Species specific IDMS for accurate quantification of carboplatin in urine by LC-ESI-TOFMS and LC-ICP-QMS

Abstract

For the first time, ¹⁹⁴Pt isotopically enriched carboplatin has been synthesized, serving for species specific isotope dilution analysis. Carboplatin was determined in urine, implementing both LC-ICP-QMS and LC-ESI-TOFMS. Hence, a separation method was applied, compatible with both ICP-QMS and ESI-MS detection. IDMS quantification was evaluated in terms of measurement uncertainty, comparing the two mass spectrometric detection methods. The procedural limits of detection of HPLC-ICP-QMS and HPLC-ESI-TOFMS were 0.1 ng g^–1 and 15 ng g^–1, respectively. Both methodologies were successfully applied to the quantification of carboplatin in the urine of a chemotherapy patient. The obtained carboplatin concentrations agreed within their uncertainties. Uncertainty budgeting calculations revealed that the accuracy of quantification was primarily limited by the precision of isotope ratio measurement, governing the determination of the ratios R_x, R_y and R_b of the IDMS equation. For ICP-QMS, a total combined uncertainty of 5.7% (coverage factor 2) was assessed. In accordance to the impaired precision of blend ratio determination in ESI-TOFMS (precision of 5% compared to 1% in ICP-QMS), this method revealed a total combined uncertainty of 23%.

---

Introduction

Isotope dilution mass spectrometry (IDMS) is applied if the accuracy of results is of predominant analytical importance. IDMS is recognized as a primary method of measurement.¹ As a key advantage, after isotope equilibration, the method compensates for partial loss of the analytes during sample preparation and storage.² Moreover, unwanted sources of error, such as long-term drift or signal suppression, are eliminated due to the fact that only relative measurements (i.e. the signal ratio at two different isotopic masses) are necessary.^3,4 The quantification strategy is applicable for both methods of detection, i.e.ICP-MS and ESI-MS. So far, the majority of applications employing IDMS have been performed in the field of elemental mass spectrometry by inductively coupled plasma mass spectrometry (ICP-MS) and thermal ionization mass spectrometry (TIMS).⁵ However, hyphenated techniques with ICP-IDMS suffer both from lack of commercially available isotope-labeled spike compounds for species-specific isotope dilution and from the more complicated system set-up required for species-unspecific ICP-IDMS analysis.^6–8

The advantages of employing isotopically enriched compounds for quantification have been recognized in molecular mass spectrometry in specific applications, such as gas chromatography mass spectrometry (GC-MS) of persistent organic pollutants (e.g. dioxines, furans, pesticides, polycyclic aromatic hydrocarbons)^9–11 and, more recently, in liquid chromatography mass spectrometry (LC-MS), predominantly in the context of environmental and pharmaceutical analysis, but also in proteomic studies.^12,13

The integration of enriched isotopes into quantification algorithms of elemental (ICP-MS, TIMS) and molecular MS (LC-MS, GC-MS) follows substantially different concepts. The general equation of isotope dilution³ is

If the isotopic composition of the regarded element is constant in nature, eqn (1) can be simplified and the number of measured parameters can be reduced according to eqn (2):³This approach requires the measurement of four measurands, i.e. the mass of the sample (m_x) and the spike (m_y) used for preparation of the blend (gravimetric determination), the signal ratio R_b of the reference isotope and the spike isotope, and the factor for mass bias correction (K) which is determined according to eqn (3) (linear correction)

R^obs = R^true(1 + εΔM) (3)

where R^obs is the measured ratio, R^true the true ratio, ε the mass bias per mass unit and ΔM the mass difference of the measured isotopes.^14,15

The remaining parameters in eqn (2) are constants or certified values: c_y is the certified concentration of the spike isotope in the spike solution; R_x represents the constant natural isotope ratio of the reference isotope and the spike isotope; R_y is the certified ratio of the reference isotope and the spike isotope in the spike solution; f_x is the constant abundance of the spike isotope in the sample and f_y is the certified abundance of the spike isotope in the spike solution. In cases where no certified spike solution is available, the spike has to be characterized by reverse IDMS using a certified elemental standard of which the concentration is traceable to NIST, BCR or other certification bodies.^3,4

The application of this “classic” IDMS approach using eqn (1) and (2) is in most cases restricted to elemental MS and has, to our knowledge, not been employed in molecular MS. Here, the isotopically enriched spike isotope is conventionally used conceptually as an internal standard, either for quantification of the compound with natural isotopic composition or for quantification of a group of similar analytes. The concentration of the analyte in the sample is calculated by external calibration, with internal standardization by the isotopically enriched spike isotope.^16,17

