Radiation-induced changes in phosphorus T1values in human melanoma xenografts studied by 31P-MRS

Magnetic

ELSEVIER

Resonance

Imaging, Vol. 0 1997 Elsevier

15, No. 10, pp. 1187-1192, 1997 Science Inc. All rights reserved. Printed in the USA. 0730725x/97 $17.00 + .OO

PI1 SO730-725X(97)00181-1

0 Original Contribution RADIATION-INDUCED CHANGES IN PHOSPHORUS TJALUES IN HUMAN MELANOMA XENOGRAFTS STUDIED BY 31P-MRS DAG

R. OLSEN,* STEFFEN B. PETERSEN,? AND EINAR K. ROFSTAD$

*Department of Medical Physics, The Norwegian Radium Hospital, Montebello, 0310 Oslo, Norway, tMR-Center, SIN’IBFAJNIMED, 7034 Trondheim, Norway, and *Department of Biophysics, Institute for Cancer Research, The Norwegian Radium Hospital, Montebello, 0310 Oslo, Norway 3tP-magnetic resonance spectroscopy (MRS) has been shown to be a promising method for monitoring tumor response to radiation therapy. The purpose of the work reported here was to investigate whether the usefulness of 31P-MRS might be enhanced by measurement of spin-lattice relaxation times (Tts) in addition to resonance ratios. The work was based on the hypothesis that tumors having a high probability of being controlled locally would show shortened T,s during the treatment course due to reoxygenation and development of necrosis. BEX-t human melanoma xenografts, which show efficient reoxygenation and development of necrosis following single dose irradiation, were used as tumor models. Tumors were treated with single doses of 5.0 or 15.0 Gy and the T,s of the inorganic phosphate and nucleoside triphosphate /3 resonances were measured as a function of time after irradiation by using the superfast inversion recovery method. Fractional tumor water content was determined by drying excised tumors at 50°C until a constant weight was reached. The T,s in irradiated tumors were either longer than or not significantly different from those in unirradiated control tumors. The increase in the T,s following irradiation coincided in time with a radiation-induced increase in tumor water content, suggesting a causal relationship. The effects of reoxygenation and development of necrosis on T,s were probably overshadowed by the effects of tumor water content. Consequently, the usefulness of 31P-MRS in monitoring tumor response to radiation therapy might not be significantly enhanced by measurement of T,s. 0 1997 Elsevler Science Inc. Keywords: Magnetic resonancespectroscopy;Melanomaxenografts; PhosphorusT,s; Radiation therapy; Tumor

water content.

INTRODUCTION

/3]) and the inorganic phosphate resonance (Pi) provide information on the physiological state of tumors, whereas the phospholipid resonances (phosphomonoesters [PME] and phosphodiesters[PDE]) provide information on tumor growth.2-4 Radiation-induced changes in the resonance ratios of 31P-MR spectra of experimental tumors have been reported by several groups. 5-9 Irradiated tumors with unimpaired oxidative metabolism are characterised by increased PCr and NTPP resonances relative to the Pi resonance. This increase in bioenergetic status is a result of increased blood supply and hence reflects reoxygenation. 829The magnitude of the increase is dependent on a

The radiation therapy of cancer might be improved if a sensitive method for monitoring tumor response during a fractionated course of treatment is developed.’ An adequate method should be capable of recognizing potential treatment failures sufficiently‘ early that changes in the treatment strategy can be introduced. A non-invasive method is preferable to invasive methods since repeated measurements at short time intervals are required. 31Pmagnetic resonance spectroscopy (MRS) has been shown to be a particularly promising method.2*3 The high energy phosphate resonances (phosphocreatine [PCr] and nucleoside triphosphate y, (Y,and p [NTPy, QI, and RECEIVED

4123197;

ACCEPTED

multitude of factors including tumor line, site of implanNorwegianRadiumHospital,Montebello,0310Oslo,Norway. E-mail: [email protected]

7118197.

