High Performance Liquid Chromatography Determination of Doxorubicin and Daunorubicin in Plasma Using UV Detection and Column Switching A. Mikan JZD Agrokombinat, 763 15 Slusovice, Czechoslovakia
J. Martinez Lanao*, F. Gonzalez Lopez and A. Dominguez-Gil Hurle Universidad de Salamanca, Facultad de Farmacia, Avda Campo Charro s / n , 37 007 Salamanca, Spain
A method for the determination of doxorubicin and daunorubicin in plasma is described. The plasma is injected directly into a loop column and then washed with water. After switching the injection valve, the sample is separated on a phenyl column using detection at 254 nm. The detection limit is 10 ng/mL, the coefficient of variation is 7% for 100 ng/mL of doxorubicin and 4% for 200 ng/mL of daunorubicin.
INTRODUCTION Doxorubicin (adriamycin) and daunorubicin (adriamycinol) are broad spectrum antineoplastic drugs of the anthracycline group, widely used in clinical practice since 1969 and 1966, respectively. Daunorubicin differs from doxorubicin by one hydroxyl group. Their value as anticancer drugs is restricted by their high cardiotoxicity. Therapeutic drug monitoring is an integral step for the improvement of their therapeutic effect and at the same time for controlling their cardiotoxicity. The pharmacokinetic properties of doxorubicin have recently been reviewed by Speth et al. (1988). Many different analytical techniques have been employed in the past (Florey et al., 1980), but today reversed phase high performance liquid chromatography (HPLC) with fluorimetric detection is the method of choice (Eksborg et al., 1979, Brown et al., 1981). An excellent critical review on analytical methods, physicochemical and stability properties was published by Bouma et al. (1986). Israel and coworkers (1978) have pointed out that UV detection is about three times less sensitive than the fluorimetric technique and that it should not be used for analyses of biological material due to interferring endogenous substances. The sensitivity in the visible region is about three times lower than in the UV (Eksborg el al., 1978). Since we lacked a sufficiently sensitive fluorimetric detector we developed a method using UV detection that may be useful in other drug monitoring laboratories. The extraction of anthracyclines was primarily performed by liquid/liquid extraction with all the inherent drawbacks (Eksborg et al., 1979). Another way of sample purification for HPLC is protein precipitation, resulting in a loss of sensitivity (Brown et al., 1981). In the past decade solid phase extraction on C , , Sep-Pak cartridges is the most commonly used method (Robert 1980). Doxorubicin is a very unstable drug in solution (its halflife at 37 "C is only 2 h) and the samples should thus be ready for injection in the shortest time possible. The Author to whom correspondence should be addressed
same applies to daunorubicin. From this point of view a loop column extraction is an ideal way for preventing drug deterioration during sample purification. In the present work we adapted the loop column method described by Riley et al. (1985). EXPERIMENTAL Chemicals.. Doxorubicin hydrochloride and daunorubicin as Daunoblastina were supplied by Farmitalia (Carlo Erba, Italy), orthophosphoric acid, of reagent grade, by Montplet & Esteban (Barcelona, Spain) and acetonitrile, far UV HPLC grade by Scharlau (Barcelona, Spain). Instruments. Hitachi spectrophotometer, Model U-2000 (Danbury, CT, USA); Crison pH meter, Model 501 (Barcelona, Spain), equipped with an Ingold 9811 combined glass electrode (Wilmington, MA, USA); Microspin 12 high speed centrifuge, Sorval Instruments, Du Pont (Stevenage, Herts, UK); Ultrason Selecta ultrasonic bath, Model 513; Supelco 5-8068 Vacuum Ultrafiltration System with 0.45 )*m Ultipor NX membrane; Unimetrics 1 ml gastight syringe, Model TP 5001-71. Chromatographic system: Varian HPLC Pump 2010 (Sunnyvale, CA, USA) equipped with a Rheodyne 7125 injector valve; Varian 2050 variable wavelength detector with a 10 mm (8 kL) flow cell used at 254 nm and Varian 4270 Integrator. A Varichrom UV-VIS Variable Wavelength Detector was also used during the development of the method. Columns: All columns were stainless steel 4 mm ID. Loop column (4 cm) was packed with Vydac SC 201 RP, particle size 30 - 40 pm; guard column (4 cm) packed with Nucleosil 10 C,, particle size 10 km; analytical column (30 cm) was MicroBondapack Phenyl P/N 27198 (Waters, Milford, MA, USA), particle size 10 km. Nucleosil 120-10 C I Rpacked analytical column was also used during the development of the method. The loop column was attached where the standard injection loop is normally connected, as described by Riley et rl. (1985). The guard column was attached proximal to the analytical column to protect it since plasma was injected directly into the system. The mobile phase consisted of 0.01 M phosphoric acid in distilled water+ acetonitrile 20: 80 by volume for doxorubicin and 24: 76 for daunorubicin. Prior to use the mobile phase was filtered and ultrasonically degassed. The flow rate was 2 mL/min. Calculation. Since the dose response relationship is linear throughout the concentration range of interest and the internal
