- Email: [email protected]
High-performance liquid chromatographic determination of platinum (II) in plasma ultrafiltrate and urine: comparison with a flameless atomic absorption spectrometric method
65
Clinica Chimica Acfa, 136 (1984) 65-74 Elsevier
CCA 02726
High-performance liquid chromatographic d~ter~nation of platinum (II) in plasma ultrafiltrate and urine: comparison with a flameless atomic absorption spectrometric method Olaf H. Drummer
*, Alex Proudfoot,
Laurie Howes and William J. Louis
Uniuersity of Melbourne, Clinical Pharmacology and Therapeutics Unit, Austin Hospital, Heidelberg, 3084, (A ~r~alia~
Vie.,
(Received June 3rd; revision August 19th, 1983) Key worak
cis - Platinum; High performance liquid chromatography; Plasma ultrafiltrate;
Urine
Summary
A technically simple, rapid and sensitive high performance liquid chromatographic assay for cis-dic~orodiam~neplatinum (II) in human plasma ultrafiltrate and urine is described. The drug was chelated by exchange with diethyldithiocarbamate and extracted into chloroform. Nickel (II) was used as an internal standard which allows correction for the matrix effects observed with previous chromatographic and spectrometric methods. Chromatography was performed on a p-Bondapak CN column and the eluent measured spectrophotometrically at 254 nm. Precision and reproducibility were both excellent and the detection limit was less than 50 ng/ml using only 1 ml of biological fluid.
Introduction Cis-Dichlorodiammineplatinum (II) (DDP) is currently a widely used drug for the treatment of a variety of malignant solid tumours [1,2]. Measurement of this drug in biological fluids has been restricted to radioactive studies [3], flameless atomic absorption spectrometric [4-g] and X-ray fluorescence [lo] methods. Although adequate these methods require equipment out of reach of most clinical pharmacology laboratories. In addition, some recent reports have suggested that these methods are subject to matrix effects resulting in variable recoveries [ll]. * To whom correspondence and reprint requests (Air Mail) should be addressed. Non-standard abbreuiutions: DDP, cis-dichloro-diammine platinum (II); DDTC, diethylditbiocarbamate. ~9-8981/84/$03.~
0 1984 Elsevier Science Publishers B.V.
66
High-performance liquid chromatography (HPLC), a technique now available in most laboratories, lends itself to measuring DDP in biological fluids. Hincal et al [12] published an HPLC method for DDP in aqueous solutions utilising its weak native UV absorption. The limit of detection was about 20 pg/ml. Bannister et al [ll] measured DDP in urine by chelating DDP with diethyldithiocarbamate followed by chromatography. However, the method was cumbersome and no comparison was made with a standard reference atomic absorption method nor was an internal standard used to improve precision. No published method using HPLC has yet been reported to measure DDP in plasma ultra-filtrate. We report here a reliable and sensitive HPLC procedure for measuring DDP in urine and plasma ultrafiltrate from patients receiving DDP therapy. The assay is based on the formation of a diethyldithiocarbamate chelate and the use of nickel (II) as an internal standard. Plasma levels of DDP in patients receiving DDP therapy have also been measured by this HPLC procedure and the levels compared with that obtained from a flameless atomic absorption method [8]. Materials and methods Apparatus High-performance liquid chromatography was performed on a Waters Associates (Melbourne, Australia) Model 6000-A solvent delivery system, Model U6K universal injector, and a Model 440 absorbance detector operated at 254 nm. The stainless steel column (30 cm x 3.9 mm i.d.) packed with a p-Bondapak CN-bonded silica phase was obtained from Waters Associates. Flameless atomic absorption spectroscopy was performed on a Perkin Elmer graphite furnace (Melbourne, Australia) Model 460 spectrometer with a HGA 500 programmer. Ultrafiltration of plasma was performed by filtering into Centriflo tubes (part No. CTI) through Amicon Centriflo Membranes (Danvers, MA, USA; part No. CF25A) with an exclusion limit of 25 000 amu. Chemicals Cis-Dichlorodiammineplatinum (II) and trans-dichlorodiammineplatinum (II) were obtained from Bristol Myers. Chloroplatinic (IV) acid was purchased from Hopkin and Williams (Chadwell Heath, Essex, UK). Sodium diethyldithiocarbamate (DDTC) was purchased from Fluka (Basle, Switzerland) and Nickel (II) chloride from Sigma (St. Louis, MO, USA). Chloroform was Merck analytical reagent grade (Merck, Darmstadt, FRG), heptane and propan-2-01 were Waters Associates liquid chromatography grade. All other chemicals were laboratory reagent grade and were used as obtained. Collection and treatment
of biological fluids
Plasma: Blood was collected in lithium heparin tubes from patients undergoing treatment for ovarian cancer, genitourinary cancer and resistant lymphoma at 0 and 60 min following a 15 min intravenous infusion of DDP at a dose of 120 mg/m2 and
67
stored at 4°C in an ice bath. Plasma was separated by centrifugation at 1000 X g for 10 min at 4°C within 30 min of collection and immediately subjected to ultra-filtration prior to storage at - 20°C. Urine: Urine specimens were collected from the same patients at 60 min post DDP infusion and stored frozen until use. Immediately prior to an assay thawed urine was adjusted to pH 7.4 and filtered through an Argyle 5 pm disposable filter (St. Louis, MO, USA). Liquid chromatographic method Urine or plasma ultrafiltrate (1 ml) was added to glass extraction tubes. Standards were prepared by spiking DDP to 1 ml of ultrafiltrate or urine collected from volunteers not receiving any medication. Nickel chloride (100 ~1 of 1 g/l solution) and 200 pl of a 2% (w/v) solution of DDTC were added in turn to each sample. Tubes were capped and incubated at 37°C for 1 h. Chloroform (1 ml) was then added to each tube and the entire contents extracted by shaking for 10 min on a mechanical shaker. After a brief centrifugation (1000 X g for 5 mm), the lower chloroform layer was transferred to a fresh tube and evaporated to dryness under a stream of nitrogen. The residue was reconstituted with a volume of mobile phase (usually 500 ~1). It was however, possible to omit this last evaporation step and inject the chloroform layer if only lo-$ volumes were injected. Chromatography Ten to fifty microliters of the extract were injected and, depending on the sensitivity required, chromatographic conditions on the I_c-Bondapak CN column were based on an isocratic mobile phase, heptane/propan-2-01 (90 : 10, v/v). Flow rate was 4 ml/mm. Flameless atomic absorption spectrometric method Urine or plasma ultrafiltrate were assayed essentially as reported by SmeyersVerbeke et al [8]. The sample (10 ~1) was applied to the graphite furnace by an automatic sample injector. The drying, ashing and atomisation conditions are shown in Table I. Argon purge gas was turned off during the atomisation stage. Integration of the signal started 1 s before the atomisation stage and lasted throughout the 10 s TABLE I Furnace conditions for determination of Pt in biological fluids
Sample volume Drying Thermal decomposition Atomisation
Plasma ultrafiltrate
Urine
log1 150°C 1. 600°C 2.14OO”C 2700°C
10 pl 150°C (40)/120 s 1400°C (S)/lO s
8 (40) b/20 s c (lo)/75 s (S)/lO s (O)/lO s
a Final temperature; b ramp “C/s, ’ total time in s for step.
