Clinica Chimica Acta, 200 (1991) 95 - 106

95

0 1991 Elsevier Science Publishers B.V. 0009-8981/91/$03.50

ADONIS 000989819100158W

CCA 05023

A sensitive fluorescence assay for quantitation

of fludarabine and metabolites in biological fluids
Annette Kemena , Mechelle Fernandez *, John Bauman *,
Michael Keating and William Plunkett
I The University of Texas M. D. Anderson Cancer Center, Houston, TX and Berlex Biosciences,
Alameda, CA (USA)

(Received 24 December 1990; accepted 22 March 1991)

Key words: Fluorescence assay; Fludarabine; Biological fluids

Summary
Rapid and quantitative dephosphorylation
of the new anticancer nucleotide
analogue fludarabine phosphate to its nucleoside 9-/3-D-arabinofuranosyl-2-fluoroadenine (F-ara-A) renders this metabolite the target for pharmacologic investigations. At clinically effective doses of fludarabine phosphate (18-30 mg/m per
day) comprehensive pharmacokinetic analysis of F-ara-A has been limited by the
sensitivity of UV based HPLC assays. To address this problem we developed a
sensitive test based on the condensation of F-ara-A with chloroacetaldehyde
to
form the fluorescent derivative, arabinosyl-l,N6-etheno-isoguanine.
Combined with
a solid-phase extraction step prior to derivatization and separation of the reaction
products by reverse-phase
HPLC, this assay had a quantitation limit of 2 pmol
F-ara-A per ml plasma. Slightly modified, the system was also applicable to urine
specimens, with a quantitation limit of 1 nmol F-ara-A per ml urine.

Introduction
The nucleotide analogue 9-/3-D-arabinofuranosyl-2-fluoroadenine
5-monophosphate (fludarabine phosphate, F-ara-AMP, Fludara), appears to be effective in the
treatment of low grade lymphocytic malignancies at doses of 18-30 mg/m per
day for 5 days [l-5]. Nevertheless, clinical studies in acute leukemias which

Correspondence and requests for reprints to: William Plunkett, Department of Medical Oncology, Box
52, The University of Texas M. D. Anderson Cancer Center, Houston, TX 77030, USA.

employed relatively high doses either by daily i.v.-infusions (196 mg/m* per day
for 5-7 days) or continuous infusions (> 40 mg/m* per day for 5 days) were
associated with severe neurotoxicity [6-81. This therapeutic window suggested that
pharmacokinetic
analyses may provide critical information concerning dose
scheduling and the rationale for the design of combination therapies [9,10].
Rapid and quantitative dephosphorylation
of fludarabine phosphate to its
nucleoside 9-P-D-arabinofuranosyl-2-fluoroadenine
(F-ara-A) [ill renders this
metabolite the target for pharmacologic investigations. Conventional methods of
quantitating F-ara-A in plasma consist of a deproteinization procedure prior to
HPLC separation and detection by UV absorbance [ll-141. The detection limits
reported were 0.05 [13] and 0.53 [ll] nmol F-ara-A/ml plasma, or approximately
15 pmol, given as an absolute value [12,14]. The similarity of the assay systems used
suggests that the sensitivities of all four are essentially comparable. After low-dose
i.v.-infusions of fludarabine phosphate (18-25 mg/m* over 30 min) the range of
the detection limit was reached in one case at 24 h [14]; in another F-ara-A was no
longer detectable at 3 h after the end of the infusion [13]. At a dose of 80 mg/m*
given as an i.v.-bolus the detection limit was approached 30 h after treatment [ll].
To achieve increased sensitivity, we combined a solid-phase extraction for
deproteinization
and concentration of the drug metabolite with a fluorescence
detection method after HPLC separation. The final validation of the method gave
a quantitation limit of 2 pmol F-ara-A per ml plasma.
Materials and methods
Materials

