Leadln. Article
--------~----------~
Cion Pllarmacol ,R('I. 1\1 (6,: 425-4 33. 1990 0) I 2·$%)/90100 I :-04~5/S04 ~/O
o
Ad l~ InH.'rnahon;'" L,m"cd All "Ih t) rnc .. nI
c....OO)_
T hera peutic Drug Monitoring in Pregnancy Rationale a nd C urrent Status
Chrisrine Knoll and Felicity Reynolds Department of Pharmacolog). School of Pharmacy. London. and AnaesthetiC' Unit.
U MDS. 51 Thomas's Campus. London. England
J. Mon itoring Drugs in Pregnancy
ure of pharmacologlcafly aC/II'£' drug. which is also
1.1 Ratio nale
the fraction available for metabolism and renal elimination. This. of course. is rarely practical but because the free . therapeutically active drug component equilibrates with that bound to plasma proteins it has become standard practice to use plasma or serum drug concentrations to provide an index of therapeutically acti ve drug. This is valid if the degree of protein binding remains constant but there are numerous instances when protein binding changes unpredictably. incl uding renal insufficiency (Pacific et al . 1986). hepatic disease (George 1979). in patients receiving poly therapy with highly bound comedicants (Knott et al. 1982; Neuvonen et al. 1979: Patsa10s & Lascelles 1977) and with hypoalbuminaemia arising postoperatively (Elfstrom 1979). after severe burns (Bowdle et al. 1980a). in neonates and infan ts (Morsell i 1989). in pregnancy (Knott et al. 1986; Kra uer et al. 1984; Reynolds & Knott 1989). in old age (Patterson et a\. 1982) and with overdosing (personal observations). In these situations, therefore. it is no longer valid to assume that plasma drug concentrations reflect the therapeutically acti ve concentrations. and such measurements become meaningless (Kilpatrick el al. 1984; Knot! 1983). Instead, il is essential to measure the plasma unbound drug concentrations (fig. I) or, where possible, the saliva drug concentrations. which for some drugs accurately reflect them (Knott 1989),
The rationale for mOnitoring drugs
In
preg-
nancy is. of coursc. Ihe same as lhal for any patient group. It depends on a strong correlation between the unbound concentration of drug althe 'receplOr' (or other target) site and Ihe drug response. If the effects of such drug-receptor complexes can be assessed directly by clinical measurements, no furIher improvement in palicnt management is achieved by therapeutic drug monitoring (TOM). This is the case for hypnotics. anticoagulants, di· uretics. hypogl ycaemics. hypolipidaemics, vasopressors. analgesics and some hormones. Other drugs. whose effects correlate highly with eoncen· tration at the site of action rather than with dose administered. those with narrow therapeutic dose ranges (e.g. phenytoin. lithium). and drugs whose toxic effects are indistinguishable from underlying diseases. are ideal candidates for TOM. For the majority of drugs, variations in patient response are a result of phannacokinetic rather than pharmacodynamic differences. and in pregnancy these differences are often amplified.
1.2 Free Venus Total Drug Conctntrntion Measurements Ideally. the interstitial fluid bathing the target sites should be sampled to provide a dirf'Cl meas-
426
Oln. Phurmucol.lnN. 19 (6) 1Y90
2. Drug Disposition in Pregnllncy
EC'
The possibili ty of adverse effects of maternal drug administ ration on the developing fetu s may result in reduced maternal compliance (fig. 2). This may appear to reduce drug clearance, as does reduced oral drug absorption by vomiting or by antacid (Carter et al. 198 1) o r mineral ingcstion (Ca mpbell et al. 1988). which may be particularl y relevant wi th current interest in health products. Significant reductions in gastroinlestinal motility on ly occur during labour when absorption is also im pai red, and drugs may accumulate in the gul. After delivery their absorpt ion may result in greatly elevated plasma concentrations and undercstimated clearance values. However, dose requirements may genuincly increase with changing ma-
.
_-
I
--,
I
Ya
..I_
Ya
" 1nw. .... ceo.J
No
00.., __
----Do they
.
c:orr-. wIIh
Ya
~ ••1klnI1
No
Do they conwInI ....,
cor .........,,,.?
-"'No
Ya
-
-
I I
!tug ITIOI ....... 'II
Fig. 1. A'ded$lon Ire(" based o n the major faclors 10 ,o nsider when determining the yalue or Iher.ilpc"utie drug monitonn ...