0269-3879/90/0154-0156 $5.00 154 BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO 4,1990
@ John Wiley & Sons Limited, 1990
HPLC OF DOXORUBlClN A N D DAUNORUBICIN IN PLASMA
standard (daunorubicin for determination of doxorubicin and vice versa) did not improve reproducibility, we used one external standard of 200 ng/mL to determine the concentration/peak height ratio from which the unknown concentrations were calculated as directly proportional.
Procedure. The standard 100 kg/mL stock solutions of doxorubicin and of daunorubicin were prepared in 0.01 M phosphoric acid and the portions in polyethylene Eppendorf test tubes were stored at -20 "C. The stock solution was diluted prior to analysis with 0.01 M phosphoric acid at a working concentration of 200 ng/mL, injected as external standard. Blood samples were collected in heparinized test tubes, immediately centrifuged and the plasma thus obtained immediately frozen at -20 "C to prevent deterioration of doxorubicin. Prior to analysis the plasma was centrifuged for 5 min at 10 000 rpm and kept in ice-cold water. 0.5 mL of water was injected into the loop column to prepare it for the injection of 0.5 mL of clarified plasma. Proteins and other substances of n o interest were purged from the loop column by injecting 2 mL of water. The injection valve was then switched from the inject to the load position, thus introducing the sample to the guard and analytical columns.
~
~~
~
~~
~~
RESULTS AND DISCUSSION
0
I'
20
Time lmin)
Figure 1. Chromatogram of a patient's plasma containing 53 ng/mL of doxorubicin. The doxorubicin peak is marked by an arrow. There are fewer peaks between the doxorubicin peak and the main plasma peak in the chromatogram of blood donor plasma (see Fig. 2). The peaks are probably other drugs used for treatment. Separation conditions are described in the Experimental Section.
Wavelength selection
We measured the UV-VIS spectra of doxorubicin and daunorubicin in the mobile phase; in fact in 0.01 M phosphoric acid + acetonitrile (75 :25). The spectra were practically identical. There were three distinct peaks in UV region and one broad peak with two poorly separated maxima in the visible region. The absorbance of the UV peaks relative to the visible 480 nm peak were 0.7, 2.2 and 3.5 times higher at 290,254 and 233 nm respectively. On the other hand, the endogenous substances in serum start to absorb at about 300 nm, with a strong increase towards the shorter wavelengths. Our first choice was 233 nm since we wanted to achieve the highest sensitivity possible. At this wavelength the acetate buffer according to Riley et al. (1985) shows considerable absorbance and this is why we preferred to work with 0.01 M phosphoric acid. However, the absorbance of plasma components was so high that it was not possible to separate the peaks of the two anthracyclines from the plasma peak and thus UV detection failed at this wavelength. The second choice was 480 nm in order to eliminate plasma components from the chromatogram. The main plasma peak was practically eliminated in this way and the retention time of doxorubicin may be cut down to 3 - 4 min by using higher concentrations of acetonitrile (30%) in the mobile phase. The plasma concentrations to be measured for therapeutic drug monitoring are as low as 10 ng/mL. Unable to achieve this value at 480 nm due to high detector noise, we therefore returned to UV detection. At 254 nm the plasma peak was considerably smaller than at 233 nm and appeared earlier down the baseline, thus allowing separation of the peaks of doxorubicin and daunorubicin. This was achieved by reducing the acetonitrile concentration so that the anthracycline peaks did not appear earlier than 14 min after injection, @ J o h n Wiley & Sons Limited, 1990
see Fig. 1 and Fig. 2. The retention times were 14.5 and 14.3 min for doxorubicin and daunorubicin, respectively. The sensitivity at 254 nm was more than two-fold that at 480 nm. Column selection
At the beginning of our work we used an analytical column packed with Nucleosil 120-10 C,, . Occasionally there was a peak in plasma at the retention time only some 2% lower than that of the doxorubicin peak. It
0
20 Time (min)
Figure 2. Chromatogram of plasma containing 200 ng/mL of daunorubicin. Plasma from a blood bank was used since no patient was available. Separation conditions are described in the Experimental section.
BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO.4,1990 155
A. M l K A N E T A L .
was not possible to separate it from that of doxorubicin by changing the acetonitrile concentration of the mobil phase. The only remedy for avoiding this interfering peak was to change the analytical column. The peaks were well separated on the phenyl column.
Loop column performance We tested if there might be some loss of drug caused by washing the proteins out of the loop column by repeatedly injecting 0.5 mL of 200 ng/mL doxorubicin followed by 0.5, 1, 2 , 4 and 8 mL of water. There was no difference in peak areas. At the begining we had troubles with clogging of frit at the injection side since we used only 3000rev/min centrifugation. This was resolved by 10 000 rev/min centrifugation and thorough visual check of plasma prior to injection. Plasma must be clear and absolutely free of any solid particles or turbidity. Nevertheless, the frits have to be changed from time to time when the resistance of the loop column makes sample injection difficult or prevents injection at all. If the change of frits does not help, the loop column must be repacked.
Quantitative parameters We compared the recovery of 400 ng/mL doxorubicin solutions on the loop column in 0.01 M phosphoric acid and in plasma after 2 h of incubation at room temperature. The plasma peak heights were 96% those of the buffer, but this difference was insignificant at 95% probability level according to Student’s t-test. Doxorubicin can be assumed to be fully released from the proteins during the loop column extraction. The peak height/area vs concentration relationship was linear at least up to 2000 ng/mL ( r = 0.998). The within-assay reproducibility was characterized by variation coefficients of 7% for 100 ng/mL of doxorubicin and 4% for 200 ng/mL of daunorubicin. The detection limit was 10 ng/mL. Initially we used daunorubicin as internal standard for doxorubicin determination but the reproducibility without an internal standard was the same and even slightly better; we therefore stopped using the internal standard. The identification and integration of very small peaks depends on optimal tuning of integrator parameters. Despite the drawbacks of the UV detection pointed by Israel et al. (1978), we have managed to use it by suitable wavelength selection, column selection and precise mobile phase tuning.
REFERENCES Bourna. J., Beijnen, J. H., Bult, A. and Unerberg, W. J. M. (1986). Pharmaceutisch Weekblad Scientific Edition, 8, p. 109. Brown, J. E., Wilkinson, P. A. and Brown, J. R. (1981). J. Chromatogr. 226, 521. Eksborg, S., Ehrsson, H., Andersson, B. and Beran, M. (1978). J. Chromatogr. 153, 21 1. Eksborg, S., Ehrsson, H. and Andersson, I. (1979). J. Chromatogr. 164, 479. Florey. K. (1980). Analytical Profiles of Drug Substances, 9, p. 246. Academic Press, New York, USA.
156 BIOMEDICAL CHROMATOGRAPHY, VOL 4, NO 4,1990
Israel, M., Pegg, W. J., Wilkinson, P. M. and Garnick, M. B. (1978). J. Liq. Chromatogr. 1, 795. Riley, C. A,, Crorn, W. R. and Evans, W. E. (1985). Therapeutic Drug Monitoring 7 (4). 455. Robert, J. (1980). J. Liq. Chromatogr. 3, 1561. Speth. P. A. J., van Hoesel, Q. G. C. M. and Haanen, C. (1988). Clin. Pharmacokin. 15, 15.