F-ara-A was prepared by Ash Stevens Inc. (Detroit, MI). 2-Fluoroadenine

(F-Ade) and arabinosylisoguanine 5-monophosphate were synthesized at Berlex
Biosciences, Alameda, CA. 9-P-D-arabinofuranosyl-2-fluoroadenine
5-triphosphate (F-ara-ATP) was synthesized by Alina Sen, Department of Medical Oncology, University of Texas M. D. Anderson Cancer Center. F-ara-AMP and erythro9-(2-hydroxy-3nonyl)adenine
were obtained from the Drug Synthesis and Chemistry Branch of the National Cancer Institute, Bethesda, MD. Adenine (Ade),
adenosine (Ado), deoxyadenosine (dAdo), 1,N6-etheno-adenosine
(E-Ado), 1,N6etheno-deoxyadenosine
(e-dAdo), cytidine (Cyd), and deoxycytidine (dCyd) were
purchased from Sigma Chemical Co. (St. Louis, MO). Chloroacetaldehyde
(50%),
and N,N-dimethylformamide
(HPLC grade) were obtained from Aldrich Chemical
Co. (St. Louis, MO). All other chemicals used were of the highest purity available.
The SEP-PAK C,, cartridges were a product of Waters Associates (Milford, MA).
Synthesis, isolation, and spectroscopic characterization
derivative

of the fluorescent

F-ara-A

A sample of F-ara-A (0.285 g, 1.0 mmol) was suspended in a mixture of 50%

aqueous chloroacetaldehyde (3.0 ml) and potassium acetate buffer (pH 4.0, 2.0 ml)

97

and heated at 55C overnight. The mixture was concentrated under high vacuum to
obtain a glassy solid. This was triturated in 10 ml of acetonitrile and concentrated
under vacuum. The residue was resuspended in acetonitrile and allowed to stand
at room temperature overnight. The fine precipitate that formed was collected by
filtration and suspended in 3 ml of water before neutralization with saturated
aqueous sodium bicarbonate. After filtration the aqueous solution was concentrated to about 1 ml and refiltered prior to preparative HPLC. The sample was
divided in half, each portion was loaded onto a 10 X 250 mm Alltech Econosil C,,
column and eluted with 5:95 methanol: water at a flow rate of 5 ml/min and
monitored at 254 nm. Appropriate fractions were pooled, concentrated under
vacuum, and the resulting glassy residue was dissolved in 0.5 ml methanol and
diluted with 2 ml acetonitrile. The solvents were evaporated under vacuum to
obtain 40 mg of the fluorescent F-ara-A derivative as a dry but hygroscopic
powder. This material was approximately 95% pure by HPLC and was shown to be
identical to material prepared in the analytical scale experiments by HPLC
coelution and fluorescence characteristics. H-NMR (DMSO-d,, 500 MHz) 6 3.62
(dd, J = 5.0 and 11.6 Hz, H-5a), 3.69 (dd, J= 4.0 and 11.6 Hz, H-5b), 3.76 (q,
J = 4.4 Hz, H-4), 4.11 (t, J = 4.4 Hz, H-2 or H-3), 4.16 (t, J = 4.2 Hz, H-3 or
H-2), 6.16 (d, J= 4.8 Hz, H-l), 7.04 (bs, H-10 or H-111, 7.43 (bs, H-11 or H-101,
8.67 (s, H-8) (purine nucleoside numbering system, OH signals not listed); 13C-NMR
(DMSO-d,, 125 MHz) 61.3, 75.7, 76.0, 83.6, 88.3, 110.2, 112.2, 127.9, 134.1, 143.2,
144.3, 151.1 ppm; CI+-mass spectrometry (m/z) 308 (M + H)+; UV (methanol)
Amax 292 nm.
Plasma preparation and deriuatization

Blood samples (5-10 ml>, obtained through a peripheral catheter, were immediately transferred into a heparinized glass tube containing 100 ,ul 0.1 mmol/l
erythro-9-(2-hydroxy-3-nonyljadenine
to inhibit adenosine deaminase, and placed
in an ice-bath. After centrifugation for 20 min at 500 X g and 4C the supernatant
was removed and stored at -20C. Standard samples were prepared by spiking 5
ml of thawed fresh frozen plasma with a known amount of F-ara-A.
Protein removal was accomplished by solid-phase extraction. A SEP-PAR C,,
cartridge was preconditioned
with 2 ml alkaline methanol (1 ,& concentrated
ammonium hydroxide per 1 ml methanol) and subsequently rinsed with 10 ml
water. The thawed plasma was centrifuged to remove coagulates, and 1 to 5 ml was
passed through the cartridge, which was then rinsed with 5 ml water, forcing out
most of the liquid phase. The F-ara-A was eluted with 2 ml alkaline methanol. All
steps were carried out at a flow rate of about 5 ml/min using a Vat Elute SPS 24
(Analytichem International Inc.). The methanol was evaporated completely in a
Speed Vat Concentrator (Savant), and the sample was stored at - 20C. To the dry
sample 0.1 ml 0.6 mol/l citrate buffer, pH 4.0, was added and vortex-mixed to
dissolve. After addition of 0.2 ml 50% chloroacetaldehyde (final concentration 5.2
mol/l>, the sample was incubated for 24 h in a 50C water bath.