Modern speciation analysis is performed by complementary use of elemental and molecular mass spectrometry. The major advantage of combining the two techniques relies on the fact that the sensitivity and accuracy of quantification is superior for ICP-MS. Accordingly, several authors claim that substances containing a hetero-element amenable to ICP-MS analysis should be quantified by this method.^18,19 However, no comparative study has been carried out yet addressing the quantification of a species by ICP-MS and ESI-MS in terms of total combined uncertainty. In the present work we have investigated different LC-MS strategies for quantification of the cancerostatic platinum compound carboplatin (cis-diammine-1,1-cyclobutanedicarboxylatoplatinum(II)) in human urine. For the first time, carboplatin isotopically enriched in ¹⁹⁴Pt has been synthesized and employed for an IDMS study. We have utilized two different mass spectrometersi.e. an elemental quadrupole based MS (ICP-QMS) and a molecular MS (ESI-TOFMS) in combination with a HPLC system. On the one hand, we compare the uncertainty of measurement of the classic IDMS approach obtained with the two different mass spectrometric techniques. On the other hand, we evaluate the figures of merit of the classic IDMS and external calibration with internal standardization by molecular mass spectrometry. In each case, the major sources of measurement uncertainty have been determined and assessed via uncertainty budgeting.

Experimental

Reagents and standards

A British Pharmacopoeia (BP) carboplatin assay standard with a certified purity of 99.9% was purchased at LGC Promochem (Wesel, Germany). 5.3 mg L^–1 stock solutions of BP-carboplatin were prepared in 150 mmol L^–1 Cl^– using sodium chloride (p.a., VWR). All further dilutions were made gravimetrically with ultrapure water. For preparation of HPLCeluents, suprapure formic acid (VWR, Darmstadt, Germany), subboiled ammonium hydroxide (Aldrich, Milwaukee, WI, USA) and methanol (HPLC gradient grade, VWR) were used. All solutions and eluents were prepared using ultrapure subboiled water. All bottles used were made of polyethylene, except eluent containers, which were made of glass.

Synthesis of carboplatin enriched in ¹⁹⁴Pt

Metallic ¹⁹⁴Pt was purchased from Chemotrade (Duesseldorf, Germany), enrichment of ¹⁹⁴Pt was 87.34%. HCl (DonauChem, Vienna, Austria) was of chemical pure grade, HNO₃ and KHCO₃ (VWR) were of analytical and of chemical pure grade, respectively, KCl and hydrazine dihydrochloride (Fluka) were of analytical grade. All reactions were carried out under protection from light, and for stirring, a glass coated magnetic stirrer was used.

Synthesis of potassium tetrachloroplatinate (K₂PtCl₄)

¹⁹⁴Pt powder (490 mg, 2.52 mmol) was dissolved in a mixture of 10 mL concentrated HCl and 3 mL of concentrated HNO₃ at 50 °C. The red solution was reduced to ca. 2 mL; HNO₃ was removed by adding concentrated HCl (15 mL) and reducing the volume to 2 mL. This procedure was repeated five times. Eventually, the majority of HCl was withdrawn by addition of hot water (30 mL) and reduction of the volume to ca. 10 mL (six times). After cooling the solution in an ice bath, saturated aqueous KCl was added until complete precipitation of potassium hexachloroplatinate (K₂PtCl₆). KHCO₃ (ca. 20 mg) was added followed by stirring for 30 min at 0 °C. The precipitate was centrifuged, washed once with 1.5 M KCl and twice with water, and dried in vacuum to yield 986 mg of K₂PtCl₆ as a yellow powder. The product (986 mg, 2.03 mmol) was suspended in 150 mL of water; hydrazine dihydrochloride (107 mg, 1.02 mmol) was added, and the reaction mixture was heated up slowly under stirring, until the color changed from yellow to red. The solution was further heated and kept at boiling for 10 min. After cooling down to room temperature, residual solids of K₂PtCl₆ and platinum black were filtered off. The K₂PtCl₄ containing solution was divided into two parts in a ratio of 1 : 2. The greater volume (solution 1) was used for the following synthesis.

Synthesis of PtI₂(NH₃)₂

Solution 1 was adjusted to pH 7 with 1 M KOH. KI (1.200 mg, 7.23 mmol) was added and the dark suspension was stirred for 30 min at room temperature. Thereafter, 0.2 mL of 25% aqueous NH₃ was added drop wise. The reaction mixture was stirred for 3 h and stored at 4 °C overnight. The product, a greenish yellow solid, was washed with water three times and several times with acetone until the filtrate remained colorless. After drying in vacuum, 285 mg of PtI₂(NH₃)₂ as a yellow powder, could be obtained.