11
AlbJmin
'FA
1-----
J_
::::---11 ,,
""""''''''I '. ......... ~ Clearance t I . . - Enzyme Oral
-
I'
""".-
I t
Dose raquil1Ml"le!llS
ioduedon
I
Fig. 2. rnnelpil\ rX10f"S which ;nn ... ~n..., mal ~mal dow ~
eon _ _
No
I
qUlrementS In prq.nancy: arT"()\tl'S dIow lhe: dllttuon or ellanl". Increased Yolume ordlsmbul1on (Vd) ca n Iol'>'er the: pta$ma COIK't'TItnlllon and onnuence lher.ilpevl O(" drill mOf11tonn ... and onCTI:ased elearance ca n alter dose ~ul~menls. ECF • e~ traccllular nUId Yolu me: FFA - free fauy acids.
ternal physiology (fig. 2) in complian t women (Nau et at 1982). Total body clearance (el). defined as the volume of plasma cleared of drug per hour per kilogram bodyweight. may i ncrease in pregnancy. During long tenn drug administration it is usually calculated as the ratio of the fraction (f) of dose absorbed in unit time (D) to the plasma concentration at steady-state «('55) (Voteh et al. 1988J. according to the fo llowing form ula: fD CL: - -
C'.
(Eo.
I)
The use of the term 'steady-state' in pregnancy is not strictly appropriate, because pregnancy is the one condition in which the physiological state is reliably not ·steady'. However, useful information can be obtai ned from clearance data if account is taken of factors which may appear to alter ils v alue. Drug concentrations may fall with maternal enzyme induction. reflecting the balance between progesterone and estrogen concentrations (Loock
Drug MOflnonng
In
427
Pregnancy
ct al. 1988) and. to a lesser extent. the increasing contribution 10 drug metabolism by the fetal com· partmcnt (fig. 2) [Juchau 1982). Plasma dilution also lowers Ihe levels of plasma albumin and. less consISICIHI}. plasma brev/M;on.: I", - tennirlal half-life; r~ • bound fraction of drug in plasma: FPlA • fluorescence poIfIrlNtIon Immunoa,uy: AlA • r~S5'Y; EMIT • en~linked imtnunoessay: HPlC - I'Iigh pertom'Iance liquid dYomatogf~: GLe • gal liquid chrom.togr.phy: AAS • •tomic .bsorpIIon apectr~ FES • ft.",. emillion ~oecopJ; OMO • dctmethykIiuepam.
_CUI.
efficacy a nd depend on the infecting palhogen, severi ty of infection and time that the minimum inhibitory concentration (MIC) is exceeded (Nightingale \989). However, Grevel and colleagues (1989) have suggesld that greater therapeutic pre-
cision can be achieved frorp measurements of the area under the concentration-time curve (AUC) than from isolated trough concentration measurements in renal transplant patients. A knowledge of renal status is important as aminoglycosides and
Drug Monllorlng
In
429
Pr('gnanq
some cephalosporins produce nephrotoxicilY and ototoxicity at doses only slightly above the microbiologically effective concentrations. Many of these drugs arc poorly lipid soluble and prOiein bound, and are excreled largely unmelabolised at the glomerular fillration rate (White et al. 198]), which increases in pregnancy (Dunlop 1979). Some newer cephalosporins arc highly bound (Lam el al. 1988), have lower CL and Vd but similar I.·" and show saturable binding which is displaced by FAAs. bilirubin and comedicants (Bialer el al. 1988; Decroix et at 1988). Their protein bi nd ing, therefore. is likely to fall in pregnancy. In general. increased plasma clearances with either unchangcd or increased Vd and shorter ( .1, arc reported for the aminoglycosides (Lazebnik et al. 1985), the penicillin derivatives (Kjer & Ollesen 1986) and the cephalosporins (Chow & Jcwesson 1985) in pregnanc),. and plasma concentrat ions fa ll below the MIC (Nau 1987).