2700°C (O)/lO s
68
atomisation. Hydrogen background correction was used. 265.9 nm was monitored using an element lamp current Under these conditions no spattering of the sample and ashing stages and responses to a known amount of plasma ultrafiltrate or urine were equivalent.
The platinum absorption at of 10 mA. occurred during the drying DDP added to either water,
Quantitation DDP standards in urine and ultrafiltered plasma were prepared by spiking drug-free fluids with known amounts of pure DDP in aqueous solutions. For HPLC the ratio of the Pt peak to the Ni peak was calculated from each chromatogram. Quantitation was achieved by relating the peak height ratios of unknown samples to those obtained from standards. A control sample of urine or ultrafiltered plasma was routinely included in each assay to maintain a check on between-run variability. Kinetics of chelation The time course of the chelation of DDP with DDTC was assessed by incubating with DDTC for various times at 37°C with blank filtered urine (adjusted to pH 7.4) which had previously been incubated with 25 pg/ml DDP and 100 pg/ml nickel (II) chloride for 24 h at 37°C. The reaction times studied were 30 min, 60 min, 120 min and 24 h. At each specified time duplicate l-ml samples of urine were extracted with 1 ml chloroform. The chloroform layers were separated from the aqueous material by centrifugation and stored at 4°C in a stoppered glass tube prior to chromatography. This experiment was performed three times. Statistical analysis Linear regression analysis performed using the Deming
to analyse the HPLC method [14].
and flameless
AA methods
was
Results Representative liquid chromatograms for plasma ultrafiltrate are shown Fig. 1 and for filtered urine in Fig. 2. The chromatographic profile was identical for aqueous standards, plasma ultrafiltrate and urine. Under these conditions clean separation of all peaks occurred with symmetrical peaks. The retention volume for the nickel (II) chelate internal standard was 9.5 ml and for the platinum chelate 15.8 ml. The only chromatographic peak originating from the chelating agent occurred at a retention volume of 6 ml. This component did not interfere with the assay as it was completely separated from the internal standard. Time course of chelation The chelating effect of DDTC was rapid at 37’C, for both DDP and the internal standard nickel in both plasma ultrafiltrate and urine. The time course of chelation in urine is shown in Fig. 3. The value obtained at 24 h has been assigned as 100% binding. As can be seen the curves for both DDP and nickel are almost superim-
69
I
ATTENUATION 0.05
0.01
0.05
0.01
U-
--
3
3
I
ATTENUATION
0.01 uz
0.01 MM
0.20
0.20
4
3
i(l
6 I
2
-i
0.
A
11”“l 6
4
2
1”“‘J 6 4
0 TIME
Fig. 1
(
min
1
2
0
I 6
fi 1 1 1 1 I 4
2
l”“‘1
0
TIME
6
tmin
4
2
0
)
Fig. 2
Fig. 1. HPLC chromatogram of (A) a blank plasma ultrafiltrate extract and (B) a plasma ultrafiltrate extract containing 5 ag/ml cis-platinum (II). The identity of the peaks are (1) injection event, (2) (3) and (4) solvent front artefacts and excess diethyldithiocarbamate, (5) nickel chelate (internal standard) and (6) platinum chelate. Attenuation is increased to 0.01 aufs after elution of the internal standard to show the region of chromatogram corresponding to elution of platinum chelate in the blank. Fig. 2. HPLC cis-platinum nickel-chelate elution of the chelate in the
chromatogram of (A) a blank urine extract and (B) an urine extract containing 5 gg/ml (II). The identity of the peaks are (1) injection event, (2) solvent front artefact, (3) (internal standard) and (4) cis-platinum-chelate. Attenuation is increased to 0.01 aufs after internal standard to show the region of chromatogram corresponding to elution of platinum blank.
posable with 71% and 73% of maximum chelation occurring at 30 min. By 60 min these figures were 78% and 93% respectively. A 60-min reaction time at 37°C was routinely used as this gave almost complete chelation without an unnecessarily long waiting period. Reaction curves for plasma ultrafiltrate were qualitatively similar to those for urine.
70
IOOr
0’
80
-
F 4 A
60
-
.
Nickel
l
I” U
40
-
20
-
z
0
0
30
60
90
TIME
120 (min
Fig. 3. Curve to show time course of formation chelates using the described method.