Synthesis of carboplatin (C₆H₁₂N₂O₄Pt)

PtI₂(NH₃)₂ (285 mg, 0.59 mmol) was suspended in 10 mL of water; AgNO₃ (193 mg, 1.14 mmol) was added and the suspension was stirred for 24 h. AgI was filtered off and the clear solution of [Pt(NH₃)₂(H₂O)₂](NO₃)₂ was divided into two equal volumes; one part was further used. This activated platinum complex was mixed with a solution of 1,1-cyclobutane dicarboxylic acid (42 mg, 0.29 mmol) and 1.0 M NaOH (580 µl, 0.58 mmol) in 2 mL of water and stirred at room temperature for 2 h. The volume was reduced to 5 mL under reduced pressure at 40 °C and the product was stored at room temperature overnight. A white precipitate was isolated, which was washed with water twice and dried in vacuum. Yield: 66 mg (21%, refer to Pt metal).

Sampling of patient urine

Urine samples were collected at the oncology ward of the Vienna University Hospital from patients treated with cancerostatic platinum compounds. The samples were frozen immediately after collection and stored at –20 °C until analysis.

Preparation of the blend

All blends, i.e. mixtures of solutions of BP-carboplatin and the prepared carboplatin spike, as well as mixtures of diluted human urine and the prepared carboplatin spike, were prepared gravimetrically in a temperature controlled clean room at 20 °C. All dilutions of carboplatin stock solutions and human urine were made with ultrapure water.

Instrumental

HPLC-ICP-MS . An inert HPLC gradient system (Rheos 2000, Flux Instruments AG, Basel, Switzerland) in combination with a metal-free autosampler (CTC Analytics AG, Zwingen, Switzerland) was used throughout the study. Chromatographic separation of carboplatin was performed by reversed phase chromatography, based on a pentafluorophenylpropyl functionalized silica (Discovery HS F5, Supelco, Bellefonte, PA, USA). The HPLC operating parameters are listed in Table 1.

Table 1 Chromatographic conditions of carboplatin separation

HPLC column	Discovery HS-F5 150 × 2.1 mm
Eluent	A	20 mM Ammonium formiate (4 v/v% MeOH)
	B	H2O
	C	MeOH
Flow rate		0.25 mL min^–1
Injection volume		3 µL
Oven temperature		45 °C
Gradient program
Time/min	A(%)	B(%)	C(%)
0	50	50	0
2	50	50	0
6	50	40	10
8	50	40	10
8.5	50	50	0
23	50	50	0

---

Elemental speciation of platinum was performed by coupling the HPLC system to a quadrupole based ICP-MS (Elan DRC II, PESCIEX, Ontario, Canada). A detailed discussion of the chromatographic separation and the hyphenated system can be found elsewhere.²⁰ The ICP-QMS operating parameters are listed in Table 2a.

Table 2 Operation parameter for carboplatin measurement via IDMS

(a) ICP—DRCMS operation parameters
Nebulizer	PFA-ST microconcentric
Spray chamber	Cyclonic
Nebulizer gas flow	1.0 L min^–1
Aux. gas	1.15 L min^–1
Plasma gas	15 L min^–1
Ion lens voltage	7.5 V
ICP RF power	1250 W
Scan mode	Peak hopping
m/z measured	193.96, 194.97, 195.97
Dwell time per isotope	50 ms
Datapoints	6 s^–1

---

(b) ESI-TOFMS operation parameter
Nebulizer gas pressure	20 psig
Drying gas flow	10 L min^–1
Gas temperature	350 °C
Capillary voltage	4000 V
Fragmentor voltage	120 V
Skimmer voltage	60 V
Octopole RF voltage	250 V
Transients/scan	105 792
Scans	2 s^–1
Extracted mass ranges	371.0–371.1, 372.0–372.1, 373.0–373.1

---

HPLC-ESI-TOFMS. For molecular mass spectrometric measurements, a time-of-flight mass spectrometer with electrospray ionization (Agilent 6210 time-of-flightLC-MS, Agilent Technologies, Palo Alto, California, USA) was combined with the HPLC, using identical HPLC instrumental set-up and operation parameters as in the case of the HPLC-ICP-QMS coupling described above. The ESI-TOFMS operating parameters are given in Table 2b.