3.1.1 Antiarrh)'lhmics These weak bases are highly ionised at physiological pH and the)' arc absorbed erratically - a problem which ma y be worsened during pregnancy, Disopyramide. like the cephalosporins, shows saturable binding at clinically effective concentrations. For such drugs, unbound clearance values may be bener estimates as they change independently of protein binding changes (Upton & Williams 1986). In a recent study using spiked plasma samples. disopy ramide protein binding was found to be significantly lower in the second and third trimesters and significantly greater I month after delivery than in nonpregnant women (Echizen et al. 1990). Thus, regular monitoring of the free drug concentration is indicated during pregnancy and Ihe puerperium. Digoxin is poorly bound (table I). widely distributed and almost exclusively disposed of by renal elimination. It has a very narrow therapeutic dose ratio. and the correlation between plasma concentration and effect is only significant at 6 to 8 hours after dosing. Hypot hyroidism, hypokalaemia, hypomagnesaemia and hype rcalcaemia can all increase myocardial sensitivity so that toxicity may
occur at subtherapeutic concentrations (Keys & Stafford 1981), while some palients can tolerale concentrations in excess of 3 ~g/L Pharmacodynamic interactions with other antiarrhythmics, however, raise questions as \0 the value of digoxin maintenance therapy and its TOM (Spector el al. 1988) olher than for confirmation of compliance or toxicity. Maternal digoxin concentrations fall in pregnancy (Aronson 1980) as renal clearance increases. and if dose increments arc made. they must be reduced I}()sl parrum. Radioimmunoassays arc unsuitable in pregnancy because the)' show high crossreactivi ty with endogenous digoxin-like substances. found particularly in pregnancies with a high risk of complications ( Koren et al. 1988),
313 Bronchodilalors Theophylline and the anticonvulsants arc Ihe only drugs for which optimal concen tration ranges for saliva as well as plasma have been established (KnOll et al 1984) for routine TOM (table I). Agitation and tachycardia become increasingly evident above 20 mg/L. coincident with saturation of the binding of Iheophylline to albumin. T hcophylline is predominantly hydroxylatcd and demethylated. although small amounts arc excreted unchanged; its metabolism may be inhibited by cimetidine, propranolol and erythromycin. Theophylline protein binding is lo.....cr in the third trimester of pregnancy than in nonpregnant women (Connell y el al. 1990). The plasma clearance and Vd both for theophylline and for caffeine are often reduced in pregnancy (Cartcr 1986; Scarcy et al. 198]) and plasma half-lives are generally increased, thus necessitating morc frequent monitormg.
3.2.4 Antidepressants The pharmacokinetics of the tricyclic and related an tidepressants have been reviewed by Sjoqvist and 8enilsson (1984) and Potter et a1. (1984). The dosage of these highly bound drugs predicts the CS~ poorly because they are cleared at a rate genetically determined by the hydroxylator status (Sjoqvisl & 8enilsson 1984), which may be
eonvuts.ants
Cepllaiosponn,
Anhdepressan15
SalicylIC aCId
t ...
t t 1.
+-+
Benzodlazepones Xanthlnes
l ow UpkI .oIubillty
High I lplcl lOIubility
Cl
_
Vd
t
'. t
,
pt.",,~
Amlnog l1C01ideS
(anbp)'rine)
Lithium PeniCllhns
+
t t , +-+ '. + Cl
, +-+
Vd
, +-+ Low protein
bindIng
Fig. 3. PredIl::I~ chan&« m drug pharmacokmetlCS In prq,nanc), based on their ph)slochemlcal propcnles. with e.\ampks: f - 1I\ereased; •• decreased: . -- • no chance: C L • 10lal bod) plasma drug ckarance: Vd _ \oIumc of dlslrlbuuon: '. '" ttrmmal half·1Ift.
432
Re/eretrcu Albanl E. Riva R. Contin M. BarolZI A. Altoman: M, et II . Dif. fC11:n ti.1 I ... nsplactnl.ill blndin& of "alprooc ~Id: influenct of fm fally ACld$. 8ritlSh Journa l of ClinIcal Pharmacoiosy 17: 759.762. 1984 Aronson JK . Clln..:.1 phannacok"lC'IICS of dltolln 1980. Chnical I'tIarmacoklll~tlC$ 5: 137·149. 1980 BIIrdy AH . Hllinmaa VK. Tenlmo K. Neuvonen PS. Proleln blndlna: of.n,oeplk!oc dropdunna: pr-eananf)'.lIIbor Ind puer· ptnum. ThnlIpeutlC Dru, MORltonna: 12: -4()....46. 1990 8lala' M . Wu WH o Faulkner RD. SIlber BM. YaoolM A.. In vilro prol~ln blndln, Inl
--------~----------~
Cion Pllarmacol ,R('I. 1\1 (6,: 425-4 33. 1990 0) I 2·$%)/90100 I :-04~5/S04 ~/O
o
Ad l~ InH.'rnahon;'" L,m"cd All "Ih t) rnc .. nI
c....OO)_
T hera peutic Drug Monitoring in Pregnancy Rationale a nd C urrent Status
Chrisrine Knoll and Felicity Reynolds Department of Pharmacolog). School of Pharmacy. London. and AnaesthetiC' Unit.