98

Preparation and derivatization of urine samples

Urine fractions of 300-1,000 ml were collected at 8-h intervals in bags containing 100 ~1 1 mmol/l erythro-9-(2-hydroxy-3-nonyl)adenine.
Aliquots of about 2 ml
each were stored at -20C. For derivatization, thawed urine was centrifuged to
remove sediment and diluted 1: 10 with water; 0.1 ml was added to an equal
volume of 1 mol/l citrate buffer, pH 4.0 and 0.3 ml 50% chloroacetaldehyde was
added (final concentration 4.7 mol/l>. The sample was incubated for 24 h in a
50C water bath. Standard samples were prepared by spiking urine with a known
amount of F-ara-A.
Separation and detection

After derivatization, the samples were diluted 1: 10 with 0.1 mol/l citrate
buffer, pH 4.0, and 50 ~1 was injected into a Waters HPLC system consisting of 2
model 6000A pumps, a model 715 Ultra WISP sample processor, a model 470
fluorescence detector, a system interface module, and an NEC data capture system
equipped with Maxima 820 Chromatography
Software. As stationary phase a
PBondapak C,, column (Waters Associates, 3.9 mm X 30 cm, 10 km particle size)
connected with a precolumn filter cartridge (Waters Associates) was used. The
mobile phase consisted of a ternary solution of 2% methanol and 5% N,N-dimethylformamide in water. Under isocratic conditions at a flow rate of 1 ml/min the
fluorescent derivative of F-ara-A eluted with a retention time of about 10 min. An
excitation wavelength of 296 nm and an emission wavelength of 410 nm were
optimal for detection. Fluorescence signals of samples corresponding to more than
42 pmol or 12 ng of underivatized F-ara-A per injection were truncated by the
detector at the highest sensitivity setting, necessitating further dilution of these
samples. After 20-25 runs the column was regenerated with a gradient from the
ternary mobile phase to 50% methanol in water to maintain optimal separation.
Analysis of data

F-ara-A concentrations in patient samples were determined by comparison of

peak areas of the fluorescent derivative to those of standard samples spiked with
F-ara-A which had been processed simultaneously in triplicate. Pharmacokinetic
analyses were performed using the ESTRIP computer program 1151.
Results
Optimization of the reaction conditions

No fluorescent signal was detected with F-ara-A at reaction conditions reported

to be optimal for the generation of etheno-adenine compounds (0.03-0.07 mol/l
chloroacetaldehyde
at pH 4.5 to 6.1 for 10 to 40 min at 80 to lOOC>[16-201;
therefore modifications were required. About 2 mol/l chloroacetaldehyde
was

99

HO
HO
4
3
Fig. 1. Formation and structure of the fluorescent F-ara-A derivative.

needed to reach saturation of the conversion to the fluorescent derivative in 0.2

mol/l citrate buffer, pH 4.0. No inhibitory effect of the chloroacetaldehyde on the
formation of the fluorescent product was seen up to a concentration of 5.5 mol/l.
Best yields were obtained at pH 4; increasing the pH above 4.5 or decreasing it
below 3.5 resulted in the appearance of multiple fluorescent peaks in parallel with
a decrease in the F-ara-A derivative signal. Incubation at 50C for 20 h maximized
the product; increasing the incubation time to 25 h did not change signal intensity.
With [8-3H]F-ara-A, under optimal reaction conditions (5.2 mol/l chloroacetaldehyde in 0.2 mol/l citrate buffer, pH 4.0 at 50C for 24 h) 66.5 + 1.8% (n = 6) of the
radioactivity was associated with the fluorescent product.
Characterization of the fluorescent product

By analogy with other adenine nucleosides it was anticipated that derivatization

of F-ara-A (1) with chloroacetaldehyde would form the 1,N6-etheno-derivative
(2)
(Fig. 1). To confirm this hypothesis the derivatization reaction was scaled-up and
the derivative was isolated and characterized. The isolated derivative and the
derivative from the analytical-scale reactions were identical as determined by
HPLC coelution and by their fluorescence characteristics. On the basis of NMR
and mass spectral data it was determined that in the isolated fluorescent derivative
the fluorine had been replaced by a hydroxyl group. The loss of the fluorine was
most obvious from the 13C-NMR spectrum which showed seven single lines for the