Data evaluation

All ICP-QMS signals were corrected for detector dead time using the built in dead time correction feature of the ICP-MS instrument software (Ver. 3.0, PESCIEX, Ontario, Canada), following dead time determination by a procedure (Method 2) proposed by Nelms et al.²¹ Generation and export of HPLC-ICP-QMS chromatograms was carried out using Chromlink (Version 2.1, PE SCIEX) in combination with Turbochrom (Version 6.2, PE-SCIEX). Chromeleon software (Version 6.4, Dionex, Sunnyvale, CA, USA) was used for integration of all chromatographic data. The isotopic ratio (R_b) of the blend was calculated by dividing the integrated peak areas obtained for m/z 194 and m/z 195.

HPLC-ESI-TOFMS chromatograms were processed by software extraction (Analyst QS 1.1, Applied Biosystems/MDSSCIEX, Concord, Canada) of m/z 371.0–371.1 (corresponding to Ptm/z 194) and 372.0–372.1 (corresponding to Ptm/z 195). The integrated peak areas obtained for the mass range 372.0–372.1 were further corrected for the contribution of the isotopes ²H, ¹³C, ¹⁵N and ¹⁷O using eqn (4),

a^corr_m+1 = a^means_m+1 – (a^meas_m (n_HA_²H + n_CA_¹³C + n_NA_¹⁵N + n_OA_¹⁷O)) (4)

where a^corr_m+1 represents the corrected peak area measured at the mass range 371.0–371.1, and a^meas_m and a^meas_m+1 represent the peak area measured at the mass ranges 370.0–370.1 and 371.0–371.1, respectively. n is the number of respective atoms of an element in the carboplatin molecule and A is the natural abundance of the respective isotopes.

The isotope ratio (R_b) of the blend was calculated by dividing the values of a^corr_m+1 by a^meas_m.

IDMS calculations

Characterization of the prepared carboplatin spike was accomplished by reverse species specific on-line IDMS by HPLC-ICP-QMS. The concentration of the spike (c_{x(reverse IDMS)}) was calculated as follows (eqn (5)). The accurate determination of c_{x(reverse IDMS)} requires the measurement of the mass of the prepared carboplatin spike solution (m_{x(reverse IDMS)}) and the—natural—certified BP-carboplatin solution (m_{y(reverse IDMS)}) used for preparation of the blend (gravimetric determination), the signal ratio R_b of the reference isotope (¹⁹⁶Pt) and the spike isotope (¹⁹⁴Pt), the isotope ratio of the reference isotope and the spike isotope in the carboplatin spike solution (R_{x(reverse IDMS)}), as well as the ratio of the reference isotope and the spike isotope in the certified BP-carboplatin solution (R_{y(reverse IDMS)}). f_x is the measured abundance of the spike isotope in the carboplatin spike solution and f_y is the measured abundance of the spike isotope in the certified BP-carboplatin solution. c_{y(reverse IDMS)} is the concentration of the gravimetrically prepared BP-standard solution. In the case of eqn (5), no mass bias correction of the measured ratios is necessary if all isotope ratios are measured under the same conditions (blend measurement bracketed by spike and standard measurement).

The quantification of carboplatin in human urine via species specific on-line IDMS by HPLC-ICP-QMS and HPLC-ESI-TOFMS was performed according to eqn (2) using the carboplatin spike concentration determined by the reverse IDMS procedure above.

Uncertainty of measurement

The total combined uncertainty of the ICP-MS results was evaluated via the Kragten approach²² of error propagation according to EURACHEM/CITAC,²³ using the equations described above. All input quantities were associated with the standard uncertainties given in the respective certificates or with experimentally determined standard uncertainties. All gravimetric dilution steps were included in the uncertainty budgets. The GUM-workbench software (Metrodata G.m.b.H., Grenzach-Wyhlen, Germany) was applied for all uncertainty calculations and for uncertainty budgeting.

Results and discussion

Measurement of on-line isotope ratios by HPLC-ICP-QMS and HPLC-ESI-TOFMS

A fit-for-purpose chromatographic method was applied for analysis of carboplatin in human urine. The method is capable of separating carboplatin from its major metabolitediamminediaquaplatinun(II) and inorganic platinum impurities.²⁰ Under the conditions described in Table 1, carboplatin eluted at a retention time of 3 min. The void volume of the system was 1.5 min. Fig. 1 shows the HPLC-ICP-QMS (Fig. 1a) and HPLC-ESI-TOFMS (Fig. 1b) chromatograms obtained for standard solutions containing 0.53 and 5.3 µg g^–1carboplatin, respectively. For ESI-TOFMS, the masses corresponding to the isotopic pattern obtained for [M + H]⁺ were extracted from the total ion current (TIC). The extracted mass ranges are listed in Table 2. Both MS instruments were combined with the identical chromatographic set-up and methodology. ICP-QMS was superior in sensititvity by a factor of 100. A comparison of the two hyphenations in terms of peak-broadening revealed a slightly better value for the TOF system. As a matter of fact, the spray chamber causes peak asymmetry (tailing) in ICP-QMS.

