Extraction of tamoxifen and its metabolites from formalin-fixed, paraffin-embedded tissues: an innovative quantitation method using liquid chromatography and tandem mass spectrometry

Tamoxifen, tamoxifen-d5, 4-hydroxy-tamoxifen (pure Z-trans-isomer, >98 % Z), 4-hydroxy-tamoxifen-d5, N-desmethyl-tamoxifen hydrochloride, N-desmethyl-tamoxifen-d5, N-desmethyl-4-hydroxy-tamoxifen (1:1 E/Z mixture) and N-desmethyl-4-hydroxy-tamoxifen-d5 (1:1 E/Z mixture) were obtained from Toronto Research Chemicals (North York, ON, Canada). Optima-grade acetonitrile, methanol, water and certified ACS xylene were purchased from Fisher Scientific (Fair Lawn, NJ, USA). Formic acid and ammonium acetate were from Sigma-Aldrich (Oakville, ON, Canada). Bond Elut^® C2 (100 mg, 3 mL) solid-phase extraction (SPE) cartridges were obtained from Agilent Technologies (Santa Clara, CA, USA).

Stock solutions of 4-OH, endoxifen, TAM and DMT were prepared individually at 1 mg/mL in methanol. A mixture of working standards was prepared by diluting the stock solutions in methanol at 0.1–100 ng/mL for 4-OH and endoxifen, and 1–1,000 ng/mL for TAM and DMT. Standard curves were prepared by spiking known concentrations of each mixture to non-TAM FFPE tissues (paraffin tissue blocks from patients prior to TAM treatment). Concentrations were corrected for tissue mass (~25 mg) and expressed in nanogram of TAM or metabolites per gram of breast tumor tissue extracted from the tissue blocks. The concentration ranges for the standard curves were 0.4–200 ng/g for 4-OH and endoxifen, and 4–2,000 ng/g for TAM and DMT.

Deuterated internal standards (I.S.) were used for quantitation since they closely resembled the analytes in extraction and chromatographic properties, and also corrected for the loss of analytes during sample preparation. A mixture of I.S. working solution consisting of tamoxifen-d5, 4-hydroxy-tamoxifen-d5, N-desmethyl-tamoxifen-d5 and N-desmethyl-4-hydroxy-tamoxifen-d5 was prepared by diluting each individual stock solution (1 mg/mL) in methanol to a concentration of 50 ng/mL. A final concentration of 10 ng/mL (equivalent to 400 ng/g per analysis) was added to each FFPE tissue sample.

In-house quality controls (QC) were prepared at 100 ng/g for 4-OH, endoxifen and 1,000 ng/g for TAM, DMT in non-TAM FFPE tissues. The QC was processed with each batch of paraffin tissue samples to assess the accuracy of the analytical method.

This study was approved by the Conjoint Health Research Ethics Board. FFPE tissue blocks were acquired from the archives of the pathology department at Foothills Medical Centre (Calgary, AB, Canada). Patients who had estrogen-receptor positive breast cancers diagnosed from 1990 to 2010 and treated with standard doses of TAM (20 mg/day of TAM for at least 6 months until time of biopsy with a treatment plan of 5 years on TAM) were selected for investigation. Patients were divided into two groups categorized as breast tumor recurrence (n = 16) and non-recurrence (n = 8). Non-recurrent breast cancer patients were classified as those who developed suspicious breast lesions during their follow-up visit after several months of TAM treatment but had a benign pathology diagnosis upon breast tissue biopsy (range 6–26 months).

Liquid chromatography was performed on a 1200 SL series LC system consisting of a binary pump, well-plate autosampler and thermostated column compartment (Agilent Technologies). Separation of TAM and metabolites was carried out using a Zorbax SB-C18 column (150 × 2.1 mm i.d., 3.5 μm particle size, Agilent Technologies) with a column temperature of 40 °C. Mobile phase consisted of 0.2 % formic acid and acetonitrile (60:40) at a flow rate of 0.2 mL/min with a gradient of 40–90 % of acetonitrile over 5 min. About 30 μL of the sample extract was injected onto the LC system. The total run time was 12 min including column equilibration.

MS detection of the analytes was accomplished by a 6410 triple-quadrupole mass spectrometer equipped with an electrospray ion source operating in the positive ion mode (Agilent Technologies). High purity nitrogen was used as the drying gas at a flow rate of 11 L/min with a gas temperature of 350 °C, and the nebulizer pressure was set at 35 psi. The capillary voltage was set at 4,000 V for each compound; however, the fragmentor voltages and the collision energies were operated at different settings for optimization of each analyte (Table 1). Quantitation was performed in multiple-reaction monitoring (MRM) mode, and the transitions of the precursors to product ions are summarized in Table 1. Data acquisition and analysis were performed using Mass Hunter Software v.B.02.01 (Agilent Technologies).