Addresscorrespondence to: Einar K. Rofstad,Ph.D., Departmentof Biophysics,Institute for CancerResearch,The 1187

1188

Magnetic

Resonance

Imaging

tation, and host animal. The enhanced bioenergetic status is accompanied by an increase in the PME resonance in some tumor lines.2s4 As tumors regrow, spectral characteristics compatible with unperturbed growth are observed.3 Although numerous studies of radiation-induced changes in phosphorus resonance ratios in tumors have been performed, only two studies have been concerned with radiation-induced changes in phosphorus spin-lattice relaxation times (T,s).‘~~’ ’ Okunieff et al.” studied MCaIV tumors and found that the T, of the Pi resonance increased following single dose irradiation. In contrast, Li et al.,’ ’ who irradiated RIF- 1 and SCCVII tumors with single doses, found decreased T,s of the Pi and NTPy, ar, and /3 resonances and attributed the decrease to improved oxygenation in the irradiated tumors. The main purpose of the work reported here was to investigate whether the usefulness of 31P-MRS in monitoring tumor response to radiation therapy might be enhanced by measurement of T,s in addition to resonance ratios. The investigation was based on the hypothesis that tumors showing extensive reoxygenation and development of necrosis during a course of radiation therapy have a high probability of being controlled locally.‘* Such tumors might show shortened phosphorus T,s during the treatment course for three reasons. First, reoxygenation should lead to decreased T,s due to the paramagnetic properties of the oxygen molecule.‘O1’l Second, the T, of the Pi resonance should decrease after reoxygenation due to increased magnetization transfer secondary to an increased rate of the ATPase reaction.” Third, development of necrosis should lead to decreased T,s owing to an increased concentration of freely dissolved paramagnetic ions resulting from the denaturation of proteins with which the ions are complexed in intact tissue.13’14 Radiation therapy is usually given as multiple fractions with doses of 2-3 Gy per fraction. Experimental tumors show more efficient reoxygenation and more extensive necrotization following irradiation with large single doses than following irradiation with clinically relevant fractional doses when comparable total radiation doses are used.” Single dose irradiation is therefore more likely to result in shortened phosphorus T,s than fractionated irradiation. Consequently, to have a high probability of detecting potential radiation-induced decreases in phosphorus Tls, tumors were irradiated with single doses of 5.0 or 15.0 Gy in the present work.

MATERIALS

AND METHODS

Mice and Tumor Line Fifty-eight male BALBlc-nu/nu mice, B-10 weeks old, were used. They were bred at the animal department

0 Volume

15, Number

10, 1997

of our institution and kept under specific-pathogen free conditions at constant temperature (24-26°C) and humidity (30-50%). Sterilized food and tap water were given ad libitum. The BEX-t amelanotic human melanoma xenograft line, established in the athymic mouse from a metastasis of a patient admitted to The Norwegian Radium Hospital, was used as tumor model.16 The line was maintained in the same mouse strain by serial subcutaneous transplantation of tumor fragments, approximately 2 X 2 X 2 mm in size. Tumor volume (V) was calculated as V = n/6 * a * b*, where a is the longer and b the shorter of two perpendicular diameters. The diameters were measured with callipers. The tumors were 120-200 mm3 in volume at the time of irradiation. Tumor growth delay; i.e., the time required for an irradiated tumor minus the median time required for the unit-radiated control tumors to reach a volume of 1000 mm3, was used as a parameter for radiation response. Fractional tumor water content was determined by excising and drying tumors at 50°C until a constant weight was reached. Irradiation The tumors were irradiated using a Philips SL-75 linear accelerator with a photon beam energy of 5 MV. Single doses of 5.0 or 15.0 Gy, specified at a depth corresponding to the center of the tumors, were given. The tumors were covered with 1 cm of bolus material to ensure adequate dose homogeneity. The field size was 3 X 3 cm. Calibration was performed by inserting thermoluminescence dosimeters in the center of a wax phantom with outer diameters similar to those of a tumor with bolus. Magnetic Resonance Spectroscopy 31P-MRS was performed using solenoidal coils and a Bruker 4.7 T spectrometer operating at 81.025 MHz for phosphorus. l3 A Faraday shield was used to avoid signals from normal tissues adjacent to a tumor. Magnetic field homogeneity was ensured for each tumor by shimming on the water proton resonance. The line-width of the water proton resonance depended on tumor size and ranged from 0.18 to 0.37 ppm. A spectral width of 4 KHz and 1 K of data points per free induction decay were used during acquisition. The T,s of the Pi and NTPp resonances were measured using the superfast inversion recovery (SUFIR) technique.13 The SUFIR experiments were based on a reference spectrum (Sl) from the 90” pulse and a partially relaxed spectrum (S2) from the inversion part of the following sequence: (90°(S 1)-~-l 80”-T-9O”(S2)r),, where the inverting pulse was a composite pulse. Sl and S2 were recorded in blocks of 64 consecutive acquisitions. A total number of 512 acquisitions were accu-

PhosphorusT,s in Tumors 0 10000

OLSEN

1189

ET AL.

sion analysis. Statistical comparisonsof data were per* :

formed under conditions of normality and equal variance by using the Student’s t test for single comparisons and one-way analysis of variance and the Student-NewmanKeuls test for multiple comparisons. A significance criterion of p < 0.05 was used.