Fig. 1 (a) HPLC-ICP-QMS chromatogram of a carboplatin solution obtained from a British Pharmacopoeia (BP) Commission laboratory. (b) HPLC-ESI-TOFMS chromatogram of a carboplatin solution obtained from a BP Commission laboratory.

As can be observed in Table 3, the mass accuracy of the LC-ESI-TOFMS determination ranged an excellent 1.5–2 ppm for 6 replicate measurements, without using the reference mass option of the instrument. Evidently, for determination of isotopic ratios, mass stability is an absolute prerequisite.

Table 3 Mass accuracy of LC-ESI-TOFMS measurements of BP-carboplatin standard (5.3 µg g^–1)

Replicate	m/z	m/z	m/z	m/z
1	371.0501	372.0525	373.0529	375.0548
2	371.0497	372.0521	373.0520	375.0553
3	371.0487	372.0512	373.0509	375.0542
4	371.0494	372.0509	373.0516	375.0538
5	371.0489	372.0512	373.0514	375.0552
6	371.0489	372.0505	373.0514	375.0547
Average/m/z	371.0493	372.0514	373.0517	375.0547
SD/m/z	0.00055	0.00075	0.00069	0.00058
RSD(ppm)	1.5	2.0	1.8	1.5
Theoret. value/m/z	371.0497	372.0518	373.0519	375.0549
Deviation/m/z	–0.0004	–0.0004	–0.0002	–0.0002
Deviation(ppm)	–1.0	–1.0	–0.6	–0.5

---

Next, both mass spectrometric systems were evaluated for measurement of isotope ratios on transient signals. 5 repetitive injections of the carboplatin standards on the LC-ICP-QMS and the LC-ESI-TOFMS system, respectively, were evaluated in terms of isotopic ratio measurement precision (see Table 4). While the ratios of integrated peak areas revealed precisions of 5% for ESI-TOFMS, precisions of <1% could be achieved in ICP-QMS detection. The lower precision of the ESI-TOFMS system could be predominantly attributed to the lower short term stability of electrospray ionization, compared to the ICP-QMS system tested with the used analytical settings (HPLC flow, eluent composition). Since it was previously shown that the mass accuracy was 2 ppm in the worst case for an analogue experiment, the use of high mass resolution was not contributing significantly to the only moderate ratio measurement precision. The stability of ESI-TOFMS measurement could only be improved by optimizing the separation for electro-spray ionization, i.e. decreasing the eluent flow rate (with the used analytical set-up 250 µL min^–1) and using other eluents, more acidic than formiate buffer systems.

Table 4 Precision of on-line isotope ratio measurements of carboplatin by LC-ESI-TOFMS and LC-ICP-QMS. The concentration of the carboplatin standard was 0.53 and 5.3 µg g^–1 corresponding to an injected amount of 1.6 and 16 ng carboplatin, respectively

	LC-ICP-QMS		LC-ESI-TOFMS
Replicate	^195Pt/^194Pt	^196Pt/^194Pt	^372Carbo/^371Carbo	^373Carbo/^371Carbo
1	1.044	0.786	0.993	0.714
2	1.027	0.784	1.015	0.781
3	1.025	0.770	0.973	0.720
4	1.036	0.783	1.112	0.836
5	1.028	0.775	1.008	0.713
Average	1.032	0.780	1.020	0.753
SD	0.008	0.007	0.054	0.055
RSD(%)	0.8	0.9	5.3	7.3
Theor. ratio	1.026	0.766	1.026	0.766

---

Characterization of carboplatin-spike via reverse species specific on-line IDMS by HPLC-ICP-QMS

¹⁹⁴Pt isotopically enriched carboplatin was synthesized for the first time according to the protocol described in the Experimental section. Platinum metal enriched in ¹⁹⁴Pt formed the base material for carboplatin spike production. Although the metal spike was certified for its isotopic composition and purity, the synthesized carboplatin product enriched in ¹⁹⁴Pt had to be characterized separately in terms of isotopic composition and concentration, this being a prerequisite for accurate IDMS analysis.

