Purpose: Studies in oncology have implicated multiple molecular mechanisms as contributors to intrinsic and acquired tumor resistance to antifolate therapy. Here we show the utility of an 19F-labeled methotrexate (FMTX) with 19F magnetic resonance to differentiate between sensitive and resistant tumors in vivo and thus predict therapeutic response.

Experimental Design: Human sarcoma xenografts in nude mice were used in this study. The sarcoma cell lines chosen for this study (HT-1080, HS-16, and M-805) are well characterized in terms of their methotrexate sensitivity and molecular mechanisms of resistance. The pharmacokinetics of tumor uptake/washout of FMTX were monitored via in vivo19F magnetic resonance spectroscopy (pulse/acquire with surface coil localization) following an i.v. bolus injection. Response post-therapy, following leucovorin rescue, was monitored via tumor growth.

Results: The three tumor models show differences in both the peak concentrations of tumor FMTX and the dynamics of uptake/retention. These differences are most pronounced for time points late in the magnetic resonance observation period (225-279 minutes post-injection). A statistically significant linear correlation between tumor tissue concentrations of FMTX at these late time points and therapeutic response in the days/weeks post-treatment is shown (R = 0.81, F = 9.27, P < 0.001). Interestingly, a 400 mg/kg i.v. bolus injection of FMTX is a more potent cytotoxic agent in vivo against methotrexate-sensitive tumors than is the parent compound (P = 0.011).

Conclusions: In principle, the assay method described herein could be implemented in the clinic as a diagnostic tool to make decisions regarding therapeutic protocol for the treatment of osteosarcoma on a case-by-case basis.

Current standard of care for the patient with osteosarcoma involves an initial regimen of high-dose methotrexate for up to four cycles over the course of 10 weeks given in conjunction with cisplatin and doxorubicin. However, it is known that >50% of osteosarcoma tumors in patients exhibit molecular evidence of methotrexate resistance (1, 2). Additionally, high-dose methotrexate therapy is not without potential complication. A diagnostic tool capable of predicting therapeutic efficacy at an early stage would be extremely useful in managing therapy in these patients. It is with these considerations in mind that we have undertaken the present work.

A number of factors at the cellular level contribute to methotrexate resistance in human cancers (1–5). These factors can include a failure of the cancer cell to transport the drug into the intracellular space or a failure to retain the drug intracellularly. The dianionic methotrexate molecule is primarily transported into the intracellular space via the reduced folate carrier (RFC). At very high extracellular methotrexate concentrations a small diffusional contribution has also been observed in vitro(6). The cytotoxicity of this drug is further potentiated via the action of the enzyme folylpolyglutamylate synthetase (FPGS). The FPGS enzyme conjugates multiple anionic glutamate residues to methotrexate, increasing intracellular retention. Decreased RFC activity is a common intrinsic methotrexate resistance mechanism in high-grade osteosarcoma in humans (2) with decreased FPGS activity representing a common mechanism of intrinsic resistance in soft tissue sarcoma (5). These and other means of resistance can also be acquired following initial treatment with high-dose methotrexate therapy (3, 4, 7).

In the past, emphasis has been placed on monitoring plasma concentrations of methotrexate as a means to ensure therapeutic efficacy (8, 9). However, it is fundamentally important that any cytotoxic drug be delivered to the target tissue. We have developed an 19F-labeled methotrexate (FMTX, Fig. 1) with the goal of using 19F magnetic resonance (19F MR) as a means to assay tumor methotrexate sensitivity. The synthesis of this pharmacokinetic probe has recently been reported (10). It has been noted that FMTX is readily MR visible in vivo and its in vitro cytotoxicity is equivalent to that of the parent antifolate (methotrexate) against a methotrexate-sensitive cell line.

Fig. 1

The chemical structure of FMTX and its potential metabolites. The FPGS reaction catalyzes the formation of FMTX-polyglutamates (FMTX-glun). DAMFPA and 7-OH-FMTX are formed by the action of carboxypeptidase and hydroxylase enzymes, respectively.

Fig. 1

The chemical structure of FMTX and its potential metabolites. The FPGS reaction catalyzes the formation of FMTX-polyglutamates (FMTX-glun). DAMFPA and 7-OH-FMTX are formed by the action of carboxypeptidase and hydroxylase enzymes, respectively.

Close modal

Clinical 19F MR spectroscopy has been used previously to follow the tumor pharmacokinetics of 5-fluorouracil (5-FU) in a study of 57 patients, and it was shown that intratumor trapping of 5-FU is a positive predictor of therapeutic response (11). Subsequently, quantitative MR spectroscopy–based determinations of tissue concentrations of 5-FU and its metabolites in human liver were reported (12). In the case of a uniform excitation (B1) over the volume of the sample, accurate quantitation requires knowledge of the T1 relaxation times (or more precisely the ratio of TR/T1, where TR is the pulse sequence repetition time) of the metabolite and concentration reference species as well as the pulse angles experienced by the sample and reference (13). The more complex interrelationships among the factors affecting resonance intensity in the case of nonuniform excitation, as in the case of surface coil spectroscopy, have also been discussed (14). It has been emphasized that adequate signal-to-noise ratio is required for accurate determination of in vivo tissue metabolite concentrations (15). Variations and uncertainties in T1 and/or MR pulse angle are potential sources of error. In the end, the choice of acquisition variables in MR spectroscopy is always a tradeoff between the need for acceptable signal-to-noise and trying to avoid quantitation errors (16).

