Determination of Pt in Biological Fluids With ICP-MS:

Evaluation of Analytical Uncertainty

*M. Bettinelli, S. Spezia, A. Ronchi, and C. Minoia
Laboratory for Environmental and Toxicological Testing
S. Maugeri Foundation, Via Ferrata 8, 29100 Pavia, Italy

INTRODUCTION Uncertainty in Measurement (GUM)

ABSTRACT
published by ISO (7) establishes
Chrysoteraphy with Pt-based Estimation of the uncertainty general rules for evaluating and
drugs is gaining increasing impor- associated with analytical meth- expressing uncertainty for a wide
tance, especially in the treatment of ods is necessary in order to estab- range of measurements. The guide
solid tumors, testicular and ovarian lish comparability of the results. was interpreted for analytical chem-
tumors, head, neck, and bladder Methods of Pt determination in
istry by EURACHEM in 2000 (8).
carcinomas, and brain malignancies biological fluids very often lack of
information about the uncertainty The approach described in the
(1,2). GUM requires the identification of
of results. This has implications
A variety of oral analogues of when results are used to interpret all possible sources of uncertainty
known antitumor agents have been the mechanism of platinum com- associated with the procedure, the
developed in the past few years for pounds, or when the results are estimation of their magnitude from
economic, pharmacological, and considered to optimize clinical either experimental or published
practical reasons. To merit further therapies. data, and the combination of these
clinical development, the oral ana- An ICP-MS method for the individual uncertainties to give stan-
determination of Pt in biological dard and expanded uncertainties
logue should have a preclinical fluids (plasma, ultrafiltrate, and
antitumor activity and toxicity for the procedure as a whole. How-
urine) of patients treated with
comparable to those of parent com- antitumor agents has been devel- ever, the GUM principles are signifi-
pound, with only limited gastroin- oped and validated. The limits of cantly different from the methods
testinal toxicity, acceptable quantification (LOQ) in the three currently used in analytical chem-
bioavailability, safe and reproducible matrices were 1.0, 0.1, and 2.0 µg istry for estimating uncertainty
pharmacokinetic profile. L–1, respectively. which generally make use of
Intra-day and inter-day preci- "whole method" performance para-
A series of ammine/amine Pt(IV) sion and accuracy were in good meters, such as precision, recovery,
dicarboxilates of higher lipophilic- agreement with the FDA criteria and ruggedness. The purpose of
ity and stability than cisplatin and for validation of the analytical this paper is to underline that if val-
carboplatin have been developed methods. The validation study
idation studies are properly
in an effort to improve absorption was implemented by assessing
the uncertainty evaluation for Pt planned and executed, the data
and decrease the metabolic activa- produced can easily be handled in
tion through processes in the gas- determination in the different
matrices according to the the uncertainty estimation process.
trointestinal track (3). EURACHEM/CITAC Guide.
From this point of view, it is EXPERIMENTAL
essential to correctly interpret the
mechanism by which platinum Instrumentation
compounds act in order to Harmonization (ICH) (5), and the The PerkinElmer SCIEX ELAN®
optimize the therapies that have Food and Drug Administration 5000 Inductively Coupled Plasma
been proposed to date and to (FDA) (6) provide a framework for Mass Spectrometer (PerkinElmer
develop adequate analytical meth- performing such validations. In gen- SCIEX, Concord, Ontario, Canada)
ods for monitoring the Pt levels in eral, methods for regulatory submis- was designed for routine and rapid
biological fluids. sion must include studies on multielement quantitative determi-
specificity, linearity, accuracy, pre- nations of trace and ultratrace ele-
Method validation is the process cision, range, detection limit, quan- ments and isotopes (6).
of proving that an analytical titation limit, and robustness.
method is acceptable for its An IBM® PS/2 Model 70 386,
intended purpose. For pharmaceuti- Today, it is generally acknowl- equipped with an operating system
cal methods, guidelines from edged that the fitness of an analyti- Xenix® 386 (Version 2.3.4) and an
United States Pharmacopeia (USP) cal result cannot be assessed Intel® 80386 microprocessor, con-
(4), International Conference on without some estimate of the mea- trols the ELAN ICP-MS.
surement uncertainty for compari-
The instrument is equipped with
son with the confidence required.
*Corresponding author. a Cool Flow CFT-75 NESLAB chiller
E-mail: [email protected] The Guide to the Expression of
(Neslab, Portsmouth, New Hampshire,

