Extraction of Physostigmine from Biologic Fluids and Analysis by Liquid Chromatography with Electrochemical Detection

GLEN D. LAWRENCEAND NORAINI YATIM

A rapid and simple method is described for the extraction of physostigmine (Phy) and its hydrolysis product, eseroline, from plasma, whole blood, and cerebrospinal fluid (CSF) and their subsequent quantitation by high-performance liquid chromatography (HPLC) with dual electrode electrochemical detection. Phy and eseroline were extracted from biologic fluids with cyano-phase columns eluted with 0.1 M citrate buffer, pH 4 containing 20% acetonitrile. Phy recovery from citrate buffer and CSF was nearly 100%. Phy recovery from plasma was 82% when methanol was used to precipitate proteins and 62% when HC104 was used to precipitate proteins. Phy recovery from whole blood was only 17%. These results are discussed in the context of attempting to measure Phy in fluids of patients receiving this drug in clinical trials for the treatment of Alzheimer’s disease. Key Words: Physostigmine; Plasma; Cerebrospinal fluid; High-performance uid chromatography; Electrochemical detection; Cyano column extraction

liq-

INTRODUCTION Oral

doses

of Physostigmine

(Phy)

have

been

shown

to improve

memory

in

healthy individuals (Davis et al., 1978) and have been used in clinical trials to assess improvements in memory deficit associated with Alzheimer’s disease (Mohs et al., 1985; Stern et al., 1987; Thal and Fuld, 1983). Although improve

in a majority

of patients,

response

in Alzheimer’s

long-term

administration

patients

other

cognitive

receiving

memory

functions

has been found

to

do not show a positive

oral Phy (Thai et al., 1989). A study of

of Phy to Alzheimer’s

patients

has indicated

a sustained

improvement in memory in some patients for up to 1 yr (Stern et al., 1988). Phy is a highly potent and toxic drug with a maximum tolerable oral dose in the range

of 4 mg for human

treatment

of memory

adults

deficit

and a relatively

near its maximum

narrow

tolerable

therapeutic dosage

window

for

(Thai et al., 1989).

There is a large individual variability with regard to tolerance and therapeutic response to Phy. For these reasons, it would be desirable to have a reliable method for monitoring

levels of Phy and its metabolites

in the body during

drug therapy.

Phy is rapidly hydrolyzed under slightly alkaline conditions (pH > 9) to eseroline, which can undergo subsequent autoxidation above pH 6 to rubreserine. Phy is also From the Chemistry Department, Long Island University, Brooklyn, New York. reprint requests to: Glen D. Lawrence, Chemistry Department, Long Island University, Brooklyn, NY 11201-5372. Received November 27, 1989; revised and accepted March 21, 1990. Address

137 Journal of Pharmacological 0 1990 Elsevier

Science

Methods Publishing

24, 137-143 Co.,

Inc.,

(1990)

655 Avenue

0160.5402/90/$3,50 of the Americas,

New

York,

NY 10010

138

G. D. Lawrence and N. Yatim readily hydrolyzed to eseroline by acetylcholinesterase (AChE) and other esterases in plasma. Whelpton and Moore (1985) reported a highly sensitive liquid chromatographic

(LC) method

for analysis

of Phy using electrochemical

detection

and an

alkaline mobile phase. It was also shown that Phy could be extracted from plasma, blood, and cerebrospinal fluid (CSF) with organic solvents under alkaline conditions.

However,

the chemical

seek extraction phase

methods

columns.

We

report

plasma and CSF with electrochemical

a rapid and simple

detection.

AND

of Phy under

more compatible

subsequent

with trying to measure drug in clinical trials. MATERIALS

instability

that were

alkaline

method

for extraction

analysis by reverse-phase

There

is also a discussion

conditions

with Phy stability

led us to

and reverseof Phy from

LC with dual electrode

of the problems

Phy in plasma and CSF of Alzheimer’s

encountered

patients

receiving

this

METHODS

Reagents and Standards Physostigmine were

purchased

were

from

hemisulfate, from

J. T. Baker

Chemical

phases (aromatic sulfonic from J. T. Baker Chemical Nanopure

water

tions were erator.