Herein we address the utility of FMTX with 19F MR as a diagnostic tool in the early prediction of therapeutic outcome. Tumor tissue concentration can be monitored noninvasively in real time. The human tumor xenograft models studied in this work were the methotrexate-sensitive HT-1080 fibrosarcoma (5, 17); the HS-16 cell line, a mesenchymal chondrosarcoma, which is methotrexate resistant as a result of decreased FPGS activity (5, 17); and the malignant fibrohistocytoma M-805, which exhibits resistance due to decreased RFC activity (18).

By studying the kinetics of drug uptake and retention in both sensitive and resistant human tumor xenograft models in mice, we show that it is possible to differentiate sensitive tumors from resistant tumors with this method. In addition to its potential clinical diagnostic utility, FMTX shows greater therapeutic efficacy in vivo than the parent compound, methotrexate.

Cell Lines and In vitro Cytotoxicity. The human sarcoma cell lines HT-1080 (5, 17), M-805 (18), and HS-16 (5, 17) have been described previously. The HT-1080 cell line was obtained from American Type Culture Collection (Rockville, MD) and the M-805 and HS-16 cell lines were obtained from the laboratory of Dr. Joseph R. Bertino (The Cancer Institute of New Jersey, New Brunswick, NJ). The HS-16 cell line is derived from a human mesenchymal chondrosarcoma, the HT-1080, a human fibrosarcoma, and the M-805 cell line from a human malignant fibrohistocytoma.

Previously, it has been shown that the in vitro cytotoxicity of FMTX is equivalent to that of methotrexate against the HT-1080 cell line (10). In the current study, the in vitro cytotoxicities of methotrexate and FMTX were also compared against the HS-16 and M-805 cell lines. FMTX [N-(3-fluoro-4-amino-4-deoxy-N-methyl-pteroyl)-l-glutamic acid] was synthesized according to published methods (10) and methotrexate was obtained from Immunex (Seattle, WA). Cells were maintained as monolayer cultures in RPMI 1640 supplemented with 10% FCS at 37°C under a humidified 5% CO2 atmosphere. For cytotoxicity assays, monolayer cells were trypsinized and plated in 6-well culture plates (10 cm2 per well). After a period of 48 to 72 hours to allow for cell attachment and establishment of cell proliferation, the medium was aspirated and replaced withfresh medium and methotrexate or FMTX. Following a 24-hour drug exposure, culture medium was aspirated and replaced with fresh culture medium. Cells were incubated for a further 72 to 96 hours and cell viability was determined via the 2,3-bis[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide inner salt/phenazine methosulfate assay (19). 2,3-bis(2-Methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino) carbonyl]-2H-tetrazolium hydroxide and phenazine methosulfate were obtained from Sigma-Aldrich (St. Louis, MO). In vitro cytotoxicity was evaluated from drug concentration/response curves.

In vivo19F Magnetic Resonance–Observed Tumor Pharmacokinetics. Animal studies were done according to institutionally approved protocols for the safe and humane treatment of animals. Tumor xenografts were initiated by the injection of 0.1 mL of a slurry of ∼106 cells. The tumor cell slurry was inoculated s.c. into the left flanks of 6-week-old male athymic nude mice (Charles River, Boston, MA). Tumors used in this study ranged in size between 0.16 and 0.41 cc. FMTX was given via i.v. bolus tail-vein injection at a dosage of 400 mg/kg, a level comparable with but less than the highest doses used clinically in humans (8, 9, 20). Mice were unanesthetized for 19F MR experiments. The mice readily crawl into a 60-cc syringe barrel (with air holes) which is used as an animal holder with the tumor protruding through a hole into a home-built two-turn 19F MR surface coil (centered around the tumor). Tumor immobilization within the MR coil was achieved by anchoring the tumor with a 3-0 silk tie sutured into the flank skin flap just anterior to the tumor. In vivo19F MR studies at 188 MHz (Bo = 4.7 T) were done in a wide-bore (33-cm diameter) small animal imaging system (Omega-Bruker, Billerica, MA) with the use of a temperature control/susceptibility matching water bath (21).

For 19F MR experiments (pulse and acquire), acquisition variables included a 60° pulse-angle, a pulse repetition (TR) of 0.49 second with 2,048 transients per spectrum. Thus, the temporal resolution was 18 minutes/spectrum. Whereas it does not lead to the highest possible signal-to-noise ratio, this set of MR acquisition variables was chosen as the most desirable because of the relative insensitivity of signal amplitude to variations in either TR/T1 or pulse angle (13). Additionally, the choice of TRT1 localizes the MR observation volume to regions located primarily within the center of and directly adjacent to the coil with a rather uniform signal intensity profile over the excitation volume (14). MR time domain data was analyzed using jMRUI software (22). The FMTX resonance intensity was determined relative to that of an external 19F reference standard, an 18-μL glass microsphere of 0.1 mol/L trifluoroacetic acid in D2O. Intratumor FMTX concentrations were estimated from the 19F MR intensity ratios using the measured longitudinal relaxation times (T1) for trifluoroacetic acid (3.37 seconds) and FMTX (0.60 seconds, from ref. 10) at 4.7 T and 37°C, the volume-averaged pulse angle applied to the tumor and external reference. This is similar to the method of Murphy-Boesch et al. (12, 23); however, in the current study, we do not generate a B1 field map in order to calculate precisely how the applied pulse angle varies across the region of interest. Tumor size and geometry were all roughly similar in the current set of experiments; thus, an empirically determined correction factor was measured to account for nonuniform excitation across the volume of the tumor. This volume-averaged correction factor was determined from a calibration curve generated from measurements on a series of spherical FMTX phantom samples (0.8 cm OD) of varying concentration using the 19F MR acquisition variables described above. The calibration curve (data not shown) relates the observed ratio of the FMTX and external reference 19F resonance intensities to FMTX concentration in the phantom sample. Tumor tissue volume average concentrations of FMTX are reported in millimoles/liter (mM).