Received 8 January 1990; accepted 20 February 1990
@ J o h n Wiley & Sons Limited, 1990
J. Martinez Lanao*, F. Gonzalez Lopez and A. Dominguez-Gil Hurle Universidad de Salamanca, Facultad de Farmacia, Avda Campo Charro s / n , 37 007 Salamanca, Spain
A method for the determination of doxorubicin and daunorubicin in plasma is described. The plasma is injected directly into a loop column and then washed with water. After switching the injection valve, the sample is separated on a phenyl column using detection at 254 nm. The detection limit is 10 ng/mL, the coefficient of variation is 7% for 100 ng/mL of doxorubicin and 4% for 200 ng/mL of daunorubicin.
INTRODUCTION Doxorubicin (adriamycin) and daunorubicin (adriamycinol) are broad spectrum antineoplastic drugs of the anthracycline group, widely used in clinical practice since 1969 and 1966, respectively. Daunorubicin differs from doxorubicin by one hydroxyl group. Their value as anticancer drugs is restricted by their high cardiotoxicity. Therapeutic drug monitoring is an integral step for the improvement of their therapeutic effect and at the same time for controlling their cardiotoxicity. The pharmacokinetic properties of doxorubicin have recently been reviewed by Speth et al. (1988). Many different analytical techniques have been employed in the past (Florey et al., 1980), but today reversed phase high performance liquid chromatography (HPLC) with fluorimetric detection is the method of choice (Eksborg et al., 1979, Brown et al., 1981). An excellent critical review on analytical methods, physicochemical and stability properties was published by Bouma et al. (1986). Israel and coworkers (1978) have pointed out that UV detection is about three times less sensitive than the fluorimetric technique and that it should not be used for analyses of biological material due to interferring endogenous substances. The sensitivity in the visible region is about three times lower than in the UV (Eksborg el al., 1978). Since we lacked a sufficiently sensitive fluorimetric detector we developed a method using UV detection that may be useful in other drug monitoring laboratories. The extraction of anthracyclines was primarily performed by liquid/liquid extraction with all the inherent drawbacks (Eksborg et al., 1979). Another way of sample purification for HPLC is protein precipitation, resulting in a loss of sensitivity (Brown et al., 1981). In the past decade solid phase extraction on C , , Sep-Pak cartridges is the most commonly used method (Robert 1980). Doxorubicin is a very unstable drug in solution (its halflife at 37 "C is only 2 h) and the samples should thus be ready for injection in the shortest time possible. The Author to whom correspondence should be addressed
same applies to daunorubicin. From this point of view a loop column extraction is an ideal way for preventing drug deterioration during sample purification. In the present work we adapted the loop column method described by Riley et al. (1985). EXPERIMENTAL Chemicals.. Doxorubicin hydrochloride and daunorubicin as Daunoblastina were supplied by Farmitalia (Carlo Erba, Italy), orthophosphoric acid, of reagent grade, by Montplet & Esteban (Barcelona, Spain) and acetonitrile, far UV HPLC grade by Scharlau (Barcelona, Spain). Instruments. Hitachi spectrophotometer, Model U-2000 (Danbury, CT, USA); Crison pH meter, Model 501 (Barcelona, Spain), equipped with an Ingold 9811 combined glass electrode (Wilmington, MA, USA); Microspin 12 high speed centrifuge, Sorval Instruments, Du Pont (Stevenage, Herts, UK); Ultrason Selecta ultrasonic bath, Model 513; Supelco 5-8068 Vacuum Ultrafiltration System with 0.45 )*m Ultipor NX membrane; Unimetrics 1 ml gastight syringe, Model TP 5001-71. Chromatographic system: Varian HPLC Pump 2010 (Sunnyvale, CA, USA) equipped with a Rheodyne 7125 injector valve; Varian 2050 variable wavelength detector with a 10 mm (8 kL) flow cell used at 254 nm and Varian 4270 Integrator. A Varichrom UV-VIS Variable Wavelength Detector was also used during the development of the method. Columns: All columns were stainless steel 4 mm ID. Loop column (4 cm) was packed with Vydac SC 201 RP, particle size 30 - 40 pm; guard column (4 cm) packed with Nucleosil 10 C,, particle size 10 km; analytical column (30 cm) was MicroBondapack Phenyl P/N 27198 (Waters, Milford, MA, USA), particle size 10 km. Nucleosil 120-10 C I Rpacked analytical column was also used during the development of the method. The loop column was attached where the standard injection loop is normally connected, as described by Riley et rl. (1985). The guard column was attached proximal to the analytical column to protect it since plasma was injected directly into the system. The mobile phase consisted of 0.01 M phosphoric acid in distilled water+ acetonitrile 20: 80 by volume for doxorubicin and 24: 76 for daunorubicin. Prior to use the mobile phase was filtered and ultrasonically degassed. The flow rate was 2 mL/min. Calculation. Since the dose response relationship is linear throughout the concentration range of interest and the internal