100

TABLE I
Derivatization to fluorescent compounds *
compound

retention time [min]

relative sensitivity

F-ara-A
F-ara-AMP
F-ara-ATP
F-Ade
Ado
dAdo
Ade
Cyd
dCyd
arabinosylisoguanine

9.8
4.0
3.1
19.7
18.3
24.3 and 14.3
14.4
no signal up to 20 PM
no signal up to 20 PM

1.oo
1.00
0.98
0.17
0.26
0.05 and 0.12
0.15

9.8

1.36

* The listed compounds were incubated with chloroacetaldehyde for condensation

etheno-derivatives under the conditions optimized for F-ara-A (see Methods).

to the respective

aromatic carbon signals and no carbon-fluorine spin-spin coupling which would be

characteristic of a fluoroheterocycle.
The derivative has thus been identified as
arabinosyl-l,N6-ethenoiso-guanine
{4}. Additional evidence for this assignment
was obtained by derivatization of arabinosyl-isoguanine (3) with chloroacetaldehyde to form the same fluorescent product as determined by HPLC coelution and
fluorescence characteristics. This reaction proceeds under relatively mild conditions which suggests that {3} may be an intermediate in the conversion of F-ara-A
to {4). Most likely, hydrolytic removal of the deactivating fluoro-substituent
of
F-ara-A occurs prior to - and thus facilitates - the condensation with chloroacetaldehyde. However, on the basis of this evidence alone, the formation of (4) via
121 cannot be ruled out completely.

Specificity of the derivatization reaction

In addition to F-ara-A, physiological nucleosides of adenine and cytosine which
occur in plasma are known to react with chloroacetaldehyde
to form fluorescent
etheno-derivatives [21]. As these could potentially interfere with the determination
of F-ara-A, it was appropriate to identify their fluorescent derivatives by HPLC.
Table I lists the retention times and relative sensitivities of some representative
nucleoside derivatives under the assay conditions optimized for F-ara-A. Only the
adenine nucleosides Ado and dAdo could be detected. The relative sensitivity for
these nucleosides was considerably less than the sensitivity for F-ara-A, which may
be attributed to several factors. First, the derivatization conditions may be less
than optimal. For example, under the vigorous conditions required for the derivatization of F-ara-A, two fluorescent derivatives were formed from dAdo. The major
derivative was identified as e-Ade and the minor as E-dAdo by HPLC coelution
with authentic compounds. This result is consistent with the known lability of

101

2-deo~urine
nucleosides under acidic conditions. in addition, the excitation and
emission wavelengths (296 nm and 410 nm, respectively) have been optimized for
detection of the F-ara-A derivative and are clearly less than optimal for detection
of either the adenine or cytosine derivatives.
The relative sensitivity of this assay for F-ara-A and for its nucleotides was very
similar. The sensitivity for F-Ade, however, was considerably lower, perhaps due to
the fo~ation of multiple ChIoroacetaIdehyde derivatives which were not detected
by fluorescence.
Since both F-ara-A and arabinosyl-isoguanine form the same fluorescent derivative, these cannot be independently quantitated by this assay. The higher sensitivity
for arabinosyl-isoguanine
indicates that it is derivatized more efficiently than
F-ara-A. This is consistent with the 66.5% conversion rate observed for F-ara-A. It
is also consistent with the expectation that hydrolytic cleavage of F-ara-A to
arabinosyl-isoguanine is the iimiting step in the derivatization of F-ara-A.
Validation of the assay in biological samples

A linear relationship between the concentration of F-ara-A in plasma and the

fluorescent signal was found throughout the entire range tested from 2 pmot to 2

ooi
0

F-ara-A. nH

10

10

15

20

minutes

Fig. 2. Concentration-to-signal
relationship of the derivatization of F-ara-A in plasma in the low
concentration range, including the quantitation limit. Each data point represents the mean of triplicate
determinations&SD.
Standard deviations ranged from 8% tat 2 nmol/l) to 2% (at 10 nmol/l), and
therefore some SD values do not appear. 5 ml plasma was processed for each determination as
described in Methods.
Fig. 3. HPLC chromatograms of patient plasma samples obtained prior to treatment (lower panel), and
6 h after the end of a 30-min i.v.-infusion of 60 mg fludarabine phosphate (upper panel). Plasma (1 ml)
was processed in each case as described in Methods. The fluorescence signal at 9.8 min was equivalent
to an F-ara-A plasma concentration of 0.34 pmol/l.