OGY 1:::

o1

RESULTS

1000 iti 2

0

5

10

15

20

25

30

35

Time (days)

Fig. 1. Growth curvesfor unirradiatedcontroltumors(n = 6) and tumorsirradiatedwith singledosesof 5.0 Gy (n = 6) or 15.0 Gy (n = 6). The three groupswere not significantly different before irradiation @ > 0.05) and were therefore analyzedasone groupat time zero (n = 18). Pointsand bars representmeanvalues and standarderrors. The experiments were performedwith BEX-t human melanomaxenografted tumors.

mulated for each of the two spectra Sl and S2. A common r of 2.0 s was chosen to obtain optimal accuracy of the T, determinations. T, was calculated as T, = -r/ln(l - S2/Sl). The SUFIR technique has been shown to provide phosphorus T,s in tumors in vivo in good agreement with T,s measured with conventional inversion recovery techniques, and repeated T, measurements in the same tumors have shown that the reproducibility is similar for the two techniques.13 31P-MRS resonance ratios were calculated from the Sl-spectra. The acquisition parameters of the S l-spectra were similar to those usually used in 31P-MRS of experimental tumors in vivo.17 The (PCr + NTPp)/pi resonance ratio was used as a parameter for tumor bioenergetic status.” The resonance ratios were corrected for effects of partial saturation by using the relationship M, = M, * (1 - ee7/r1), where M, is the longitudinal com-

Volumetric growth curves for unirradiated control tumors and tumors irradiated with single doses of 5.0 or 15.0 Gy are shown in Fig. 1. The control tumors grew exponentially during the whole observation period. The growth of the irradiated tumors was suppressed for some days after the radiation exposure, and then the tumors showed exponential growth. The growth delays of the two groups of irradiated tumors were significantly different @ < 0.001). Median growth delay was 10 days for the tumors irradiated with 5.0 Gy and 19 days for the tumors irradiated with 15.0 Gy.

a

PME ‘l N A

I

I

I

I

I

20

10

0

-10

-20

(PPM)

20

10

0

-10

-20

@‘PM)

ponent of the magnetization and M. is the magnetization at equilibrium. l3 Spectral processing included 15-30 Hz exponential multiplication and a convolution difference

of 600 Hz. Resonance areas were calculated from the best fits of Lorentzian lineshapes to phased, resolutionenhanced, and base-line corrected spectra. Statistical Analysis Statistically significant correlations between 31P-MR parametersand time were searchedfor by linear regres-

Fig. 2. Spectraobtainedby SUFIR 31P-MRSof a 180-mm3 BEX-t tumor:(a) referencespectrum(Sl); (b) partially relaxed spectrum(S2). The resonanceassignment is: PME: phosphomonoesters;Pi: inorganic phosphate;PDE: phosphodiesters; PCr: phosphocreatine; NTPy, cr,and/3:nucleoside triphosphate Y, a, andP.

MagneticResonanceImaging0 Volume 15,Number 10, 1997

1190

creased with time @ < 0.001). The tumors irradiated with 15.0 Gy showed a higher bioenergetic status than the control tumors during the whole observation period

(p < 0.05). Figure 4 shows the T,s of the Pi and NTPp resonances vs. time after irradiation for unirradiated control tumors and tumors irradiated with 5.0 or 15.0 Gy. The control

4

tumors and the tumors irradiated with 5.0 Gy showedT,s that decreased slightly during the observation period

(p < 0.05). The T,s in the tumors irradiated with 5.0 Gy were not significantly different from those in the control 1 6 a 0 ‘I 0

I

I

I

I ’

2

4

6

6

Bys afterimdiation Fig. 3. Bioenergeticstatus;i.e., the (PCr + NTPp)/P, resonanceratio, vs. time after irradiationfor unirradiatedcontrol tumors(n = 6) andtumorsirradiatedwith singledosesof 5.0 Gy (n = 6) or 15.0 Gy (n = 6). The three groupswere not significantlydifferent beforeirradiation(p > 0.05) and were thereforeanalyzedasonegroupat time zero (n = 18).Points and barsrepresentmeanvaluesand standarderrors.The experimentswere performedwith BEX-t humanmelanomaxenograftedtumors.