Analytical
variables for HPLC
Platinum
(IT) (XI)
‘24w
1 of (a) nickel and (0) platinum
diethyldithiocarbamate
assa_y
Precision: Within-run precision of the assay was determined by processing l-ml samples of pooled plasma ultrafiltrate and urine through the procedure during a single run. The coefficients of variation (CV) were found to be 1.4% for ultrafiltrate and 5.8% for urine (Table II). Day-to-day variability, determined over a period of 2 months, was 8.1% for ultrafiltrate and 8.8% for urine (Table II). Analytical recovery Analytical recovery of known amounts of DDP added to blank plasma ultrafiltrate and urine was greater than 95%. This was determined by measuring the aqueous and chloroform layers of extracted samples by atomic absorption spectrometry tuned for measuring platinum at 265.9 nm. Amounts of platinum corresponding to less than 5% of that added were seen in the aqueous layer. The remaining fraction was observed in the organic layer suggesting complete extraction of the platinum
TABLE
II
Assay characteristics
for the liquid-chromatography Urine
Precision intra-day inter-day
5.8% 8.8%
Recovery Detection
> 95% limit
50 n&ml
of DDP-DDTC
chelate Plasma ultrafiltrate
1.4% 8.1% > 95% 50 ng/ml
71
chelate. Experiments were also conducted which demonstrated that no platinum appeared in the chloroform layer when chelating agent was absent. Linearity and detection limit Calibration curves were linear for the HPLC assay for both lower and upper concentration ranges and passed through the origin when drug-free plasma ultrafiltrate or urine fortified with DDP in the concentration range 50 ngfml to 100 pg/ml were processed. Separate calibration curves were usually constructed in the range 0.25-5 @g/ml and for the levels higher 5 mg/ml in the range 2.5-50 pg,/ml to avoid the need to change the attenuation during a chromatogram. The amount of internal standard used for the lower range was 20 pg instead of the usual 100 pg. This range represented the plasma ultrafiltrate and urine levels following a 120 mg/m* IV infusion of DDP. The detection limit for DDP was about 50 ng/ml in both urine and ultrafiltrate. This represented a peak about three times baseline noise at the highest sensitivity on the UV detector (0.005 aufs). Specificity Assay of urine and plasma ultrafiltrate spiked with chloroplatinic (IV) acid at concentrations up to 50 pg/ml did not give rise to any peaks in the chromatogram. Thus the assay appears to be specific to platinum in the II valency state. This was expected, since the Pt (IV)-DDTC chelate is likely to have the stoichiometric formula Pt (DDTC),, which would be a charged species and not extractable into chloroform. The assay did not discriminate between the cis and trans forms, nor the potential aquo metabolites which form in low chloride solutions [ll]. Both pure trans-dichlorodiammineplatinum (II) and the mono and diaquo forms of DDP formed by incubating DDP in low chloride solution gave peaks with identical retention volumes and peak heights to DDP. Interference Although the possibility existed that other metals may interfere with either nickel or platinum chelate peaks, no other metals were found with retention volumes close enough to exhibit interference. The metals tested included zinc (II), cadmium (II), cobalt (II), calcium (II), iron (II), iron (III), gold (I), silver (I), lead (II), cerium (II), uranium (II), copper (II), manganese (II), palladium (II), and mercury (II), The drugs adriamycin, bleomycin, cyclophospharnide, vinblastine, vincristine, metoclopramide and prochlorperazine, which are sometimes used in combination with DDP, did not interfere with this assay. Comparison with atomic absorption spectrometry In patients receiving DDP therapy at a dose of 120 mg/m2 there was a close correlation between results obtained with the standard flameless atomic absorption spectrometric [8] and the HPLC procedure. The Deming method of least-squares regression analysis was designed to handle data where the independent variable (x) as well as the dependent variable (y) is measured with some error 1141.Using this
12
B.
A. E .
::
URINE
FILTRATES
ULTRA
E \
60-
60
s z
50
-
40
-
50
0 lP
a 0 v) m a
g v) m a
30 20
u 5
I 0
0 I-
.a
IO
20
30
HPLC
40
50 (p/ml
60
)
2
0 IO
20 HPLC
30
40
50 (yg/ml
60
1
Fig. 4. Comparison of cis-dichlorodiammineplatinum (11) concentrations obtained by a flameless AA method and by the present HPLC method for (a) plasma ultrafiltrate, n = 19 and (b) urine, n = 15. The lines represent equality by the two methods.
statistical test it was found that the slopes for both ultrafiltered plasma and urine data were not significant from unity (slope ultrafiltrate = 0.9088, SD = 0.089; slope urine = 1.262, SD = 0.313). The correlation coefficients for ultrafiltrate and urine were 0.9373 and 0.7322, respectively. The intercepts for ultrafiltrate (y-intercept = 0.58, SD = 0.21) and urine (y-intercept = - 2.367, SD = 3.294) were also not significantly different from the origin.