Atomic Spectroscopy 103

Vol. 25(3), May/June 2004
USA) and a Model AS-90 Autosam- 185 PLUS purifier (Millipore, Bed- three different concentrations: 5,
pler (PerkinElmer Life and Analyti- ford, MA, USA). 40, 160 µg L–1; 3 µg L–1 Ir used as an
cal Sciences, Shelton, CT, USA). internal standard was added to all
A platinum chloride stock solu-
To optimize the ICP-MS signal, the samples employed for the
tion of 1 g/L was diluted (1:10 v/v)
a standard solution (Number preparation of calibration curves
in ultrapure water (Milli-RO 10
N812–2014) containing 10 µg/L and quality control samples.
PLUS, Milli-Q 185 PLUS, Millipore)
of three elements that covered the to obtain working solutions for the Sampling Procedure
entire mass range (for example, preparation of calibration curves
24Mg, 103Rh, and 208Pb) was used.
and quality control samples. Five-mL blood samples were col-
A fourth isotope, 197Au, was used to lected in a tube containing EDTA at
monitor background noise in the In order to reduce the matrix the following times: T = 0, 0.5, 1, 2,
system. Typically, the intensity val- effect, 193Ir was used as an internal 3, 4, 6, 8, 12, 24 hours; all blood
ues obtained for 10 µg/L were as standard. The iridium chloride samples were immediately
follows: 24Mg ≥ 5,000 ions/sec., stock solution of 1 g L–1 was diluted centrifuged at 2000 rpm for 15 min
103Rh ≥ 30,000 ions/sec., 208Pb ≥ (1:10 v/v) in ultrapure water to at 4oC and the plasma removed.
5,000 ions/sec., and 197Au ≤ 20 obtain the working solution.
The plasma was separated and
ions/sec. Preparation of Calibration divided in two portions: one for the
The instrumental specifications Curves and Quality Control determination of total Pt (TPt) and
and the analytical conditions are Samples the other for the preparation of free
reported in Table I. Pt species (UPt, unbound Pt).
Aliquots of the working
Separation of UPt species was per-
solutions were spiked into blank
Reagents and Standard formed by centrifugal ultrafiltration
pooled human plasma [diluted 1:30
Solutions using a modified version of the
(v/v) with 1% HNO3 (v/v)] to repro-
- Platinum (PtCl2, 10% HCl) and method of Bannister et al. (6).
duce concentrations of 1, 10, 50,
iridium (IrCl3*3H2O,10 % HCl) were 100, 200 µg L–1, and into plasma A portion of 2 mL from each
provided by BDH Laboratory Sup- ultrafiltrate samples [diluted 1:30 plasma sample was placed into a
plies, England. (v/v) with 1% HNO3 (v/v)] to repro- microconcentrator Centricon-10
duce concentrations of 0.01, 0.1, 1, (Amicon Corp., Dan, MA, USA) hav-
- Nitric acid 65% (m/v)
10, 20 µg L–1. These samples were ing a cut-off value of 30000
Suprapur“ was provided by Merck,
used as calibration standards. daltons, and was centrifuged at
Darmstadt, Germany.
2000 rpm for 30 min at 4oC.
The same working solutions
- Water used for the ICP-MS
were used for the preparation of Two urine samples (fraction 0–8
analysis was prepared in-house
the quality control samples (QC) at hours and 8–24 hours) were col-
using a Milli-RO 10 PLUS, Milli-Q™
three different concentrations: 2, lected during the treatment on days
41, 152 µg L–1 for blank human 1 and 14, the volume measured and
TABLE I plasma and 0.02, 6, 18 µg L–1 for stored at –20oC until analysis.
Instrumental Specifications and blank plasma ultrafiltrate.
Sample Handling
Analytical Conditions One µg L–1 Ir was used as an
195Pt internal standard and was added to Plasma, ultrafiltrate, and urine
Element/mass samples were thawed in a warm
Internal standard 193Ir all samples employed for the prepa-
ration of calibration curves and bath immediately before analysis.
Replicate time 900 ms
quality control samples. Plasma and ultrafiltrate samples
Dwell time 300 ms were diluted [1:30 (v/v)] in
Scanning mode Peak hopping Aliquots of the working
solutions were spiked into blank polypropylene tubes, with 1%
Sweeps/reading 3 pooled human urine [diluted 1:50 HNO3 (v/v). After addition of 1 µg
Number of replicates 10 (v/v) with 1% HNO3 (v/v)] to repro- L–1 Ir as an internal standard, the
Points/spectral peak 1 duce concentrations of 2, 10, 50, samples were analyzed by ICP-MS.
100, 200 µg L–1. These samples The sample uptake rate was of 1.0
Resolution Normal mL/min.
Power 1100 W were used as calibration standards.
The same working solutions Urine samples were diluted
Plasma argon 12 L/min
were used for the preparation of [1:50 (v/v)] in polypropylene tubes,
Auxiliary argon 0.80 L/min with 1% HNO3 (v/v). After addition
the quality control samples (QC) at
Nebulizer argon 0.99 L/min of 3 µg L–1 Ir as internal standard,