Species specific on-line reverse IDMS was performed using the previously described LC-ICP-QMS set-up. Fig. 2 shows the HPLC-ICP-QMS chromatogram obtained for the synthesized carboplatin spike enriched in ¹⁹⁴Pt dissolved in 150 mM NaCl solution. The chromatogram shows the signals of enriched carboplatin (3.0 min), as well as those of two minor impurities (1.8 min; 4.5 min) enriched in ¹⁹⁴Pt. As published elsewhere, carboplatin is stable in 150 mM NaCl,²⁴ accordingly these impurities are by-products of carboplatin synthesis.

Fig. 2 HPLC-ICP-QMS chromatogram of a spike solution synthesized according to the procedure given in the experimental section.

The quantification of the spike followed eqn (5). As a reverse standard, certified carboplatin from a BP Commission Laboratory was employed. HPLC-ICP-QMS measurement of the spike, the gravimetrically prepared blend (mixture of spike solution and BP-standard solution) and the BP-standard solution gave the values of R_{y(reverse IDMS)}, R_{b(reverse IDMS)} and R_{x(reverse IDMS)}, respectively. According to eqn (5), the spike concentration, c_{x(reverse IDMS)}, was 1916 ± 28 ng g^–1carboplatin. A purity of 86.54 ± 1.26% (n = 3 independently prepared blend solutions) was determined for the ¹⁹⁴carboplatin spike.

Quantification of carboplatin in human urine via species specific on-line IDMS by HPLC-ICP-QMS and HPLC-ESI-TOFMS

The two methods were compared in terms of their analytical figures of merit for quantification of carboplatin in human urine. The limits of detection of the two methods were quantified as the three-fold standard deviation of the blank signal measured at m/z 194 and 371.0–371.1 in the retention window of carboplatin. The LODs of HPLC-ICP-QMS and HPLC-ESI-TOFMS were 0.1 ng g^–1 and 15 ng g^–1carboplatin, which corresponds to absolute LODs of 0.8 fmol and 120 fmol, respectively.

The measurement of a gravimetrically prepared solution containing 5.24 µg g^–1BP-carboplatin by HPLC-ICP-MS revealed an excellent IDMS result of 5.19 ±0.1 µg g^–1 (n = 6), demonstrating the accuracy of the SSIDMS approach.

To evaluate the contribution of interferences to the carboplatin signal in human urine, carboplatin-free urine was measured. The high mass accuracy and small mass window of ESI-TOFMS and the separation power of HPLC reduce the occurrence of unwanted interferences in ESI-MS. Accordingly, we did not observe any interferences leading to false positive detection ofcarboplatin.

Fig. 3 shows two chromatograms obtained for diluted, spiked urine from a chemotherapy patient. The sample was diluted 1 : 2000 (Fig. 3a) and 1 : 30 (Fig. 3b) for HPLC-ICP-QMS and HPLC-ESI-MS analysis, respectively.

Fig. 3 (a) HPLC-ICP-QMS chromatogram of diluted, spiked patient urine containing 118 ng g^–1carboplatin eluting at 3.0 min. The solution represents the blend of the on-line IDMS procedure. (b) HPLC-ESI-TOFMS chromatogram of diluted, spiked patient urine containing 8.33 µg g^–1carboplatin. The solution represents the blend of the on-line IDMS procedure.

The areas of the integrated carboplatin peak were divided to obtain R_b for further IDMS calculations. The carboplatin concentrations in the measured solutions ranged at 0.1 (Fig. 3a) and 8 µg g^–1 (Fig. 3b). In Fig. 3a the platinum signal at 1.8 min represents an impurity of the synthesized carboplatin spike, whereas the peak at 3.5 min represents diamminediaquaplatinun(II), which is the major degradation product of carboplatin. The impurity at 1.8 min showed the isotope ratio of the carboplatin spike. The isotope ratio of ¹⁹⁶Pt/¹⁹⁴Pt of diamminediaquaplatinun(II) corresponded to a blend isotope ratio different from the blend ratio of the carboplatin peak. Accordingly, diamminediaquaplatinun(II) was present in the sample and, more importantly, partially formed after addition of the isotopically enriched carboplatin standard and dilution. Degradation occurred during autosampler dwell time, since the samples were diluted and then measured in an autosampler sequence routine lasting several hours. Hence, in this specific application IDMS provides the only accurate quantification strategy, since the accuracy is not compromised by degradation upon sample dilution and storage.

On-line IDMS quantification of three independently processed blends revealed the carboplatin concentrations listed in Table 5. The average values obtained by the two MS-procedures correspond within their standard uncertainties. Compared to ESI-TOFMS detection, ICP-QMS was superior in terms of repeatability, a fact which is directly related to the differing precisions obtained for R_b (Table 4). For N = 3 determinations, standard uncertainties of 1.3% and 7% were obtained for ICP-QMS and ESI-TOFMS, respectively.