102

nmol/ml (r = 0.9997). Figure 2 shows the concentration-to-signal

relationship for
the concentration range between 2 and 10 pmol F-ara-A per ml plasma. The limit
of quantitation of F-ara-A in plasma was 2 pmol per ml (coefficient of variation =
8.1%, n = 3) when 5 ml of plasma was processed. This is equivalent to 0.57 ng
F-ara-A per ml plasma, or 167 fmol per 50 ~1 HPLC injection. Plasma samples
spiked with 0.1 nmol F-ara-A per ml, a concentration present 24 to 72 h after
fludarabine phosphate infusion (vide infra), had a precision of 3.0% (n = 6).
Investigation of the stability of the fluorescent F-ara-A derivative in the assay over
8 h showed a linear decay of the signal of 0.5% per h at 25C in the dark.
The sensitivity for F-ara-A in plasma samples could be improved by removal of
the plasma proteins and by sample concentration. This was accomplished by
solid-phase extraction which was shown to remove over 99.8% of the protein and
to allow quantitative recovery of the F-ara-A. The recovery was determined by
comparing peak areas from water and plasma samples spiked with F-ara-A; only
the latter were processed by the solid-phase extraction. Comparison of the corresponding samples showed no significant differences in peak areas over the concentration range from 0.05 to 2 nmol F-ara-A per ml plasma. In addition, dialysis of a
plasma sample which had been spiked with [8-3H]F-ara-A showed that < 0.03% of
the original radioactivity was associated with the plasma proteins.
Processing urine samples by solid-phase extraction did not improve the
signal : noise value or lower the detection limit of the fluorescent F-ara-A derivative. Therefore unprocessed urine specimens were employed in the derivatization
step. A linear relationship with fluorescence response was shown from the quantitation limit of 1 nmol F-ara-A per ml urine (equivalent to 0.29 Kg/ml, or 100 fmol
per HPLC injection; coefficient of variation 5.6%, n = 3) up to 200 nmol/ml
(r = 0.9993). This range includes F-ara-A concentrations found in urine up to 72 h
after fludarabine phosphate infusion.
Quantitation of F-ara-A in patient samples
After obtaining informed consent, patients with refractory chronic lymphocytic
leukemia were treated with 60 mg of fludarabine phosphate infused i.v. over 30
min. Representative chromatograms of plasma before and 6 h after fludarabine
phosphate infusion are shown in Fig. 3. There were no interfering peaks at a
retention time around 10 min in this and other pretreatment samples. For the 6-h
chromatogram the peak at 9.8 min corresponded to an F-ara-A plasma concentration of 0.34 pmol/l. The peak at a retention time of about 18 min was identified
as e-Ado by coelution with an authentic sample, and by the fact that the peak
could be eliminated by treatment of the sample with adenosine deaminase before
protein removal.
The kinetics of F-ara-A in the plasma of patients after a 30-min i.v.-infusion of
60 mg fludarabine phosphate are shown in Fig. 4. The highest plasma concentration, measured 5 min after the end of the infusion, was 2.54 f 0.98 pmol/l (n = 3).
The data for individuals were fit to a triexponential equation giving elimination
half-lives of 9 f 7 min for the a-phase, 1.7 &-0.8 h for the P-phase, and 33.5 Z!Z
5h

103

.OlL
0

I
20

40

60

24

40

72

hours after treatment

hours
after
treatment
Fig. 4. F-ara-A plasma elimination after a 30-min i.v.-infusion of 60 mg fludarabine phosphate. The
data points are mean values f SD of elimination kinetic studies from three patients. Plasma (1 to 4 ml)
was processed at each time point according to the procedure described in Methods.

Fig. 5. Excretion of F-ara-A in urine after a 30-min i.v.-infusion of 60 mg fludarabine phosphate. The
data represent mean values f SD of three patients; 10 ~1 of urine from each 8-h collection period was
processed as described in Methods.

for the terminal phase. The F-ara-A plasma level was 62 & 19 nmol/l 72 h after
the end of the infusion. A total body clearance of 15.0 k 9.7 l/h was calculated
from these data. Quantitation of F-ara-A in urine collected in 8-h intervals for 24 h
after fludarabine phosphate infusion indicated that 37 f 5% of the dose had been
excreted during that time (Fig. 5). When values from the interval 64-72 h after
infusion were included, excretion of 57 f 7% of the dose was estimated with a
renal clearance of 5.4 k 1.6 I/h.