U MDS. 51 Thomas's Campus. London. England
J. Mon itoring Drugs in Pregnancy
ure of pharmacologlcafly aC/II'£' drug. which is also
1.1 Ratio nale
the fraction available for metabolism and renal elimination. This. of course. is rarely practical but because the free . therapeutically active drug component equilibrates with that bound to plasma proteins it has become standard practice to use plasma or serum drug concentrations to provide an index of therapeutically acti ve drug. This is valid if the degree of protein binding remains constant but there are numerous instances when protein binding changes unpredictably. incl uding renal insufficiency (Pacific et al . 1986). hepatic disease (George 1979). in patients receiving poly therapy with highly bound comedicants (Knott et al. 1982; Neuvonen et al. 1979: Patsa10s & Lascelles 1977) and with hypoalbuminaemia arising postoperatively (Elfstrom 1979). after severe burns (Bowdle et al. 1980a). in neonates and infan ts (Morsell i 1989). in pregnancy (Knott et al. 1986; Kra uer et al. 1984; Reynolds & Knott 1989). in old age (Patterson et a\. 1982) and with overdosing (personal observations). In these situations, therefore. it is no longer valid to assume that plasma drug concentrations reflect the therapeutically acti ve concentrations. and such measurements become meaningless (Kilpatrick el al. 1984; Knot! 1983). Instead, il is essential to measure the plasma unbound drug concentrations (fig. I) or, where possible, the saliva drug concentrations. which for some drugs accurately reflect them (Knott 1989),
The rationale for mOnitoring drugs
In
preg-
nancy is. of coursc. Ihe same as lhal for any patient group. It depends on a strong correlation between the unbound concentration of drug althe 'receplOr' (or other target) site and Ihe drug response. If the effects of such drug-receptor complexes can be assessed directly by clinical measurements, no furIher improvement in palicnt management is achieved by therapeutic drug monitoring (TOM). This is the case for hypnotics. anticoagulants, di· uretics. hypogl ycaemics. hypolipidaemics, vasopressors. analgesics and some hormones. Other drugs. whose effects correlate highly with eoncen· tration at the site of action rather than with dose administered. those with narrow therapeutic dose ranges (e.g. phenytoin. lithium). and drugs whose toxic effects are indistinguishable from underlying diseases. are ideal candidates for TOM. For the majority of drugs, variations in patient response are a result of phannacokinetic rather than pharmacodynamic differences. and in pregnancy these differences are often amplified.
1.2 Free Venus Total Drug Conctntrntion Measurements Ideally. the interstitial fluid bathing the target sites should be sampled to provide a dirf'Cl meas-
426
Oln. Phurmucol.lnN. 19 (6) 1Y90
2. Drug Disposition in Pregnllncy
EC'
The possibili ty of adverse effects of maternal drug administ ration on the developing fetu s may result in reduced maternal compliance (fig. 2). This may appear to reduce drug clearance, as does reduced oral drug absorption by vomiting or by antacid (Carter et al. 198 1) o r mineral ingcstion (Ca mpbell et al. 1988). which may be particularl y relevant wi th current interest in health products. Significant reductions in gastroinlestinal motility on ly occur during labour when absorption is also im pai red, and drugs may accumulate in the gul. After delivery their absorpt ion may result in greatly elevated plasma concentrations and undercstimated clearance values. However, dose requirements may genuincly increase with changing ma-
.
_-
I
--,
I
Ya
..I_
Ya
" 1nw. .... ceo.J
No
00.., __
----Do they
.
c:orr-. wIIh
Ya
~ ••1klnI1
No
Do they conwInI ....,
cor .........,,,.?
-"'No
Ya
-
-
I I
!tug ITIOI ....... 'II
Fig. 1. A'ded$lon Ire(" based o n the major faclors 10 ,o nsider when determining the yalue or Iher.ilpc"utie drug monitonn ...