Theoretical Modeling of 19F Magnetic Resonance Chemical Shift. Quantum chemical calculations were undertaken to obtain an estimate of the change in 19FMTX chemical shift with metabolism (see Fig. 1). The calculations were intended to ascertain whether or not the single observed in vivo19F MR resonance arises from FMTX alone or whether some contribution could be due to FMTX-polyglutamates; the less potent FMTX metabolite, 7-hydroxy-FMTX (7-OH-FMTX, ref. 24); and/or the inactive metabolite, 2,4-diamino-methylfluoropteroic acid (DAMFPA, ref. 25). Theoretical calculations of the isotropic 19F MR chemical shifts for FMTX, 7-OH-FMTX, FMTX-glutamate, and DAMFPA were done using Gaussian 03 Revision C.02 software (Gaussian, Inc., Wallingford, CT). The molecular geometry was first optimized via a density functional B3LYP/6-31G* calculation, and the resulting molecular configuration was used in a restricted Hartree-Fock electronic structure calculation using the expanded 6-311+G** basis set (26). Calculations for each molecule were done on a PC with 2.4 GHz Pentium 4 CPU and required ∼200 hours for completion. Isotropic 19F MR chemical shifts (in ppm) are reported relative to that for 19FMTX.

Leucovorin Rescue and Tumor Response. At 6 hours post-FMTX administration, mice received i.p. hydration (1.0 cc normal saline) and at 24 hours received leucovorin (Bedford Laboratories, Bedford, OH) rescue therapy (27). This protocol allows the mice to survive what would otherwise be a fatal dosage of FMTX and is analogous to that used in high-dose methotrexate therapy in humans (28). Tumor growth was monitored post-therapy. The three perpendicular axes of the tumor were measured with a micrometer and the tumor volume modeled as a spheroid [V = (π/6) abc, where a, b, c are the dimensions of the tumor axes in cm]. Tumor growth curves post-therapy were Gompertzian and hence in vivo surviving fraction (the fraction of cancer cells in the pretreatment tumor that survived therapy), SF, is reported based on the tumor volume growth variables:

\[\mathrm{SF}\ =\ \left(\frac{1}{2}\right)^{\mathrm{TGD/DT}_{\mathrm{o}}}\]

The variable DTo is the mean doubling time of the untreated, control group for each xenograft tumor model and TGD is the tumor growth delay for the treated tumor (the difference in tumor doubling time between the treated tumor and DTo). An assumption inherent in this model for determining SF is that the tumor growth rate is the same in the untreated tumor as in the post-treatment tumor during the regrowth phase, which is consistent with the observed tumor growth curves (see e.g., Fig. 2). For untreated growth the tumor volume doubling times DTo were 2.9 ± 0.1, 25.8 ± 5.5, and 23.4 ± 1.9 days for the HT-1080 (n = 8), HS-16 (n = 4), and M-805 (n = 6), respectively.

Fig. 2

HT-1080 control and post-therapy growth curves (normalized tumor volume, Vinit = 1).

Fig. 2

HT-1080 control and post-therapy growth curves (normalized tumor volume, Vinit = 1).

Close modal

Plasma Pharmocokinetics. In a separate set of experiments, the time course of plasma pharmacokinetics was studied in mice. Each mouse received a 400 mg/kg i.v. bolus injection of FMTX. Plasma levels of FMTX were determined for the time points 0.5, 1, 2, 3, 4, 5, and 24 hours post-injection (n = 3-4 per time point). Ten minutes before blood collection, mice were anesthetized with ketamine/xylazine. Blood was collected via cardiac puncture into heparinized vials and centrifuged. Typical collection volumes were 0.5 to 1 cc and hence this was a terminal procedure. To prepare the samples for high-performance liquid chromatography analysis 20 μL of 50% trichloroacetic acid was added to 200 μL of plasma. Samples were allowed to stand on ice for several hours after which the precipitate was removed by centrifugation and the supernatant collected for high-performance liquid chromatography analysis without further modification. An Agilent 1100 high-performance liquid chromatography system (Agilent, Palo Alto, CA) was used. Twenty microliters of the supernatant were injected onto an Econoshpere C18 5 μm 4.6 × 250 mm column (Econosphere, Deerfield, IL) with an eluent of 15% acetonitrile/85% 50 mmol/L potassium dihydrogen phosphate at a flow rate of 1 mL/min and monitored at 313 nm. A standard curve of spiked plasma was linear from 0.31 to 10 μg/mL. Samples with higher drug concentrations were diluted to bring them to within the analytic range.

Statistical Analysis. The single factor two-sided ANOVA/Tukey test procedure (29) was done to check for statistically significant differences between mean values [including intratumor FMTX concentrations, area under the curve (AUC), and SF] for the three tumor models. Significance of linear regression was verified using the F test. Mean values are reported as mean ± SE unless otherwise indicated.

Plasma 19F-Labeled Methotrexate. In nude mice, plasma levels of FMTX follow biexponential decay kinetics after an i.v. bolus injection of a 400 mg/kg dosage (Fig. 3). The data were fit to Eq. B:

Fig. 3

FMTX plasma pharmacokinetics following an i.v. bolus dosage of 400 mg/kg FMTX.

Fig. 3

FMTX plasma pharmacokinetics following an i.v. bolus dosage of 400 mg/kg FMTX.