104
 Vol. 25(3), May/June 2004

the samples were analyzed by ICP- TABLE II

MS. The sample uptake rate was of Natural Abundance and Potential Interferences of Pt Isotopes
1.0 mL/min. Pt Mass Abundance Potential Interferences
RESULTS AND DISCUSSION 189.961 0.013% Os, YbO, HfO
191.961 0.780% Os, YbO, HfO, LuO
Specificity 193.961 32.900% YbO, HfO
The two major sources of analyt- 194.961 33.800% HfO
ical inaccuracy during ICP-MS analy- 195.969 25.300% Hg, HfO, WO, TaO
ses are matrix-induced signal
suppression and spectral interfer- 197.969 7.210% Hg, HfO, WO, TaO
ences.
In the case of matrix-induced
signal suppression, the ion intensity
of an analyte element is somewhat
dependent upon the total composi-
tion of the sample. For a matrix of
unknown composition, the method
of standard addition calibration, in
which known concentrations of
standards are added directly to a
sample solution, compensate for
the difference in sensitivity
between samples and standards.
Spectral interferences may origi-
nate from several sources: isobaric
overlaps, plasma-induced
polyatomic ions, matrix solvent-
induced polyatomic ions, and
matrix-induced polyatomic ions (7). Fig. 1. Spectrum of human blank plasma sample spiked with 1 µg/L Ir.
Considering the natural abundance No interfering signal was present in the mass range of interest (194 – 196 amu).
and potential interferences of Pt
isotopes reported in Table II and
preliminary studies not reported in
the present paper, the 195Pt was
identified as the isotope with the
lowest number of interferences.
Under the described experimen-
tal conditions, no significant inter-
ference on Pt in human urine,
plasma and ultrafiltrate matrix was
observed in the drug-free samples,
as shown for plasma by a represen-
tative ICP-MS spectrum (Figure 1).
An ICP-MS spectrum of the
"blank" plasma sample enriched
with 10 µg/L PtCl2 is given in Fig-
ure 2; Pt-measured isotopes corre-
spond to the signal at 194, 195 and
196 amu.
Fig. 2. Spectrum of human blank plasma sample enriched with 10 µg/L Pt.
Figure 3 shows a spectrum of a
real sample obtained from a subject
treated with an anticancer agent.

105
Linearity
The standard addition method
was used for plasma, ultrafiltrate,
and urine matrices.
The calibration curves were
established by plotting the current
intensity (ions/sec) ratio between
195Pt and the internal standard

(193Ir) against the concentration of

Pt. Slopes and intercepts were
determined by linear regression
analysis.