These stock solutions

was added

under

Extraction

system.

and neostigmine methanol

columns

Physostigmine

in 0.01 N HCI and stored

Eseroline was prepared or salicylate

Co.

salicylate,

Co. HPLC grade

with

bromide

and acetonitrile various

stationary

acid, carboxylic acid, octadecyl, and cyano phases) were Co. Distilled water was further purified with a Barnstead

purification

prepared

physostigmine

Sigma Chemical

were

hydrolysis

atmosphere.

to 1.98 mL distilled

stock

bottles

solu-

in the refrig-

stable for at least 1 mo.

by base-catalyzed

a nitrogen

and neostigmine

in dark brown

water,

of physostigmine

Phy stock solution

the

reaction

hemisulfate

(20 PL, 100 pg/mL)

vial stoppered

with

a rubber

septum, deaerated with water saturated nitrogen gas for IO min, and 2.00 mL of deaerated 0.10 N NaOH added with a gas tight syringe. A stream of nitrogen passed over the solution for IO min and the resulting hydrolysate acidified with 4.00 mL of 0.10 N HCI to prevent autoxidation. The hydrolysis was complete within 10 min in dilute

NaOH,

indicated

by the loss of the Phy peak and appearance

of the eseroline

peak in the LC system. This eseroline stock solution was stable for several days in the refrigerator and was diluted with mobile phase to the desired concentration for LC analysis. Authentic eseroline, search Biochemicals, Inc. (Natick, prepared

by base hydrolysis

eseroline

from

as the fumarate salt, was later obtained from ReMA, USA) and found to coelute with the eseroline

and showed

Phy by this hydrolysis

there

was essentially

100%

recovery

of

procedure.

Rubreserine was prepared by autoxidation of eseroline under alkaline condition and was detected as a separate peak in the LC system. Authentic rubreserine was not available for chromatographic comparison, and no attempt was made to quantitate the rubreserine prepared by this technique.

Chromatography The chromatographic system consisted system, a Gilson model 231 autoinjector,

of a Waters model 6000K solvent delivery a 4.6 mm x 10 cm, 3-pm spherical Cls

Physostigmine in Biologic Fluids reverse phase analytical column with 4.6 x 30 mm C18 guard column upstream from the analytical column. The ESA model 5100A Coulochem electrochemical detector (ESA, Inc., Bedford, MA, USA) had a guard cell placed between the column and the analytical cell, which operated at -0.40 V (reducing). The upstream electrode (detector 1) of the analytical cell operated at + 0.80 V to oxidize Phy and the downstream electrode (detector 2) of the analytical cell operated at -0.20 V to reduce the oxidized intermediate. A Kipp and Zonen strip chart recorder monitored the output signal from detector 2. The mobile phase was 0.10 M sodium citrate buffer, pH 4.0 with 0.02% sodium octylsulfate (w/v), 0.05% octylamine (v/v), and 8% acetonitrile (v/v). The ion-pairing agents were necessary in the mobile phase to prevent tailing of the Phy peak. The mobile phase flow rate was 1.0 mUmin. The autoinjector was programmed for 40 ~J,Linjections. Rubreserine and salicylate eluted after Phy in the chromatographic system and were well resolved from the Phy peak. Physostigmine

Extraction from Plasma and CSF

Plasma (usually 1.0 mL) from healthy volunteers was spiked with 5-100 ng Phy/ mL, made 0.3 N in perchloric acid, and centrifuged at 12,000 x g for 15 min to remove protein. The clear supernantant was transferred with a Pasteur pipet to a 3-mL cyano extraction column (Baker Chemical Co.) that had been preconditioned with 0.01 M sodium citrate buffer, pH 4.0. After all of the sample entered the solid phase, the column was washed with 2.0 mL of 0.01 M citrate buffer, followed by 1.0 mL of 8% acetonitrile in 0.10 M citrate buffer, pH 4.0. Then 1.5 mL of 20% acetonitrile in 0.10 M citrate buffer was added to elute the Phy from the column. Plasma (usually 1 mL) was also treated with an equal volume of methanol to precipitate protein, and the clear supernatant after centrifugation at 12,000 x g was evaporated under a stream of nitrogen gas for 20 min to remove methanol. The remaining aqueous residue was transferred with a Pasteur pipet to a 3-mL cyano extraction

column,

perchloric

acid treated

CSF from

followed

Phy-free

CSF sample

added

by column

wash

and elution

individuals directly

was spiked

with IO-50

to the preconditioned

teers was spiked with Phy (51 ng/mL recovery studies in whole blood. Blood was collected clotting

cells as soon

for

blood)

from Phy treated

and immediately

as possible

after

blood

in vacutainer

of plasma

(usually

for Phy

tubes containing

on ice. Plasma was separated

was collected

column.

column in the same way de(4.0 mL) from healthy volun-

prior to separation

patients

placed

ng Phy/mL and the spiked

3-mL cyano extraction

The Phy was washed and eluted from the extraction scribed for plasma supernatant above. Whole blood

to prevent

steps as described

plasma.

IO-20

min).

EDTA

from

red

In some

cases, neostigmine (I-IO kg/mL) was added to whole blood prior to separation of plasma. It was felt that high concentrations of neostigmine would compete with Phy for binding to the active site of AChE, inhibiting the hydrolysis of Phy during the plasma separation procedure and possibly displacing Phy from other potential binding sites in whole blood.