Discussion
The determination of F-ara-A in plasma at fludarabine phosphate doses in the
range of X3-30 mg/m2 per day has been limited by the sensitivity of UV based
HPLC assays [12-141. To study the F-ara-A pharmacokinetics more precisely at
these low doses, we developed a more sensitive assay based on fluorescence
detection after HPLC. The condensation of adenine nucleosides with chloroacetaldehyde produces fluorescent 1,N6-etheno-derivatives
[16-18,21-261, and assays employing the detection of these compounds by fluorescence have shown a loto 30-fold greater sensitivity than the corresponding UV based methods [l&19].
Treatment of F-ara-A with chloroacetaldehyde under conditions suitable for the
derivatization of other adenine nucleosides [16-201 produced no fluorescent
derivative. The unique structural characteristic of this adenine nucleoside analogue

104

is the fluorine attached to the 2-carbon of the adenine moiety (Fig. 1, (1)) which
renders it a poor target for degradation by adenosine deaminase [29,30]. Substitution of adenine nucleosides with a halogen atom at C-2 also considerably reduces
their reactivity with chloroacetaldehyde
[23]. It was therefore necessary to modify
the derivatization conditions. We found that by decreasing the pH and temperature of the reaction while increasing the reaction time and concentration of
chloroacetaldehyde,
a fluorescent derivative of F-ara-A could be produced in
66.5% yield. The use of bromoacetaldehyde,
which has been reported to provide
superior results in some cases 1251, led to the generation of two fluorescent
derivatives from F-ara-A and was therefore not suitable.
Purification and structural characterization revealed that the fluorine had been
replaced by a hydroxy group during derivatization, so that the fluorescent derivative was actually arabi~osyl-l,~~-etheno-isoguanine
(Fig. 1, (4)) rather than
etheno-F-ara-A (2). The identity of the compound was supported by conversion of
authentic arabinosyl-isoguanine to its etheno-derivative, which coeluted by HPLC
with the F-ara-A reaction product. Thus, the derivatization reaction to a fluorescent product would also be of value for investigations of arabinosyl-isoguanine 2311
and of the respective monophosphate.
The optimized assay for the quantitation of F-ara-A in plasma was sensitive
enough to monitor the pharmacokinetics of the drug metabolite over a 72-h period
after the end of a 30-min i.v.-infusion of 60 mg fludarabine phosphate (Fig. 4).
Minimum plasma levels during this time were still 20 to 40 times above the limit of
quantitation of the assay. The triphasic F-ara-A elimination profile with a terminal
half-life of 33.5 + 5 h was in contrast to a previous report, in which the less
sensitive UV detection method was employed and a biphasic elimination profile
with a mean terminal half-life of 9.2 h was found under comparable treatment
conditions (e.g. 18-25 mg/m infusion over 30 min) [14]. At 50 mg/m2 fludarabine phosphate infused over 30 min, similar pharmacokinetic parameters were
reported, but the terminal half-lives varied from 7.8 h to over 24 h [13]. After an
i.v.-bolus of 80 mg/m, plasma elimination was triphasic with a terminal half-life
of 10.9 + 2.9 h ill]. Additional studies utilizing this fluorescence assay should
clarify the long-term elimination kinetics of F-ara-A.
The reaction conditions which provide the maximum yield of the fluorescent
F-ara-A derivative may also be applied to the mono- and triphosphates and F-Ade
(Table I). There are several potential applications of the derivatization reaction for
these compounds. First, it may permit the quantitation of very low levels of
F-ara-AMP in human plasma (or tissue), because the nucleotide is very rapidly
dephosphorylated
and is not measurable by UV detection only minutes after the
end of i.v.-infusions [11,13,14]. Second, it may be possible to detect F-adenine, a
potential metabolite of F-ara-A, which is highly toxic [32,33], and has been found
in animal tissue and urine after treatment with 40-400 mg/m* [3H]F-ara-A
[12,34]. Third, quantitation of very small intracellular F-ara-ATP concentrations
might be achieved by this method. Finally, it is likely that this procedure has
adequate sensitivity to facilitate quantitation of F-ara-AMP incorporated into
DNA. The feasibility of all these applications is being pursued.