Accuracy and Precision

As CRMs are not available for Pt
determination in blood and urine
matrices, the accuracy and preci- Fig. 3. Spectrum of real human plasma sample obtained from a subject treated
sion were evaluated on spiked sam- with anticancer agent.
ples at three different
concentrations. Intra-day and inter-day accuracy
tory degree of accuracy (deviation
Intra-day and inter-day precision from nominal value ≤20%) and pre- and precision of the method for the
and accuracy were assessed by cision (CV≤20%) to provide pharma- determination of platinum in
replicate determinations of Pt con- cokinetically useful results." In the human plasma and ultrafiltrate sam-
centration added as PtCl2 to the case of plasma and urine analysis, ples are reported in Tables III and
blank human plasma, human considering that the concentrations IV, respectively.
plasma ultrafiltrate, and the urine of interest were one order of magni-
Intra-day precision and accuracy
on the same day and on three dif- tude higher than those in ultrafil-
for Pt determination in human
ferent days, respectively. The preci- trate. the LOQ was set at 1 and 2 µg
plasma samples ranged,
sion of the method was calculated L–1, respectively.
respectively, between 0.14 and
as the coefficient of variation of the 2.01, and 99.7 and 103.2 (C.V.%),
mean value found at each concen- Plasma and Ultrafiltrate
whereas in the ultrafiltrate they
tration level (one near the lower Matrix-matched calibration was ranged between 0.03 and 5.22, and
limit of quantitation, one near the achieved by adding dilute Pt stan- 97.5 and 108.0 (C.V.%). Slightly bet-
middle, and one near the upper dard solution to pooled plasma or ter ranges of inter-day precision and
boundary of the calibration curve). ultrafiltrate samples and kept at accuracy were achieved for plasma
The accuracy, expressed as the per- –20oC until use. with respect to ultrafiltrate, proba-
cent of recovery, was calculated as bly due to the higher Pt concentra-
the ratio of the concentration For plasma, the mean slope and
intercept values of the calibration tions investigated.
found to the spiked concentration
value. curves (0–200 µg L–1) obtained over The instrumental detection limit
a three-day validation period were: (LOD) was 0.002 µg L–1 with a mini-
Limit of Detection (LOD) and slope = 0.199 ± 0.008 (CV= 3.86%) mum quantifiable platinum concen-
Limit of Quantification (LOQ) and intercept = –0.055 ± 0.043 with tration (LOQ) for ultrafiltrate equal
a correlation coefficient better than to 0.01 µg L–1. In the case of plasma
The instrumental detection limit
0.9996. analysis, the concentrations of
(LOD) was evaluated on the basis of
the 3σ criterion by analyzing five For ultrafiltrate plasma, the statis- interest were one order of magni-
"blanks" of plasma, ultrafiltrate, and tical parameters for the calibration tude higher than those in ultrafil-
urine on three non-consecutive curves (0–20 µg L–1) obtained over a trate samples; therefore, the LOQ
days. three-day validation period were: was set at 1 µg L–1.
slope = 0.376 ± 0.042 (CV = 11.1%)
For the LOQ, we adopted the
and intercept = –0.005 ± 0.006 with
criterion reported by the FDA (8)
a correlation coefficient better than
"as the lowest concentration that
0.9989.
can be determined with a satisfac-

106
 Vol. 25(3), May/June 2004

TABLE III
Accuracy and Precision in the Determination of Pt in Human Plasma Samples,
Evaluated on Three Different Days
Nominal Concentration (µg L ) QC1 (2 µg L–1)
–1
QC2 (41 µg L–1) QC3 (152 µg L–1) LOQ (1 µg L–1)
1st day
Found Concentrations (µg L–1)
Intra-day Mean ± S.D. 2.06 ± 0.02 41.14 ± 0.29 152.16 ± 0.55 1.032 ± 0.005
Intra-day C.V. (%) 0.87 0.70 0.36 0.46
Intra-day Accuracy (%) 103.2 100.3 100.1 103.2
2nd day
Intra-day Mean ± S.D. 2.01 ± 0.04 40.88 ± 0.06 152.14 ±0.32 1.032 ± 0.028
Intra-day C.V. (%) 2.01 0.14 0.21 2.69
Intra-day Accuracy (%) 100.5 99.7 100.1 103.2
3rd day
Intra-day Mean ± S.D. 1.99 ± 0.03 41.17 ± 0.70 151.93 ±0.50 1.037 ± 0.017
Intra-day C.V. (%) 1.57 1.70 0.33 1.64
Intra-day Accuracy (%) 99.7 100.4 99.9 103.7
Inter-day Accuracy and Precision
Mean (µg L–1) 2.02 ± 0.04 41.06 ± 0.40 152.08 ± 0.42 1.034 ± 0.017
C.V. (%) 2.05 0.98 0.28 1.61
Accuracy (%) 101.1 100.2 100.1 104.6