139

140

G. D. Lawrence and N. Yatim RESULTS AND Eseroline

DISCUSSION

and Phy had retention

chromatographic

conditions

signal at these min under

detector

times of 3.9 and 6.0 min, respectively,

described

potentials.

these chromatographic

(see Figure 1). Neostigmine

Salicylate conditions.

from

some

under

Phy standards

The peak height

the

did not give any elutes

response

at 9

was found

to be linear (r2> 0.99) over a range of 0.5-10 ng Phy injected, with a lower limit of detection of about 0.2 ng/mL. Eseroline gave a sharper peak in the chromatogram, resulting

in greater

assay sensitivity

for this hydrolysis

product.

Several stationary phases were tested for Phy extraction. In summary, aromatic sulfonic acid and carboxylic acid ion exchange resins gave poor recovery and required

extreme

umns.

Extremes

changes

in pH or salt concentrations

in pH or salt concentration

to elute

resulted

Phy from

in distortion

these

col-

of the Phy peak

in the HPLC system. Reverse-phase C18 resins required high levels of organic solvent in the elution medium to remove Phy, and even then the Phy eluted in a large

A

B

Phy

.i

C

Phy

L-J a

a

Is,. 12

I. 0

I

min

..-

1

4

0

FIGURE 1. (A) Chromatogram of a mixture of physostigmine and eseroline prepared in situ from physostigmine hemisulfate (see Methods), corresponding to 1.63 ng Phy and 1.29 ng eseroline. (B) Chromatogram of spiked plasma (51 ng Phy/mL) following the extraction procedure, using methanol to precipitate proteins, as described in the text. (C) Chromatogram of plasma without added Phy following the extraction procedure as described in (B). Es = eseroline; Phy = physostigmine.

Physostigmine in Biologic Fluids volume pairing

of extraction

buffer,

resulting

agents to the extraction

poor reproducibility. Of the solid phases

tested,

suitable for Phy extraction. was 99 ? 7% from citrate Eseroline

cyano

Methanol perchloric

extraction

of Phy. Addition gave variable

columns

were

of ion

results and

found

to be most

in CSF that had been spiked with eseroline,

in CSF of Phy treated was found

recoveries

for Cls columns

Recovery of Phy by the cyano column extraction method buffer solutions and 102 ? 4% from CSF (see Table I).

could be quantitated

was detected

in poor

medium

to be a better

acid for Phy recovery.

but none

patients. plasma protein

However,

precipitating

evaporation

agent than 0.3 N

of methanol

from

plasma

supernatant was necessary for the subsequent cyano column extraction of Phy because this analyte did not bind to the cyano extraction columns in the presence of high methanol

concentrations

in the sample

medium.

There was 82 ? 4% recovery

of Phy from 1.0 mL spiked plasma (102 ng Phy/mL) treated with an equal volume of methanol to precipitate proteins and 62 + 4% recovery of Phy from 1.0 mL spiked plasma

treated

with

perchloric

acid to precipitate

proteins

(see Table

1). Plasma

contained substances that interfered with eseroline quantitation, which were not removed by any of the extraction procedures used for Phy, and coeluted with eseroline in the chromatographic It appears that Phy is bound

system (see Figure 1). to plasma proteins and may be trapped

in the pellet

when proteins are precipitated. Unni and Somani (1985) have demonstrated Phy binding to human and rat plasma proteins and reported a K, of 1.6 PM for Phy binding

to crystalline

50% methanol are differences

serum albumin.

The difference

versus 0.3 N perchloric in Phy partition

between

these denaturing agents. There was only 17 + 5% recovery

in plasma Phy recoveries

acid precipitation the soluble

of proteins and particulate

of Phy from spiked whole

(IO kg/mL) was added to the whole blood prior to spiking of Phy increased to 20 & 4% (see Table 1).

blood.

% PHY Citrate

buffer,

Cerebrospinal

EXTRACTED

Plasma (methanol Plasma (HCIO, Whole

blood

Whole

blood

102 + 4

and calculation eluent.

82 2 4

pptn)

62 -t 4

pptn)

17 2 5 (3 mg/mL

See text for conditions are for 5-102

99 ? 7

pH 4.0 fluid

neostigmine

and procedures.

ng Phy added/ml

biologic

of the total obtained

Percentages

added)

are mean

20 2 4 Recoveries

fluid or buffer

in the extraction

2 SD for 3-5

phases

with

If neostigmine

it with Phy, the recovery

TABLE 1 Physostigmine Recoveries from Various Media by the Cyano Column Extraction Method

MEDIUM

with

suggests there

samples.