Close modal
\[\left[\mathrm{FMTX}\right]_{\mathrm{plasma}}\ =\ \mathit{A}\ {\cdot}\ \mathrm{e}^{{-}\mathit{k}_{1}\ {\cdot}\mathit{t}}\ +\ \mathit{B}\ {\cdot}\ \mathrm{e}^{{-}\mathit{k}_{2}{\cdot}\mathit{t}}\]

The resulting fit variables were A = 0.480 mmol/L with rate constant, k1 = 1.74 h−1 (t1/2 = 0.399 hour) for the fast-washout component and B = 2.47 × 10−3 mmol/L and rate constant k2 = 0.11 h−1 (t1/2 = 6.3 hours) for the second, slow-washout component (R = 0.91). The fit results in an estimate of [FMTX]plasma at t = 0 minute of 0.482 mmol/L, whereas the value at the first measurement time point, 30 minutes post-injection was 0.32 ± 0.02 mmol/L. By 4 hours post-injection, plasma concentrations of FMTX fell to 1.7 ± 0.3 μmol/L and at 24 hours decreased further to 0.18 ± 0.09 μmol/L.

19F-Labeled Methotrexate Is more Potent than Methotrexate In vivo against a Methotrexate-Sensitive Tumor Xenograft. Tumor growth curves indicate that FMTX is considerably more potent than methotrexate in vivo against the methotrexate-sensitive HT-1080 tumor xenograft model (Fig. 2). This is somewhat surprising because in vitro, both agents have equivalent cytotoxic action against both sensitive and resistant cell lines (data not shown). Analysis of the tumor growth post-therapy allows for a comparison of the efficacy of methotrexate and MTX in terms of SF, the fraction of cells in the pretreatment tumor surviving therapy (Eq. A). The volume doubling time for the untreated HT-1080 tumor is 2.9 ± 0.1 days (n = 8). Following a 400 mg/kg i.v. bolus (FMTX or methotrexate), in the HT-1080 model, SFFMTX = 0.29 ± 0.06 (n = 7), whereas SFmethotrexate = 0.56 ± 0.07 (n = 6), a statistically significant difference (P = 0.011). Even a dosage of 200 mg/kg FMTX was more potent than a 400 mg/kg methotrexate dosage, but not significantly (P = 0.051). At the 200 mg/kg FMTX dosage level, SF = 0.42 ± 0.04 (n = 5), whereas for 200 mg/kg methotrexate it was 0.64 ± 0.08(n = 6).

Tumor Tissue Pharmacokinetics via 19F Magnetic Resonance. Tumor tissue concentrations of FMTX are readily MR visible with 18-minute temporal resolution following the 400 mg/kg i.v. bolus dosage (Fig. 4). In vivo tumor tissue pharmacokinetic data acquired by 19F MR show variation in drug uptake/retention between the xenograft models investigated in this study (Fig. 5). Theoretical quantum chemical calculations of the isotropic 19F MR chemical shifts for FMTX and its metabolites indicate that metabolized species would likely contribute to broadening of the observed in vivo tumor 19F MR resonance (Table 1). Separate signals from each metabolite would probably not be well resolved. The three tumor models (HT-1080, HS-16, and M-805) show differences both in the peak concentrations of tumor FMTX achieved and the dynamics of uptake/retention. The methotrexate-sensitive HT-1080 tumor model achieves peak tissue concentrations at 234 minutes post-injection (0.181 ± 0.054 mmol/L), whereas in the other two models, peak tumor concentrations are lower in magnitude and are achieved at earlier times preceding washout from the tumor tissue. In the M-805 tumor model, wherein, we would expect reduced uptake secondary to decreased RFC expression, peak tissue concentrations of FMTX occurs early (72 minutes post-injection) and are considerably lower (0.047 ± 0.021 mmol/L) than the maximum levels observed in the HS-16 and HT-1080 tumor models. The HS-16 tumor shows some ability to accumulate FMTX in the tumor tissue, but these levels peak at 126 minutes post-administration (0.077 ± 0.037 mmol/L) and are followed by drug efflux from the tumor as FMTX clears from the plasma, consistent with the FPGS-deficient status of this cell line. Although the mean maximum intratumor FMTX concentrations are greatest in the HT-1080 tumor, comparison of the maximum concentrations (HT-1080 at 234 minutes, M-805 at 72 minutes, and HS-16 at 126 minutes) does not reveal any statistically significant differences between methotrexate-sensitive and -resistant xenografts.

Fig. 4

Stacked 19F MR spectra with 18-minute temporal resolution from a 0.23 cc, HS-16 tumor following an i.v. bolus dosage of 400 mg/kg FMTX.

Fig. 4

Stacked 19F MR spectra with 18-minute temporal resolution from a 0.23 cc, HS-16 tumor following an i.v. bolus dosage of 400 mg/kg FMTX.

Close modal
Fig. 5

FMTX tumor tissue pharmacokinetics as measured via 19F MR. Bars, ±SE for the last three time points.

Fig. 5

FMTX tumor tissue pharmacokinetics as measured via 19F MR. Bars, ±SE for the last three time points.

Close modal
Table 1

Theoretical 19F MR chemical shifts for FMTX and metabolites

Chemical speciesPredicted 19F MR chemical shift (ppm)*
FMTX 0.00 
7-OH-FMTX 0.33 
FMTX-glutamate 0.19 
DAMFPA 0.36 
Chemical speciesPredicted 19F MR chemical shift (ppm)*
FMTX 0.00 
7-OH-FMTX 0.33 
FMTX-glutamate 0.19 
DAMFPA 0.36 
*

Chemical shifts are reported relative to the calculated chemical shift for FMTX.