All experimental data were processed using Microsoft^® Excel 2007 software and expressed as mean ± SEM. The coefficient of variation (CV) was expressed as the ratio of the standard deviation to the mean.

Statistical analysis was performed using two-tail unpaired Student’s t test at 95 % confidence interval (GraphPad Prism v.2.0) for comparison between the two groups of breast cancer patients. All results were considered significant at p values of less than 0.05.

The assay was linear over the range of 0.4–200 ng/g for 4-OH/endoxifen and 4–2,000 ng/g for TAM/DMT. Mean correlation coefficients (R ²) were between 0.9995 and 0.9999 (n = 3). The slopes of the standard curves for each compound were consistent with coefficient of variation (CV) between 2.1 and 8.7 % (data not shown). Quantitation of the analytes in FFPE tissues obtained from TAM-treated patients was achieved using peak area ratios of each individual compound to I.S. and was interpolated from the standard curves.

The MRM technique used in our LC–MS/MS assay provides a high degree of sensitivity and specificity. Interference from substances other than those being analyzed is not likely to occur due to the choice of specific parent–daughter fragmentation ion and retention time monitoring. Figure 1 demonstrates that TAM and its metabolites spiked in non-TAM FFPE tissues at the LOQ and QC levels were highly distinguishable from the non-spiked. The LOQ of our assay was set at 4 ng/g for TAM and DMT, but 0.4 ng/g was used as the lowest quantitation point for 4-OH and endoxifen since these two metabolites have lower concentration in vivo [20]. LOD of the assay was obtained based on the concentrations that produced a signal-to-noise (S/N) ratio of ≥3 to deduce the presence of the analytes in FFPE tissue. Our results show that 2 ng/g was the LOD for TAM and DMT, whereas 0.2 ng/g was observed for 4-OH and endoxifen (data not shown).

It should be noted that two chromatographic peaks were observed for endoxifen in the spiked FFPE tissue because the reference compound of endoxifen was a racemic mixture of E-endoxifen and Z-endoxifen. Additionally, a small peak of E-4-OH was identified probably due to the impurity of cis-isomer in the standard (Fig. 1). Although the peaks were well resolved with different retention time in our assay, only Z-endoxifen and Z-4-OH were quantitated since the drug given to patients was a trans-isomer of Z-TAM. Also, previous studies have shown that Z-isomers of endoxifen and 4-OH had stronger ER activity than E-isomers (E-isomers are weak agonists with <1 % of the affinity to the ER as compared to the Z-isomers) [22‐25], and thus, the focus of this study was primarily on the Z-isomers as they are considered as the major active metabolites in TAM treatment.

Table 2 summarizes the results for precision and accuracies for each compound. Our results show that this assay has good intra- and inter-day precision since the %CV are less than 10 %. Additionally, the accuracies of each analyte were within ±20 % of the true concentrations.

The determination of extraction efficiencies for each analyte was performed in triplicates. Recoveries were good and consistent, ranging between 83 and 88 %.

TAM and its metabolite peaks were not observed in either the blank solvent or the non-TAM FFPE tissue samples after running the high concentration standard, and the calculated concentration of the QC was within ±20 % of the expected value.

Formalin fixation with paraffin embedding is the standard procedure for tissue preservation and stabilization in pathology laboratories prior to histological evaluation by pathologists. Previous studies have suggested that the fixation, paraffin processing conditions are harsh, and the prolonged FFPE tissue storage times have made retrospective biological studies difficult since the efficiency of protein recovery was influenced by the fixation protocols, fixation time and sample age [26, 27]. However, numerous groups have demonstrated that the quantity and quality of proteins identified by LC tandem MS from FFPE tissue samples are not significantly impacted by fixation and tissue processing when compared with matched frozen tissues [28‐30]. In addition, the study from Scicchitano et al. [31] showed that the number of proteins extracted from FFPE tissue was, indeed, more than the frozen samples. Furthermore, Crockett et al. [32] demonstrated similar results of protein yields, and Palmer-Toy et al. [33] reported a 30 % greater yield of identified proteins from FFPE tissue compared with frozen samples. Clearly, the bulk of evidence would support the view that formalin fixation did not necessarily affect protein recovery and molecules that are bound to the proteins.

In the current study, TAM and its metabolites are the target analytes that bound to proteins in FFPE tissue. Apparently, ethanol that was used as a dehydrating agent in preparation for paraffin wax infiltration did not fully deplete TAM and its metabolites from the tumor tissues because these molecules are hydrophobic; they tend to bind to proteins and remain in the resin until xylene was added to the FFPE tissues. Xylene, which acts as a tissue clearing agent following dehydration, is non-polar and paraffin miscible. This solvent dissolves the paraffin and extracts lipids, proteins as well as the non-polar analytes from the tissues. Once the tissues were spun down, all the hydrophobic components remain in the organic solvent as supernatant.