141

142

G. D. Lawrence and N. Yatim

The red cell membrane is known to be extremely rich in AChE activity (Rosenberry and Scoggin, 1984). Brossi et al. (1986) found 50% inhibition of purified electric eel AChE activity at 4.0 nM Phy. Fifty percent inhibition of mouse brain AChE activity was found at 60 nM Phy in earlier studies (Arnal et al., in press). Considering the relative affinity of Phy for plasma albumin and red-cell AChE, most of the Phy in the blood would be expected to be associated with the red-cell AChE and lost in that fraction during plasma separation at the Phy concentrations used in spiked whole-blood samples in this study. Attempts to measure Phy in plasma and CSF of patients receiving oral Phy (up to 4 mg) have not been successful in our laboratory to date. We believe this is due to the rapid and efficient sequestering of Phy by erythrocyte AChE and plasma proteins. Whelpton and Moore (1985) have reported measuring Phy in blood of a single healthy volunteer receiving a 4-mg oral dose of Phy, although the drug could not be measured when the same individual received a lower dose (1 or 2 mg). The cyano extraction method reported here allows a simple and rapid extraction of Phy from CSF and plasma. It is possible to achieve an 8-IO-fold increase in sample concentration by increasing the volume of sample applied to the extraction column (up to 15 mL) relative to the volume of eluent (1.5 mL used in this study). When more than 15 mL of Phy solution was added to the extraction columns, the recoveries began to decrease. This cyano extraction procedure results in similar recoveries, yet offers advantages over the ether extraction method of Whelpton and Moore (1985) by avoiding alkaline conditions. Their chromatographic system had a greater sensitivity than ours (0.05 ng/mL versus 0.2 ng/mL, respectively), which may account for their success in measuring plasma Phy after an oral dose in one healthy patient and our lack of success in several Alzheimer’s patients receiving this drug in clinical trials. lsaksson and Kissinger (1987) have recently reported similar recoveries (62%) of Phy from spiked plasma using C 18 extraction columns, but their procedure required high concentrations of ion-pairing agents that resulted in nonlinear responses in their chromatographic system. It is clear from the present study that erythrocyte AChE activity and plasma protein binding of Phy will confound measurements of Phy in clinical trials of patients receiving this drug. The authors numerous

thank

Dr. Lucien J. Cote of Columbia

discussions

Yatim thanks

the Parkinson’s

study. C. D. Lawrence Federation

and guidance

for granting

Disease

and providing Foundation

thanks the Research time from teaching

University,

for a summer

Time Awards duties

College

his laboratory

research

Committee

to participate

of Physicians

facilities

for much

stipend

and Surgeons

for

of this project.

N.

to perform

part of this

of the Long Island University

in this research

Faculty

project.

REFERENCES Brossi A, Schonenberger Inhibition eel by

(-)-

compounds.

B, Clark OE, Ray R (1986)

of acetylcholinesterase and (+)-physostigmine FEBS Lett 201:190-192.

from

electric

and related

Davis KL, Mohs Hollister provement normal

RC, Tinklenberg

LE, Kopell

of long-term

humans.

JR, Pfefferbaum

BS (1978) Physostigmine:

Science

memory

processes

201:272-274.

A, Imin

Physostigmine in Biologic Fluids lsaksson K, Kissinger P (1987) Determination of physostigmine in plasma by liquid chromatography with dual electrode amperometric detection. / Liq Chromatogr 10:2213-2229. Mohs RC, Davis BM, Johns CA, Mathe AA, Greenwald BS, Horvath TB, Davis KL (1985) Oral physostigmine treatment of patients with Alzheimer’s disease. Am / Psychiatry 142:28-33. Rosenberry TL, Scoggin DM (1984) Structure of human erythrocyte acetylcholinesterase. Characterization of intersubunit disulfide bonding and detergent interaction. 1 Biol Chem 259:56435652. Stern Y, Sano M, Mayeux R (1987) Effects of oral physostigmine in Alzheimer’s disease. Ann Neuro/ 22:306-310.

Stern Y, Sano M, Mayeux R (1988) Long-term administration of oral physostigmine in Alzheimer’s disease. Neurology 38:1837-1841. Thal LJ, Fuld PA (1983) Memorv enhancement with oral physostigmine in Alzheimer’s disease. N fngl / Med 308:720-721. Thal LJ, Masur DM, Blau AD, Fuld PA, Klauber MR (1989) Chronic oral physostigmine without lecithin improves memory in Alzheimer’s disease. / Am Geriatr Sot 37~42-48. Unni LK, Somani SM (1985) Binding of physostigmine to rat and human plasma and crystalline serum albumins. Life Sci 36:1389-1396. Whelpton R, Moore T (1985) Sensitive liquid chromatographic method for physostigmine in biological fluids using dual-electrode electrochemical detection. / Chromatog 341:361-371.

143