11
AlbJmin
'FA
1-----
J_
::::---11 ,,
""""''''''I '. ......... ~ Clearance t I . . - Enzyme Oral
-
I'
""".-
I t
Dose raquil1Ml"le!llS
ioduedon
I
Fig. 2. rnnelpil\ rX10f"S which ;nn ... ~n..., mal ~mal dow ~
eon _ _
No
I
qUlrementS In prq.nancy: arT"()\tl'S dIow lhe: dllttuon or ellanl". Increased Yolume ordlsmbul1on (Vd) ca n Iol'>'er the: pta$ma COIK't'TItnlllon and onnuence lher.ilpevl O(" drill mOf11tonn ... and onCTI:ased elearance ca n alter dose ~ul~menls. ECF • e~ traccllular nUId Yolu me: FFA - free fauy acids.
ternal physiology (fig. 2) in complian t women (Nau et at 1982). Total body clearance (el). defined as the volume of plasma cleared of drug per hour per kilogram bodyweight. may i ncrease in pregnancy. During long tenn drug administration it is usually calculated as the ratio of the fraction (f) of dose absorbed in unit time (D) to the plasma concentration at steady-state «('55) (Voteh et al. 1988J. according to the fo llowing form ula: fD CL: - -
C'.
(Eo.
I)
The use of the term 'steady-state' in pregnancy is not strictly appropriate, because pregnancy is the one condition in which the physiological state is reliably not ·steady'. However, useful information can be obtai ned from clearance data if account is taken of factors which may appear to alter ils v alue. Drug concentrations may fall with maternal enzyme induction. reflecting the balance between progesterone and estrogen concentrations (Loock
Drug MOflnonng
In
427
Pregnancy
ct al. 1988) and. to a lesser extent. the increasing contribution 10 drug metabolism by the fetal com· partmcnt (fig. 2) [Juchau 1982). Plasma dilution also lowers Ihe levels of plasma albumin and. less consISICIHI}. plasma brev/M;on.: I", - tennirlal half-life; r~ • bound fraction of drug in plasma: FPlA • fluorescence poIfIrlNtIon Immunoa,uy: AlA • r~S5'Y; EMIT • en~linked imtnunoessay: HPlC - I'Iigh pertom'Iance liquid dYomatogf~: GLe • gal liquid chrom.togr.phy: AAS • •tomic .bsorpIIon apectr~ FES • ft.",. emillion ~oecopJ; OMO • dctmethykIiuepam.
_CUI.
efficacy a nd depend on the infecting palhogen, severi ty of infection and time that the minimum inhibitory concentration (MIC) is exceeded (Nightingale \989). However, Grevel and colleagues (1989) have suggesld that greater therapeutic pre-
cision can be achieved frorp measurements of the area under the concentration-time curve (AUC) than from isolated trough concentration measurements in renal transplant patients. A knowledge of renal status is important as aminoglycosides and
Drug Monllorlng
In
429
Pr('gnanq
some cephalosporins produce nephrotoxicilY and ototoxicity at doses only slightly above the microbiologically effective concentrations. Many of these drugs arc poorly lipid soluble and prOiein bound, and are excreled largely unmelabolised at the glomerular fillration rate (White et al. 198]), which increases in pregnancy (Dunlop 1979). Some newer cephalosporins arc highly bound (Lam el al. 1988), have lower CL and Vd but similar I.·" and show saturable binding which is displaced by FAAs. bilirubin and comedicants (Bialer el al. 1988; Decroix et at 1988). Their protein bi nd ing, therefore. is likely to fall in pregnancy. In general. increased plasma clearances with either unchangcd or increased Vd and shorter ( .1, arc reported for the aminoglycosides (Lazebnik et al. 1985), the penicillin derivatives (Kjer & Ollesen 1986) and the cephalosporins (Chow & Jcwesson 1985) in pregnanc),. and plasma concentrat ions fa ll below the MIC (Nau 1987).