The differences in intratumor FMTX concentrations in sensitive (HT-1080) versus resistant (M-805 and HS-16) xenografts are most pronounced at later time points. Comparison of intratumor FMTX concentrations for the time point centered at 234 minutes post-injection (19F MR spectra acquired over the interval 225-243 minutes post-injection) indicate statistically significant differences between resistant and sensitive tumor models. Intratumor 19F-MR observable concentrations at this time point were 0.181 ± 0.054, 0.026 ± 0.020, and 0.025 ± 0.012 mmol/L for the HT-1080, HS-16, and M-805 tumor xenograft models, respectively. These concentrations are significantly higher for the HT-1080 than the M-805 (P < 0.001) and HS-16 (P < 0.001) tumors, whereas these last two groups do not differ significantly from each other.

Correlating Pharmacokinetic Variables with Therapeutic Efficacy. The tumor 19F MR kinetic data suggests the use of late time point intratumor FMTX concentrations as a means of differentiating between sensitive and resistant tumors. The AUC was estimated for the period from 225 to 279 minutes post-injection using the 19F MR intratumor concentration data (Fig. 5) and the trapezoidal rule. The calculated value is denoted as AUC225-279 (in mmol/L minutes). This quantity is significantly higher for the methotrexate-sensitive HT-1080 at 6.60 ± 1.42 mmol/L minutes than for either of the two resistant tumor models (P < 0.05). For the HS-16 and M-805 models, the values are 0.87 ± 0.60 and 0.75 ± 0.35 mmol/L minutes, respectively, and are not significantly different.

Plotting AUC225-279 versus log10(SF) yields a statistically significant linear correlation across the three tumor models investigated (Fig. 6). The resulting linear least squares fit to the full data set (n = 19) is:

Fig. 6

Tumor therapeutic response log10(SF) as a function of AUC225-279. Data fit to Eq. C.

Fig. 6

Tumor therapeutic response log10(SF) as a function of AUC225-279. Data fit to Eq. C.

Close modal
\[\mathrm{log}_{10}(\mathrm{SF})\ =\ {-}0.015\ {-}\ 0.111\ {\cdot}\ \mathrm{AUC}_{225{-}279}\ (\mathrm{R}\ =\ 0.81,\mathrm{F}\ =\ 9.27,\mathrm{P}\ {<}\ 0.001).\]

This fit very nearly passes through the origin (SF = 1 for AUC225-279 = 0 mmol/L minutes).

In oncology, MR imaging is routinely used in the clinical decision-making process. MR spectroscopy, on the other hand, is currently used primarily as a research tool, but is poised to play an ever-increasing role in the clinical setting (30–32). We note that 19F MR spectroscopic studies in humans have been reported for the antineoplastic agent 5-FU (11, 12, 33–35). Because they are typically localized to the extremities, osteosarcomas in the clinical setting are amenable to MR interrogation via surface coil with resulting benefits in terms of measurement sensitivity (36). The feasibility of obtaining well-defined MR spectra from bone tumors was initially unclear, but Nidecker et al. did 31P MR spectroscopic measurements of bone tumors (37) and later studies showed that 31P MR-observable tumor metabolic variables observed as little as 2 days post-therapy correlated with therapeutic response to chemotherapy (38).

Methotrexate is used in the treatment of many tumors, including acute leukemias, head and neck cancers, bone sarcomas, bladder tumors, and in some lymphoma patients. It is part of the standard chemotherapy regimen that has converted osteogenic sarcoma from an incurable tumor to its present status as a highly treatable and curable cancer. Thus, it remains in wide use despite the development of newer targeted agents and therefore a method to predict tumor sensitivity would have significant clinical importance.

Here we have shown that 19F MR spectroscopy of FMTX in vivo can be used to noninvasively differentiate between a methotrexate-sensitive tumor model (HT-1080) and two methotrexate-resistant xenografts with distinct mechanisms of resistance (HS-16 and M-805). The most pronounced differences occur at ∼4 hours post-administration of an i.v. bolus of 400 mg/kg FMTX. The significant correlation between the area under the tissue concentration/time curve for the time points 225 to 279 minutes post-injection and the resulting therapeutic response indicates that this pharmacokinetic variable or some variation thereof may serve as a predictor of therapeutic efficacy.

Methotrexate is metabolized via a number of different pathways. The hydroxylation of methotrexate to form 7-OH-methotrexate occurs in the liver (39), whereas deactivation to form the inactive 2,4-diaminomethylpteroic acid (for which DAMFPA is the fluorine-containing analogue) results from metabolism of methotrexate by intestinal flora (40). As already mentioned, antifolate polyglutamates are formed in situ through the action of the FPGS enzyme. Theoretical quantum chemical calculations indicate that the FMTX metabolites, 7-OH-FMTX, DAMFPA, and FMTX-glu, if present, would probably lead to a broadening of the observed tumor 19F MR resonance. The predicted shift in the 19F resonance upon metabolism via these routes is of the order of 0.2 to 0.35 ppm (Table 1). Quantum chemical calculations of the type we have employed are very reliable for the prediction of 13C, 15N, and 17O chemical shifts (26), but results for 19F should be approached with caution (41). However, in the case of fluorobenzenes (spanning a chemical shift range of ∼63 ppm) the agreement between theory and experiment has been shown to be excellent (42). From a molecular modeling point of view FMTX is simply a complex fluorobenzene.