TABLE IV
Accuracy and Precision in the Determination of Pt in Human Plasma Ultrafiltrate Samples,
Evaluated on Three Different Days
Nominal Concentration (µg L ) QC1( 0.02 µg L–1)
–1
QC2 ( 6 µg L–1) QC3 (18 µg L–1) LOQ (0.001 µg L–1)
1st day
Found Concentrations (µg L–1)
Intra-day Mean ± S.D. 0.019 ± 0.001 6.024 ± 0.142 17.78 ± 0.01 0.0098 ± 0.0008
Intra-day C.V. (%) 3.36 2.36 0.03 8.10
Intra-day Accuracy (%) 97.50 100.40 98.78 98.0
2nd day
Intra-day Mean ± S.D. 0.022 ± 0.001 5.975 ± 0.070 18.02 ± 0.14 0.0108 ± 0.0007
Intra-day C.V. (%) 5.22 1.18 0.77 6.52
Intra-day Accuracy (%) 108.00 99.59 100.13 107.7
3rd day
Intra-day Mean ± S.D. 0.0201 ± 0.0003 5.981 ± 0.061 18.06 ± 0.24 0.0108 ± 0.0006
Intra-day C.V. (%) 1.55 1.02 1.33 5.56
Intra-day Accuracy (%) 103.83 99.69 100.31 108.0
Inter-day Accuracy and Precision
Mean ± S.D. 0.021 ± 0.001 5.993 ± 0.088 17.95 ± 0.19 0.0105 ± 0.0008
C.V. (%) 5.51 1.47 1.06 7.49
Accuracy (%) 103.11 99.89 99.74 104.6

107
 TABLE V
Accuracy and Precision in the Determination of Pt in Human Urine Samples,
Evaluated on Three Different Days
–1
Nominal Concentration (µg L ) QC1( 5 µg L–1) QC2 (40 µg L–1) QC3 (160 µg L–1) LOQ (2 µg L–1)
1st day
Found Concentrations (µg L-1)
Intra-day Mean ± S.D. 5.037 ± 0.010 40.07 ± 0.09 160.12 ± 0.42 2.091 ± 0.012
Intra-day precision (C.V.%) 0.19 0.23 0.26 0.58
Intra-day Accuracy (mean found/added %) 100.7 100.2 100.1 104.6
2nd day
Intra-day Mean ± S.D. 5.009 ± 0.004 40.05 ± 0.05 159.96 ± 0.94 2.068 ± 0.017
Intra-day precision (C.V.%) 0.08 0.11 0.59 0.81
Intra-day Accuracy (mean found/added %) 100.2 100.1 99.9 103.4
3rd day
Intra-day Mean ± S.D. 5.021 ± 0.083 39.98 ± 0.08 159.94 ± 0.33 2.119 ± 0.012
Intra-day precision (C.V.%) 1.66 0.21 0.21 0.57
Intra-day Accuracy (mean found/added %) 100.4 99.9 100.0 105.9
Inter-day Accuracy and Precision
Mean ± S.D. 5.023 ± 0.044 40.032 ± 0.078 160.01 ± 0.55 2.093 ± 0.025
Precision (C.V.%) 0.87 0.20 0.34 1.20
Accuracy (mean found/added %) 100.46 100.08 100.00 104.6

Urine 0.59 (with only one value higher Uncertainty Evaluation From
Matrix-matched calibration was than 1%), and intra-day accuracy, Validation Data
achieved by adding dilute Pt stan- calculated as the percentage recov- The validation study of the pre-
dard solution to pooled urine sam- ery of the spiked samples, ranged sent assay was implemented by
ples, previously centrifuged to on average between 99.9 and assessing the uncertainty evaluation
remove the sediment, and kept at 100.7%. Precision lower than 1% for Pt in the different matrices at
–20oC until use. and accuracy very close to 100% the four levels of concentration.
were achieved during the inter-day According to the EURACHEM/
The mean slope and intercept evaluation. Accuracy and precision CITAC Guide (23), the combined
values of the calibration curves for urine samples at the LOQ con- uncertainty (ucombined) has been cal-
(0–200 µg/L) obtained over a three- centration are reported in Table V. culated by estimating the compo-
day validation period were: slope = nents associated with the
0.128 ± 0.04 (CV= 2.97%) and inter- The results obtained in the pre-
sent study meet to the FDA criteria repeatability (urepetability), the instru-
cept=0.033 ± 0.041 with a correla- mental calibration (ucalibration), and
tion coefficient better than 0.9999. for the validation of analytical meth-
ods (3). For each concentration the sampling (usampling). The uncer-
The detection limit (LOD) was level, the mean value is within tainty component due to the recov-
0.002 µg L–1 with a minimum quan- ±15% of the nominal value (at the ery (urecovery) was not considered
tifiable platinum concentration LOQ within ±20%), and the coeffi- because the calibration curve was
(LOQ) set at 2 µg L–1. cient of variation (CV) around the performed by standard addition
mean value is lower than ±15% (at method (in the same matrix as the
Intra-day and inter-day accuracy sample), determining the analytes
the LOQ ±20% for the CV is
and precision of the method for the according to the procedure
accepted).
determination of platinum in described in the Experimental sec-
human urine samples are reported These results prove that, in tion.
in Table V. keeping with international
standards, the method is adequate Consequently, the equation that
Intra-day precision, expressed as best represents the mathematical
for the assay of platinum in human
CV%, ranged between 0.08 and model upon which the relative
plasma and urine samples.