Discussion The measurement of the anti-neoplastic drug DDP has been largely restricted to atomic absorption spectrometric methods [4-91, owing to the poor UV-visible absorption properties of this co-ordination complex. In this paper the bidentate ligand DDTC has been used to chelate DDP using an adaption of Amore’s method [15] to exchange the chloro and ammine ligands. Our procedure, in contrast to that described by Bannister and co-workers [ll], does not require the presence of sodium nitrate or strongly basic conditions nor does it involve a large incubation volume. Despite these changes the formation of the DDTC chelate of platinum (II) is essentially complete after a 60-min incubation at 37°C. The detection limit for DDP using the method reported here was about 50 ng/ml in plasma ultrafiltrate and urine. This is less sensitive than existing flameless atomic absorption spectrometric methods [5,8] which quote detection limits of around 20 ng/ml, and for a previous HPLC procedure (25 ng/ml) [ll]. However, with this latter procedure 9 ml of urine were required to achieve this sensitivity. Other published HPLC procedures utilising
13
the weak native UV absorption of DDP were only able to measure levels down to about 20 pg/ml in aqueous solutions [12]. Therapeutic plasma ultrafiltrate or urine levels were always much higher than 50 ng/ml and usually in the range 0.25 to 50 pg/ml. Higher sensitivity could be obtained by using less mobile phase (100 ~1) to reconstitute the residue after evaporation of the chloroform. Reproducibility studies in both plasma ultrafiltrate and urine suggested that recoveries were sometimes variable depending on the clarity and protein content of the fluid. Using this procedure there was an occasional variability in the height of the platinum peak with some samples similar to that reported by Bannister et al [ll]. The use of an internal standard (nickel) was able to compensate for variability in the extraction and chromatography steps. Although the use of nickel as internal standard may not compensate for variability in the complexity of Pt (II), the excellent precision obtained both within and inter assay suggest that this step may not be the major source of variability encountered. The chromatograms obtained from plasma ultrafiltrate and filtered buffered urine confirmed the presence of an early peak immediately after the solvent front due to some unreacted chelating agent., This did not interfere with either the nickel- or platinum-DDTC peaks and allowed flow rates of up to 4 ml/mm. Under these conditions the retention times of nickel and platinum were 2.4 min and 3.9 min, respectively. Thus, chromatograms took less than 5 min to develop, enabling 12 samples to be run through the HPLC in 1 h. This method also lends itself to automatic sample processing for large batches. No interference was noted with the platinum peak in any patient samples studied nor with a number of metals likely to form complexes with DDTC or drugs co-administered with DDP. A further feature of the method was the ability to measure other metals beside platinum in biological fluids without the need of atomic absorption spectrometry. Examples include nickel and a number of other transition metals with an oxidation state of II such as zinc and mercury. Validity of this HPLC method was shown by a good correlation for both plasma ultrafiltrate and urine with a reference flameless atomic absorption spectrometric method [8]. The experiments confirmed the precision of the HPLC assay and suggested that both procedures are indeed measuring the same platinum species. Like the other reported procedures this method will not distinguish between DDP and its aquo metabolites. There is some evidence in the literature [13] that aquo metabolites of DDP might form in cells as the highly labile chloro groups are successively displaced by water molecules. In our own laboratories stability studies in which DDP was incubated in a low chloride medium confirmed that these aquo species are formed in vitro and are measured by the HPLC procedure. Although the trans isomer of DDP is also measured by the HPLC procedure, there is no evidence in the literature that this compound occurs as a metabolite of DDP. In conclusion, a rapid, sensitive and clinically useful HPLC method has been developed for measuring unbound DDP from human plasma ultrafiltrate and urine samples. The method includes an internal standard for added precision which largely circumvents matrix effects observed with a previously described HPLC method and
74
existing atomic absorption procedures. The procedure has been validated against a standard atomic absorption method for pIatinum levels in the range 50 ng/ml to 50 pg/ml in plasma ultrafiltrate and urine samples from patients receiving DDP therapy. The method requires minimal sample preparation and lends itself to automation for processing of large numbers of samples. Acknowledgements This work was supported by the National Wealth and Medical Research Council of Australia. The helpful advice of Mr. Brian Stevens from the Royal Melbourne Institute of Technology and the use of his atomic absorption spectrophotometer was greatly appreciated. References 1 Einhorn LH, Donohue JP. Cis-D~aminned~chloroplatinum, vinblastine, and bfeomycin combination chemotherapy in disseminated testicular cancer. Ann intern Med 1977; 86: 8033812. 2 Bush H, Thatcher N, Barnard R. Chemotherapy in the management of invasive bladder cancer. Cancer Chemother Pharmacol 1979; 3: 87-96. 3 De Conti RC, Toftness BR, Lange RC, Creasey WA. Clinical and pharmacological studies with cis-diamminedichloroplatinum (II). Cancer Res 1973; 33: 1310-1315. 4 Pera MF Jr, Harder HC. Analysis for platinum in biological material by flameless atomic absorption spectrometry. CIin Chem 1977; 23: 1245-1249. 5 Bannister SJ, Chang Y, Sternson LA, Repta AJ. Atomic absorption spectrometry of free circulating platinum species in plasma derived from cis-dichlorodiammineplatinum (II). Clin Chem 1978; 24: 877-880. 6 Litterst CL, Gram TE, Dedrick RL. Leroy AF, Guarino AM. Distribution and disposition of platinum following intravenous administration of cis-diamminedichloroplatinum (II) (NSC 119875) to dogs. Cancer Res 1976; 36: 2340-2344. 7 Hull DA, Muhammad N, Lanese JG, Reich SD, Finkelstein IT, Fandrich S. Determination of platinum in serum and ultrafiltrate by flameless atomic absorption spectrophotometry. J Pharm Sci 1981; 70: 500-502. 8 Smeyers-Verbeke J, Detaevemicr MR. Denis L, Massant DL. The determination of platinum in biological fluids by means of graphite furnace atomic absorption spectrometry. CIin Chim Acta 1981; 113: 329-333. 9 Preisner D, Sternson LA, Repta AJ. Analysis of total platinum in tissue samples by flameless atomic absorption spectrophotometry. Elimination of the need for sample digestion. Anal Lett 1981: 14: 1255-1268. 10 Bannister SJ, Stemson LA, Repta AJ, James GW. Measurement of free-circulating cis-dichlorodiam~neplatinum (II) in plasma. Clin Chem 1977; 23: 2258-2262. 11 Bannister SJ, Sternson LA, Repta AJ. Urine analysis of platinum species derived from cis-dichlordiam~nepIat~num (II) by high performance liquid ~bromatography following derivatization with sodium diethyIdithi~~b~ate. J Chromatogr 1979; 173: 332-342. 12 Hincal AA, Long DF, Repta AJ. Cis-platin stability in aqueous parenteral vehicles. J Parenteral Drug Assoc 1979; 33: 107-116. 13 Repta AJ, Long DF. Reactions of cisplatin with human plasma and plasma fractions. In: Prestayko AW, Crook ST, Carter SK, eds. Cisplatin current status and new developments. New York, NY: Academic Press, 1980: 285-304. 14 Combleet PJ, Gochman N. Incorrect least-squares regression coefficients in method-comparison analysis. Clin Chem 1979; 25: 432-438. 15 Amore F. Determination of calcium, lead, thallium and nickel in blood by atomic absorption spectrometry. Anal Chem 1974; 46: 1597-1599.