Because FMTX and its metabolites would likely not be well resolved via in vivo19F MR, it is necessary to consider the potential contributions of the metabolites to the MR resonance intensity, which is taken as an indicator of in situ cytotoxic potential. Specifically, we concern ourselves with the clinical scenario. A contribution by FMTX-polyglutamates to the 19F signal would be correctly interpreted as a positive predictor of therapeutic response, hence it is not problematic. The inactive metabolite 2,4-diaminomethypteroic acid is only occasionally detectable (∼0.1μmol/L limit of detection) in the serum of patients following the administration of high-dose methotrexate therapy (43, 44). Thus, a spurious contribution of DAMFPA to the observed tumor 19F MR resonance is unlikely. However, plasma concentrations of 7-OH-methotrexate following high-dose MTX administration in humans are not negligible. This metabolite is cytotoxic, but considerably less so than the parent compound, methotrexate (39). The levels of plasma 7-OH-methotrexate show a marked dependence on the methotrexate dosage as well as the infusion protocol (43–45). In one study, immediately after a 6-hour infusion of high-dose methotrexate, the ratio of concentrations of methotrexate to 7-OH-methotrexate in plasma showed a range of ∼10:1 to ∼20:1 in patients (43). At later time points, with continued metabolism of methotrexate, the proportion of plasma antifolate species was increasingly skewed in favor of 7-OH-methotrexate. Thus, a contribution of the fluorine-containing analogue of this metabolite to the 19F MR resonance could be problematic at these later time points. Whereas we are uncertain about the levels of plasma 7-OH-FMTX in the current studies, it is worth reemphasizing that a statistically significant correlation between the tumor 19F MR signal intensity and therapeutic response was observed (Fig. 6) as described by Eq. C.

The ability of HT-1080 tumors to steadily accumulate FMTX is remarkable. In this methotrexate-sensitive human sarcoma xenograft, tissue concentrations of FMTX reach a mean value of 0.18 mmol/L at 234 minutes, by which time plasma levels have fallen to ∼1.7 μmol/L. Qualitatively, the shapes of the pharmacokinetic curves of model tumor FMTX uptake/retention can be understood in terms of the molecular mechanism of resistance. In the M-805 tumor model, decreased RFC activity leads to reduced tissue uptake, whereas in the HS-16 tumor, drug uptake is rapid with high tissue concentrations achieved but followed by rapid egress of FMTX from the tumor due to decreased FPGS activity. However, as we were unable to discern any statistically significant differences between the pharmacokinetics in the HS-16 and M-805 tumor models, it seems that the method outlined here may not be a definitive diagnostic of the resistance mechanism at the molecular level. The two tumor models, HS-16 and M-805, represent prevalent mechanisms of intrinsic resistance to methotrexate. There are other important molecular methotrexate resistance mechanisms, including active excretion pathways (3, 7); hence, FMTX pharmacokinetic behavior in these tumors might be expected to bear some resemblance to that of FPGS deficiency.

A surprising finding in this work has been the increased efficacy of FMTX, compared with methotrexate, against the methotrexate-sensitive HT-1080 tumor xenograft in vivo for the 400 mg/kg dosage. Molecular modeling indicates only slightly more favorable binding of FMTX in the active site of two of the key target enzymes, dihydrofolate reductase and thymidine synthetase (10). In vitro, the two agents are equipotent against the three cell lines investigated in this study. It is possible that the distinction in vivo arises as a result of differences in the rate of production of the less cytotoxic metabolite 7-OH-FMTX or the inactive DAMFPA, but we have not investigated this hypothesis.

The studies reported herein were done at 4.7 T, a magnetic field strength only modestly higher than the 3.0 and 4.0 T clinical MR systems currently being installed at many medical centers, suggesting the feasibility of performing similar studies on patients if biological data were compelling. Numerous studies have shown the clinical feasibility of monitoring pharmacokinetics in vivo via 19F MR. Reports that monitored 5-FU uptake, retention and metabolism in tumor and liver indicate that a temporal resolution on the order of 20 minutes per spectrum and a spatial resolution of 4 × 4 × 4 cm is not an unreasonable expectation at 1.5 T (12, 34, 46, 47). As already noted, sarcomas are often superficial; thus, MR sensitivity would not be hampered by the tumor to detection coil distance. The approach we use of labeling a parent antineoplastic agent with one or even multiple 19F atoms (for increased sensitivity) may be applicable for monitoring the tumor pharmacokinetics of other drugs as well. Strategic positioning of the label atom could make it possible to monitor not just tissue concentration, but also metabolism in situ.

With regard to the feasibility of performing 19F MR with FMTX in humans, it is important to note that methotrexate is routinely given, not as a bolus, but as a 4- to 6-hour slow infusion (24); thus, the tumor tissue pharmacokinetics are likely to be different. Also, plasma levels of methotrexate in the range of 0.5 to 1.7 mmol/L (43) for a sustained period of several hours with these slow infusion protocols would likely lead to greater signal-to-noise in the MR measurement. For practical/financial reasons, it would not be possible to routinely follow the full drug uptake time course in tumors in the clinic. However, the data presented here suggest that the ability of methotrexate-sensitive tumors to concentrate and maintain elevated tissue levels of FMTX for longer periods may be a hallmark of therapeutic responsiveness to this antifolate. A single late time point 19F MR measurement, at or delayed from the conclusion of drug infusion, may be useful as a diagnostic test in the clinic. The results of such a test could potentially be used to direct the therapeutic strategy for each patient. A newly diagnosed osteosarcoma patient might be treated with methotrexate or trimetrexate based on the results of an initial FMTX 19F MR assay. Trimetrexate (24) is currently used in the treatment of patients with relapsed osteosarcoma because it can overcome methotrexate transport resistance, which is likely to be present in this setting. An early diagnosis (i.e., at the time of initial treatment) of a failure to accumulate 19FMTX within the tumor tissue could indicate the necessity of using trimetrexate at an earlier stage.