105

Acknowledgements
The authors thank Dr. Abdallah Cherif for his help in the synthesis of bromoacetaldehyde and for valuable discussions concerning the chemical aspects of this
work. We are grateful to Richard Willson and Jean Nowlin of Shell Westhollow
Research Center, Houston TX, for their assistance in obtaining the NMR and
mass spectrometry data of the fluorescent F-ara-A derivative. We would also like
to express our gratitude to Walter J. Page1 for excellent editorial assistance in the
preparation of the manuscript. This work was supported in part by grant CA32839
from the National Cancer Institute, DHHS. A. Kemena received a grant from the
German Research Association (DFG).
References
1 Grever MR, Kopecky KJ, Coltman CA, et al. Fludarabine monophosphate: a potentially useful
agent in chronic lymphocytic leukemia. Nouv Rev Fr Hematol 1988;30:457-459.
2 Keating MJ, Kantarjian H, Talpaz M, et al. Fludarabine: A new agent with major activity against
chronic lymphocytic leukemia. Blood 1989;74:19-25.
3 Leiby JM, Snider KM, Kraut EH, Metz EN, Malspeis L, Grever MR. Phase II trial of 9-/?-Darabinofuranosyl-2-fluoroadenine
5-monophosphate in non-Hodgkins lymphoma: prospective comparison of response with deoxycytidine kinase activity. Cancer Res 1987;47:2719-2722.
4 Von Hoff DD, Dahlberg S, Hartstock RJ, Eyre HJ. Activity of fludarabine monophosphate in
patients with advanced mycosis fungoides: a Southwest Oncology Group study. J Natl Cancer Inst
1990;82:1353-1355.
5 Kantarjian HM, Alexanian R, Koller CA, Kurzrock R, Keating MJ. Fludarabine therapy in
macroglobulinemic lymphoma. Blood 1990;75:1928-1931.
6 Chun HG, Leyland-Jones BR, Caryk SM, Hoth DF. Central nervous system toxicity of fludarabine
phosphate. Cancer Treat Rep 1986;70:1225-1228.
7 Spriggs DR, Stopa E, Mayer RJ, Schoene W, Kufe DW. Fludarabine phosphate (NSC 312887)
infusions for the treatment of acute leukemia: phase I and neuropathological study. Cancer Res
1986;46:5953-5958.
8 Warrell Jr RP, Berman E. Phase I and II study of fludarabine phosphate in leukemia: therapeutic
efficacy with delayed central nervous system toxicity. J Clin Oncol 1986;4:74-79.
9 Gandhi V, Plunkett W. Modulation of arabinosylnucleoside metabolism by arabinosylnucleotides in
human leukemia cells. Cancer Res 1988;48:329-334.
10 Gandhi V, Nowak B, Keating MJ, Plunkett W. Modulation of arabinosylcytosine metabolism by
arabinosyl-2-fluoroadenine
in lymphocytes from patients with chronic lymphocytic leukemia: Implications for combination therapy. Blood 1989;74:2070-2075.
11 Malspeis L, Grever MR, Staubus AE, Young D. Pharmacokinetics of 2-F-ara-A (9$-Darabinofuranosyl-2-fluoroadenine)
in cancer patients during the phase I clinical investigation of
fludarabine phosphate. Sem Oncol 1990;17:18-32.
12 Noker PE, Duncan GF, El Dareer SM. Hill DL. Disposition of 9-p-D-arabinofuranosyl-2-fluoroadenine 5-phosphate in mice and dogs. Cancer Treat Rep 1983;67:445-456.
13 Danhauser L, Plunkett W, Keating M, Cabanillas F. 9-p-D-Arabinofuranosyl-2-fluoroadenine
5-monophosphate pharmacokinetics in plasma and tumor cells of patients with relapsed leukemia
and lymphoma. Cancer Chemother Pharmacol 1986;18:145-152.
14 Hersh MR, Kuhn JG, Phillips JL, Clark G, Ludden TM, Von Hoff DD. Pharmacokinetic study of
fludarabine phosphate (NSC 312887). Cancer Chemother Pharmacol 1986;17:277-280.
15 Brown RD, Manno JE. ESTRIP, a BASIC computer program for obtaining initial polyexponential
parameter estimates. J Pharmaceut Sci 1978;67:1687-1691.