Table 5 Concentrations and precisions of carboplatin analysis in human urine obtained by species specific IDMS employing LC-ESI-TOFMS and LC-ICP-QMS and the IDMS procedure described in the introduction (eqn 2)

	Carboplatin concentration/µg g^–1
	LC-ICP-QMS	LC-ESI-TOFMS
	235	270
	236	235
	230	245
Average	234	250
SD	3	18
RSD(%)	1.3	7

---

Uncertainty of measurement of the applied quantification strategies

Uncertainty determination and budgeting was performed for comparison of the two different measurement techniques (HPLC-ICP-QMS, HPLC-ESI-MS) considering the standard uncertainties of all input quantities given in eqn (2), including all gravimetric dilution steps. The concepts of EURACHEM/CITAC for quantifying and assessing measurement uncertainty were applied using the Kragten approach of error propagation.²² Uncertainty budget calculations offer an ideal tool for understanding which input quantities contribute primarily to the total combined measurement uncertainty. Hence, it can be used as tool to understand whether the accuracy of quantification can be improved by optimization of sample pretreatment steps or is limited by instrumental parameters, such as isotopic ratio measurement precision.

The total combined uncertainty of carboplatin IDMS via HPLC-ICP-QMS was 5.7% (coverage factor 2). The concentration of the spike, c_y, the precision of R_b and the precision of the mass bias correction factor K were identified as major sources of error. Similar to the uncertainty of R_b, the two remaining sources of error can be related to the precision of isotope ratio measurement. The uncertainty of c_y is mainly related to the ratio precisions of reverse IDMS, whereas the uncertainty of K is dependent on the precision of R_x measured in the BP-carboplatin standard. The contribution of gravimetric dilution steps was found to be negligible.

IDMS quantification by ESI-TOFMS was carried out in analogy to the ICP-QMS procedure. Again, c_y determined by reverse IDMS (ICP-QMS) was considered for the calculation. The ratios R_x and R_y in the IDMS equation were derived from the natural abundance of the respective isotopes and from ICP-QMS determination after mass bias correction, respectively. R_b was determined with a precision of 5% and mass bias corrected in analogy to the ICP-QMS procedure (linear mass bias correction). The quantification of carboplatinvia HPLC-ESI-TOFMS revealed a total combined uncertainty of 23% (coverage factor 2). The higher uncertainty of the method was in accordance to the lower procedural repeatability, as listed in Table 5. After uncertainty budgeting, the identical sources of error in ICP-QMS analysis were identified. Again, the precision of isotopic ratios determined the overall uncertainty. As a consequence of the lower precision of isotope measurement achievable in ESI-TOFMS, R_b and K were the main contributing input quantities. The contribution of c_y is minor since this parameter was determined by the more precise ICP-QMS. As a consequence, the accuracy of quantification by ESI-TOFMS can only be improved by improving the signal stability. As already mentioned before, this could be accomplished by optimizing the electrospray ionization; i.e. by optimizing the flow and the eluent composition with regard to signal stability.

Generally, the key advantage of isotopically enriched standards in molecular mass spectrometry, making them a prerequisite for accurate analysis on a routine basis, is the correction of long term variations. Conventionally, IDMS in combination with ESI-MS implements isotopically enriched standards as internal standards. This is due to the fact that such standards constitute the most accurate possibility of internal standardization in ESI-MS, featuring identical molecular structures as the target analytes and hence equivalent ionization yield. The selection of suitable internal standards in ICP-QMS is more straightforward in this regard. Accordingly, in a next step, we assessed the total combined uncertainty of an external calibration by LC-ESI-TOFMS using the isotopically enriched carboplatin as the internal standard. The external calibration comprised 6 standards. A total combined uncertainty of 20% (coverage factor 2) was calculated. As can be seen in Fig. 4, the precision of the normalized signal regarding the standard and the sample are the major contributions to the total combined uncertainty.

Fig. 4 Expanded uncertainty of measurement and assessment of major input quantities obtained.

This finding shows that accuracy of quantification is limited by the isotope ratio measurement precision, regardless which IDMS calculation strategy is applied.