0269-3879/90/0154-0156 $5.00 154 BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO 4,1990
@ John Wiley & Sons Limited, 1990
HPLC OF DOXORUBlClN A N D DAUNORUBICIN IN PLASMA
standard (daunorubicin for determination of doxorubicin and vice versa) did not improve reproducibility, we used one external standard of 200 ng/mL to determine the concentration/peak height ratio from which the unknown concentrations were calculated as directly proportional.
Procedure. The standard 100 kg/mL stock solutions of doxorubicin and of daunorubicin were prepared in 0.01 M phosphoric acid and the portions in polyethylene Eppendorf test tubes were stored at -20 "C. The stock solution was diluted prior to analysis with 0.01 M phosphoric acid at a working concentration of 200 ng/mL, injected as external standard. Blood samples were collected in heparinized test tubes, immediately centrifuged and the plasma thus obtained immediately frozen at -20 "C to prevent deterioration of doxorubicin. Prior to analysis the plasma was centrifuged for 5 min at 10 000 rpm and kept in ice-cold water. 0.5 mL of water was injected into the loop column to prepare it for the injection of 0.5 mL of clarified plasma. Proteins and other substances of n o interest were purged from the loop column by injecting 2 mL of water. The injection valve was then switched from the inject to the load position, thus introducing the sample to the guard and analytical columns.
~
~~
~
~~
~~
RESULTS AND DISCUSSION
0
I'
20
Time lmin)
Figure 1. Chromatogram of a patient's plasma containing 53 ng/mL of doxorubicin. The doxorubicin peak is marked by an arrow. There are fewer peaks between the doxorubicin peak and the main plasma peak in the chromatogram of blood donor plasma (see Fig. 2). The peaks are probably other drugs used for treatment. Separation conditions are described in the Experimental Section.
Wavelength selection
We measured the UV-VIS spectra of doxorubicin and daunorubicin in the mobile phase; in fact in 0.01 M phosphoric acid + acetonitrile (75 :25). The spectra were practically identical. There were three distinct peaks in UV region and one broad peak with two poorly separated maxima in the visible region. The absorbance of the UV peaks relative to the visible 480 nm peak were 0.7, 2.2 and 3.5 times higher at 290,254 and 233 nm respectively. On the other hand, the endogenous substances in serum start to absorb at about 300 nm, with a strong increase towards the shorter wavelengths. Our first choice was 233 nm since we wanted to achieve the highest sensitivity possible. At this wavelength the acetate buffer according to Riley et al. (1985) shows considerable absorbance and this is why we preferred to work with 0.01 M phosphoric acid. However, the absorbance of plasma components was so high that it was not possible to separate the peaks of the two anthracyclines from the plasma peak and thus UV detection failed at this wavelength. The second choice was 480 nm in order to eliminate plasma components from the chromatogram. The main plasma peak was practically eliminated in this way and the retention time of doxorubicin may be cut down to 3 - 4 min by using higher concentrations of acetonitrile (30%) in the mobile phase. The plasma concentrations to be measured for therapeutic drug monitoring are as low as 10 ng/mL. Unable to achieve this value at 480 nm due to high detector noise, we therefore returned to UV detection. At 254 nm the plasma peak was considerably smaller than at 233 nm and appeared earlier down the baseline, thus allowing separation of the peaks of doxorubicin and daunorubicin. This was achieved by reducing the acetonitrile concentration so that the anthracycline peaks did not appear earlier than 14 min after injection, @ J o h n Wiley & Sons Limited, 1990
see Fig. 1 and Fig. 2. The retention times were 14.5 and 14.3 min for doxorubicin and daunorubicin, respectively. The sensitivity at 254 nm was more than two-fold that at 480 nm. Column selection
At the beginning of our work we used an analytical column packed with Nucleosil 120-10 C,, . Occasionally there was a peak in plasma at the retention time only some 2% lower than that of the doxorubicin peak. It
0
20 Time (min)
Figure 2. Chromatogram of plasma containing 200 ng/mL of daunorubicin. Plasma from a blood bank was used since no patient was available. Separation conditions are described in the Experimental section.
BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO.4,1990 155
A. M l K A N E T A L .
was not possible to separate it from that of doxorubicin by changing the acetonitrile concentration of the mobil phase. The only remedy for avoiding this interfering peak was to change the analytical column. The peaks were well separated on the phenyl column.
Loop column performance We tested if there might be some loss of drug caused by washing the proteins out of the loop column by repeatedly injecting 0.5 mL of 200 ng/mL doxorubicin followed by 0.5, 1, 2 , 4 and 8 mL of water. There was no difference in peak areas. At the begining we had troubles with clogging of frit at the injection side since we used only 3000rev/min centrifugation. This was resolved by 10 000 rev/min centrifugation and thorough visual check of plasma prior to injection. Plasma must be clear and absolutely free of any solid particles or turbidity. Nevertheless, the frits have to be changed from time to time when the resistance of the loop column makes sample injection difficult or prevents injection at all. If the change of frits does not help, the loop column must be repacked.
Quantitative parameters We compared the recovery of 400 ng/mL doxorubicin solutions on the loop column in 0.01 M phosphoric acid and in plasma after 2 h of incubation at room temperature. The plasma peak heights were 96% those of the buffer, but this difference was insignificant at 95% probability level according to Student’s t-test. Doxorubicin can be assumed to be fully released from the proteins during the loop column extraction. The peak height/area vs concentration relationship was linear at least up to 2000 ng/mL ( r = 0.998). The within-assay reproducibility was characterized by variation coefficients of 7% for 100 ng/mL of doxorubicin and 4% for 200 ng/mL of daunorubicin. The detection limit was 10 ng/mL. Initially we used daunorubicin as internal standard for doxorubicin determination but the reproducibility without an internal standard was the same and even slightly better; we therefore stopped using the internal standard. The identification and integration of very small peaks depends on optimal tuning of integrator parameters. Despite the drawbacks of the UV detection pointed by Israel et al. (1978), we have managed to use it by suitable wavelength selection, column selection and precise mobile phase tuning.
REFERENCES Bourna. J., Beijnen, J. H., Bult, A. and Unerberg, W. J. M. (1986). Pharmaceutisch Weekblad Scientific Edition, 8, p. 109. Brown, J. E., Wilkinson, P. A. and Brown, J. R. (1981). J. Chromatogr. 226, 521. Eksborg, S., Ehrsson, H., Andersson, B. and Beran, M. (1978). J. Chromatogr. 153, 21 1. Eksborg, S., Ehrsson, H. and Andersson, I. (1979). J. Chromatogr. 164, 479. Florey. K. (1980). Analytical Profiles of Drug Substances, 9, p. 246. Academic Press, New York, USA.
156 BIOMEDICAL CHROMATOGRAPHY, VOL 4, NO 4,1990
Israel, M., Pegg, W. J., Wilkinson, P. M. and Garnick, M. B. (1978). J. Liq. Chromatogr. 1, 795. Riley, C. A,, Crorn, W. R. and Evans, W. E. (1985). Therapeutic Drug Monitoring 7 (4). 455. Robert, J. (1980). J. Liq. Chromatogr. 3, 1561. Speth. P. A. J., van Hoesel, Q. G. C. M. and Haanen, C. (1988). Clin. Pharmacokin. 15, 15.
Received 8 January 1990; accepted 20 February 1990
@ J o h n Wiley & Sons Limited, 1990