Clinica Chimica Acfa, 136 (1984) 65-74 Elsevier
CCA 02726
High-performance liquid chromatographic d~ter~nation of platinum (II) in plasma ultrafiltrate and urine: comparison with a flameless atomic absorption spectrometric method Olaf H. Drummer
*, Alex Proudfoot,
Laurie Howes and William J. Louis
Uniuersity of Melbourne, Clinical Pharmacology and Therapeutics Unit, Austin Hospital, Heidelberg, 3084, (A ~r~alia~
Vie.,
(Received June 3rd; revision August 19th, 1983) Key worak
cis - Platinum; High performance liquid chromatography; Plasma ultrafiltrate;
Urine
Summary
A technically simple, rapid and sensitive high performance liquid chromatographic assay for cis-dic~orodiam~neplatinum (II) in human plasma ultrafiltrate and urine is described. The drug was chelated by exchange with diethyldithiocarbamate and extracted into chloroform. Nickel (II) was used as an internal standard which allows correction for the matrix effects observed with previous chromatographic and spectrometric methods. Chromatography was performed on a p-Bondapak CN column and the eluent measured spectrophotometrically at 254 nm. Precision and reproducibility were both excellent and the detection limit was less than 50 ng/ml using only 1 ml of biological fluid.
Introduction Cis-Dichlorodiammineplatinum (II) (DDP) is currently a widely used drug for the treatment of a variety of malignant solid tumours [1,2]. Measurement of this drug in biological fluids has been restricted to radioactive studies [3], flameless atomic absorption spectrometric [4-g] and X-ray fluorescence [lo] methods. Although adequate these methods require equipment out of reach of most clinical pharmacology laboratories. In addition, some recent reports have suggested that these methods are subject to matrix effects resulting in variable recoveries [ll]. * To whom correspondence and reprint requests (Air Mail) should be addressed. Non-standard abbreuiutions: DDP, cis-dichloro-diammine platinum (II); DDTC, diethylditbiocarbamate. ~9-8981/84/$03.~
0 1984 Elsevier Science Publishers B.V.
66
High-performance liquid chromatography (HPLC), a technique now available in most laboratories, lends itself to measuring DDP in biological fluids. Hincal et al [12] published an HPLC method for DDP in aqueous solutions utilising its weak native UV absorption. The limit of detection was about 20 pg/ml. Bannister et al [ll] measured DDP in urine by chelating DDP with diethyldithiocarbamate followed by chromatography. However, the method was cumbersome and no comparison was made with a standard reference atomic absorption method nor was an internal standard used to improve precision. No published method using HPLC has yet been reported to measure DDP in plasma ultra-filtrate. We report here a reliable and sensitive HPLC procedure for measuring DDP in urine and plasma ultrafiltrate from patients receiving DDP therapy. The assay is based on the formation of a diethyldithiocarbamate chelate and the use of nickel (II) as an internal standard. Plasma levels of DDP in patients receiving DDP therapy have also been measured by this HPLC procedure and the levels compared with that obtained from a flameless atomic absorption method [8]. Materials and methods Apparatus High-performance liquid chromatography was performed on a Waters Associates (Melbourne, Australia) Model 6000-A solvent delivery system, Model U6K universal injector, and a Model 440 absorbance detector operated at 254 nm. The stainless steel column (30 cm x 3.9 mm i.d.) packed with a p-Bondapak CN-bonded silica phase was obtained from Waters Associates. Flameless atomic absorption spectroscopy was performed on a Perkin Elmer graphite furnace (Melbourne, Australia) Model 460 spectrometer with a HGA 500 programmer. Ultrafiltration of plasma was performed by filtering into Centriflo tubes (part No. CTI) through Amicon Centriflo Membranes (Danvers, MA, USA; part No. CF25A) with an exclusion limit of 25 000 amu. Chemicals Cis-Dichlorodiammineplatinum (II) and trans-dichlorodiammineplatinum (II) were obtained from Bristol Myers. Chloroplatinic (IV) acid was purchased from Hopkin and Williams (Chadwell Heath, Essex, UK). Sodium diethyldithiocarbamate (DDTC) was purchased from Fluka (Basle, Switzerland) and Nickel (II) chloride from Sigma (St. Louis, MO, USA). Chloroform was Merck analytical reagent grade (Merck, Darmstadt, FRG), heptane and propan-2-01 were Waters Associates liquid chromatography grade. All other chemicals were laboratory reagent grade and were used as obtained. Collection and treatment
of biological fluids
Plasma: Blood was collected in lithium heparin tubes from patients undergoing treatment for ovarian cancer, genitourinary cancer and resistant lymphoma at 0 and 60 min following a 15 min intravenous infusion of DDP at a dose of 120 mg/m2 and
67
stored at 4°C in an ice bath. Plasma was separated by centrifugation at 1000 X g for 10 min at 4°C within 30 min of collection and immediately subjected to ultra-filtration prior to storage at - 20°C. Urine: Urine specimens were collected from the same patients at 60 min post DDP infusion and stored frozen until use. Immediately prior to an assay thawed urine was adjusted to pH 7.4 and filtered through an Argyle 5 pm disposable filter (St. Louis, MO, USA). Liquid chromatographic method Urine or plasma ultrafiltrate (1 ml) was added to glass extraction tubes. Standards were prepared by spiking DDP to 1 ml of ultrafiltrate or urine collected from volunteers not receiving any medication. Nickel chloride (100 ~1 of 1 g/l solution) and 200 pl of a 2% (w/v) solution of DDTC were added in turn to each sample. Tubes were capped and incubated at 37°C for 1 h. Chloroform (1 ml) was then added to each tube and the entire contents extracted by shaking for 10 min on a mechanical shaker. After a brief centrifugation (1000 X g for 5 mm), the lower chloroform layer was transferred to a fresh tube and evaporated to dryness under a stream of nitrogen. The residue was reconstituted with a volume of mobile phase (usually 500 ~1). It was however, possible to omit this last evaporation step and inject the chloroform layer if only lo-$ volumes were injected. Chromatography Ten to fifty microliters of the extract were injected and, depending on the sensitivity required, chromatographic conditions on the I_c-Bondapak CN column were based on an isocratic mobile phase, heptane/propan-2-01 (90 : 10, v/v). Flow rate was 4 ml/mm. Flameless atomic absorption spectrometric method Urine or plasma ultrafiltrate were assayed essentially as reported by SmeyersVerbeke et al [8]. The sample (10 ~1) was applied to the graphite furnace by an automatic sample injector. The drying, ashing and atomisation conditions are shown in Table I. Argon purge gas was turned off during the atomisation stage. Integration of the signal started 1 s before the atomisation stage and lasted throughout the 10 s TABLE I Furnace conditions for determination of Pt in biological fluids
Sample volume Drying Thermal decomposition Atomisation
Plasma ultrafiltrate
Urine
log1 150°C 1. 600°C 2.14OO”C 2700°C
10 pl 150°C (40)/120 s 1400°C (S)/lO s
8 (40) b/20 s c (lo)/75 s (S)/lO s (O)/lO s
a Final temperature; b ramp “C/s, ’ total time in s for step.