”

I

1

0

;

c

Days after irradiation

The general quality of the 31P-MRS data is illustrated 5 ,

in Fig. 2, which shows the reference spectrum and the partially relaxed spectrum of an unirradiated BEX-t tumor with a volume of 180 mm’. The reference spectrum is in good agreement with spectra of BEX-t tumors of

A 0

b 4

.

similar size recorded previously by using a conventional

90” pulse. l7 The spectra of irradiated tumors were qualitatively similar to those of unirradiated tumors. Repetitive acquisition of spectra from the same tumors and repetitive analysesof the samespectra showed that the

Q

OGY ~GY 15Gy

32 I= 2-

experimental uncertainties were small compared with the biological differences between individual tumors. The

reproducibility of our 31P-MRS method has been reported in detail previously. l7 Figure 3 shows bioenergetic status; i.e., the (PCr + NTPP)/P, resonance ratio, vs. time after irradiation for unirradiated control tumors and tumors irradiated with 5.0 or 15.0 Gy. The bioenergetic status of the control tumors decreased slightly but significantly during the

observation period @ < 0.05). The tumors irradiated with 5.0 Gy did not differ significantly from the control tumors in bioenergetic status at any time after the irradiation (p > 0.05). The bioenergetic statusof the tumors

irradiated with 15.0 Gy increased during the first day after the radiation exposure @ < 0.001) and then de-

11, 0

2

6 4 Days after irradiation

6

Fig. 4. ?“~of the Pi (a) andthe NTPp (b) resonancevs. time after irradiationfor nnirradiatedcontrol tumors(n = 6) and tumorsirradiatedwith singledosesof 5.0 Gy (II = 6) or 15.0 Gy (n = 6). The threegroupswere not significantlydifferent before irradiation(p > 0.05) and werethereforeanalyzedas onegroupat timezero (n = 18).Pointsandbarsrepresentmean valuesand standarderrors.The experimentswere performed with BEX-t humanmelanomaxenograftedtumors.

Phosphorus T,s in Tumors 0

OLSEN

1191

ET AL.

a

77 r+ 'i E 8 ii 2.

88 86 84 82 80 0

5

0

15

5

15

Dose (Gy)

Dose (Gy)

Fig. 5. Fractional water content at day 1 (a) andday 3 (b) after irradiationfor unirradiatedcontrol tumors(n = ‘7in a and6 in b)

andtumorsirradiatedwith singledosesof 5.0 Gy (n = 6 in a andb) or 15.0Gy (n = 7 in a and8 in b). Columnsandbarsrepresent meanvaluesand standarderrors.The experimentswereperformedwith BEX-t humanmelanomaxenograftedtumors.

tumors at any time after the radiation exposure (p > 0.05). The tumors irradiated with 15.0 Gy showed T,s that increasedto a maximum at day 3 after the irradiation and then decreased.At day 3, the T,s in these tumors were significantly higher than the T,s in the control tumors (p < 0.001). The longest T,s in the tumors irradiated with 15.0 Gy occurred 2 days later than the maximum in bioenergetic status. The water content of unirradiated control tumors and tumors irradiated with 5.0 or 15.0 Gy was measuredin separateexperiments (Fig. 5). At day 3 after the irradiation, the tumors irradiated with 15.0 Gy showed a significantly higher water content than the unirradiated control tumors (p < 0.05) and the tumors irradiated with 5.0 Gy @ < 0.05). In contrast, the water contents of the tumors irradiated with 15.0 Gy, the tumors irradiated with 5.0 Gy, and the unirradiated control tumors were not significantly different at day 1 after the irradiation (p > 0.05). The increasein water content of the tumors irradiated with 15.0 Gy coincided with the increasein the T,s of the Pi and NTPp resonances. DISCUSSION The changesin bioenergetic status induced by single dose irradiation in BEX-t human melanomaxenografted tumors were qualitatively similar to those reported previously for experimental murine tumors. One to 7 days after the radiation exposure, the BEX-t tumors given 15.0 Gy showed a higher bioenergetic status than the unirradiated control tumors, whereas the bioenergetic status of the tumors irradiated with 5.0 Gy was not significantly different from that of the control tumors. Similarly, most murine tumors treated with single doses

of 15.0 Gy or more showed an elevated bioenergetic status at and beyond 24 h after the radiation exposure, whereastreatment with singledosesof 2.0 to 5.0 Gy was not sufficient to induce changesin the bioenergetic status