Although it is noteworthy to compare both archival samples and frozen tissues from the same individual, for our archival samples this option was not available. Nevertheless, our results show that the mean concentrations of TAM, 4-OH, endoxifen and DMT extracted from FFPE tumor recurrent tissues were 122 ± 25.2, 2.13 ± 0.604, 19.2 ± 5.08 and 324 ± 103 ng/g, respectively (Table 3). Our findings are comparable with those of Furlanut et al. [34], who showed that frozen tissues (~2 mg) in breast cancer patients receiving a single dose of 20 mg of TAM (samples were taken 5 h postdrug administration) contained an average concentration of 118 ng/g TAM, 4.90 ng/g 4-OH and 23.3 ng/g DMT (ranging from 0 to 643 ng/g for DMT) (no data for endoxifen). Our data demonstrate that TAM and its metabolites were still recovered from FFPE tissues despite multiple uses of organic solvents for washing and removal of formalin during sample processing.

To verify whether TAM and its metabolites were fully extracted from the de-waxed tumor samples, FFPE tissue spiked with the analytes was homogenized with methanol after deparaffinization followed by solid-phase extraction. Our results show that only 3–4 % of TAM and its metabolites were recovered from the homogenized tissues comparing to a spiked solvent extracted under the same conditions (data not shown), suggesting that most analytes from the tissue went into the xylene fraction during the deparaffinization process. Similarly, the tissue homogenates and the xylene fractions from the breast tumor patients were tested to ensure optimal extraction efficiency.

FFPE tissues from breast cancer patients obtained in the archives of the pathology department were investigated to demonstrate the applicability of the assay. Figure 2 contains representative MRM chromatograms of a single tissue specimen from a patient without tumor recurrence comparing to one taken from a patient developing recurrence. It should be noted that only the absolute abundances of TAM and its metabolites are shown in Fig. 2 to illustrate the level of each compound found in the two groups of patients. The true recovery of the analytes in the sample was normalized by the I.S. (data not shown) as area ratios and the concentrations were calculated with respect to the standard curves. Indeed, Table 3 summarizes the mean concentrations of TAM (147 ± 51.7 ng/g), 4-OH (9.75 ± 4.62 ng/g), endoxifen (94.6 ± 41.0 ng/g) and DMT (1,256 ± 282.8 ng/g) in patients without tumor recurrence, whereas in tumor recurrent patients, the average concentrations were 122 ± 25.2, 2.13 ± 0.604, 19.2 ± 5.08 and 324 ± 103 ng/g for TAM, 4-OH, endoxifen and DMT, respectively. Interestingly, the levels of the TAM metabolites were significantly higher in tumor non-recurrent patients (p < 0.05 for 4-OH and endoxifen, and p < 0.01 for DMT) than those with tumor recurrence (Fig. 3a). One could argue that the concentration difference between the two groups of patients could be associated with the dosage problem of TAM because they were given 20 mg/day of TAM regardless of their weights. Presumably, the levels of metabolites found in each patient should be comparable if TAM was dosed accurately. It is speculated that heavier patients might not have received enough dose of TAM, thereby producing lower levels of metabolites. Also, the duration of TAM treatment before biopsy could affect the levels of metabolites. A longer period of TAM treatment could lead to higher concentrations of metabolites found in the tumor non-recurrent patients. However, our study shows that the duration of TAM treatment for the two groups of patients was on average 26 and 22 months for tumor recurrent and non-recurrent patients, respectively. Moreover, the mean values of body mass index (BMI) for recurrent patients were 26.4 versus 28.6 for non-recurrent subjects, suggesting that a dosing problem of TAM was very unlikely (Table 3) because the length of treatment (p = 0.2572) and the patients’ BMI (p = 0.4260) were not significantly different. Another interesting phenomenon observed in our study is that 4-OH and endoxifen were not found in some of the tumor recurrent patients (data not shown), suggesting that TAM was possibly not metabolized in these patients or below the detection limit of the assay.

To further investigate whether the P450 isoforms from each group of patients could possibly affect the levels of TAM metabolites, the concentration ratios of each analyte from the two groups were plotted. Figure 3b demonstrates that the concentration ratios of endoxifen to 4-OH and DMT to TAM were significantly different between the two groups (p < 0.01), suggesting that individual differences in TAM metabolism may contribute to the development of cancer recurrence and could be attributed to inter-individual variation in cytochrome P450 activity. However, it should be stressed that patients who had tumor recurred might not comply to take tamoxifen. In our experience, by 5 years only 50–60 % of patients still take the drug even though it is prescribed. Also, patients who recurred and had biopsies might have stopped taking tamoxifen or tamoxifen might have been stopped and substituted by another hormone even before biopsy confirmation. Clearly, unknown possibility such as non-compliance or therapy termination could be a cause of the concentration differences.