2700°C (O)/lO s
68
atomisation. Hydrogen background correction was used. 265.9 nm was monitored using an element lamp current Under these conditions no spattering of the sample and ashing stages and responses to a known amount of plasma ultrafiltrate or urine were equivalent.
The platinum absorption at of 10 mA. occurred during the drying DDP added to either water,
Quantitation DDP standards in urine and ultrafiltered plasma were prepared by spiking drug-free fluids with known amounts of pure DDP in aqueous solutions. For HPLC the ratio of the Pt peak to the Ni peak was calculated from each chromatogram. Quantitation was achieved by relating the peak height ratios of unknown samples to those obtained from standards. A control sample of urine or ultrafiltered plasma was routinely included in each assay to maintain a check on between-run variability. Kinetics of chelation The time course of the chelation of DDP with DDTC was assessed by incubating with DDTC for various times at 37°C with blank filtered urine (adjusted to pH 7.4) which had previously been incubated with 25 pg/ml DDP and 100 pg/ml nickel (II) chloride for 24 h at 37°C. The reaction times studied were 30 min, 60 min, 120 min and 24 h. At each specified time duplicate l-ml samples of urine were extracted with 1 ml chloroform. The chloroform layers were separated from the aqueous material by centrifugation and stored at 4°C in a stoppered glass tube prior to chromatography. This experiment was performed three times. Statistical analysis Linear regression analysis performed using the Deming
to analyse the HPLC method [14].
and flameless
AA methods
was
Results Representative liquid chromatograms for plasma ultrafiltrate are shown Fig. 1 and for filtered urine in Fig. 2. The chromatographic profile was identical for aqueous standards, plasma ultrafiltrate and urine. Under these conditions clean separation of all peaks occurred with symmetrical peaks. The retention volume for the nickel (II) chelate internal standard was 9.5 ml and for the platinum chelate 15.8 ml. The only chromatographic peak originating from the chelating agent occurred at a retention volume of 6 ml. This component did not interfere with the assay as it was completely separated from the internal standard. Time course of chelation The chelating effect of DDTC was rapid at 37’C, for both DDP and the internal standard nickel in both plasma ultrafiltrate and urine. The time course of chelation in urine is shown in Fig. 3. The value obtained at 24 h has been assigned as 100% binding. As can be seen the curves for both DDP and nickel are almost superim-
69
I
ATTENUATION 0.05
0.01
0.05
0.01
U-
--
3
3
I
ATTENUATION
0.01 uz
0.01 MM
0.20
0.20
4
3
i(l
6 I
2
-i
0.
A
11”“l 6
4
2
1”“‘J 6 4
0 TIME
Fig. 1
(
min
1
2
0
I 6
fi 1 1 1 1 I 4
2
l”“‘1
0
TIME
6
tmin
4
2
0
)
Fig. 2
Fig. 1. HPLC chromatogram of (A) a blank plasma ultrafiltrate extract and (B) a plasma ultrafiltrate extract containing 5 ag/ml cis-platinum (II). The identity of the peaks are (1) injection event, (2) (3) and (4) solvent front artefacts and excess diethyldithiocarbamate, (5) nickel chelate (internal standard) and (6) platinum chelate. Attenuation is increased to 0.01 aufs after elution of the internal standard to show the region of chromatogram corresponding to elution of platinum chelate in the blank. Fig. 2. HPLC cis-platinum nickel-chelate elution of the chelate in the
chromatogram of (A) a blank urine extract and (B) an urine extract containing 5 gg/ml (II). The identity of the peaks are (1) injection event, (2) solvent front artefact, (3) (internal standard) and (4) cis-platinum-chelate. Attenuation is increased to 0.01 aufs after internal standard to show the region of chromatogram corresponding to elution of platinum blank.
posable with 71% and 73% of maximum chelation occurring at 30 min. By 60 min these figures were 78% and 93% respectively. A 60-min reaction time at 37°C was routinely used as this gave almost complete chelation without an unnecessarily long waiting period. Reaction curves for plasma ultrafiltrate were qualitatively similar to those for urine.
70
IOOr
0’
80
-
F 4 A
60
-
.
Nickel
l
I” U
40
-
20
-
z
0
0
30
60
90
TIME
120 (min
Fig. 3. Curve to show time course of formation chelates using the described method.