3.1.1 Antiarrh)'lhmics These weak bases are highly ionised at physiological pH and the)' arc absorbed erratically - a problem which ma y be worsened during pregnancy, Disopyramide. like the cephalosporins, shows saturable binding at clinically effective concentrations. For such drugs, unbound clearance values may be bener estimates as they change independently of protein binding changes (Upton & Williams 1986). In a recent study using spiked plasma samples. disopy ramide protein binding was found to be significantly lower in the second and third trimesters and significantly greater I month after delivery than in nonpregnant women (Echizen et al. 1990). Thus, regular monitoring of the free drug concentration is indicated during pregnancy and Ihe puerperium. Digoxin is poorly bound (table I). widely distributed and almost exclusively disposed of by renal elimination. It has a very narrow therapeutic dose ratio. and the correlation between plasma concentration and effect is only significant at 6 to 8 hours after dosing. Hypot hyroidism, hypokalaemia, hypomagnesaemia and hype rcalcaemia can all increase myocardial sensitivity so that toxicity may
occur at subtherapeutic concentrations (Keys & Stafford 1981), while some palients can tolerale concentrations in excess of 3 ~g/L Pharmacodynamic interactions with other antiarrhythmics, however, raise questions as \0 the value of digoxin maintenance therapy and its TOM (Spector el al. 1988) olher than for confirmation of compliance or toxicity. Maternal digoxin concentrations fall in pregnancy (Aronson 1980) as renal clearance increases. and if dose increments arc made. they must be reduced I}()sl parrum. Radioimmunoassays arc unsuitable in pregnancy because the)' show high crossreactivi ty with endogenous digoxin-like substances. found particularly in pregnancies with a high risk of complications ( Koren et al. 1988),
313 Bronchodilalors Theophylline and the anticonvulsants arc Ihe only drugs for which optimal concen tration ranges for saliva as well as plasma have been established (KnOll et al 1984) for routine TOM (table I). Agitation and tachycardia become increasingly evident above 20 mg/L. coincident with saturation of the binding of Iheophylline to albumin. T hcophylline is predominantly hydroxylatcd and demethylated. although small amounts arc excreted unchanged; its metabolism may be inhibited by cimetidine, propranolol and erythromycin. Theophylline protein binding is lo.....cr in the third trimester of pregnancy than in nonpregnant women (Connell y el al. 1990). The plasma clearance and Vd both for theophylline and for caffeine are often reduced in pregnancy (Cartcr 1986; Scarcy et al. 198]) and plasma half-lives are generally increased, thus necessitating morc frequent monitormg.
3.2.4 Antidepressants The pharmacokinetics of the tricyclic and related an tidepressants have been reviewed by Sjoqvist and 8enilsson (1984) and Potter et a1. (1984). The dosage of these highly bound drugs predicts the CS~ poorly because they are cleared at a rate genetically determined by the hydroxylator status (Sjoqvisl & 8enilsson 1984), which may be
eonvuts.ants
Cepllaiosponn,
Anhdepressan15
SalicylIC aCId
t ...
t t 1.
+-+
Benzodlazepones Xanthlnes
l ow UpkI .oIubillty
High I lplcl lOIubility
Cl
_
Vd
t
'. t
,
pt.",,~
Amlnog l1C01ideS
(anbp)'rine)
Lithium PeniCllhns
+
t t , +-+ '. + Cl
, +-+
Vd
, +-+ Low protein
bindIng
Fig. 3. PredIl::I~ chan&« m drug pharmacokmetlCS In prq,nanc), based on their ph)slochemlcal propcnles. with e.\ampks: f - 1I\ereased; •• decreased: . -- • no chance: C L • 10lal bod) plasma drug ckarance: Vd _ \oIumc of dlslrlbuuon: '. '" ttrmmal half·1Ift.
432
Re/eretrcu Albanl E. Riva R. Contin M. BarolZI A. Altoman: M, et II . Dif. fC11:n ti.1 I ... nsplactnl.ill blndin& of "alprooc ~Id: influenct of fm fally ACld$. 8ritlSh Journa l of ClinIcal Pharmacoiosy 17: 759.762. 1984 Aronson JK . Clln..:.1 phannacok"lC'IICS of dltolln 1980. Chnical I'tIarmacoklll~tlC$ 5: 137·149. 1980 BIIrdy AH . Hllinmaa VK. Tenlmo K. Neuvonen PS. Proleln blndlna: of.n,oeplk!oc dropdunna: pr-eananf)'.lIIbor Ind puer· ptnum. ThnlIpeutlC Dru, MORltonna: 12: -4()....46. 1990 8lala' M . Wu WH o Faulkner RD. SIlber BM. YaoolM A.. In vilro prol~ln blndln, Inl