5.6,8,9,19,20

On the other hand, the changesin the T,s of the Pi and NTPP resonancesinduced by single dose irradiation in the BEX-t tumors differed from those reported previously for most experimental murine tumors. The tumors given 15.0 Gy showed longer T,s than the unirradiated control tumors at day 3 after the radiation exposure. In contrast, the RIF-1 and SCCVII tumors showed significantly shorter T,s of the Pi and NTPP resonancesat day 3 after treatment with a single doseof 15.0 Gy. ” The T, of the Pi resonancein the FSaII tumor did not increase after exposure to large single dosesof radiation either, whether the tumors were irradiated in air-breathing mice or in mice breathing 100% 02.10 However, MCaIV tumors irradiated with 30 Gy in air-breathing mice showed an increased T, of the Pi resonance 36-48 h after the treatment.lo The bioenergetic status of the RIF-1, SCCVII, FSaII, and MCaIV tumors was enhanced after the radiation exposure,‘O,” in agreement with the data for the BEX-t tumors. The mechanismbehind the radiation-induced increase in the T,s of the Pi and NTPP resonancesin BEX-t tumors remains to be determined, but water molecules appear to be involved. Thus, the irradiated tumors showed increased T,s when the water content was elevated; i.e., at day 3 after the 15.0 Gy treatment, and unchangedT, s when the water content was similar to that in unirradiated control tumors; i.e., at days 1 and 3 after the 5.0 Gy treatment and at day 1 after the 15.0 Gy

1192

Magnetic Resonance Imaging 0 Volume 15, Number 10, 1997

treatment. Tumors with increased water content might show increased phosphorus T,s for at least two reasons. First, an increase in the water content might lead to decreased concentrations of paramagnetic substances and hence to increased T,s. Second, the mobility of the phosphates might increase with the water content, resulting in increased T,s. This interpretation is consistent with the data for the RIF-1 and SCCVII tumors reported by Li et al. ‘i These tumors showed decreased T,s of the Pi and NTPP resonances at day 3 after a 15.0 Gy treatment, but also a decrease in the water content from 82 to 78%. The hypothesis of the work reported here was that tumors having a high probability of being controlled locally by radiation therapy would show decreased phosphorus T,s during the treatment course due to extensive reoxygenation and development necrosis. Human melanoma xenografts develop necrosis and reoxygenate efficiently during the first days after treatment with single radiation doses of 5.0-15.0 GY. I5 However, the Tl s in the BEX-t tumors increased rather than decreased after irradiation, probably because the effects of increased water content overshadowed the effects of reoxygenation and development of necrosis. Radiation therapy can cause increased or decreased water content in tumors depending on the tumor type, as illustrated by the present study and the study reported by Li et al.” Consequently, the potential usefulness of 31P-MRS in monitoring tumor response to radiation therapy might not be significantly enhanced by measurement of T,s in addition to resonance ratios. Acknowledgmen&-The skillful technical assistance of Berit Mathiesen is gratefully acknowledged. Financial support was received from The Norwegian Cancer Society.

REFERENCES 1. Peters, L.J.; Brock, W.A.; Chapman, J.D.; Wilson, G. Predictive assays of tumor radiocurability. Am. J. Clin. Oncol. 11:275-287; 1988. 2. Rofstad, E.K. NMR spectroscopy in prediction and monitoring of radiation response of tumours in vivo. Int. J. Radiat. Biol. 57:1-5; 1990. 3. Evanochko, W.T.; Ng, T.C.; Glickson, J.D. Application of in vivo NMR spectroscopy to cancer. Magn. Reson. Med. 1508-534; 1984. 4. Daly, P.F.; Cohen, J.S. Magnetic resonance spectroscopy of tumors and potential in vivo clinical applications: A review. Cancer Res. 49:770-779; 1989. 5. Sijens, P.E.; Bovee, W.M.M.J.; Seijkens, D.; Koole, P.; Los, G.; van Rijssel, R.H. Murine mammary tumor response to hyperthermia and radiotherapy evaluated by in vivo 31P nuclear magnetic resonance spectroscopy. Cancer Res. 47:6467-6473; 1987. 6. Kimura, H.; Itoh, S.; Kawamura, Y.; Nakatsugawa, S.; Ishi, Y. Metabolic alterations in implanted human tumors after combined radiation and hyperthermia therapy measured by in vivo 31P MRS. Magn. Reson. Imaging 12:109-119; 1994.

