
`r0ò
`9.2§
`(rò
`6.2§
`)rò
`6.6§
`))ò
`5.2‘][
`rõ
`10.4§
`*%ò
`l6§
`)ò
`27§
`r0ò
`l 8§
`40-60#
`1ò
`%*)*ö
`l2§
`rò
`15§
`ò
`
`fact, the package insert for warfarin does not recom-
`
`
`mend a change in dose with real or anticipated changes
`
`
`in fu. This recommendation is based on clinical experi-
`
`
`
`ence rather than on pharmacokinetic principles, but we
`
`ª
`have just shown that it is solidly grounded in the analy-
`
`sis of drug exposure.
`
`There are, of course, as we have shown, situations in
`
`which fu becomes a determinant of AUC“. The first is
`
`
` k,
`high extraction ratio drugs that are eliminated primar-
`
`
`
`ily by hepatic metabolism when they are administered
`
`
`
`
`intravenously (equation 16). The second case is high
`
`extraction ratio drugs given either orally or intra-
`
`venously when the liver is not the main route of sys-
`
`
`temic elimination, as follows:
`
`
` u
`AUCUIV = [fu - DoseIV]/QR
` k,
`+¹º»¿i
`01'
`
`
`(17)
`§0¨
`
`(18)
`AUCum-al = [Fabs ' FG ' fu ' DoseomyQR
`
` k, º»/¿/n¿¿i
`§1¨
`in which we have assumed renal elimination and used
`
`
`
`renal blood flow (QR) in the organ clearance equation
`
`
`(equation 2).
`§
`An examination of 456 drugs14,15 revealed that none
`
` %) %
`which are administered orally come close to meeting
`
`
`
`
`the criteria of nonhepatic elimination (>50% excreted
`
` §Â*æ
`unchanged), significant protein binding (>70% bound
`
`to plasma proteins), or high nonhepatic extraction ratio
`
`
`clearance (>0.5 QR; >8.5 mllmin per kilogram). They
`
`
`
`¨
`do not meet the criteria even when the high nonhepatic
`
`
`extraction ratio clearance cutoff is lowered to >0.28 QR,
`
`or >4.8 mllmin per kilogram. However, there are drugs
` Â%1
`
`
`!
`that are administered by nonoral routes which meet the
`
`
`
`criteria of having significant protein binding (>70%)
`
`and having either a high hepatic or a high nonhepatic
`
`
`
`
`CLINICAL PHARMACOLOGY 8: THERAPEUTICS
`VOLUME 71, NUMBER 3
`22929Ò92
`¡9¢×9£9¤
`
`Benet and Homer
`1 19
`ÍÎ]ÎÏ^\]Ð^ÑdÎ]Î`
`
`Table ]]I. Drugs for which changes in protein binding have been thought to be important and reasons these changes
`¶ØÙÚÛ·³³³Ü·i
`
`are not clinically relevant
`
`Drug
`ËÞ
`Carbamazepine
`,
`-
`Ceftriaxone
`,
`Chlorpropamide
`,
`
`Diazepam
`i-
`
`Ketoprofen
`Methotrexate
`o
`©
`Phenytoin
`Tolbutamide
`j
`
`
`
`Valproic acid
`Warfarin
`¹
`&
`
`Reason
` y~Êz
`Only given orally; low hepatic extraction ratio (0.08)
`q
`Low hepatic extraction ratio (0.01)
`v
`Very low hepatic extraction ratio (0.001)
`¹
`Low hepatic extraction ratio (0.02)
`v
`Only given orally; low hepatic extraction ratio (0.06); probably long equilibration time
`q
`
`Low protein binding (46%); low hepatic extraction ratio (0.06); probably long equilibration time
`v
`
`Low extraction ratio (~0.03 in linear range, decreases with higher saturation concentrations)
`v
`Only given orally; low hepatic extraction ratio (0.01); long equilibration time
`q
`
`Very low hepatic extraction ratio (0.005)
`¹
`Only given orally; very low hepatic extraction ratio (0.002); long equilibration time
`q
`
`
`extraction ratio clearance (Table II). To be as inclusive
`
`as possible, in Table II we list therapeutic agents that
`
`have protein binding of 270% and an extraction ratio
`
`of 20.2814a15; that is, hepatic clearance 26.0 ml/min
`½*r1%s
`
`
`per kilogram or renal clearance 24.8 ml/min per kilo-
`
`
`
`
`gram. We expanded the extraction ratio criteria far
`
`&
`beyond those usually considered to be “high.” We did
`
`
`this in part because the correct extraction ratio calcula-
`
`tion should be organ blood clearance divided by organ
`
`blood flow, whereas plasma clearances are given in
`
`
`
`Table II for all of the drugs except amitriptyline and
`
`
`propranolol.14a15 In Table II we included drugs that are
` %+
`given intravenously, intramuscularly, or intranasally.
`
`
`However, drugs administered intramuscularly or
`!
`
`
`intranasally may be incompletely absorbed from the
`
`
`
`
`site of administration, and therefore the equations for
`
`
`AUC (equation 15) and AUC11 (equation 16) will need
` k,§
`to be adjusted for that possibility.
`
`THERAPEUTIC lNDEX AND KINETIC-
`¶¬µ®«µÆ¶³°·³´ûµÅ·´û·²³´µ¶³°ü
`DYNAMIC EQUILIBRATION TIME
`ûý´¯³°·µþƳԳ ®¶³±´·¶³¯µ
`The listing in Table 11 suggests that only 25 drugs
`
`have the potential for exhibiting changes in clinical
`
`response, with changes in protein binding caused either
`
`by drug interactions or by disease states. As stated, a
`
`number of these are definitely borderline cases. Fur-
`
`
`thermore, this list would be even shorter if we were to
`
`
`consider the therapeutic index of each drug because if
`
`a drug has a wide therapeutic index (eg, propranolol),
`
`changes in free drug concentrations that result from
`
`
`protein-binding changes will have negligible clinical
`
`effects.
`
`However, there is another pharmacokinetic-pharma—
`!
`ª
`
`codynamic parameter that can expand the list of drugs
`
`
`
`beyond those in Table II. This parameter, developed by
`
`
`
`Sheiner et al17 and Holford and Sheiner,18 describes the
`#
`delay between drug effects and drug concentrations in
`
`
`terms of a pharmacokinetic-pharmacodynamic equili-
`
`
`ª
`
`
`bration half-time. Changes in protein binding caused
`
`
`
`by a drug interaction are assumed to instantaneously
`
`
`change free drug concentrations. Thus there should be
`
`a transient change in free concentrations while the body
`
`re-equilib