Analytical
variables for HPLC
Platinum
(IT) (XI)
‘24w
1 of (a) nickel and (0) platinum
diethyldithiocarbamate
assa_y
Precision: Within-run precision of the assay was determined by processing l-ml samples of pooled plasma ultrafiltrate and urine through the procedure during a single run. The coefficients of variation (CV) were found to be 1.4% for ultrafiltrate and 5.8% for urine (Table II). Day-to-day variability, determined over a period of 2 months, was 8.1% for ultrafiltrate and 8.8% for urine (Table II). Analytical recovery Analytical recovery of known amounts of DDP added to blank plasma ultrafiltrate and urine was greater than 95%. This was determined by measuring the aqueous and chloroform layers of extracted samples by atomic absorption spectrometry tuned for measuring platinum at 265.9 nm. Amounts of platinum corresponding to less than 5% of that added were seen in the aqueous layer. The remaining fraction was observed in the organic layer suggesting complete extraction of the platinum
TABLE
II
Assay characteristics
for the liquid-chromatography Urine
Precision intra-day inter-day
5.8% 8.8%
Recovery Detection
> 95% limit
50 n&ml
of DDP-DDTC
chelate Plasma ultrafiltrate
1.4% 8.1% > 95% 50 ng/ml
71
chelate. Experiments were also conducted which demonstrated that no platinum appeared in the chloroform layer when chelating agent was absent. Linearity and detection limit Calibration curves were linear for the HPLC assay for both lower and upper concentration ranges and passed through the origin when drug-free plasma ultrafiltrate or urine fortified with DDP in the concentration range 50 ngfml to 100 pg/ml were processed. Separate calibration curves were usually constructed in the range 0.25-5 @g/ml and for the levels higher 5 mg/ml in the range 2.5-50 pg,/ml to avoid the need to change the attenuation during a chromatogram. The amount of internal standard used for the lower range was 20 pg instead of the usual 100 pg. This range represented the plasma ultrafiltrate and urine levels following a 120 mg/m* IV infusion of DDP. The detection limit for DDP was about 50 ng/ml in both urine and ultrafiltrate. This represented a peak about three times baseline noise at the highest sensitivity on the UV detector (0.005 aufs). Specificity Assay of urine and plasma ultrafiltrate spiked with chloroplatinic (IV) acid at concentrations up to 50 pg/ml did not give rise to any peaks in the chromatogram. Thus the assay appears to be specific to platinum in the II valency state. This was expected, since the Pt (IV)-DDTC chelate is likely to have the stoichiometric formula Pt (DDTC),, which would be a charged species and not extractable into chloroform. The assay did not discriminate between the cis and trans forms, nor the potential aquo metabolites which form in low chloride solutions [ll]. Both pure trans-dichlorodiammineplatinum (II) and the mono and diaquo forms of DDP formed by incubating DDP in low chloride solution gave peaks with identical retention volumes and peak heights to DDP. Interference Although the possibility existed that other metals may interfere with either nickel or platinum chelate peaks, no other metals were found with retention volumes close enough to exhibit interference. The metals tested included zinc (II), cadmium (II), cobalt (II), calcium (II), iron (II), iron (III), gold (I), silver (I), lead (II), cerium (II), uranium (II), copper (II), manganese (II), palladium (II), and mercury (II), The drugs adriamycin, bleomycin, cyclophospharnide, vinblastine, vincristine, metoclopramide and prochlorperazine, which are sometimes used in combination with DDP, did not interfere with this assay. Comparison with atomic absorption spectrometry In patients receiving DDP therapy at a dose of 120 mg/m2 there was a close correlation between results obtained with the standard flameless atomic absorption spectrometric [8] and the HPLC procedure. The Deming method of least-squares regression analysis was designed to handle data where the independent variable (x) as well as the dependent variable (y) is measured with some error 1141.Using this
12
B.
A. E .
::
URINE
FILTRATES
ULTRA
E \
60-
60
s z
50
-
40
-
50
0 lP
a 0 v) m a
g v) m a
30 20
u 5
I 0
0 I-
.a
IO
20
30
HPLC
40
50 (p/ml
60
)
2
0 IO
20 HPLC
30
40
50 (yg/ml
60
1
Fig. 4. Comparison of cis-dichlorodiammineplatinum (11) concentrations obtained by a flameless AA method and by the present HPLC method for (a) plasma ultrafiltrate, n = 19 and (b) urine, n = 15. The lines represent equality by the two methods.
statistical test it was found that the slopes for both ultrafiltered plasma and urine data were not significant from unity (slope ultrafiltrate = 0.9088, SD = 0.089; slope urine = 1.262, SD = 0.313). The correlation coefficients for ultrafiltrate and urine were 0.9373 and 0.7322, respectively. The intercepts for ultrafiltrate (y-intercept = 0.58, SD = 0.21) and urine (y-intercept = - 2.367, SD = 3.294) were also not significantly different from the origin.