108
 Vol. 25(3), May/June 2004

uncertainty is evaluated is taken to

be:

u(Comb.) = (u(Repeat))2 (u(Calib.))2 (u(Samp.))2

+ +
(Comb.) √ ((Repeat.)) ((Calib.)) ((Samp.))

.
u(Comb.) =
. . .
√(u(Repeat))2+(u(Calib.))2+(u(Samp.))2

where:
.
u(Comb) = combined relative uncer-
tainty
.
u(Repeat.) = relative uncertainty of
repeatability
.
u(Calib.)= relative uncertainty of cali-
bration
. Fig. 4. Relative Uncertainty Values (%) obtained at different concentration levels
for Pt in human plasma.
u(samp.) = relative uncertainty of sam-
pling
In order to investigate the contri-
butions of all components to the
combined relative uncertainty as a
function of Pt concentration (in
plasma, ultrafiltrate, and urine),
these uncertainties were evaluated
and plotted in Figures 4, 5, and 6,
respectively.
Whit respect to Pt in plasma and
urine, the graphs show that the rela-
tive uncertainty ranges from 4–7% at
1–2 µg L–1 and go down to 2–3% for
the higher concentrations (40–150
µg L–1).
While the uncertainty component
associated with the repeatability is
constant (0.5–2%), overall the differ-
ent concentration levels and the dif-
ferent days, the component due to
the instrumental calibration
increases significantly (7–8%) rela-
tive to the lowest concentrations of
analyte.
With respect to Pt in the plasma Fig. 5. Relative Uncertainty Values (%) obtained at different concentration levels
ultrafiltrate, Figure 5 shows that the for Pt in human plasma ultrafiltrate.
relative uncertainty ranges from
4–5% at 8–16 µg L–1 and increases up

109
to 20–30% for the lower concentra-
tions (at 0.01–0.02 µg L–1).
This occurrence means that the
quantitative determination of low
concentrations of Pt in the ultrafil-
trate is greatly affected by the
uncertainty assignable to the instru-
mental calibration as done in the
present study (0–20 µg L–1). Further
tests performed show that by
reducing the calibration range but
keeping the same number of con-
centration levels (n=6) gives better
values of combined uncertainty as a
consequence of a significant
improvement of the uncertainty
component due to calibration. The
mathematical function (9) used for
the calculation of uncertainty of
(xpred) due to calibration is the fol-
lowing: _
s2y/q 1 1 (xpred – xCal)2
u(xpred)= + + _
b √ m n Σ(xi – xCal)2
Fig. 6. Relative Uncertainty Values (%) obtained at different concentration levels
for Pt in human urine.
where:
s2y/q is the residual variance of our study, with a calibration range regression line or make more than
the regression model of 0–20 µg L–1 and a centroid corre- one measurement and use the mean
b is the slope of the regression .
sponding to (5.18; 1.84), the
u(Calib.) for a predicted mean value
value in the calculation of the
predicted value.
curve
of 0.36 µg L–1 Pt in the ultrafiltrate
m is the number of repetitions (n=3 replicates) was about 21.9%; Analysis of Real Samples
from which is derived the the same results became lower The present method was
predicted value (0.22%) when the calibration range employed to determine the pharma-
was set at 0–2 µg L–1 with the same cokinetic profiles of total Pt and
n is the number of calibration number of calibration points and
points on the regression line ultrafiltrable Pt in patients who
with a centroid corresponding to received JM216 (an oral bis
x(pred) is the predicted value (0.527, 0.163). (acetate) amine dichloro cyclohexy-
These results show the impor- lammine platinum (IV) analogue)
x(Cal) is the mean value of the tance of calibration in evaluating brought into clinical development
concentration levels used in the the measurement uncertainty of for phase II evaluation. Forty-six
calibration curve low Pt concentrations in biological patients were treated at doses rang-
Inspection of the previously fluids. Keeping the repeatability of ing from 10 mg/m2/d to 50
reported equation confirms that measurements constant as, in the mg/m2/d and 39 patients were eval-
as xpred approaches xCal, the third case of a wider calibration range, uated for hematologic toxicity over
term under the square root we should accept larger uncertainty 74 cycles. The pharmacokinetics of
u(xpred)approaches zero and thus of calibration which will reflect sig- total and ultrafilterable Pt were
approaches a minimum value. In a nificantly on the combined uncer- studied on days 1 and 14 of the first
practical analysis, therefore, a cali- tainty. cycle, and the results were recently
bration of this type will give the published by Sessa (3). Figure 7
If we wished to improve (i.e., shows, as example, the pharmacoki-
most precise results when the mea- narrow) the confidence limits of
sured instrument signal corresponds netics profiles of four patients who
calibration, we could reduce the received 45 mg/m2/d of agent.
to a point close to the centroid of calibration range, increase the num-
the regression line (xCal, yCal, ). In ber of calibration points on the