Conclusion

IDMS is the quantification method of choice in the case of metallodrugs in urine. Otherwise, accuracy of quantification is obfuscated by the fact that these analytes already show degradation upon simple sample preparation steps, such as dilution. Summarizing both LC-ICP-QMS and LC-ESI-TOFMS can be applied for the quantitative analysis of carboplatin in human urine. Isotope dilution could be implemented for both detection methods. However, it became clear that in speciation analysis dealing with compounds amenable to ICP-MS analysis, quantification by ICP-MS will always be the method of choice; especially in cases were the accuracy of quantification is the analytical focus. For the tested species specific IDMS application, LC-ICP-QMS was clearly superior not only in sensitivity (by a factor of 100) but also in terms of measurement uncertainty. We think that even if in the future the sensitivity of ESI-MS instrumentation is improved, accuracy of ICP-MS determination will remain superior, since the poor precision of isotopic ratio measurement is an inherent feature of electrospray ionization.

Acknowledgements

Financial support was granted through the Austrian Science Fund (FWF-Project P16089-N03: “Speciation of cancerostatic Pt compounds in the environment”). Prof. Mader (Vienna University Hospital) is gratefully acknowledged for sampling of patient urine.

References

1. Comité Consultatif pour la Quantité de Matiere, Rapport de la 1er session, BIPM, Sèvre, France, 1995.
2. J. Vogl, Calibration strategies and quality assurance [for mass spectrometers], in Inductively Couples Plasma Mass Spectrometry Handbook, ed. S.M. Nelms, Blackwell Publishing Ltd, Oxford UK, 2005, pp. 147–181.
3. F. Vanhaecke, L. Moens and P. Taylor, Use of ICP-MS for isotope ratio measurements, in Inductively coupled plasma spectrometry and its applications, ed. S. Hill, Sheffield Academic Press, Sheffield, England, 1999, pp. 145–207.
4. K. G. Heumann, Isotope dilution mass spectrometry, in Inorganic Mass spectrometry, ed. F. Adams, Wiley, New York, 1988, pp. 301–376.
5. I. T. Platzner, Modern isotope ratio mass spectrometry, J. Wiley & Sons, West Sussex, England, 1997, 405–409.
6. K. G. Heumann, Anal. Bioanal. Chem., 2004, 378, 318–329.
7. P. Rodriguez-Gonzalez, J. M. Marchante-Gayon, I. J. Garcia Alonso and A. Sanz-Medel, Spectrochim. Acta, Part B, 2005, 60, 151–207.
8. G. Koellensperger S. Hann, J. Nurmi, T. Prohaska and G. Stingeder, J. Anal. At. Spectrom., 2003, 18, 1047–1055.
9. USEPA Office of Water Regulations & Standards, Method 1613, Tetra- through Octachlorinated Dioxins & Furans by Isotope Dilution HRGC/HRMS, US Government Printing Office, Washington, DC, 1990, Revision A.
10. D. B. Barr, J. R. Barr, V. L. Maggio, R. D. Whitehead, Jr, M. A. Sadowski, R. M. Whyatt and L. L. Needham, J. Chromatogr., B, 2002, 778, 99–111.
11. B. Bushby, A. Fernandes, D. Wallace and M. Kibblewhite, Sci. Total Environ., 1993, 135, 81–94.
12. B. J. Vanderford and S. A. Snyder, Environ. Sci. Technol., 2006, 40, 7312–7320.
13. V. Mayya and D. K. Han, Exp. Rev. Prot., 2006, 3, 597–610.
14. P. D. P. Taylor, P. de Bièvre, A. J. Walder and A. Entwistle, J. Anal. At. Spectrom., 1995, 395–398.
15. L. Yang and R. E. Sturgeon, J. Anal. At. Spectrom., 2003, 18, 1452–1457.
16. T. Delatour, Anal. Bioanal. Chem., 2004, 380, 515–523.
17. P. G. Jung, B. Kim, S.-R. Park, H.-Y. So, L. H. Shi and Y. Kim, Anal. Bioanal. Chem., 2004, 380, 782–788.
18. R. Lobinski, D. Schaumloffel and J. Szpunar, Mass Spectrom. Rev., 2006, 25, 255–289.
19. A. Sanz-Medel, M. Montes-Bayon and M. Luisa Fernandez Sanchez, Anal. Bioanal. Chem., 2003, 377, 236–247.
20. S. Hann, Z. Stefánka, K. Lenz and G. Stingeder, Anal. Bioanal. Chem., 2005, 381, 405–412.
21. S. Nelms, C. Quetel, T. Prohaska, J. Vogl and P. Taylor, J. Anal. At. Spectrom., 2001, 16, 333–338.
22. J. Kragten, Analyst, 1994, 119, 2161–2166.
23. EURACHEM, Quantifying Uncertainty in Analytical Measurement, London, UK, 2nd edn, 2000, pp. 4–31.
24. U. Frey, J. D. Ranford and P. J. Sadler, Inorg. Chem., 1993, 32, 1333–1340.

---