Discussion The measurement of the anti-neoplastic drug DDP has been largely restricted to atomic absorption spectrometric methods [4-91, owing to the poor UV-visible absorption properties of this co-ordination complex. In this paper the bidentate ligand DDTC has been used to chelate DDP using an adaption of Amore’s method [15] to exchange the chloro and ammine ligands. Our procedure, in contrast to that described by Bannister and co-workers [ll], does not require the presence of sodium nitrate or strongly basic conditions nor does it involve a large incubation volume. Despite these changes the formation of the DDTC chelate of platinum (II) is essentially complete after a 60-min incubation at 37°C. The detection limit for DDP using the method reported here was about 50 ng/ml in plasma ultrafiltrate and urine. This is less sensitive than existing flameless atomic absorption spectrometric methods [5,8] which quote detection limits of around 20 ng/ml, and for a previous HPLC procedure (25 ng/ml) [ll]. However, with this latter procedure 9 ml of urine were required to achieve this sensitivity. Other published HPLC procedures utilising
13
the weak native UV absorption of DDP were only able to measure levels down to about 20 pg/ml in aqueous solutions [12]. Therapeutic plasma ultrafiltrate or urine levels were always much higher than 50 ng/ml and usually in the range 0.25 to 50 pg/ml. Higher sensitivity could be obtained by using less mobile phase (100 ~1) to reconstitute the residue after evaporation of the chloroform. Reproducibility studies in both plasma ultrafiltrate and urine suggested that recoveries were sometimes variable depending on the clarity and protein content of the fluid. Using this procedure there was an occasional variability in the height of the platinum peak with some samples similar to that reported by Bannister et al [ll]. The use of an internal standard (nickel) was able to compensate for variability in the extraction and chromatography steps. Although the use of nickel as internal standard may not compensate for variability in the complexity of Pt (II), the excellent precision obtained both within and inter assay suggest that this step may not be the major source of variability encountered. The chromatograms obtained from plasma ultrafiltrate and filtered buffered urine confirmed the presence of an early peak immediately after the solvent front due to some unreacted chelating agent., This did not interfere with either the nickel- or platinum-DDTC peaks and allowed flow rates of up to 4 ml/mm. Under these conditions the retention times of nickel and platinum were 2.4 min and 3.9 min, respectively. Thus, chromatograms took less than 5 min to develop, enabling 12 samples to be run through the HPLC in 1 h. This method also lends itself to automatic sample processing for large batches. No interference was noted with the platinum peak in any patient samples studied nor with a number of metals likely to form complexes with DDTC or drugs co-administered with DDP. A further feature of the method was the ability to measure other metals beside platinum in biological fluids without the need of atomic absorption spectrometry. Examples include nickel and a number of other transition metals with an oxidation state of II such as zinc and mercury. Validity of this HPLC method was shown by a good correlation for both plasma ultrafiltrate and urine with a reference flameless atomic absorption spectrometric method [8]. The experiments confirmed the precision of the HPLC assay and suggested that both procedures are indeed measuring the same platinum species. Like the other reported procedures this method will not distinguish between DDP and its aquo metabolites. There is some evidence in the literature [13] that aquo metabolites of DDP might form in cells as the highly labile chloro groups are successively displaced by water molecules. In our own laboratories stability studies in which DDP was incubated in a low chloride medium confirmed that these aquo species are formed in vitro and are measured by the HPLC procedure. Although the trans isomer of DDP is also measured by the HPLC procedure, there is no evidence in the literature that this compound occurs as a metabolite of DDP. In conclusion, a rapid, sensitive and clinically useful HPLC method has been developed for measuring unbound DDP from human plasma ultrafiltrate and urine samples. The method includes an internal standard for added precision which largely circumvents matrix effects observed with a previously described HPLC method and
74
existing atomic absorption procedures. The procedure has been validated against a standard atomic absorption method for pIatinum levels in the range 50 ng/ml to 50 pg/ml in plasma ultrafiltrate and urine samples from patients receiving DDP therapy. The method requires minimal sample preparation and lends itself to automation for processing of large numbers of samples. Acknowledgements This work was supported by the National Wealth and Medical Research Council of Australia. The helpful advice of Mr. Brian Stevens from the Royal Melbourne Institute of Technology and the use of his atomic absorption spectrophotometer was greatly appreciated. References 1 Einhorn LH, Donohue JP. Cis-D~aminned~chloroplatinum, vinblastine, and bfeomycin combination chemotherapy in disseminated testicular cancer. Ann intern Med 1977; 86: 8033812. 2 Bush H, Thatcher N, Barnard R. Chemotherapy in the management of invasive bladder cancer. Cancer Chemother Pharmacol 1979; 3: 87-96. 3 De Conti RC, Toftness BR, Lange RC, Creasey WA. Clinical and pharmacological studies with cis-diamminedichloroplatinum (II). Cancer Res 1973; 33: 1310-1315. 4 Pera MF Jr, Harder HC. Analysis for platinum in biological material by flameless atomic absorption spectrometry. CIin Chem 1977; 23: 1245-1249. 5 Bannister SJ, Chang Y, Sternson LA, Repta AJ. Atomic absorption spectrometry of free circulating platinum species in plasma derived from cis-dichlorodiammineplatinum (II). Clin Chem 1978; 24: 877-880. 6 Litterst CL, Gram TE, Dedrick RL. Leroy AF, Guarino AM. Distribution and disposition of platinum following intravenous administration of cis-diamminedichloroplatinum (II) (NSC 119875) to dogs. Cancer Res 1976; 36: 2340-2344. 7 Hull DA, Muhammad N, Lanese JG, Reich SD, Finkelstein IT, Fandrich S. Determination of platinum in serum and ultrafiltrate by flameless atomic absorption spectrophotometry. J Pharm Sci 1981; 70: 500-502. 8 Smeyers-Verbeke J, Detaevemicr MR. Denis L, Massant DL. The determination of platinum in biological fluids by means of graphite furnace atomic absorption spectrometry. CIin Chim Acta 1981; 113: 329-333. 9 Preisner D, Sternson LA, Repta AJ. Analysis of total platinum in tissue samples by flameless atomic absorption spectrophotometry. Elimination of the need for sample digestion. Anal Lett 1981: 14: 1255-1268. 10 Bannister SJ, Stemson LA, Repta AJ, James GW. Measurement of free-circulating cis-dichlorodiam~neplatinum (II) in plasma. Clin Chem 1977; 23: 2258-2262. 11 Bannister SJ, Sternson LA, Repta AJ. Urine analysis of platinum species derived from cis-dichlordiam~nepIat~num (II) by high performance liquid ~bromatography following derivatization with sodium diethyIdithi~~b~ate. J Chromatogr 1979; 173: 332-342. 12 Hincal AA, Long DF, Repta AJ. Cis-platin stability in aqueous parenteral vehicles. J Parenteral Drug Assoc 1979; 33: 107-116. 13 Repta AJ, Long DF. Reactions of cisplatin with human plasma and plasma fractions. In: Prestayko AW, Crook ST, Carter SK, eds. Cisplatin current status and new developments. New York, NY: Academic Press, 1980: 285-304. 14 Combleet PJ, Gochman N. Incorrect least-squares regression coefficients in method-comparison analysis. Clin Chem 1979; 25: 432-438. 15 Amore F. Determination of calcium, lead, thallium and nickel in blood by atomic absorption spectrometry. Anal Chem 1974; 46: 1597-1599.