110
 Vol. 25(3), May/June 2004

Fig. 7. Pharmacokinetics profiles of four patients treated with 45 mg/m2/d of agent. Plasma levels of total platinum (Pt) and
plasma ultrafiltrate platinum (UPt).

JM216 was rapidly absorbed, The present study was implemented REFERENCES
with Pt and UPt detectable in by assessing the uncertainty evalua- 1. J.A. Gottlieb and B. Drewinko, Cancer
plasma as early as 30 min after the tion for Pt determination in the dif- Treat. Rep. 59, 621 (1975).
administration. On day 1, Pt and ferent matrices according to the 2. A.B. Khan, B.J. D’Souza, M.D.
UPt peak levels were achieved in EURACHEM/CITAC Guide. The Wharam,L.A.A. Champion, L.F. Sinks,S.Y.
combined uncertainty, calculated Woo, D.C. McCullough, and B.C. Leven-
most cases within 3 hours with thal, Cancer Treat. Rep. 66, 2013 (1982).
median Tmax values of 2.5 and 2, by estimating the components asso- 3. C. Sessa, C. Minoia A. Ronchi, M. Zuc-
respectively. Elimination was vari- ciated with the repeatability, the chetti, J. Bauer, M. Corner, M., J. de Joung,
able and slow with Pt and UPt con- instrumental calibration and the O. Pagani, J. Renard, C. Weil, C., and M.
centrations of, respectively, 100 µg sampling, was always lower than DÅfincalci, Annals of Oncology 9, 1315
(1988).
L–1 and 20 µg L–1 and still 10% for Pt concentrations higher
4. U.S. Pharmacopeia 23, 1982-84, United
detectable two weeks after the last than 1 µg/L. States Pharmacopeial Convention, Inc.,
dosing. (1994).
While the uncertainty compo-
5. International Conference on Harmoniza-
nent associated to the repeatability tion, Draft Guidance on Validation of Ana-
CONCLUSION
was constant overall, the different lytical Procedures: Definitions and
An ICP-MS method for the deter- concentration levels and the sam- Terminology, Federal Register, Volume
60, pp. 1260 (March 1, 1995)
mination of Pt in biological fluids pling uncertainty was negligible,
6. FDA – Guidance for Industries, Bioanalyti-
(plasma, ultrafiltrate, and urine) of the component due to the instru- cal Method Validation, U.S. Department of
patients treated with antitumor mental calibration increased signifi- Health and Human Services Food and
agents has been developed and vali- cantly in relation to the lowest Drug Administration (May 2001).
dated. The limits of quantification concentrations of analyte. 7. ISO, Guide to the Expression of Uncer-
tainty in Measurement, International Stan-
(LOQ) in the three matrices were dards Organization, Geneva, Switzerland
1.0, 0.1, and 2.0 µg L–1, (1993).
respectively. Received December 23, 2003. 8. EURACHEM/CITAC Guide (2000), Quanti-
fying Uncertainty in Analytical Measure-
Intra-day and inter-day precision ment, 2nd Edition.
and accuracy was in good agree- 9. S.J. Bannister, Y. Chang, L.A. Sterson, and
ment with the FDA criteria for the A.J. Repta, Clin. Chem 24, 877 (1978).
validation of analytical methods. 10. E.H. Evans and J.J. Giglio, J. Anal. At.
Spectrom. 8, 1 (1993).
11. V.J. Berwick and S.L.R. Ellison, Accred.
Qual. Assur. 5, 47 (2000).
111