Grant support: NIH grants R24CA83084 and P50CA86438.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: W. Spees is currently at the Department of Chemistry, Washington University, Campus Box 1134, St. Louis, MO 63130. W. Tong and W. Bornmann are currently at the Department of Experimental Diagnostic Imaging, M.D. Anderson Cancer Center, 1515 Holcombe Blvd., Box 0057, Houston, TX 77030. R. Gorlick is currently at the Department of Pediatrics, Children's Hospital of Montefiore, 3415 Bainbridge Ave., Rosenthal 4th Floor, Bronx, NY 10467.

1
Yang R, Sowers R, Mazza B, et al. Sequence alterations in the reduced folate carrier are observed in osteosarcoma tumor samples.
Clin Cancer Res
2003
;
9
:
837
–44.
2
Guo W, Healey JH, Meyers PA, et al. Mechanisms of methotrexate resistance in osteosarcoma.
Clin Cancer Res
1999
;
5
:
621
–7.
3
Gorlick R, Goker E, Trippet T, Waltham M, Banerjee D, Bertino JR. Intrinsic and acquired resistance to methotrexate in acute leukemia.
N Engl J Med
1996
;
335
:
1041
–8.
4
Gorlick R, Goker E, Trippet T, et al. Defective transport is a common mechanism of acquired methotrexate resistance in acute lymphocytic leukemia and is associated with decreased reduced folate carrier expression.
Blood
1997
;
89
:
1013
–8.
5
Li WW, Lin JT, Schweitzer BI, Tong WP, Niedzwiecki D, Bertino JR. Intrinsic resistance to methotrexate in human soft tissue sarcoma cell lines.
Cancer Res
1992
;
52
:
3908
–13.
6
Hakala MT. On the nature of permeability of sarcoma-180 cells to amethopterin in vitro.
Biochim Biophys Acta
1965
;
102
:
210
–25.
7
Chen ZS, Robey RW, Belinsky MG, et al. Transport of methotrexate, methotrexate polyglutamates, and 17β-estradiol 17-(β-d-glucuronide) by ABCG2: effects of acquired mutations at R482 on methotrexate transport.
Cancer Res
2003
;
63
:
4048
–54.
8
Pignon T, Lacarelle B, Duffaud F, et al. Pharmacokinetics of high-dose methotrexate in adult osteogenic sarcoma.
Cancer Chemother Pharmacol
1994
;
33
:
420
–4.
9
Rousseau A, Sabot C, Delepine N, Debord J, Lachatre G, Marquet P. Bayesian estimation of methotrexate pharmacokinetic parameters and area under the curve in children and young adults with localised osteosarcoma.
Clin Pharmacokinet
2002
;
41
:
1095
–104.
10
Spees WM, Yang G, Veach D, Rubio M, Koutcher JA, Bornmann W. A fluorine-labeled methotrexate as a probe for monitoring tumor antifolate pharmacokinetics: synthesis, in vitro cytotoxicity, and pilot in vivo19F magnetic resonance spectra.
Mol Cancer Ther
2003
;
2
:
933
–9.
11
Presant CA, Wolf W, Waluch V, et al. Association of intratumoral pharmacokinetics of fluorouracil with clinical response.
Lancet
1994
;
343
:
1184
–7.
12
Li CW, Negendank WG, Padavic-Shaller KA, O'Dwyer PJ, Murphy-Boesch J, Brown TR. Quantitation of 5-fluorouracil catabolism in human liver in vivo by three-dimensional localized 19F magnetic resonance spectroscopy.
Clin Cancer Res
1996
;
2
:
339
–45.
13
Becker ED, Ferreti JA, Gambhir PN. Selection of optimum parameters for pulse Fourier transform nuclear magnetic resonance.
Anal Chem
1979
;
51
:
1413
–20.
14
Evelhoch JL, Crowley MG, Ackerman JJH. Signal-to-noise optimization and observed volume localization with circular surface coils.
J Magn Reson
1984
;
56
:
110
–24.
15
Zakian KL, D'Angelica M, Matei C, et al. A quantitative assessment of liver metabolites during jaundice using three dimensional phosphorus chemical shift imaging.
Magn Reson Imaging
2000
;
18
:
181
–7.
16
Galbán CJ, Spencer RGS. Optimized pulse parameters for reducing quantitation errors due to saturation factor changes in magnetic resonance spectroscopy.
J Magn Reson
2002
;
156
:
161
–70.
17
Li WW, Cordon-Cardo C, Chen Q, Jhanwar SC, Bertino JR. Establishment, characterization and drug sensitivity of four new human soft tissue sarcoma cell lines.
Int J Cancer
1996
;
68
:
514
–9.
18
Li WW, Takahashi N, Jhanwar S, et al. Sensitivity of soft tissue sarcoma cell lines to chemotherapeutic agents: identification of ecteinascidin-743 as a potent cytotoxic agent.
Clin Cancer Res
2001
;
7
:
2908
–11.
19
Scudiero DA, Shoemaker RH, Paull KD, et al. Evaluation of a soluble tetrazolium/formazan assay for cell growth and drug sensitivity in culture using human and other tumor cell lines.
Cancer Res
1988
;
48
:
4827
–33.
20
Zaharko DS, Dedrick RL, Bischoff KB, Longstreth JA, Oliverio VT. Methotrexate tissue distribution: prediction by a mathematical model.
J Natl Cancer Inst
1971
;
46
:
775
–84.
21
Ballon D, Mahmood U, Jakubowski A, Koutcher JA. Resolution enhanced NMR spectroscopy in biological systems via magnetic susceptibility matched sample immersion chambers.