7. Koutcher, J.A.; Okunieff, P.; Neuringer, L.; Suit, H.; Brady, T. Size dependent changes in tumor phosphate metabolism after radiation therapy as detected by 31P NMR spectroscopy. Int. J. Radiat. Oncol. Biol. Phys. 13: 18511855; 1987. 8. Tozer, G.M.; Bhujwalla, Z.M.; Griffiths, J.R.; Maxwell, R.J. Phosphorus-3 1 magnetic resonance spectroscopy and blood perfusion of the RIF-1 tumor following X-irradiation. Int. J. Radiat. Oncol. Biol. Phys. 16:155-164; 1989. 9. Koutcher, J.A.; Alfieri, A.A.; Devitt, M.L.; Rhee, J.G.; Komblith, A.B.; Mahmood, U.; Merchant, T.E.; Cowbum, D. Quantitative changes in tumor metabolism, partial pressure of oxygen, and radiobiological oxygenation status postradiation. Cancer Res. 52:4620-4627; 1992. 10 Okunieff, P.; Ramsay, J.; Tokuhiro, T.; Hitzig, B.M.; Rummeny, E.; McFarland, E.; Neuringer, L.J.; Suit, H. Estimation of tumor oxygenation and metabolic rate using 31P MRS: Correlation of longitudinal relaxation with tumor growth rate and DNA synthesis. Int. J. Radiat. Oncol. Biol. Phys. 14:1185-l 195; 1988. 11 Li, S.J.; Jin, G.Y.; Moulder, J.E. Effects of local irradiation on spin-lattice relaxation time of phosphate metabolites in mouse tumors monitored by 31P magnetic resonance spectroscopy. Magn. Reson. Med. 23:302-310; 1992. 12. Kallman, R.F. The phenomenon of reoxygenation and its implications for fractionated radiotherapy. Radiology 105: 135-142; 1972. 13. Olsen, D.R.; Lyng, H.; Southon, T.E.; Rofstad, E.K. 31Pnuclear magnetic resonance spectroscopy in vivo of four human melanoma xenograft lines: Spin-lattice relaxation times. Radiother. Oncol. 32:54-62; 1994. 14. Jakobsen, I.; Kaalhus, 0.; Lyng, H.; Rofstad, E.K. Detection of necrosis in human tumour xenografts by proton-magnetic resonance imaging. Br. J. Cancer 71:456-461; 1995. 15. Rofstad, E.K. Hypoxia and reoxygenation in human melanoma xenografts. Int. J. Radiat. Oncol. Biol. Phys. 17:8189; 1989. 16. Rofstad, E.K.; Wahl, A.; Stokke, T.; Nesland, J.M. Establishment and characterization of six human melanoma xenograft lines. Acta Pathol. Microbial. Immunol. Stand. 98:945-953; 1990. 17. Lyng, H.; Olsen, D.R.; Southon, T.E.; Rofstad, E.K. 31Pnuclear magnetic resonance spectroscopy in vivo of six human melanoma xenograft lines: Tumour bioenergetic status and blood supply. Br. J. Cancer 68:1061-1070; 1993. 18. Rofstad, E.K.; DeMuth, P.; Fenton, B.M.; Sutherland, R.M. 31P nuclear magnetic resonance spectroscopy studies of tumor energy metabolism and its relationship to intracapillary oxyhemoglobin saturation and tumor hypoxia. Cancer Res. 48:5440-5446; 1988. 19. Sostman, H.D.; Arm&age, I.M.; Fischer, J.J. High resolution spectroscopy of tumors. Magn. Reson. Imaging 2: 265-278; 1984. 20. Mahmood, U.; Alfieri, A.A.; Thaler, H.; Cowbum, D.; Koutcher, J.A. Radiation dose-dependent changes in tumor metabolism measured by 31P nuclear magnetic resonance spectroscopy. Cancer Res. 54:4885-4891; 1994.