106
16 Avigad G, Damle S. Fluorimetric assay of adenine and its derivatives. Anaiyt Biochem 1972;50:321323.
17 Perrett D. Determination of adenosine ribo- and deoxyribonucleotides as their 1-Nh-etheno derivatives by reversed-phase ion-pair high-performance liquid chromatography. J Chromatogr 1987;386:
289-296.
18 Sonoki S, Tanaka Y, Hisamatsu S, Kobayashi T. High-performance liquid chromatographic analysis
of fluorescent derivatives of adenine and adenosine and its nucleotides. Optimization of derivatization with chloroacetaldehyde and chromatographic procedures. J Chromatogr 1989;475:311-319.
19 Mills GC, Schmalstieg FC, Trimmer KB, Goldman AS, Goldblum RM. Purine metabolism in
adenosine deaminase deficiency. Proc Nat1 Acad Sci USA 1976;73:2867-2871.
20 Slegel P, Kitagawa H, Maguire MH. Determination of adenosine in fetal perfusates of human
placental cotyledons using fluorescence derivatization and reversed-phase high-performance liquid
chromatography. Analyt Biochem 1988;171:124-134.
21 Barrio JR, Secrist III JA, Leonard NJ. Fluorescent adenosine and cytidine derivatives. Biochem
Biophys Res Commun 1972;46:597-604.
22 Kochetkov NK, Shibaev VW, Kost AA. New reaction of adenine and cytosine derivatives, potentially useful for nucleic acids modification. Tetrahedron Lett 1971;22:1993-1996.
23 Bhat GA, Schram KH, Townsend LB. The synthesis of certain fluorescent imidazo[l,2-c]pyrrolo[3,2elpyrimidine nucleoside derivatives related to e-adenosine. J Carbohydrates Nucleosides Nucleotides 1980;7:333-345.
24 Kovai J, Turanek J, ZabrSova H, Kahle V. Liquid chromatographic separations of fluorescent
1,N6-ethenoadenosine derivatives of AMP, ATP and NAD. J Chromatogr 1986;360:185-191.
2.5 Yoshioka M, Nishidate K, Iizuka H, et al. Sensitive fluorimetry of adenine-containing compounds
with high-performance liquid chromatography. J Chromatogr 1984;309:63-71.
26 Secrist III JA, Barrio JR, Leonard NJ, Weber G. Fluorescent modification of adenosine-containing
coenzymes. Biological activities and spectroscopic properties. Biochemistry 1972;11:3499-3506.
27 Wojcik WJ, Neff NH. Adenosine measurement by a rapid HPLC-fluorimetric method: Induced
changes of adenosine content in regions of rat brain. J Neurochem 1982;39:280-282.
28 Jacobson MK, Hemingway LM, Farrell TA, Jones CE. Sensitive and selective assay for adenosine
using high-pressure liquid chromatography with tluorometry. Am J Physiol 1983;245:H887-890.
29 Brockman RW, Schabel Jr FM, Montgomery JA. Biologic activity of 9-p-D-arabinofuranosyl-2-fluoroadenine, a metabolically stable analog of 9-P-D-arabinofuranosyladenine.
Biochem Pharmacol
1977;26:2193-2196.
30 Bennett LL Jr, Chang C-H, Rose L, et al. Metabolism and metabolic effects of halopurine
nucleosides in tumor cells in culture. Nucleosides Nucleotides 1985;4:107-116.
31 Elion GB, Rideout JL, deMiranda P, Collins P, Bauer DJ. Biological activities of some purine
arabinosides. Ann N Y Acad Sci 1975;255:468-480.
32 Skipper HE, Montgomery JA, Thomson JR, Schabel Jr FM. Structure-activity relationships and
cross-resistance observed on evaluation of a series of purine analogs against experimental neoplasms. Cancer Res 1959;19:425-437.
33 Bennett Jr LL, Vail MH, Chumley S, Montgomery JA. Activity of adenosine analogs against a cell
culture line resistant to 2fluoroadenine. Biochem Pharmacol 1966;15:1719-1728.
34 Struck RF, Shortnacy AT, Kirk MC, et al. Identification of metabolites of 9-P-D-arabinofuranosyl2-fluoroadenine, an antitumor and antiviral agent. Biochem Pharmacol 1982;31:1975-1978.