Magn Reson Med
1993
;
30
:
754
–8.
22
Naressi A, Couturier C, Devos JM, et al. Java-based graphical user interface for the MRUI quantitation package.
MAGMA
2001
;
12
:
141
–52.
23
Murphy-Boesch J, Jiang H, Stoyanova R, Brown TR. Quantification of phosphorus metabolites from chemical shift imaging spectra with corrections for point spread effects and B1 inhomogeneity.
Magn Reson Med
1998
;
39
:
429
–38.
24
Chu E, Takimoto CH. Antimetabolities. In: DeVita VT, Hellman S, Rosenberg SA, editors. Cancer: principles and practice of oncology. Vol 1. 4th ed. Philadelphia (PA): Lippincott Williams & Wilkins; 1993. p. 358–74.
25
Donehower RC, Hande KR, Drake JC, Chabner BA. Presence of 2,4-diamino-N10-methylpteroic acid after high-dose methotrexate.
Clin Pharmacol Ther
1979
;
26
:
64
–72.
26
Cheeseman JR, Trucks GW, Keith TA, Frisch MJ. A comparison of models for calculating nuclear magnetic resonance shielding tensors.
J Chem Phys
1996
;
104
:
5497
–509.
27
Sirotnak FM, Otter GM, Schmid FA. Markedly improved efficacy of edatrexate compared to methotrexate in a high-dose regimen with leucovorin rescue against metastatic murine solid tumors.
Cancer Res
1993
;
53
:
587
–91.
28
Treon SP, Chabner BA. Concepts in use of high-dose methotrexate therapy.
Clin Chem
1996
;
42
:
1322
–9.
29
Zar JH. Biostatistical analysis. Englewood Cliffs (NJ): Prentice Hall; 1984.
30
Nelson SJ. Multivoxel magnetic resonance spectroscopy of brain tumors.
Mol Cancer Ther
2003
;
2
:
497
–507.
31
Salibi N, Brown MA. Clinical MR Spectroscopy: first principles. New York: Wiley-Liss; 1998.
32
McSheehy PM. On the role of MRS in drug development.
NMR Biomed
1999
;
12
:
402
–3.
33
Wolf W, Waluch V, Presant CA. Non-invasive 19F-NMRS of 5-fluorouracil in pharmacokinetics and pharmacodynamic studies.
NMR Biomed
1998
;
11
:
380
–7.
34
Murphy-Boesch J, Li CW, He L, Padavic-Shaller KA, Negendank W, Brown TR. Proton-decoupled 19F spectroscopy of 5-FU catabolites in human liver.
Magn Reson Med
1997
;
37
:
321
–6.
35
Schlemmer HP, Becker M, Bachert P, et al. Alterations of intratumoral pharmacokinetics of 5-fluorouracil in head and neck carcinoma during simultaneous radiochemotherapy.
Cancer Res
1999
;
59
:
2363
–9.
36
Ackerman JJH, Grove TH, Wong GG, Gadian DG, Radda GK. Mapping of metabolites in whole animals by 31P NMR using surface coils.
Nature
1980
;
283
:
167
–70.
37
Nidecker AC, Müller S, Aue WP, et al. Extremity bone tumors: evaluation by P-31 MR spectroscopy.
Radiology
1985
;
157
:
167
–74.
38
Ross B, Helsper JT, Cox J, et al. Osteosarcoma and other neoplasms of bone.
Arch Surg
1987
;
122
:
1464
–9.
39
Fabre G, Goldman ID. Formation of 7-hydroxymethotrexate polyglutamyl derivatives and their cytotoxicity in human chronic myelogenous leukemia cells, in vitro.
Cancer Res
1985
;
45
:
80
–5.
40
Widemann BC, Sung E, Anderson L, et al. Pharmacokinetics and metabolism of the methotrexate metabolite 2,4-diamino-N10-methylpteroic acid.
J Pharmacol Exp Ther
2000
;
294
:
894
–901.
41
Chestnut DB, Rusiloski BE, Moore KD, Egolf DA. Use of locally dense basis sets for nuclear magnetic resonance shielding calculations. J Comput Chem 1993;1364–75.
42
de Dios AC, Oldfield E. Evaluating 19F chemical shielding in fluorobenzenes: implications for chemical shifts in proteins. J Am Chem Soc 1994;7453–4.
43
Breithaupt H, Küenzlen E. Pharmacokinetics of methotrexate and 7-hydroxymethotrexate following infusions of high-dose methotrexate.
Cancer Treat Rep
1982
;
66
:
1733
–41.
44
Wolfrom C, Hepp R, Hartmann R, Breithaupt, Henze G. Pharmacokinetic study of methotrexate, folinic acid and their serum metabolites in children treated with high-dose methotrexate therapy.
Eur J Clin Pharmacol
1990
;
39
:
377
–83.
45
Borsi JD, Sagen E, Romslo I, Moe PJ. Comparative study on the pharmacokinetics of 7-hydroxy-methotrexate after administration of methotrexate in the dose range of 0.5-33.6 g/m2 to children with acute lymphoblastic leukemia.
Med Pediatr Oncol
1990
;
18
:
217
–24.
46
van Laarhoven HWM, Klomp DWJ, Kamm YLJ, Punt CJA, Heerschap A. In vivo monitoring of capecitabine metabolism in human liver by 19Fluorine magnetic resonance spectroscopy at 1.5 and 3 Tesla field strength.
Cancer Res
2003
;
63
:
7609
–12.
47
Gonen O, Murphy-Boesch J, Li CW, Padavic-Shaller K, Negendank WG, Brown TR. Simultaneous 3D NMR spectroscopy of proton-decoupled fluorine and phosphorus in human liver during 5-fluorouracil chemotherapy.
Magn Reson Med
1997
;
37
:
164
–9.